Datasets:

---

acmc

/

beamit-annotated_full_texts_dataset

like 0

[ "Ambulatory Care", "Ambulatory Care Facilities", "Humans", "Pilot Projects", "Prospective Studies", "Referral and Consultation", "Telemedicine" ]

[ "Hintergrund und Fragestellung", "Studiendesign und Untersuchungsmethoden", "Ergebnisse", "Diskussion" ]

[ "Im Rahmen der COVID-19-Pandemie wurden, im Sinne des Infektionsschutzgesetzes, kontaktminimierende Maßnahmen beschlossen [11, 20, 21]. Dies betraf auch Krankenhäuser, um Kapazitäten für infizierte Patienten zu schaffen [1, 10, 21]. Mit Beginn der Kontaktbeschränkung, dem sog. Lockdown im März 2020 [21, 22], wurden auch die ambulanten Behandlungsmöglichkeiten zunehmend eingeschränkt. Entsprechend den Regelungen des American College of Surgeons (ACS) und der Deutschen Gesellschaft für Orthopädie und Unfallchirurgie (DGOU) wurden elektive Eingriffe abgesetzt, um entsprechende Bettenkapazität vorhalten zu können [1, 9, 10]. Auch im Rahmen der Hochschulambulanzen wurde eine Reduktion der Patienten notwendig [12]. Als Alternative boten sich videogestützte Formate, die sog. Videosprechstunden, an. Durch die Politik gefördert, konnten bereits ab 2015 erste Versuche einer Implementierung digitaler Sprechstunden in die Chirurgie erfolgen [6, 7, 29]. Auch in Deutschland erfolgten Vorschläge, eine digitale Sprechstunde in den Praxisalltag zu integrieren [13]. Durch die kassenärztlichen Vereinigungen wurden schließlich 2018 die weiteren Weichen zur Förderung einer digital basierten Vorstellung von Patienten gestellt [3]. Die Videosprechstunde bietet hierbei Flexibilität, erspart Wege und verhindert insbesondere in Zeiten der Pandemie einen direkten Arzt-Patient-Kontakt mit erhöhtem Infektionsrisiko [26]. Zwar fehlen bei einer virtuellen Konsultation die spezifischen klinischen orthopädischen Tests als wesentliches Element der klinischen Untersuchung, aber eine inspektorische Erfassung der Gelenkbeweglichkeit und eine Selbstpalpation des Gelenkes durch den Patienten sind möglich [4]. In Studien wurde auf die Gleichwertigkeit der virtuellen Sprechstunde mit der realen Sprechstunde unter gewissen Voraussetzungen hingewiesen [16, 24]. Unmittelbar mit Beginn des ersten Lockdowns der Coronakrise am 16.03.2020 erfolgte im Bereich für arthroskopische und spezielle Gelenkchirurgie/Sportverletzungen der Klinik für Orthopädie, Unfallchirurgie und Plastische Chirurgie am Universitätsklinikum Leipzig die Umstellung auf eine internetbasierte Videosprechstunde. Auch nach Beendigung der Maßnahmen wurde das Angebot der Videosprechstunde weitergeführt.\nZiel der folgenden prospektiven Pilotstudie ist die Darstellung und Beurteilung der Effektivität einer videobasierten Sprechstunde bezüglich Akzeptanz, technischer Machbarkeit und Performance sowie bezüglich Steuerung von Patientenströmen sowohl unter Lockdown-Bedingungen sowie in der Zeit danach.", "Seit dem 13.03.2020 erfolgte die Onlinesprechstunde als freiwilliges Angebot für die Patienten. Die hier vorgestellte Gelenkspezialsprechstunde findet einmal wöchentlich statt. Die Auswertung erfolgte bis zum Stichtag der einschränkenden Maßnahmen im Rahmen der zweiten Welle der Coronapandemie am 14.12.2020. Im hier vorgestellten Zeitraum wurden 2 Anbieter für videobasierte Sprechstunden genutzt. Initial erfolgte die Sprechstunde mit „Sprechstunde.online“ (Fa. Zava Sprechstunde Online GmbH, Essen, Deutschland). Im weiteren Verlauf erfolgte ein Anbieterwechsel zu „Samedi.de“ (Fa. Samedi GmbH, Berlin). Parallel zur Erfassung klinisch-inspektorischer Befunde und Bewegungsausmaße in der Patientenakte erfolgte die prospektive Dokumentation von Parametern zur Bestimmung der Qualität der Sprechstunde. Die Patienten wurden vorab nach telefonischer Terminabstimmung gemäß einem spezifischen Algorithmus für die Videosprechstunde terminiert (Abb. 1). Die Datenerhebung erfolgte auf Basis des § 34 des Sächsischen Krankenhausgesetzes. Bei Zustandekommen des Termins wurden die Dauer des einzelnen Gesprächs sowie die Qualität des Bildes und des Tones mittels dichotomer Fragen dokumentiert. Weiterhin wurden der Grund der Vorstellung erfasst, ob es sich um eine Erst- oder eine Wiedervorstellung handelte, und die diagnostisch-therapeutischen Konsequenzen der Videokonsultation dokumentiert. Hierbei wurden folgende 4 Kategorien zusammengefasst:Wiedervorstellung des Patienten nur bei Bedarf empfohlen, bzw. es wurde eine Weiterbehandlung durch einen niedergelassenen Kollegen initiiert.Erneute Vorstellung zur Verlaufskontrolle in der Videosprechstunde indiziert.Eine Einweisung zur Operation erfolgte bei eindeutigen Befunden und vollständiger Bildgebung. Hier wurden die Patienten am Tag vor der eigentlichen Operation in personam einbestellt, klinisch vollständig untersucht und die Indikation zur Operation letztmalig geprüft.Vorstellung in der realen Sprechstunde zur exakteren klinischen Evaluation bei unklaren Befunden oder zur Durchführung von Röntgenbildern.\nWiedervorstellung des Patienten nur bei Bedarf empfohlen, bzw. es wurde eine Weiterbehandlung durch einen niedergelassenen Kollegen initiiert.\nErneute Vorstellung zur Verlaufskontrolle in der Videosprechstunde indiziert.\nEine Einweisung zur Operation erfolgte bei eindeutigen Befunden und vollständiger Bildgebung. Hier wurden die Patienten am Tag vor der eigentlichen Operation in personam einbestellt, klinisch vollständig untersucht und die Indikation zur Operation letztmalig geprüft.\nVorstellung in der realen Sprechstunde zur exakteren klinischen Evaluation bei unklaren Befunden oder zur Durchführung von Röntgenbildern.", "Insgesamt erfolgten im oben genannten Zeitraum 236 videobasierte Patientenvorstellungen. Das Durchschnittsalter der Teilnehmer betrug 37 Jahre (Min. 15, Max. 70), 92 (39 %) waren weiblich. Vierzehn (6 %) der ursprünglich terminierten Patienten konnten weder per Videosprechstunde noch telefonisch erreicht werden und nahmen nicht an der Sprechstunde teil. Insgesamt konnten von allen durchgeführten Gesprächen 182 (82 %) ohne Einschränkungen durchgeführt werden (Abb. 2). Die durchschnittliche reine Gesprächszeit betrug 8:27 min (Min. 1:20; Max. 23:21). Bei 6 (3 %) Kontakten gab es Einschränkungen in der Qualität des Videos, bei 9 (4 %) eine geminderte Qualität des Tones. Bei 29 (13 %) Vorstellungen musste die Konsultation mittels eines Telefons durchgeführt werden, da ein internetbasierter Kontakt nicht möglich bzw. die Videoqualität nicht ausreichend war. Bei 47 (21 %) Konsultationen handelte es sich um Erstvorstellungen und bei 175 (79 %) um eine Wiedervorstellung nach stationärem Aufenthalt oder ambulanter Vorbehandlung. Das am häufigsten betroffene Gelenk bei der Vorstellung in der Videosprechstunde war mit 108 (49 %) das Kniegelenk (Abb. 3). Bei 74 (33 %) Vorstellungen handelte es sich um Beschwerden im Bereich der Schulter. Bei 25 (11 %) Konsultationen war das Hüftgelenk und bei 12 (5 %) der Ellenbogen betroffen. Bei 3 (1 %) Patienten handelte es sich um proximale Ausrisse der Hamstring-Sehnen bzw. um einen Einriss der Peronäalsehne.\nAls Konsequenz wurde bei 41 (18 %) Patienten eine erneute Vorstellung bei Bedarf bzw. eine Weiterbehandlung durch einen niedergelassenen Kollegen vereinbart (Abb. 4). In 36 (16 %) der Fälle wurde eine erneute Wiedervorstellung zur Verlaufskontrolle in der Videosprechstunde geplant. Bei 105 (47 %) Patienten erfolgte die Wiedervorstellung zur besseren Diagnosefindung in der realen ambulanten Sprechstunde zur Verifizierung der klinischen Befunde. Insgesamt konnte bei 36 (16 %) der Patienten aufgrund der vorhandenen Vorbefunde und der eindeutigen Klinik eine direkte Einweisung zur Operation erfolgen. Hierbei waren bei 20 (56 %) der Patienten Beschwerden an den Kniegelenken führend. Bei 12 (33 %) waren Beschwerden an der Schulter führend, und bei 4 (11 %) erfolgte die Indikationsstellung zur Operation aufgrund von Beschwerden in der Hüfte.", "In Deutschland wird die Zahl der Arztbesuche zunehmend kritisch diskutiert [27]. Hierbei erfolgen durchschnittlich 10 Arztbesuche pro Patient und Jahr [15]. In Schweden etwa werden zur Reduktion der Arztbesuche vermehrt E‑Health-Lösungen verwendet [15]. In den USA wurden für orthopädische Sprechstunden in Zeiten von COVID-19 spezielle Fragebogen zur Untersuchung der Patienten mittels Videosprechstunde entwickelt [17, 18, 25]. In Deutschland hat selbst im Rahmen der Coronapandemie das Angebot von Onlinesprechstunden wie Chats oder Videosprechstunden zu Beginn nur gering zugenommen [19].\nIm Verlauf der Pandemie wurden auch in Deutschland zunehmend Formate und Untersuchungstechniken evaluiert und entwickelt [23, 28]. Buvik et al. konnten zeigen, dass die Videosprechstunde von norwegischen Orthopäden als gleichwertig mit der normalen Sprechstunde angesehen wird [7, 8].\nDie hier vorgestellte Patientenzahl zeigt eine gute Akzeptanz von Angeboten wie der Videosprechstunde. Das gesamte Kollektiv der hier gewählten Sprechstunde ist mit einem Durchschnittsalter von 37 Jahren ein jüngeres Kollektiv und im Umgang mit modernen Techniken versiert und aufgeschlossen. Aber auch ältere Patienten sind der Integration von modernen Kommunikationsmitteln nicht unaufgeschlossen [24]. So zeigt die hier vorliegende Untersuchung, dass auch Patienten mit höherem Alter ein Interesse an der digitalen Sprechstunde haben.\nMit einer durchschnittlichen Behandlungszeit (Patient-Arzt-Gespräch) von 8 min zeigt sich kein großer Unterschied zu der durchschnittlichen Behandlungszeit eines deutschen Hausarztes [14, 30]. Hierbei ist kritisch anzumerken, dass weder die Vorbereitung noch die Nachbereitung samt Dokumentation berücksichtigt wurden. Insgesamt kann zumindest für den Arzt nicht von einer Verkürzung der aufzuwendenden Zeit pro Patient ausgegangen werden.\nNegativ anzumerken ist die aktuell immer noch fehlende Regelung bei der Kostenübernahme insbesondere für Hochschulambulanzen [2]. Dies betrifft u. a. die Ausstellung von Verordnungen bei konservativer Therapie.\nDurch die mittlerweile verbesserte Netzabdeckung ist im vorliegenden Set-up eine ausreichende Netzgeschwindigkeit vorhanden, sodass auch in einem relativen Flächenland wie Sachsen eine ausreichende Qualität der Infrastruktur vorliegt, um eine videobasierte Sprechstunde durchzuführen [5]. Auch bei schlechter Bildqualität war teilweise eine weitere Sprechstunde nur mittels Tonübertragung möglich. Nur bei schlechter Tonqualität musste auf eine telefonische Beratung zurückgegriffen werden.\nIn 21 % der Vorstellungen handelte es sich um Erstvorstellungen. Hierbei können insbesondere extern einzuholende Befunde wie die Schnittbildgebung oder konsiliarische Untersuchungen (z. B. ENG/EMG) auf Vollständigkeit überprüft und ggf. indiziert werden. So lässt sich sicherlich auch die hohe Anzahl an Wiedervorstellungen in der ambulanten Sprechstunde erklären. Ein wesentlicher Vorteil der Videosprechstunde besteht darin, dass die Zahl der Patienten mit unvollständiger Diagnostik in der realen Sprechstunde minimiert werden kann und so Mehrfachvorstellungen vermeiden werden.\nBei 40 % der Patienten konnte bereits in der Videosprechstunde eine definitive Entscheidung zum weiteren Prozedere gestellt werden. Entweder erfolgte die Einleitung einer konservativen funktionellen Therapie oder eine Indikation zur Operation.\nBei der hier vorgestellten Arbeit handelt es sich um eine prospektive reine Beobachtungsstudie ohne Intervention oder Kontrollgruppe. Eine Qualitätsbeurteilung der Bild- und Tonqualität erfolgte nur anhand dichotomer Fragen. Eine differenzierte quantitative Aussage zur Qualität ist somit nicht möglich. Jedoch lässt sich feststellen, dass bei stabiler Verbindung die Qualität der Sprechstunde ausreichend war, um eine adäquate Anamnese durchzuführen. Weiterhin ist anzumerken, dass durch die anonymisierte Datenerhebung eine Mehrfachvorstellung einzelner Patienten nicht ausgeschlossen werden kann." ]

[ null, null, null, null ]

[ "Hintergrund und Fragestellung", "Studiendesign und Untersuchungsmethoden", "Ergebnisse", "Diskussion", "Fazit für die Praxis" ]

[ "Im Rahmen der COVID-19-Pandemie wurden, im Sinne des Infektionsschutzgesetzes, kontaktminimierende Maßnahmen beschlossen [11, 20, 21]. Dies betraf auch Krankenhäuser, um Kapazitäten für infizierte Patienten zu schaffen [1, 10, 21]. Mit Beginn der Kontaktbeschränkung, dem sog. Lockdown im März 2020 [21, 22], wurden auch die ambulanten Behandlungsmöglichkeiten zunehmend eingeschränkt. Entsprechend den Regelungen des American College of Surgeons (ACS) und der Deutschen Gesellschaft für Orthopädie und Unfallchirurgie (DGOU) wurden elektive Eingriffe abgesetzt, um entsprechende Bettenkapazität vorhalten zu können [1, 9, 10]. Auch im Rahmen der Hochschulambulanzen wurde eine Reduktion der Patienten notwendig [12]. Als Alternative boten sich videogestützte Formate, die sog. Videosprechstunden, an. Durch die Politik gefördert, konnten bereits ab 2015 erste Versuche einer Implementierung digitaler Sprechstunden in die Chirurgie erfolgen [6, 7, 29]. Auch in Deutschland erfolgten Vorschläge, eine digitale Sprechstunde in den Praxisalltag zu integrieren [13]. Durch die kassenärztlichen Vereinigungen wurden schließlich 2018 die weiteren Weichen zur Förderung einer digital basierten Vorstellung von Patienten gestellt [3]. Die Videosprechstunde bietet hierbei Flexibilität, erspart Wege und verhindert insbesondere in Zeiten der Pandemie einen direkten Arzt-Patient-Kontakt mit erhöhtem Infektionsrisiko [26]. Zwar fehlen bei einer virtuellen Konsultation die spezifischen klinischen orthopädischen Tests als wesentliches Element der klinischen Untersuchung, aber eine inspektorische Erfassung der Gelenkbeweglichkeit und eine Selbstpalpation des Gelenkes durch den Patienten sind möglich [4]. In Studien wurde auf die Gleichwertigkeit der virtuellen Sprechstunde mit der realen Sprechstunde unter gewissen Voraussetzungen hingewiesen [16, 24]. Unmittelbar mit Beginn des ersten Lockdowns der Coronakrise am 16.03.2020 erfolgte im Bereich für arthroskopische und spezielle Gelenkchirurgie/Sportverletzungen der Klinik für Orthopädie, Unfallchirurgie und Plastische Chirurgie am Universitätsklinikum Leipzig die Umstellung auf eine internetbasierte Videosprechstunde. Auch nach Beendigung der Maßnahmen wurde das Angebot der Videosprechstunde weitergeführt.\nZiel der folgenden prospektiven Pilotstudie ist die Darstellung und Beurteilung der Effektivität einer videobasierten Sprechstunde bezüglich Akzeptanz, technischer Machbarkeit und Performance sowie bezüglich Steuerung von Patientenströmen sowohl unter Lockdown-Bedingungen sowie in der Zeit danach.", "Seit dem 13.03.2020 erfolgte die Onlinesprechstunde als freiwilliges Angebot für die Patienten. Die hier vorgestellte Gelenkspezialsprechstunde findet einmal wöchentlich statt. Die Auswertung erfolgte bis zum Stichtag der einschränkenden Maßnahmen im Rahmen der zweiten Welle der Coronapandemie am 14.12.2020. Im hier vorgestellten Zeitraum wurden 2 Anbieter für videobasierte Sprechstunden genutzt. Initial erfolgte die Sprechstunde mit „Sprechstunde.online“ (Fa. Zava Sprechstunde Online GmbH, Essen, Deutschland). Im weiteren Verlauf erfolgte ein Anbieterwechsel zu „Samedi.de“ (Fa. Samedi GmbH, Berlin). Parallel zur Erfassung klinisch-inspektorischer Befunde und Bewegungsausmaße in der Patientenakte erfolgte die prospektive Dokumentation von Parametern zur Bestimmung der Qualität der Sprechstunde. Die Patienten wurden vorab nach telefonischer Terminabstimmung gemäß einem spezifischen Algorithmus für die Videosprechstunde terminiert (Abb. 1). Die Datenerhebung erfolgte auf Basis des § 34 des Sächsischen Krankenhausgesetzes. Bei Zustandekommen des Termins wurden die Dauer des einzelnen Gesprächs sowie die Qualität des Bildes und des Tones mittels dichotomer Fragen dokumentiert. Weiterhin wurden der Grund der Vorstellung erfasst, ob es sich um eine Erst- oder eine Wiedervorstellung handelte, und die diagnostisch-therapeutischen Konsequenzen der Videokonsultation dokumentiert. Hierbei wurden folgende 4 Kategorien zusammengefasst:Wiedervorstellung des Patienten nur bei Bedarf empfohlen, bzw. es wurde eine Weiterbehandlung durch einen niedergelassenen Kollegen initiiert.Erneute Vorstellung zur Verlaufskontrolle in der Videosprechstunde indiziert.Eine Einweisung zur Operation erfolgte bei eindeutigen Befunden und vollständiger Bildgebung. Hier wurden die Patienten am Tag vor der eigentlichen Operation in personam einbestellt, klinisch vollständig untersucht und die Indikation zur Operation letztmalig geprüft.Vorstellung in der realen Sprechstunde zur exakteren klinischen Evaluation bei unklaren Befunden oder zur Durchführung von Röntgenbildern.\nWiedervorstellung des Patienten nur bei Bedarf empfohlen, bzw. es wurde eine Weiterbehandlung durch einen niedergelassenen Kollegen initiiert.\nErneute Vorstellung zur Verlaufskontrolle in der Videosprechstunde indiziert.\nEine Einweisung zur Operation erfolgte bei eindeutigen Befunden und vollständiger Bildgebung. Hier wurden die Patienten am Tag vor der eigentlichen Operation in personam einbestellt, klinisch vollständig untersucht und die Indikation zur Operation letztmalig geprüft.\nVorstellung in der realen Sprechstunde zur exakteren klinischen Evaluation bei unklaren Befunden oder zur Durchführung von Röntgenbildern.", "Insgesamt erfolgten im oben genannten Zeitraum 236 videobasierte Patientenvorstellungen. Das Durchschnittsalter der Teilnehmer betrug 37 Jahre (Min. 15, Max. 70), 92 (39 %) waren weiblich. Vierzehn (6 %) der ursprünglich terminierten Patienten konnten weder per Videosprechstunde noch telefonisch erreicht werden und nahmen nicht an der Sprechstunde teil. Insgesamt konnten von allen durchgeführten Gesprächen 182 (82 %) ohne Einschränkungen durchgeführt werden (Abb. 2). Die durchschnittliche reine Gesprächszeit betrug 8:27 min (Min. 1:20; Max. 23:21). Bei 6 (3 %) Kontakten gab es Einschränkungen in der Qualität des Videos, bei 9 (4 %) eine geminderte Qualität des Tones. Bei 29 (13 %) Vorstellungen musste die Konsultation mittels eines Telefons durchgeführt werden, da ein internetbasierter Kontakt nicht möglich bzw. die Videoqualität nicht ausreichend war. Bei 47 (21 %) Konsultationen handelte es sich um Erstvorstellungen und bei 175 (79 %) um eine Wiedervorstellung nach stationärem Aufenthalt oder ambulanter Vorbehandlung. Das am häufigsten betroffene Gelenk bei der Vorstellung in der Videosprechstunde war mit 108 (49 %) das Kniegelenk (Abb. 3). Bei 74 (33 %) Vorstellungen handelte es sich um Beschwerden im Bereich der Schulter. Bei 25 (11 %) Konsultationen war das Hüftgelenk und bei 12 (5 %) der Ellenbogen betroffen. Bei 3 (1 %) Patienten handelte es sich um proximale Ausrisse der Hamstring-Sehnen bzw. um einen Einriss der Peronäalsehne.\nAls Konsequenz wurde bei 41 (18 %) Patienten eine erneute Vorstellung bei Bedarf bzw. eine Weiterbehandlung durch einen niedergelassenen Kollegen vereinbart (Abb. 4). In 36 (16 %) der Fälle wurde eine erneute Wiedervorstellung zur Verlaufskontrolle in der Videosprechstunde geplant. Bei 105 (47 %) Patienten erfolgte die Wiedervorstellung zur besseren Diagnosefindung in der realen ambulanten Sprechstunde zur Verifizierung der klinischen Befunde. Insgesamt konnte bei 36 (16 %) der Patienten aufgrund der vorhandenen Vorbefunde und der eindeutigen Klinik eine direkte Einweisung zur Operation erfolgen. Hierbei waren bei 20 (56 %) der Patienten Beschwerden an den Kniegelenken führend. Bei 12 (33 %) waren Beschwerden an der Schulter führend, und bei 4 (11 %) erfolgte die Indikationsstellung zur Operation aufgrund von Beschwerden in der Hüfte.", "In Deutschland wird die Zahl der Arztbesuche zunehmend kritisch diskutiert [27]. Hierbei erfolgen durchschnittlich 10 Arztbesuche pro Patient und Jahr [15]. In Schweden etwa werden zur Reduktion der Arztbesuche vermehrt E‑Health-Lösungen verwendet [15]. In den USA wurden für orthopädische Sprechstunden in Zeiten von COVID-19 spezielle Fragebogen zur Untersuchung der Patienten mittels Videosprechstunde entwickelt [17, 18, 25]. In Deutschland hat selbst im Rahmen der Coronapandemie das Angebot von Onlinesprechstunden wie Chats oder Videosprechstunden zu Beginn nur gering zugenommen [19].\nIm Verlauf der Pandemie wurden auch in Deutschland zunehmend Formate und Untersuchungstechniken evaluiert und entwickelt [23, 28]. Buvik et al. konnten zeigen, dass die Videosprechstunde von norwegischen Orthopäden als gleichwertig mit der normalen Sprechstunde angesehen wird [7, 8].\nDie hier vorgestellte Patientenzahl zeigt eine gute Akzeptanz von Angeboten wie der Videosprechstunde. Das gesamte Kollektiv der hier gewählten Sprechstunde ist mit einem Durchschnittsalter von 37 Jahren ein jüngeres Kollektiv und im Umgang mit modernen Techniken versiert und aufgeschlossen. Aber auch ältere Patienten sind der Integration von modernen Kommunikationsmitteln nicht unaufgeschlossen [24]. So zeigt die hier vorliegende Untersuchung, dass auch Patienten mit höherem Alter ein Interesse an der digitalen Sprechstunde haben.\nMit einer durchschnittlichen Behandlungszeit (Patient-Arzt-Gespräch) von 8 min zeigt sich kein großer Unterschied zu der durchschnittlichen Behandlungszeit eines deutschen Hausarztes [14, 30]. Hierbei ist kritisch anzumerken, dass weder die Vorbereitung noch die Nachbereitung samt Dokumentation berücksichtigt wurden. Insgesamt kann zumindest für den Arzt nicht von einer Verkürzung der aufzuwendenden Zeit pro Patient ausgegangen werden.\nNegativ anzumerken ist die aktuell immer noch fehlende Regelung bei der Kostenübernahme insbesondere für Hochschulambulanzen [2]. Dies betrifft u. a. die Ausstellung von Verordnungen bei konservativer Therapie.\nDurch die mittlerweile verbesserte Netzabdeckung ist im vorliegenden Set-up eine ausreichende Netzgeschwindigkeit vorhanden, sodass auch in einem relativen Flächenland wie Sachsen eine ausreichende Qualität der Infrastruktur vorliegt, um eine videobasierte Sprechstunde durchzuführen [5]. Auch bei schlechter Bildqualität war teilweise eine weitere Sprechstunde nur mittels Tonübertragung möglich. Nur bei schlechter Tonqualität musste auf eine telefonische Beratung zurückgegriffen werden.\nIn 21 % der Vorstellungen handelte es sich um Erstvorstellungen. Hierbei können insbesondere extern einzuholende Befunde wie die Schnittbildgebung oder konsiliarische Untersuchungen (z. B. ENG/EMG) auf Vollständigkeit überprüft und ggf. indiziert werden. So lässt sich sicherlich auch die hohe Anzahl an Wiedervorstellungen in der ambulanten Sprechstunde erklären. Ein wesentlicher Vorteil der Videosprechstunde besteht darin, dass die Zahl der Patienten mit unvollständiger Diagnostik in der realen Sprechstunde minimiert werden kann und so Mehrfachvorstellungen vermeiden werden.\nBei 40 % der Patienten konnte bereits in der Videosprechstunde eine definitive Entscheidung zum weiteren Prozedere gestellt werden. Entweder erfolgte die Einleitung einer konservativen funktionellen Therapie oder eine Indikation zur Operation.\nBei der hier vorgestellten Arbeit handelt es sich um eine prospektive reine Beobachtungsstudie ohne Intervention oder Kontrollgruppe. Eine Qualitätsbeurteilung der Bild- und Tonqualität erfolgte nur anhand dichotomer Fragen. Eine differenzierte quantitative Aussage zur Qualität ist somit nicht möglich. Jedoch lässt sich feststellen, dass bei stabiler Verbindung die Qualität der Sprechstunde ausreichend war, um eine adäquate Anamnese durchzuführen. Weiterhin ist anzumerken, dass durch die anonymisierte Datenerhebung eine Mehrfachvorstellung einzelner Patienten nicht ausgeschlossen werden kann.", "Zusammenfassend lässt sich feststellen, dass die Videosprechstunde im Lockdown und auch in der Zeit danach ein effektives Verfahren bezüglich Akzeptanz sowie technischer Machbarkeit und Perfomance ist. Die vom Arzt aufzuwendende Zeit wird nicht reduziert. Das Patientenaufkommen kann insofern gesteuert werden, dass Mehrfachvorstellungen aufgrund fehlender Vorbefunde sowie routinemäßige Kontrolluntersuchungen reduziert werden können. Die Videosprechstunde ist somit als ergänzende Maßnahme einzuschätzen, die den direkten Arzt-Patient-Kontakt mit der klinischen Untersuchung jedoch kaum ersetzt.\nAufgrund der positiven Ergebnisse und der hohen Akzeptanz der Videosprechstunde durch die Patienten, insbesondere aus entfernteren Regionen, wird die Videosprechstunde auch nach Pandemiezeiten in unserer Klinik weiterhin angeboten werden." ]

[ null, null, null, null, "conclusion" ]

[ "Videosprechstunde", "Telemedizin", "COVID-19", "Klinische Untersuchung", "Digitalisierung", "Video consultation", "Telemedicine", "COVID-19", "Clinical examination", "Digitalisation" ]

[ 393, 387, 465, 605 ]

[ "der", "die", "und", "bei", "eine", "patienten", "zur", "sprechstunde", "videosprechstunde", "von" ]

[ "wiedervorstellungen der ambulanten", "orthopädie und unfallchirurgie", "durchschnittlichen behandlungszeit patient", "spezifischen klinischen orthopädischen", "der kontaktbeschränkung" ]

[ "Adult", "Aged", "Aged, 80 and over", "Cost of Illness", "Costs and Cost Analysis", "Female", "Follow-Up Studies", "Hospitalization", "Humans", "Male", "Managed Care Programs", "Medicare", "Middle Aged", "Morbidity", "Patient Compliance", "Pulmonary Disease, Chronic Obstructive", "Retrospective Studies", "Treatment Outcome", "United States" ]

[ "Introduction", "Study design", "Identification of populations", "Medication utilization", "High-complexity criteria for medication pattern reporting", "Adherence", "Identification of COPD population", "Medication utilization", "Medications of interest", "Medication patterns", "High-complexity criteria for medication pattern reporting", "Treatment measures", "Adherence", "Conclusion" ]

[ "The World Health Organization estimates chronic obstructive pulmonary disease (COPD) to be the fourth leading cause of death, accounting for at least 5% of all deaths worldwide (about 3.02 million).1 This is likely an underestimate of COPD mortality, given that COPD patients often have a high number of comorbidities and complications,2,3 and airflow obstruction is an important contributor to other causes of morbidity and mortality.4,5\nAdvances in pharmacotherapy and improvements in COPD disease management have brought about the realization that COPD is a preventable and treatable disease.2 These advances have also resulted in the development of COPD diagnosis and treatment guidelines,6–9 which are widely available to health care practitioners. Treatment guidelines aim to increase awareness of COPD and improve patient management and outcomes. For example, the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines recommend that an effective COPD management plan should include assessment and monitoring of the disease, reduction in risk factors, management of stable COPD, and management of exacerbations.9,10\nDespite the widespread availability of evidence-based guidelines, survey results suggest that practitioners have major gaps in their knowledge of the core elements of COPD management, with at least 50% reportedly unaware of guidelines for COPD diagnosis and treatment.11 This lack of knowledge of recommended treatment may lead to suboptimal management of COPD patients in the primary care setting. COPD is a progressive illness with worsening symptoms, and therefore patients need to be actively prescribed appropriate therapies and to have ongoing assessments to manage their COPD and improve their health status.\nIn this retrospective analysis of managed care patients, we sought to document and evaluate patterns of medication utilization and medication-related assessments recorded for COPD patients. Medication utilization, adherence, and indicators of treatment and care were analyzed to assess treatment of COPD patients based on recognized guidelines. A subgroup of COPD patients with the most severe or complex COPD were identified and analyzed to determine if COPD severity changes the patterns of medication use in this cohort of commercial and Medicare patients.", "Continuously eligible populations from commercial, Medicare, and Medicaid membership of large, national US health plans including 12.4 million covered lives were identified from the PharMetrics database (Watertown, MA). The database included 19 health plans across the US: 3.2 million from the Northeast, 6.4 million from the Midwest, 1.8 million from the South, and 0.7 million from the West. The plans varied in size: six were <200,000 covered lives, nine were between 200,001 and one million covered lives, and four were over one million covered lives.12 Pharmacy and medical claims from the 7.79 million members who were continuously eligible during the study period (July 2004 and June 2005) were retrospectively analyzed for COPD cohorts. Medicaid (a government-funded program primarily for indigent younger adults and children) claims data were analyzed, but are not reported here because only 485 of 83,007 patients were identified as having COPD. Patient stratification is outlined in Figure 1. The commercial population represented employees and their eligible dependants from employer-based health insurance product offerings, including health maintenance organization, preferred provider organization, and point of service plans. Medicare is a government-sponsored health care program for patients aged ≥65 years and for others with certain disabilities. The Medicare population in this analysis includes persons who chose to have this benefit managed by a private insurance company. Additional details of the populations studied in this analysis have been reported elsewhere.12", "Patients were identified as having COPD if they were aged ≥40 years and had any one of the following:\nOne inpatient hospitalization or one emergency room encounter with a COPD diagnosis (2004 International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 491.x [chronic bronchitis], 492.x [emphysema], or 496 [chronic airway obstruction]) listed in any position as a discharge diagnosis\nTwo professional claims, with different dates of services, with a COPD diagnosis listed in any position\nA COPD-related surgical procedure (eg, lung volume reduction) listed on either a professional or facility claim. COPD-related surgical procedures included open excision plication of bullae of lung, open excision plication of emphysematous lung(s) for lung volume reduction, and thoracoscopy with excision plication of bullae of lung.", "Use of COPD medication(s) was analyzed during the one-year period based on pharmacy claims in the datasets. A patient was considered to be taking a medication if they had at least one filled prescription for that medication product during the one-year study period. The following medications were considered maintenance COPD pharmacotherapy: long-acting β2-agonist bronchodilators (LABA), short-acting or long-acting anticholinergic bronchodilators (SAAC and LAAC, respectively), methylxanthine bronchodilators (theophylline), and inhaled corticosteroids. As per the GOLD guidelines, short-acting β2-agonists (SABAs) were considered symptomatic medications or rescue drugs.10 Medication utilization is reported in two ways:\nMedications of interest, which represent the number of COPD patients who filled at least one prescription for a defined medication class (or fixed-dose combination) of a respiratory or nonrespiratory medication (eg, antimicrobials, narcotic analgesics, diuretics, antidepressants, anxiolytics/sedatives, gastrointestinal antisecretory/antiulcer agents, β-blockers, antihistamines [not part of cough/cold preparations], smoking deterrents/nicotine replacements, gastrointestinal promotility agents, antiarrhythmia agents) during the one-year study period. Reporting in this manner is not mutually exclusive and patients could be receiving more than one medication of interest.\nMedication patterns, which represent the number of COPD patients who filled at least one prescription for each defined respiratory drug class or drug class combination during the one-year study period. Reporting in this manner is mutually exclusive and patients could only be classified as receiving one medication pattern. Medication patterns are used to evaluate the proportion of patients who received single or multiple maintenance COPD medications and for identifying patients receiving single or multiple therapies during the one-year study period as well as those receiving no therapy or SABA therapy alone.\nTo illustrate the differences between these two methods of reporting, a patient who received an inhaled anticholinergic and an inhaled corticosteroid during the study period was reported in both classes in medications of interest, but only in the anticholinergic + inhaled corticosteroid category of medication patterns.\nIn order to characterize further medication use in this population of COPD patients, the following were also assessed: claims for smoking cessation therapy in COPD patients documented to be current smokers (based on current tobacco use using ICD-9-CM codes); use of a maintenance COPD pharmacotherapy within 45 days following a hospitalization for a COPD exacerbation;13 and proportion of patients vaccinated against influenza.", "The guidelines are clear that patients with more severe disease should receive maintenance COPD pharmacotherapy.10 We therefore conducted an analysis of a subset with likely more severe disease. Claims data do not contain pulmonary function test values, and it was therefore not possible to assess COPD severity according to the GOLD guidelines.10 Instead, a claims-based classification of COPD complexity was used to serve as a surrogate for COPD disease severity. The methodology for this has been described fully elsewhere.12 For this analysis, comorbid respiratory conditions and medical procedures occurring during the study period were used to assign patients to a high complexity subset, based on selected diagnostic, therapeutic, procedures, and services codes (2004 ICD-9-CM, Current Procedural Terminology, Fourth Edition, and Healthcare Common Procedure Coding System).12 For example, a patient with a claim for tuberculosis, malignant neoplasm, or cor pulmonale was classified as high complexity. A medications patterns analysis is reported for this high-complexity subset.", "Adherence to prescribed medications was estimated using pharmacy claims by calculating a medication possession ratio (MPR), defined as the total days of medication supplied in all prescription fills/total days between the first and last fill + days supplied in last prescription. MPR is a commonly used metric to assess adherence/compliance.14 Adherence with noninhaled medications (tablets, capsules, liquids) was calculated using the actual reported days of supply from the claims data. However, because the days of supply reported in pharmacy claims for inhaled medication prescriptions is often unreliable, the number of “days supplied” for inhaled medications was assigned a value based on average adult doses (in number of puffs) for each product and the number of puffs per container. Adherence to each respiratory medication is reported as the average MPR for all patients receiving two or more prescriptions of a relevant medication. MPR was only calculated for maintenance COPD medications.", "Of the 7,869,677 patients in the overall dataset, 42,565 commercial and 8507 Medicare patients were identified as having COPD (Figure 1). Among these two groups, 21.4% of commercial (9121/42,565) and 27.6% of Medicare patients (2350/8507) were categorized as high-complexity, based on comorbid respiratory and nonrespiratory conditions and claims for procedures and services.12", " Medications of interest The proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.\nThe proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.\n Medication patterns Figure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.\nFigure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.\n High-complexity criteria for medication pattern reporting In order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.\nIn order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.", "The proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.", "Figure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.", "In order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.", "The evaluation of other measures of treatment also highlighted that COPD in both cohorts was often managed suboptimally. For example, smoking deterrents/nicotine replacement products were prescribed to less than 10% of patients in either population (commercial 9.4% [3990/42,565] versus Medicare 2.8% [234/8507]). Indeed, of the commercial and Medicare patients diagnosed as smokers, only 18.0% (n = 1466/8104) and 9.8% (n = 62/636), respectively, had a claim for a smoking cessation intervention (medication or behavioral therapy) during the one-year study period. Among patients hospitalized for COPD, less than half of the commercial cohort (40.2% [1912/4757] were prescribed bronchodilator therapy within 45 days following their hospitalization, and fewer still (29.9% [359/1200]) of the older Medicare cohort. Furthermore, despite guidelines recommending vaccination against influenza as a risk reduction strategy for all patients with COPD, only 7070 patients (16.6%) in the commercial cohort and 1996 patients (23.5%) in the Medicare cohort were documented to have received the vaccine.", "Medication adherence (reported as mean MPR by drug class) is provided in Table 3. While adherence was above 50% for each respiratory medication class for both cohorts, acceptable adherence (MPR ≥ 80%)14 was only achieved for the oral drug classes (leukotriene modifiers and theophylline). Among the inhaled agents, LAAC and LABA had the highest average adherence, while SAAC had the lowest.", "This study highlights marked undertreatment of COPD in both commercial and Medicare managed care populations. The majority of patients with COPD, even those with a high number of comorbidities, were untreated with respiratory medications. Adherence with COPD therapies was suboptimal, and related therapies, such as smoking cessation and influenza vaccinations, were underprescribed. These findings suggest that there is significant opportunity to improve the lives of patients with COPD with appropriate treatment. A major educational effort is needed to disseminate evidence-based guidelines supported by recent landmark trials to health care providers, and educate providers and patients on the long-term benefits of appropriately treating COPD." ]

[ null, "methods", null, null, null, null, null, null, null, null, null, null, null, null ]

[ "Introduction", "Materials and methods", "Study design", "Identification of populations", "Medication utilization", "High-complexity criteria for medication pattern reporting", "Adherence", "Data collection and outcome measures", "Results", "Identification of COPD population", "Population characteristics", "Medication utilization", "Medications of interest", "Medication patterns", "High-complexity criteria for medication pattern reporting", "Treatment measures", "Adherence", "Discussion", "Conclusion" ]

[ "The World Health Organization estimates chronic obstructive pulmonary disease (COPD) to be the fourth leading cause of death, accounting for at least 5% of all deaths worldwide (about 3.02 million).1 This is likely an underestimate of COPD mortality, given that COPD patients often have a high number of comorbidities and complications,2,3 and airflow obstruction is an important contributor to other causes of morbidity and mortality.4,5\nAdvances in pharmacotherapy and improvements in COPD disease management have brought about the realization that COPD is a preventable and treatable disease.2 These advances have also resulted in the development of COPD diagnosis and treatment guidelines,6–9 which are widely available to health care practitioners. Treatment guidelines aim to increase awareness of COPD and improve patient management and outcomes. For example, the Global Initiative for Chronic Obstructive Lung Disease (GOLD) guidelines recommend that an effective COPD management plan should include assessment and monitoring of the disease, reduction in risk factors, management of stable COPD, and management of exacerbations.9,10\nDespite the widespread availability of evidence-based guidelines, survey results suggest that practitioners have major gaps in their knowledge of the core elements of COPD management, with at least 50% reportedly unaware of guidelines for COPD diagnosis and treatment.11 This lack of knowledge of recommended treatment may lead to suboptimal management of COPD patients in the primary care setting. COPD is a progressive illness with worsening symptoms, and therefore patients need to be actively prescribed appropriate therapies and to have ongoing assessments to manage their COPD and improve their health status.\nIn this retrospective analysis of managed care patients, we sought to document and evaluate patterns of medication utilization and medication-related assessments recorded for COPD patients. Medication utilization, adherence, and indicators of treatment and care were analyzed to assess treatment of COPD patients based on recognized guidelines. A subgroup of COPD patients with the most severe or complex COPD were identified and analyzed to determine if COPD severity changes the patterns of medication use in this cohort of commercial and Medicare patients.", " Study design Continuously eligible populations from commercial, Medicare, and Medicaid membership of large, national US health plans including 12.4 million covered lives were identified from the PharMetrics database (Watertown, MA). The database included 19 health plans across the US: 3.2 million from the Northeast, 6.4 million from the Midwest, 1.8 million from the South, and 0.7 million from the West. The plans varied in size: six were <200,000 covered lives, nine were between 200,001 and one million covered lives, and four were over one million covered lives.12 Pharmacy and medical claims from the 7.79 million members who were continuously eligible during the study period (July 2004 and June 2005) were retrospectively analyzed for COPD cohorts. Medicaid (a government-funded program primarily for indigent younger adults and children) claims data were analyzed, but are not reported here because only 485 of 83,007 patients were identified as having COPD. Patient stratification is outlined in Figure 1. The commercial population represented employees and their eligible dependants from employer-based health insurance product offerings, including health maintenance organization, preferred provider organization, and point of service plans. Medicare is a government-sponsored health care program for patients aged ≥65 years and for others with certain disabilities. The Medicare population in this analysis includes persons who chose to have this benefit managed by a private insurance company. Additional details of the populations studied in this analysis have been reported elsewhere.12\nContinuously eligible populations from commercial, Medicare, and Medicaid membership of large, national US health plans including 12.4 million covered lives were identified from the PharMetrics database (Watertown, MA). The database included 19 health plans across the US: 3.2 million from the Northeast, 6.4 million from the Midwest, 1.8 million from the South, and 0.7 million from the West. The plans varied in size: six were <200,000 covered lives, nine were between 200,001 and one million covered lives, and four were over one million covered lives.12 Pharmacy and medical claims from the 7.79 million members who were continuously eligible during the study period (July 2004 and June 2005) were retrospectively analyzed for COPD cohorts. Medicaid (a government-funded program primarily for indigent younger adults and children) claims data were analyzed, but are not reported here because only 485 of 83,007 patients were identified as having COPD. Patient stratification is outlined in Figure 1. The commercial population represented employees and their eligible dependants from employer-based health insurance product offerings, including health maintenance organization, preferred provider organization, and point of service plans. Medicare is a government-sponsored health care program for patients aged ≥65 years and for others with certain disabilities. The Medicare population in this analysis includes persons who chose to have this benefit managed by a private insurance company. Additional details of the populations studied in this analysis have been reported elsewhere.12\n Identification of populations Patients were identified as having COPD if they were aged ≥40 years and had any one of the following:\nOne inpatient hospitalization or one emergency room encounter with a COPD diagnosis (2004 International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 491.x [chronic bronchitis], 492.x [emphysema], or 496 [chronic airway obstruction]) listed in any position as a discharge diagnosis\nTwo professional claims, with different dates of services, with a COPD diagnosis listed in any position\nA COPD-related surgical procedure (eg, lung volume reduction) listed on either a professional or facility claim. COPD-related surgical procedures included open excision plication of bullae of lung, open excision plication of emphysematous lung(s) for lung volume reduction, and thoracoscopy with excision plication of bullae of lung.\nPatients were identified as having COPD if they were aged ≥40 years and had any one of the following:\nOne inpatient hospitalization or one emergency room encounter with a COPD diagnosis (2004 International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 491.x [chronic bronchitis], 492.x [emphysema], or 496 [chronic airway obstruction]) listed in any position as a discharge diagnosis\nTwo professional claims, with different dates of services, with a COPD diagnosis listed in any position\nA COPD-related surgical procedure (eg, lung volume reduction) listed on either a professional or facility claim. COPD-related surgical procedures included open excision plication of bullae of lung, open excision plication of emphysematous lung(s) for lung volume reduction, and thoracoscopy with excision plication of bullae of lung.\n Medication utilization Use of COPD medication(s) was analyzed during the one-year period based on pharmacy claims in the datasets. A patient was considered to be taking a medication if they had at least one filled prescription for that medication product during the one-year study period. The following medications were considered maintenance COPD pharmacotherapy: long-acting β2-agonist bronchodilators (LABA), short-acting or long-acting anticholinergic bronchodilators (SAAC and LAAC, respectively), methylxanthine bronchodilators (theophylline), and inhaled corticosteroids. As per the GOLD guidelines, short-acting β2-agonists (SABAs) were considered symptomatic medications or rescue drugs.10 Medication utilization is reported in two ways:\nMedications of interest, which represent the number of COPD patients who filled at least one prescription for a defined medication class (or fixed-dose combination) of a respiratory or nonrespiratory medication (eg, antimicrobials, narcotic analgesics, diuretics, antidepressants, anxiolytics/sedatives, gastrointestinal antisecretory/antiulcer agents, β-blockers, antihistamines [not part of cough/cold preparations], smoking deterrents/nicotine replacements, gastrointestinal promotility agents, antiarrhythmia agents) during the one-year study period. Reporting in this manner is not mutually exclusive and patients could be receiving more than one medication of interest.\nMedication patterns, which represent the number of COPD patients who filled at least one prescription for each defined respiratory drug class or drug class combination during the one-year study period. Reporting in this manner is mutually exclusive and patients could only be classified as receiving one medication pattern. Medication patterns are used to evaluate the proportion of patients who received single or multiple maintenance COPD medications and for identifying patients receiving single or multiple therapies during the one-year study period as well as those receiving no therapy or SABA therapy alone.\nTo illustrate the differences between these two methods of reporting, a patient who received an inhaled anticholinergic and an inhaled corticosteroid during the study period was reported in both classes in medications of interest, but only in the anticholinergic + inhaled corticosteroid category of medication patterns.\nIn order to characterize further medication use in this population of COPD patients, the following were also assessed: claims for smoking cessation therapy in COPD patients documented to be current smokers (based on current tobacco use using ICD-9-CM codes); use of a maintenance COPD pharmacotherapy within 45 days following a hospitalization for a COPD exacerbation;13 and proportion of patients vaccinated against influenza.\nUse of COPD medication(s) was analyzed during the one-year period based on pharmacy claims in the datasets. A patient was considered to be taking a medication if they had at least one filled prescription for that medication product during the one-year study period. The following medications were considered maintenance COPD pharmacotherapy: long-acting β2-agonist bronchodilators (LABA), short-acting or long-acting anticholinergic bronchodilators (SAAC and LAAC, respectively), methylxanthine bronchodilators (theophylline), and inhaled corticosteroids. As per the GOLD guidelines, short-acting β2-agonists (SABAs) were considered symptomatic medications or rescue drugs.10 Medication utilization is reported in two ways:\nMedications of interest, which represent the number of COPD patients who filled at least one prescription for a defined medication class (or fixed-dose combination) of a respiratory or nonrespiratory medication (eg, antimicrobials, narcotic analgesics, diuretics, antidepressants, anxiolytics/sedatives, gastrointestinal antisecretory/antiulcer agents, β-blockers, antihistamines [not part of cough/cold preparations], smoking deterrents/nicotine replacements, gastrointestinal promotility agents, antiarrhythmia agents) during the one-year study period. Reporting in this manner is not mutually exclusive and patients could be receiving more than one medication of interest.\nMedication patterns, which represent the number of COPD patients who filled at least one prescription for each defined respiratory drug class or drug class combination during the one-year study period. Reporting in this manner is mutually exclusive and patients could only be classified as receiving one medication pattern. Medication patterns are used to evaluate the proportion of patients who received single or multiple maintenance COPD medications and for identifying patients receiving single or multiple therapies during the one-year study period as well as those receiving no therapy or SABA therapy alone.\nTo illustrate the differences between these two methods of reporting, a patient who received an inhaled anticholinergic and an inhaled corticosteroid during the study period was reported in both classes in medications of interest, but only in the anticholinergic + inhaled corticosteroid category of medication patterns.\nIn order to characterize further medication use in this population of COPD patients, the following were also assessed: claims for smoking cessation therapy in COPD patients documented to be current smokers (based on current tobacco use using ICD-9-CM codes); use of a maintenance COPD pharmacotherapy within 45 days following a hospitalization for a COPD exacerbation;13 and proportion of patients vaccinated against influenza.\n High-complexity criteria for medication pattern reporting The guidelines are clear that patients with more severe disease should receive maintenance COPD pharmacotherapy.10 We therefore conducted an analysis of a subset with likely more severe disease. Claims data do not contain pulmonary function test values, and it was therefore not possible to assess COPD severity according to the GOLD guidelines.10 Instead, a claims-based classification of COPD complexity was used to serve as a surrogate for COPD disease severity. The methodology for this has been described fully elsewhere.12 For this analysis, comorbid respiratory conditions and medical procedures occurring during the study period were used to assign patients to a high complexity subset, based on selected diagnostic, therapeutic, procedures, and services codes (2004 ICD-9-CM, Current Procedural Terminology, Fourth Edition, and Healthcare Common Procedure Coding System).12 For example, a patient with a claim for tuberculosis, malignant neoplasm, or cor pulmonale was classified as high complexity. A medications patterns analysis is reported for this high-complexity subset.\nThe guidelines are clear that patients with more severe disease should receive maintenance COPD pharmacotherapy.10 We therefore conducted an analysis of a subset with likely more severe disease. Claims data do not contain pulmonary function test values, and it was therefore not possible to assess COPD severity according to the GOLD guidelines.10 Instead, a claims-based classification of COPD complexity was used to serve as a surrogate for COPD disease severity. The methodology for this has been described fully elsewhere.12 For this analysis, comorbid respiratory conditions and medical procedures occurring during the study period were used to assign patients to a high complexity subset, based on selected diagnostic, therapeutic, procedures, and services codes (2004 ICD-9-CM, Current Procedural Terminology, Fourth Edition, and Healthcare Common Procedure Coding System).12 For example, a patient with a claim for tuberculosis, malignant neoplasm, or cor pulmonale was classified as high complexity. A medications patterns analysis is reported for this high-complexity subset.\n Adherence Adherence to prescribed medications was estimated using pharmacy claims by calculating a medication possession ratio (MPR), defined as the total days of medication supplied in all prescription fills/total days between the first and last fill + days supplied in last prescription. MPR is a commonly used metric to assess adherence/compliance.14 Adherence with noninhaled medications (tablets, capsules, liquids) was calculated using the actual reported days of supply from the claims data. However, because the days of supply reported in pharmacy claims for inhaled medication prescriptions is often unreliable, the number of “days supplied” for inhaled medications was assigned a value based on average adult doses (in number of puffs) for each product and the number of puffs per container. Adherence to each respiratory medication is reported as the average MPR for all patients receiving two or more prescriptions of a relevant medication. MPR was only calculated for maintenance COPD medications.\nAdherence to prescribed medications was estimated using pharmacy claims by calculating a medication possession ratio (MPR), defined as the total days of medication supplied in all prescription fills/total days between the first and last fill + days supplied in last prescription. MPR is a commonly used metric to assess adherence/compliance.14 Adherence with noninhaled medications (tablets, capsules, liquids) was calculated using the actual reported days of supply from the claims data. However, because the days of supply reported in pharmacy claims for inhaled medication prescriptions is often unreliable, the number of “days supplied” for inhaled medications was assigned a value based on average adult doses (in number of puffs) for each product and the number of puffs per container. Adherence to each respiratory medication is reported as the average MPR for all patients receiving two or more prescriptions of a relevant medication. MPR was only calculated for maintenance COPD medications.\n Data collection and outcome measures DTEC™ software (Version 3.3, Pfizer, New York, NY) was used to integrate administrative data and claims files, identify and stratify patients with COPD, as well as to characterize demographics, comorbidities (respiratory and nonrespiratory), and utilization of medications. All analyses were specified prior to the study and programmed in the software.12 These analyses, including rationale for complexity stratification, were developed by a panel of experts including pulmonologists, outcomes researchers, and claims-based research consultants.12 They were developed based on information from accepted guidelines,9,15 but also incorporate the previous experiences of the panel in claims-based research. While DTEC (a proprietary software program) was used for these analyses, the algorithms for the data reported have been specifically outlined and published previously,12 in order that they may be used in other claims querying systems.\nClaims data during the 1-year study period were analyzed and are presented as means with standard deviations. Categorical data are presented as numbers and percentages. The database was compiled in accordance with all aspects of the Health Information Portability and Accountability Act of 1996.\nDTEC™ software (Version 3.3, Pfizer, New York, NY) was used to integrate administrative data and claims files, identify and stratify patients with COPD, as well as to characterize demographics, comorbidities (respiratory and nonrespiratory), and utilization of medications. All analyses were specified prior to the study and programmed in the software.12 These analyses, including rationale for complexity stratification, were developed by a panel of experts including pulmonologists, outcomes researchers, and claims-based research consultants.12 They were developed based on information from accepted guidelines,9,15 but also incorporate the previous experiences of the panel in claims-based research. While DTEC (a proprietary software program) was used for these analyses, the algorithms for the data reported have been specifically outlined and published previously,12 in order that they may be used in other claims querying systems.\nClaims data during the 1-year study period were analyzed and are presented as means with standard deviations. Categorical data are presented as numbers and percentages. The database was compiled in accordance with all aspects of the Health Information Portability and Accountability Act of 1996.", "Continuously eligible populations from commercial, Medicare, and Medicaid membership of large, national US health plans including 12.4 million covered lives were identified from the PharMetrics database (Watertown, MA). The database included 19 health plans across the US: 3.2 million from the Northeast, 6.4 million from the Midwest, 1.8 million from the South, and 0.7 million from the West. The plans varied in size: six were <200,000 covered lives, nine were between 200,001 and one million covered lives, and four were over one million covered lives.12 Pharmacy and medical claims from the 7.79 million members who were continuously eligible during the study period (July 2004 and June 2005) were retrospectively analyzed for COPD cohorts. Medicaid (a government-funded program primarily for indigent younger adults and children) claims data were analyzed, but are not reported here because only 485 of 83,007 patients were identified as having COPD. Patient stratification is outlined in Figure 1. The commercial population represented employees and their eligible dependants from employer-based health insurance product offerings, including health maintenance organization, preferred provider organization, and point of service plans. Medicare is a government-sponsored health care program for patients aged ≥65 years and for others with certain disabilities. The Medicare population in this analysis includes persons who chose to have this benefit managed by a private insurance company. Additional details of the populations studied in this analysis have been reported elsewhere.12", "Patients were identified as having COPD if they were aged ≥40 years and had any one of the following:\nOne inpatient hospitalization or one emergency room encounter with a COPD diagnosis (2004 International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 491.x [chronic bronchitis], 492.x [emphysema], or 496 [chronic airway obstruction]) listed in any position as a discharge diagnosis\nTwo professional claims, with different dates of services, with a COPD diagnosis listed in any position\nA COPD-related surgical procedure (eg, lung volume reduction) listed on either a professional or facility claim. COPD-related surgical procedures included open excision plication of bullae of lung, open excision plication of emphysematous lung(s) for lung volume reduction, and thoracoscopy with excision plication of bullae of lung.", "Use of COPD medication(s) was analyzed during the one-year period based on pharmacy claims in the datasets. A patient was considered to be taking a medication if they had at least one filled prescription for that medication product during the one-year study period. The following medications were considered maintenance COPD pharmacotherapy: long-acting β2-agonist bronchodilators (LABA), short-acting or long-acting anticholinergic bronchodilators (SAAC and LAAC, respectively), methylxanthine bronchodilators (theophylline), and inhaled corticosteroids. As per the GOLD guidelines, short-acting β2-agonists (SABAs) were considered symptomatic medications or rescue drugs.10 Medication utilization is reported in two ways:\nMedications of interest, which represent the number of COPD patients who filled at least one prescription for a defined medication class (or fixed-dose combination) of a respiratory or nonrespiratory medication (eg, antimicrobials, narcotic analgesics, diuretics, antidepressants, anxiolytics/sedatives, gastrointestinal antisecretory/antiulcer agents, β-blockers, antihistamines [not part of cough/cold preparations], smoking deterrents/nicotine replacements, gastrointestinal promotility agents, antiarrhythmia agents) during the one-year study period. Reporting in this manner is not mutually exclusive and patients could be receiving more than one medication of interest.\nMedication patterns, which represent the number of COPD patients who filled at least one prescription for each defined respiratory drug class or drug class combination during the one-year study period. Reporting in this manner is mutually exclusive and patients could only be classified as receiving one medication pattern. Medication patterns are used to evaluate the proportion of patients who received single or multiple maintenance COPD medications and for identifying patients receiving single or multiple therapies during the one-year study period as well as those receiving no therapy or SABA therapy alone.\nTo illustrate the differences between these two methods of reporting, a patient who received an inhaled anticholinergic and an inhaled corticosteroid during the study period was reported in both classes in medications of interest, but only in the anticholinergic + inhaled corticosteroid category of medication patterns.\nIn order to characterize further medication use in this population of COPD patients, the following were also assessed: claims for smoking cessation therapy in COPD patients documented to be current smokers (based on current tobacco use using ICD-9-CM codes); use of a maintenance COPD pharmacotherapy within 45 days following a hospitalization for a COPD exacerbation;13 and proportion of patients vaccinated against influenza.", "The guidelines are clear that patients with more severe disease should receive maintenance COPD pharmacotherapy.10 We therefore conducted an analysis of a subset with likely more severe disease. Claims data do not contain pulmonary function test values, and it was therefore not possible to assess COPD severity according to the GOLD guidelines.10 Instead, a claims-based classification of COPD complexity was used to serve as a surrogate for COPD disease severity. The methodology for this has been described fully elsewhere.12 For this analysis, comorbid respiratory conditions and medical procedures occurring during the study period were used to assign patients to a high complexity subset, based on selected diagnostic, therapeutic, procedures, and services codes (2004 ICD-9-CM, Current Procedural Terminology, Fourth Edition, and Healthcare Common Procedure Coding System).12 For example, a patient with a claim for tuberculosis, malignant neoplasm, or cor pulmonale was classified as high complexity. A medications patterns analysis is reported for this high-complexity subset.", "Adherence to prescribed medications was estimated using pharmacy claims by calculating a medication possession ratio (MPR), defined as the total days of medication supplied in all prescription fills/total days between the first and last fill + days supplied in last prescription. MPR is a commonly used metric to assess adherence/compliance.14 Adherence with noninhaled medications (tablets, capsules, liquids) was calculated using the actual reported days of supply from the claims data. However, because the days of supply reported in pharmacy claims for inhaled medication prescriptions is often unreliable, the number of “days supplied” for inhaled medications was assigned a value based on average adult doses (in number of puffs) for each product and the number of puffs per container. Adherence to each respiratory medication is reported as the average MPR for all patients receiving two or more prescriptions of a relevant medication. MPR was only calculated for maintenance COPD medications.", "DTEC™ software (Version 3.3, Pfizer, New York, NY) was used to integrate administrative data and claims files, identify and stratify patients with COPD, as well as to characterize demographics, comorbidities (respiratory and nonrespiratory), and utilization of medications. All analyses were specified prior to the study and programmed in the software.12 These analyses, including rationale for complexity stratification, were developed by a panel of experts including pulmonologists, outcomes researchers, and claims-based research consultants.12 They were developed based on information from accepted guidelines,9,15 but also incorporate the previous experiences of the panel in claims-based research. While DTEC (a proprietary software program) was used for these analyses, the algorithms for the data reported have been specifically outlined and published previously,12 in order that they may be used in other claims querying systems.\nClaims data during the 1-year study period were analyzed and are presented as means with standard deviations. Categorical data are presented as numbers and percentages. The database was compiled in accordance with all aspects of the Health Information Portability and Accountability Act of 1996.", " Identification of COPD population Of the 7,869,677 patients in the overall dataset, 42,565 commercial and 8507 Medicare patients were identified as having COPD (Figure 1). Among these two groups, 21.4% of commercial (9121/42,565) and 27.6% of Medicare patients (2350/8507) were categorized as high-complexity, based on comorbid respiratory and nonrespiratory conditions and claims for procedures and services.12\nOf the 7,869,677 patients in the overall dataset, 42,565 commercial and 8507 Medicare patients were identified as having COPD (Figure 1). Among these two groups, 21.4% of commercial (9121/42,565) and 27.6% of Medicare patients (2350/8507) were categorized as high-complexity, based on comorbid respiratory and nonrespiratory conditions and claims for procedures and services.12\n Population characteristics Patients with COPD from the commercial and Medicare data sets had a mean age of 54.7 years and 74.8 years, respectively, and approximately half were male (Table 1). The most common comorbidity was hypertension in both data sets (55.2% commercial; 71.6% Medicare) followed by dyslipidemia (48.2% commercial; 47.3% Medicare). The majority of patients recorded at least one office visit/consultation (98.7% commercial; 96.4% Medicare) with a mean of 11.3 visits in the commercial group and 11.5 visits in the Medicare group during the one-year study period. More than half (53.9%) of the Medicare cohort and nearly 40% of the commercial cohort (39.8%) were hospitalized at least once for any reason during the study period. In total, 1651 Medicare patients (19.4%) and 5922 commercial patients (13.9%) were hospitalized at least once due to COPD. Additional details describing these COPD populations have been published in full elsewhere.12\nPatients with COPD from the commercial and Medicare data sets had a mean age of 54.7 years and 74.8 years, respectively, and approximately half were male (Table 1). The most common comorbidity was hypertension in both data sets (55.2% commercial; 71.6% Medicare) followed by dyslipidemia (48.2% commercial; 47.3% Medicare). The majority of patients recorded at least one office visit/consultation (98.7% commercial; 96.4% Medicare) with a mean of 11.3 visits in the commercial group and 11.5 visits in the Medicare group during the one-year study period. More than half (53.9%) of the Medicare cohort and nearly 40% of the commercial cohort (39.8%) were hospitalized at least once for any reason during the study period. In total, 1651 Medicare patients (19.4%) and 5922 commercial patients (13.9%) were hospitalized at least once due to COPD. Additional details describing these COPD populations have been published in full elsewhere.12\n Medication utilization Medications of interest The proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.\nThe proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.\n Medication patterns Figure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.\nFigure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.\n High-complexity criteria for medication pattern reporting In order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.\nIn order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.\n Medications of interest The proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.\nThe proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.\n Medication patterns Figure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.\nFigure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.\n High-complexity criteria for medication pattern reporting In order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.\nIn order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.\n Treatment measures The evaluation of other measures of treatment also highlighted that COPD in both cohorts was often managed suboptimally. For example, smoking deterrents/nicotine replacement products were prescribed to less than 10% of patients in either population (commercial 9.4% [3990/42,565] versus Medicare 2.8% [234/8507]). Indeed, of the commercial and Medicare patients diagnosed as smokers, only 18.0% (n = 1466/8104) and 9.8% (n = 62/636), respectively, had a claim for a smoking cessation intervention (medication or behavioral therapy) during the one-year study period. Among patients hospitalized for COPD, less than half of the commercial cohort (40.2% [1912/4757] were prescribed bronchodilator therapy within 45 days following their hospitalization, and fewer still (29.9% [359/1200]) of the older Medicare cohort. Furthermore, despite guidelines recommending vaccination against influenza as a risk reduction strategy for all patients with COPD, only 7070 patients (16.6%) in the commercial cohort and 1996 patients (23.5%) in the Medicare cohort were documented to have received the vaccine.\nThe evaluation of other measures of treatment also highlighted that COPD in both cohorts was often managed suboptimally. For example, smoking deterrents/nicotine replacement products were prescribed to less than 10% of patients in either population (commercial 9.4% [3990/42,565] versus Medicare 2.8% [234/8507]). Indeed, of the commercial and Medicare patients diagnosed as smokers, only 18.0% (n = 1466/8104) and 9.8% (n = 62/636), respectively, had a claim for a smoking cessation intervention (medication or behavioral therapy) during the one-year study period. Among patients hospitalized for COPD, less than half of the commercial cohort (40.2% [1912/4757] were prescribed bronchodilator therapy within 45 days following their hospitalization, and fewer still (29.9% [359/1200]) of the older Medicare cohort. Furthermore, despite guidelines recommending vaccination against influenza as a risk reduction strategy for all patients with COPD, only 7070 patients (16.6%) in the commercial cohort and 1996 patients (23.5%) in the Medicare cohort were documented to have received the vaccine.\n Adherence Medication adherence (reported as mean MPR by drug class) is provided in Table 3. While adherence was above 50% for each respiratory medication class for both cohorts, acceptable adherence (MPR ≥ 80%)14 was only achieved for the oral drug classes (leukotriene modifiers and theophylline). Among the inhaled agents, LAAC and LABA had the highest average adherence, while SAAC had the lowest.\nMedication adherence (reported as mean MPR by drug class) is provided in Table 3. While adherence was above 50% for each respiratory medication class for both cohorts, acceptable adherence (MPR ≥ 80%)14 was only achieved for the oral drug classes (leukotriene modifiers and theophylline). Among the inhaled agents, LAAC and LABA had the highest average adherence, while SAAC had the lowest.", "Of the 7,869,677 patients in the overall dataset, 42,565 commercial and 8507 Medicare patients were identified as having COPD (Figure 1). Among these two groups, 21.4% of commercial (9121/42,565) and 27.6% of Medicare patients (2350/8507) were categorized as high-complexity, based on comorbid respiratory and nonrespiratory conditions and claims for procedures and services.12", "Patients with COPD from the commercial and Medicare data sets had a mean age of 54.7 years and 74.8 years, respectively, and approximately half were male (Table 1). The most common comorbidity was hypertension in both data sets (55.2% commercial; 71.6% Medicare) followed by dyslipidemia (48.2% commercial; 47.3% Medicare). The majority of patients recorded at least one office visit/consultation (98.7% commercial; 96.4% Medicare) with a mean of 11.3 visits in the commercial group and 11.5 visits in the Medicare group during the one-year study period. More than half (53.9%) of the Medicare cohort and nearly 40% of the commercial cohort (39.8%) were hospitalized at least once for any reason during the study period. In total, 1651 Medicare patients (19.4%) and 5922 commercial patients (13.9%) were hospitalized at least once due to COPD. Additional details describing these COPD populations have been published in full elsewhere.12", " Medications of interest The proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.\nThe proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.\n Medication patterns Figure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.\nFigure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.\n High-complexity criteria for medication pattern reporting In order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.\nIn order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.", "The proportion of COPD patients in the commercial and Medicare populations receiving different classes of respiratory and nonrespiratory drugs are shown in Table 2. Oral corticosteroids and SABAs were the most commonly filled respiratory medications in commercial (30.1%; 30.0%) and Medicare (20.9%; 25.5%) populations, respectively. Systemic antibiotics or antimicrobials were the most common non-COPD medications used in both data sets, prescribed to 57.7% and 40.2% of the commercial and Medicare populations, respectively.", "Figure 2 shows patterns of medication use, and indicates that the majority of patients in both cohorts did not receive any maintenance COPD pharmacotherapy. Figure 2A shows that 66.3% of commercial patients (n = 28,206) received no maintenance COPD pharmacotherapy, including 25,157/42,565 (59.1%) prescribed no COPD medications and 3049/42,565 (7.2%) prescribed only inhaled SABAs. Figure 2B shows that 70.9% of Medicare patients (n = 6376) received no maintenance COPD pharmacotherapy, including 5615/8507 (66.0%) prescribed no COPD medications, and 416/8507 (4.9%) prescribed SABA only.", "In order to evaluate medication utilization in patients with a high number of comorbidities who were more likely to be severe and compromised, we performed a similar evaluation for COPD patients who met criteria for high complexity (as a surrogate for COPD severity)12 and found similar trends. More than half of all high-complexity patients were not prescribed maintenance COPD pharmacotherapy. For the commercial population, 5358 (58.7%) of high-complexity patients did not receive any maintenance COPD pharmacotherapy, including 4756 (52.1%) receiving no COPD medications and 602 (6.6%) receiving SABAs only. Similarly, 68.8% (1616/2350) high-complexity Medicare patients did not receive any maintenance COPD pharmacotherapy, including 1504/2350 (64.0%) receiving no COPD medication and 112/2350 (4.8%) receiving SABAs only.", "The evaluation of other measures of treatment also highlighted that COPD in both cohorts was often managed suboptimally. For example, smoking deterrents/nicotine replacement products were prescribed to less than 10% of patients in either population (commercial 9.4% [3990/42,565] versus Medicare 2.8% [234/8507]). Indeed, of the commercial and Medicare patients diagnosed as smokers, only 18.0% (n = 1466/8104) and 9.8% (n = 62/636), respectively, had a claim for a smoking cessation intervention (medication or behavioral therapy) during the one-year study period. Among patients hospitalized for COPD, less than half of the commercial cohort (40.2% [1912/4757] were prescribed bronchodilator therapy within 45 days following their hospitalization, and fewer still (29.9% [359/1200]) of the older Medicare cohort. Furthermore, despite guidelines recommending vaccination against influenza as a risk reduction strategy for all patients with COPD, only 7070 patients (16.6%) in the commercial cohort and 1996 patients (23.5%) in the Medicare cohort were documented to have received the vaccine.", "Medication adherence (reported as mean MPR by drug class) is provided in Table 3. While adherence was above 50% for each respiratory medication class for both cohorts, acceptable adherence (MPR ≥ 80%)14 was only achieved for the oral drug classes (leukotriene modifiers and theophylline). Among the inhaled agents, LAAC and LABA had the highest average adherence, while SAAC had the lowest.", "This analysis of more than 51,000 patients highlights significant undertreatment of COPD in typical US commercial and Medicare populations. The majority of diagnosed COPD patients remain undertreated with no maintenance COPD pharmacotherapy for their COPD over the one-year study period.\nInhaled bronchodilators are the accepted f irst-line therapies for symptomatic COPD management, because regular use can increase exercise capacity and improve health status.9,16,17 However, in the present analysis, more than two-thirds of COPD patients in the commercial population and >70% of the Medicare population were not prescribed any long-term maintenance COPD medications. Because patients with mild COPD may appropriately not be prescribed long-term therapy according to GOLD guidelines, we also evaluated subsets of patients classified as “high-complexity,” where complexity served as a surrogate for disease severity and represented patients with the highest number of comorbid conditions and complications.12 More than half of high-complexity commercial and 69% of high-complexity Medicare patients failed to receive any maintenance COPD pharmacotherapies. Although clinicians may be less likely to prescribe chronic COPD pharmacotherapy to older patients receiving multiple medications who may have cognitive impairment or difficulty using inhalers, the observation that 70% of older patients, such as those represented by the Medicare cohort, and 60% of high-complexity Medicare patients, did not receive even one long-term COPD pharmacotherapy is a cause for concern.\nThe effectiveness of recommended treatment regimens is dependent on how well patients adhere with them. Adherence to COPD medications in the present study based on medication refills was only about 50%, which is comparable with other studies,18,19 but still suboptimal for patients to derive benefit from therapies being prescribed.\nOur findings are particularly noteworthy given recommended guidelines outlining a clear strategy for COPD and the management of associated exacerbations.7,9,10,20 In addition, recent large, multiyear studies have shown therapies such as LAACs and LABA/inhaled corticosteroid combinations used chronically offer significant benefits to COPD patients, including a reduction in exacerbation frequency and possibly a reduction in mortality.21,22 The undermanagement highlighted here is in agreement with a survey of primary care physicians, which reported that only 35% of physicians chose a long-acting bronchodilator (recommended by GOLD)10 if use of a short-acting agent had failed to manage a patient’s COPD symptoms.23 Furthermore, although 55% of physicians were aware of major COPD guidelines (GOLD or American Thoracic Society/European Respiratory Society guidelines), only 25% used them to guide decision-making.23\nIn addition to highlighting the suboptimal prescription of COPD medications, we also found suboptimal use of certain non-COPD maintenance pharmacotherapies. Less than 20% of COPD patients in this study were documented to have received an influenza vaccine, despite vaccination being recommended for all COPD patients to reduce exacerbation frequency.9 In addition, 82% of patients reported to be current smokers in the commercial dataset did not receive any smoking cessation intervention. Similarly, 90% of current smokers in the Medicare population did not receive smoking cessation interventions. The majority of COPD cases can be attributed to tobacco smoking,2 and smoking cessation has been shown to be the most effective method of slowing the progression of COPD.9,10,24 The evidence indicates that smoking cessation interventions, including non-nicotine pharmacotherapy,25,26 or nicotine replacement therapy,27 combined with counseling, are both effective and cost-effective.28–30\nOur data should be interpreted in light of certain limitations. As with any retrospective database study, the COPD diagnosis for patients in the present study was established by the presence of claims coded for this diagnosis from health care providers, and no spirometry results are available to verify diagnosis or assess disease severity. We compensated for the latter using a surrogate, ie, COPD complexity, for severity using service and procedure codes and comorbid conditions selected by a panel of experts. It is possible that there is under-reporting of medication utilization because the study was limited to a one-year period, and newly diagnosed patients near the end of the inclusion period may not have had time to begin drug therapy. However, our sensitivity analyses using high-complexity patients showed similar underuse of COPD medications in this subgroup as per the whole cohort. Some diagnoses and therapies are consistently under-reported in claims data, including smoking and antismoking medications, since many health care plans do not cover these medications. Therefore, the proportion of patients recorded as smokers or taking smoking cessation therapies in the current study may actually be an under-representation. While this study primarily evaluates treatment with pharmacotherapy, other therapies are important in the treatment of COPD. Other health care utilization in this population including hospitalization and outpatient services such as respiratory-related equipment and supplies (including oxygen) and respiratory therapy services (including assisted ventilation) have been reported elsewhere.12 A detailed analysis of oxygen therapy in a US managed care population has also been reported previously.31\nMedication adherence was estimated using reasonable assumptions for patients who filled more than two prescriptions. Patients who filled only one prescription and could be, therefore, nonadherent, were not included in the calculation. Our method of estimating the number of “days supplied” for inhaled drugs based upon recommended doses and number of puffs/inhaler,32 could be inaccurate in patients on unusual doses. While these assumptions are reasonable for the population as a whole, unusual dosing in a specific patient could result in inaccurate MPRs. We cannot account for prescriptions that were filled but not taken by the patient, or for medication samples that may have been dispensed by the health plan physicians, because these are not accounted for in claims data. Because the study was a cross-sectional analysis, no cause and effect analyses could be conducted. Analysis of the management of comorbid conditions was beyond the scope of this analysis.\nOur study used data from 2004 to 2005, prior to the publication of landmark studies, such as the Towards a Revolution in COPD Health (TORCH) and the Understanding Potential Long-Term Impacts on Function with Tiotropium (UPLIFT®) studies,21,22 which demonstrated significant benefits of providing maintenance pharmacotherapy for COPD.33 Although we hope that the undertreatment documented in the present study has improved with the publication of these studies, the magnitude of our findings and their consistency with other reports34,35 make it very likely that a significant problem with the undermanagement of COPD patients still exists.", "This study highlights marked undertreatment of COPD in both commercial and Medicare managed care populations. The majority of patients with COPD, even those with a high number of comorbidities, were untreated with respiratory medications. Adherence with COPD therapies was suboptimal, and related therapies, such as smoking cessation and influenza vaccinations, were underprescribed. These findings suggest that there is significant opportunity to improve the lives of patients with COPD with appropriate treatment. A major educational effort is needed to disseminate evidence-based guidelines supported by recent landmark trials to health care providers, and educate providers and patients on the long-term benefits of appropriately treating COPD." ]

[ null, "materials|methods", "methods", null, null, null, null, "methods", "results", null, "intro", null, null, null, null, null, null, "discussion", null ]

[ "managed care", "chronic obstructive pulmonary disease", "health care utilization", "quality of care" ]

[ 366, 265, 156, 462, 178, 171, 66, 720, 92, 108, 149, 204, 73, 118 ]

[ "copd", "patients", "medication", "medicare", "commercial", "medications", "maintenance", "high", "maintenance copd", "complexity" ]

[ "copd improve patient", "copd pharmacotherapies clinicians", "retrospectively analyzed copd", "chronic copd pharmacotherapy", "symptomatic copd management" ]

[ "Academic Performance", "Adolescent", "Child", "Cross-Sectional Studies", "Female", "Humans", "Iran", "Male", "Schools", "Sleep", "Students", "Surveys and Questionnaires", "Time Factors" ]

[ "Introduction", "STUDY DESIGN", "STUDY TOOL", "STATISTICAL ANALYSIS", "Results", "Discussion", "Conclusions" ]

[ "Sleep is a periodic and natural state of human which during that body and mind is at rest, the eyes usually close and consciousness is partially or completely reduced [1]. Body movements reduce during sleep and body does not respond to external stimuli. Almost a third of humans’ life is spent in sleep [2, 3]. Sleep is a biological behavior of human that regulation of it is based on a complex biological pattern in the brain. In most cases, disruption of the sleep cycle is an early sign of physical and mental illnesses [4]. Good quality of sleep is necessary to perform routine daily function including metabolic activities, hormonal processes, and proper regulation of appetite [5, 6]. Chronic sleep deprivation induces many detrimental effects of physical health including impaired carbohydrate metabolism, increase risk of diabetes, and dysfunction of appetite regulation hormones such as leptin and ghrelin. In some documents suggested that development of obesity and diabetes in elder people is due to poor quality of sleep in this group [7]. It is important that inadequate sleep (insufficient quantity or poor quality of sleep) is epidemic in our modern societies, and many people suffer from it [8-10]. Empirical evidence has shown that children need an average of 9 hours of sleep at night. According to the researches, 45% of them sleep less than 8 hours at night [10, 11]. Sleep problems affect the academic performance of students [12-14]. Obtaining adequate sleep is essential for good performance of children in school. Inadequate or poor sleep qualities are common problems in adolescent that effect on their learning, memory and school performance [12]. Other studies have shown that insufficient sleep, fragmentation of sleep and sleeping late have detrimental effects of academic performance of teenagers [13-15]. In a study in America, which was performed between 88 students, it was proved that better quality, longer duration, and greater consistency of sleep were correlated with the better grades in the lessons [16]. In a research in Pakistan, 64.24% of students with global Pittsburgh sleep quality index (PSQI) score ≥ 5 have poor sleep quality. The mean grade point average (GPA) of poor sleepers was 2.92 ± 1.09 which was significantly lower than that of good sleepers. Poor sleep quality had a negative impact on the academic performance and adequate sleep had a positive impact on the refresh of students every day; adequate sleep helps them in learning and memory processing [5]. In Iran, study of 407 students found that 9.1, 36.1, 39.3 and 13.5% of them had excellent, good, satisfactory and poor daily sleep quality. In this research, appropriate sleep duration and adequate sleep period had a positive effect on the academic performance including educational achievement, high scores in the exams and freshness in the classroom [17]. Given the importance of sleep for students, the present study aimed to determine the association between sleep hours and academic performance in young adults.", "This cross-sectional study was designed as a descriptive-analytic study. A sample of adolescents of 14-17 years of age in the first, second and third degree of high school (two schools were chosen, randomly) in Qazvin city were enrolled. Sample size was calculated with considering 8% precision, 95% confidence interval and 80% power about 700 students. Participants were selected using cluster sampling, so that 700 students were chosen from 10 schools in 5 different parts of city (two schools were selected from each area of city). Pre-university students were excluded from study because of the stress due to university entrance exam and its effects on sleep pattern. After selecting the desired school, some students were selected randomly from them.", "Study’s questionnaires were distributed among them. Data collection tools were two questionnaires which their validity and reliability has been confirmed in previous studies. The Pediatric Sleep Questionnaire consist 22 questions was designed to evaluate sleep problems in children. Its sensitivity and specificity have a range between 0.81 to 0.85, and 0.87, respectively compared to polysomnographic results. Also, Cronbach’s alpha coefficient of questionnaire was 0.77 in this study for PSQ. Also, the BEARS questionnaire developed by Owen used for all students to screen comprehensively major sleeps problems in them. Five sleep domain evaluated by this questionnaire including bedtime problems, excessive daytime sleepiness, awakening during the night, regularity and duration of sleep, and snoring. Previously, Mohammadi and colleague were assessed validity and reliability of Persian version of this questionnaire. The BEARS internal consistency in our study was high with a Cronbach’s alpha of 0.79. A total of 700 questionnaires were distributed. Twenty of questionnaires were excluded because of the incomplete filling of the questions. Also, 27 of students were excluded due to suffering from diseases that effect on their sleep parameters [18-21].", "Finally, data collected from 653 cases were confirmed for use in the analysis. Statistical package for social sciences (SPSS) version 16 was used for data analysis. Chi-square test, t-test, analysis of variance (ANOVA), and correlation were performed to determine the relationship between the data (P < 0.001).", "In this study, 653 adolescents 14-18 years were surveyed, of which 261 (40%) were male and 392 (60%) were female. Average age of our sample was 15.7 ± 0.9. Education levels in students were as follow: 154 (23.6%) in the first year, 242 (37.1%) in the second year of high school and 257 (39.4%) in the third year of high school. Information on age, education grade, hours of sleep during the night, and delay amount in the sleep onset are mentioned (Tab. I). The average sleep duration was significantly higher in boys than girls (P < 0.001). The hours of starting sleep at night and waking in the morning was significantly different between boys and girls (P < 0.05). Sleep onset delay was significantly higher in girls than boys (P < 0.001).\nHours of sleep on weekdays and weekends between boys and girls were compared (Tab. II). Sleep duration was classified in three different categories (less than 7 hours, 7-9 hours, and more than 9 hours during night). Sleep insufficient was significantly higher in girls than boys (P < 0.001). In contrast, the rate of oversleeping was higher in boys. Also, hours of sleep on weekends were higher than weekdays. Pearson analysis showed that time of falling sleep at weekend nights (ρ = 0.83, P = 0.03) and weekday nights (ρ = 0.14, P<0.001) have positively correlation with age.\nAcademic performance of students was divided into three categories based on their GPA. Students with GPA equal or less than 15 was classified in poor academic performance. Also, students with GPA between 15.01-17.99 and equal or more than 18 were classified as medium and good academic performance, respectively. Results showed that academic performance was significantly better in girls than boys (Tab. III).\nStudents who sufferings from difficulty to falling asleep have significantly lower academic performance (P < 0.001) (Tab. IV). Also, there was a significant relationship between students’ sleep hours in weekends and weekdays with academic performance. The average hours of sleep among students with better academic performance was less than students with medium and poor academic performance (P < 0.001). The amount of delay in falling asleep was more in students who had lower academic performance (P = 0.002) and total sleep time during weekend (P < 0.001) and weekdays (P < 0.001) were significantly higher in students with better academic performance (Tab. V).", "In this study, the mean total sleep duration was 42.8 hours during weekdays, and this amount was significantly lower in girls than boys. Also, a significant difference found between boys and girls in terms of the time to go to bed and waking up in the morning. Another result showed that the average number of sleep hours was lower in students who had better academic performance than students with average and poor academic performance. According to the center for disease control and prevention (CDC) in America, the optimal duration of sleep recommended for teenagers are more than 8.5 hours at night [19]. In our sample, duration of sleep for 24 hours was 8.42 ± 1.6 and 10.4 ± 1.8 for weekdays and weekends, respectively. Total duration of sleep was more in boys than girls, and girls went later to sleep than boys at both of the weekends and weekdays. In a study in America the average length of sleep on the weekend was about 9 hours in boys and girls. But, this amount was reduced to 7.1 hours for boys and 5.2 hours for girls during weekdays. These results were different from current study that in our study students’ sleep duration on weekend was 5.1 hours more than American teenagers. Also, the duration of sleep in weekdays was lower for both American girls and boys about 2.1 and 1.8 hours, respectively. So, in total during the whole week, Iranian teenagers slept more than American teenagers. In this study, there was not significant correlation between age and sleep duration. Total sleep duration decreases with increasing age [19]. In another study in Canada, it was found that seventy percent of students aged 14 to 18 years old sleep less than 8.5 hours during the night [10]. In this study, sixty percent of students reported that they sleep less than 8.5 hours during night at weekdays. Also, we found a wide gap in sleep duration between weekends and weekdays (Yo-yo sleeping), so that our samples slept two hours more in weekends compared to weekdays. These results show that students’ sleep was not enough during weekdays, therefore they sleep more in weekends to compensate their sleep insufficiency. This large difference will cause a negative impact on students’ performance in school [20]. This amounts in studies conducted in other countries is as follow: America 2.1 hours [19], Taiwan 1 hour [21], China 2.5 hours [22], and Australia 16 minutes [20]. There was no difference in sleep duration between weekdays and weekend in Switzerland [23]. Current study showed that with increasing students’ age, the time for going to bed were delayed. Similar results were obtained in other studies to confirm our results [21, 24]. In this study, total sleep duration was more in boys than girls in all days of the week. Similar results were obtained in a study conducted in Taiwan [21], but a survey in Australia showed that teenager girls sleep more than boys [20]. Recent findings have shown that sleep is important for the proper function of learning and memory [13, 25]. The possibility of delay in sleep onset increases with the onset of puberty that can lead to daytime sleepiness and negative effects on academic performance in students. A study in America showed that delay in sleep initiation more than 30 minutes for more than one night during the week has a significant association with increased rates of academic failure in students [26]. In another study conducted in China, the average delay in starting sleep was 30 minutes in students [27]. This amount was calculated in the present study 36.77 minutes. A review study in 2010 showed that increasing hours of sleep is associated with better academic performance in students [12]. But in our study, students who have fewer hours of sleep during night had better academic performance. On the other hand, another study did not report a significant relationship between sleep duration and academic performance in students [28]. In a study in Iran, 102 students completed PSQI. Based on the results there was no significant difference between students with high grades and those with low grades. But there were moderate and sometimes severe sleep disturbances in both groups. Also, there was no significant difference between sleep quality and academic achievement [29]. In another hand, in our results significant difference between sleep quality and academic performances between students was existed. A research between 341 selected students in Iran showed that 59.1% of them had poor sleep quality. Also, there was a significant negative relationship between sleep quality with academic interaction and academic vitality. There was a significant positive relationship between academic motivations with academic vitality. Also, a significant difference between male and female students in academic vitality was observed [30]. Different reasons such as level of family income, family size, intake of supplements and vitamins, social media dependency, addiction to social networks and social issues can affect the academic success in the different students [29, 31]. Improve adolescent sleep including delaying school start times, providing sleep education, and utilizing light therapy to improve the health, wellbeing and academic performance of sleepy teenagers are in the programs of researchers in this field in different countries [30]. The results from this study indicated that sleeping time in Iranian students is more than students in other countries. In this study, sleep duration, sleep onset delay, sleep insufficient and rate of oversleeping was associated with better academic performance in students. Also, difficulty in falling asleep was associated with weaker academic performance. The overall result was that in students without difficulty in falling asleep, a positive influence on the academic performance was observed.", "The result of this study showed that some of sleep characteristics such as sleep duration, sleep onset delay, sleep insufficient and rate of oversleeping of students could be a significant influence on adolescents’ academic performance. Further studies are needed to objectively determine the effect of sleep variables on adolescents’ performance.\nLIMITATIONS First, the design is cross sectional. Therefore, it may be difficult to confirm a cause-effect relationship. Moreover, the selected students were found randomly. Also, findings may not be applicable to all students in other geographical locations.\nFirst, the design is cross sectional. Therefore, it may be difficult to confirm a cause-effect relationship. Moreover, the selected students were found randomly. Also, findings may not be applicable to all students in other geographical locations." ]

[ null, null, null, null, null, null, null ]

[ "Introduction", "Materials and methods", "STUDY DESIGN", "STUDY TOOL", "STATISTICAL ANALYSIS", "Results", "Discussion", "Conclusions" ]

[ "Sleep is a periodic and natural state of human which during that body and mind is at rest, the eyes usually close and consciousness is partially or completely reduced [1]. Body movements reduce during sleep and body does not respond to external stimuli. Almost a third of humans’ life is spent in sleep [2, 3]. Sleep is a biological behavior of human that regulation of it is based on a complex biological pattern in the brain. In most cases, disruption of the sleep cycle is an early sign of physical and mental illnesses [4]. Good quality of sleep is necessary to perform routine daily function including metabolic activities, hormonal processes, and proper regulation of appetite [5, 6]. Chronic sleep deprivation induces many detrimental effects of physical health including impaired carbohydrate metabolism, increase risk of diabetes, and dysfunction of appetite regulation hormones such as leptin and ghrelin. In some documents suggested that development of obesity and diabetes in elder people is due to poor quality of sleep in this group [7]. It is important that inadequate sleep (insufficient quantity or poor quality of sleep) is epidemic in our modern societies, and many people suffer from it [8-10]. Empirical evidence has shown that children need an average of 9 hours of sleep at night. According to the researches, 45% of them sleep less than 8 hours at night [10, 11]. Sleep problems affect the academic performance of students [12-14]. Obtaining adequate sleep is essential for good performance of children in school. Inadequate or poor sleep qualities are common problems in adolescent that effect on their learning, memory and school performance [12]. Other studies have shown that insufficient sleep, fragmentation of sleep and sleeping late have detrimental effects of academic performance of teenagers [13-15]. In a study in America, which was performed between 88 students, it was proved that better quality, longer duration, and greater consistency of sleep were correlated with the better grades in the lessons [16]. In a research in Pakistan, 64.24% of students with global Pittsburgh sleep quality index (PSQI) score ≥ 5 have poor sleep quality. The mean grade point average (GPA) of poor sleepers was 2.92 ± 1.09 which was significantly lower than that of good sleepers. Poor sleep quality had a negative impact on the academic performance and adequate sleep had a positive impact on the refresh of students every day; adequate sleep helps them in learning and memory processing [5]. In Iran, study of 407 students found that 9.1, 36.1, 39.3 and 13.5% of them had excellent, good, satisfactory and poor daily sleep quality. In this research, appropriate sleep duration and adequate sleep period had a positive effect on the academic performance including educational achievement, high scores in the exams and freshness in the classroom [17]. Given the importance of sleep for students, the present study aimed to determine the association between sleep hours and academic performance in young adults.", "STUDY DESIGN This cross-sectional study was designed as a descriptive-analytic study. A sample of adolescents of 14-17 years of age in the first, second and third degree of high school (two schools were chosen, randomly) in Qazvin city were enrolled. Sample size was calculated with considering 8% precision, 95% confidence interval and 80% power about 700 students. Participants were selected using cluster sampling, so that 700 students were chosen from 10 schools in 5 different parts of city (two schools were selected from each area of city). Pre-university students were excluded from study because of the stress due to university entrance exam and its effects on sleep pattern. After selecting the desired school, some students were selected randomly from them.\nThis cross-sectional study was designed as a descriptive-analytic study. A sample of adolescents of 14-17 years of age in the first, second and third degree of high school (two schools were chosen, randomly) in Qazvin city were enrolled. Sample size was calculated with considering 8% precision, 95% confidence interval and 80% power about 700 students. Participants were selected using cluster sampling, so that 700 students were chosen from 10 schools in 5 different parts of city (two schools were selected from each area of city). Pre-university students were excluded from study because of the stress due to university entrance exam and its effects on sleep pattern. After selecting the desired school, some students were selected randomly from them.\nSTUDY TOOL Study’s questionnaires were distributed among them. Data collection tools were two questionnaires which their validity and reliability has been confirmed in previous studies. The Pediatric Sleep Questionnaire consist 22 questions was designed to evaluate sleep problems in children. Its sensitivity and specificity have a range between 0.81 to 0.85, and 0.87, respectively compared to polysomnographic results. Also, Cronbach’s alpha coefficient of questionnaire was 0.77 in this study for PSQ. Also, the BEARS questionnaire developed by Owen used for all students to screen comprehensively major sleeps problems in them. Five sleep domain evaluated by this questionnaire including bedtime problems, excessive daytime sleepiness, awakening during the night, regularity and duration of sleep, and snoring. Previously, Mohammadi and colleague were assessed validity and reliability of Persian version of this questionnaire. The BEARS internal consistency in our study was high with a Cronbach’s alpha of 0.79. A total of 700 questionnaires were distributed. Twenty of questionnaires were excluded because of the incomplete filling of the questions. Also, 27 of students were excluded due to suffering from diseases that effect on their sleep parameters [18-21].\nStudy’s questionnaires were distributed among them. Data collection tools were two questionnaires which their validity and reliability has been confirmed in previous studies. The Pediatric Sleep Questionnaire consist 22 questions was designed to evaluate sleep problems in children. Its sensitivity and specificity have a range between 0.81 to 0.85, and 0.87, respectively compared to polysomnographic results. Also, Cronbach’s alpha coefficient of questionnaire was 0.77 in this study for PSQ. Also, the BEARS questionnaire developed by Owen used for all students to screen comprehensively major sleeps problems in them. Five sleep domain evaluated by this questionnaire including bedtime problems, excessive daytime sleepiness, awakening during the night, regularity and duration of sleep, and snoring. Previously, Mohammadi and colleague were assessed validity and reliability of Persian version of this questionnaire. The BEARS internal consistency in our study was high with a Cronbach’s alpha of 0.79. A total of 700 questionnaires were distributed. Twenty of questionnaires were excluded because of the incomplete filling of the questions. Also, 27 of students were excluded due to suffering from diseases that effect on their sleep parameters [18-21].\nSTATISTICAL ANALYSIS Finally, data collected from 653 cases were confirmed for use in the analysis. Statistical package for social sciences (SPSS) version 16 was used for data analysis. Chi-square test, t-test, analysis of variance (ANOVA), and correlation were performed to determine the relationship between the data (P < 0.001).\nFinally, data collected from 653 cases were confirmed for use in the analysis. Statistical package for social sciences (SPSS) version 16 was used for data analysis. Chi-square test, t-test, analysis of variance (ANOVA), and correlation were performed to determine the relationship between the data (P < 0.001).", "This cross-sectional study was designed as a descriptive-analytic study. A sample of adolescents of 14-17 years of age in the first, second and third degree of high school (two schools were chosen, randomly) in Qazvin city were enrolled. Sample size was calculated with considering 8% precision, 95% confidence interval and 80% power about 700 students. Participants were selected using cluster sampling, so that 700 students were chosen from 10 schools in 5 different parts of city (two schools were selected from each area of city). Pre-university students were excluded from study because of the stress due to university entrance exam and its effects on sleep pattern. After selecting the desired school, some students were selected randomly from them.", "Study’s questionnaires were distributed among them. Data collection tools were two questionnaires which their validity and reliability has been confirmed in previous studies. The Pediatric Sleep Questionnaire consist 22 questions was designed to evaluate sleep problems in children. Its sensitivity and specificity have a range between 0.81 to 0.85, and 0.87, respectively compared to polysomnographic results. Also, Cronbach’s alpha coefficient of questionnaire was 0.77 in this study for PSQ. Also, the BEARS questionnaire developed by Owen used for all students to screen comprehensively major sleeps problems in them. Five sleep domain evaluated by this questionnaire including bedtime problems, excessive daytime sleepiness, awakening during the night, regularity and duration of sleep, and snoring. Previously, Mohammadi and colleague were assessed validity and reliability of Persian version of this questionnaire. The BEARS internal consistency in our study was high with a Cronbach’s alpha of 0.79. A total of 700 questionnaires were distributed. Twenty of questionnaires were excluded because of the incomplete filling of the questions. Also, 27 of students were excluded due to suffering from diseases that effect on their sleep parameters [18-21].", "Finally, data collected from 653 cases were confirmed for use in the analysis. Statistical package for social sciences (SPSS) version 16 was used for data analysis. Chi-square test, t-test, analysis of variance (ANOVA), and correlation were performed to determine the relationship between the data (P < 0.001).", "In this study, 653 adolescents 14-18 years were surveyed, of which 261 (40%) were male and 392 (60%) were female. Average age of our sample was 15.7 ± 0.9. Education levels in students were as follow: 154 (23.6%) in the first year, 242 (37.1%) in the second year of high school and 257 (39.4%) in the third year of high school. Information on age, education grade, hours of sleep during the night, and delay amount in the sleep onset are mentioned (Tab. I). The average sleep duration was significantly higher in boys than girls (P < 0.001). The hours of starting sleep at night and waking in the morning was significantly different between boys and girls (P < 0.05). Sleep onset delay was significantly higher in girls than boys (P < 0.001).\nHours of sleep on weekdays and weekends between boys and girls were compared (Tab. II). Sleep duration was classified in three different categories (less than 7 hours, 7-9 hours, and more than 9 hours during night). Sleep insufficient was significantly higher in girls than boys (P < 0.001). In contrast, the rate of oversleeping was higher in boys. Also, hours of sleep on weekends were higher than weekdays. Pearson analysis showed that time of falling sleep at weekend nights (ρ = 0.83, P = 0.03) and weekday nights (ρ = 0.14, P<0.001) have positively correlation with age.\nAcademic performance of students was divided into three categories based on their GPA. Students with GPA equal or less than 15 was classified in poor academic performance. Also, students with GPA between 15.01-17.99 and equal or more than 18 were classified as medium and good academic performance, respectively. Results showed that academic performance was significantly better in girls than boys (Tab. III).\nStudents who sufferings from difficulty to falling asleep have significantly lower academic performance (P < 0.001) (Tab. IV). Also, there was a significant relationship between students’ sleep hours in weekends and weekdays with academic performance. The average hours of sleep among students with better academic performance was less than students with medium and poor academic performance (P < 0.001). The amount of delay in falling asleep was more in students who had lower academic performance (P = 0.002) and total sleep time during weekend (P < 0.001) and weekdays (P < 0.001) were significantly higher in students with better academic performance (Tab. V).", "In this study, the mean total sleep duration was 42.8 hours during weekdays, and this amount was significantly lower in girls than boys. Also, a significant difference found between boys and girls in terms of the time to go to bed and waking up in the morning. Another result showed that the average number of sleep hours was lower in students who had better academic performance than students with average and poor academic performance. According to the center for disease control and prevention (CDC) in America, the optimal duration of sleep recommended for teenagers are more than 8.5 hours at night [19]. In our sample, duration of sleep for 24 hours was 8.42 ± 1.6 and 10.4 ± 1.8 for weekdays and weekends, respectively. Total duration of sleep was more in boys than girls, and girls went later to sleep than boys at both of the weekends and weekdays. In a study in America the average length of sleep on the weekend was about 9 hours in boys and girls. But, this amount was reduced to 7.1 hours for boys and 5.2 hours for girls during weekdays. These results were different from current study that in our study students’ sleep duration on weekend was 5.1 hours more than American teenagers. Also, the duration of sleep in weekdays was lower for both American girls and boys about 2.1 and 1.8 hours, respectively. So, in total during the whole week, Iranian teenagers slept more than American teenagers. In this study, there was not significant correlation between age and sleep duration. Total sleep duration decreases with increasing age [19]. In another study in Canada, it was found that seventy percent of students aged 14 to 18 years old sleep less than 8.5 hours during the night [10]. In this study, sixty percent of students reported that they sleep less than 8.5 hours during night at weekdays. Also, we found a wide gap in sleep duration between weekends and weekdays (Yo-yo sleeping), so that our samples slept two hours more in weekends compared to weekdays. These results show that students’ sleep was not enough during weekdays, therefore they sleep more in weekends to compensate their sleep insufficiency. This large difference will cause a negative impact on students’ performance in school [20]. This amounts in studies conducted in other countries is as follow: America 2.1 hours [19], Taiwan 1 hour [21], China 2.5 hours [22], and Australia 16 minutes [20]. There was no difference in sleep duration between weekdays and weekend in Switzerland [23]. Current study showed that with increasing students’ age, the time for going to bed were delayed. Similar results were obtained in other studies to confirm our results [21, 24]. In this study, total sleep duration was more in boys than girls in all days of the week. Similar results were obtained in a study conducted in Taiwan [21], but a survey in Australia showed that teenager girls sleep more than boys [20]. Recent findings have shown that sleep is important for the proper function of learning and memory [13, 25]. The possibility of delay in sleep onset increases with the onset of puberty that can lead to daytime sleepiness and negative effects on academic performance in students. A study in America showed that delay in sleep initiation more than 30 minutes for more than one night during the week has a significant association with increased rates of academic failure in students [26]. In another study conducted in China, the average delay in starting sleep was 30 minutes in students [27]. This amount was calculated in the present study 36.77 minutes. A review study in 2010 showed that increasing hours of sleep is associated with better academic performance in students [12]. But in our study, students who have fewer hours of sleep during night had better academic performance. On the other hand, another study did not report a significant relationship between sleep duration and academic performance in students [28]. In a study in Iran, 102 students completed PSQI. Based on the results there was no significant difference between students with high grades and those with low grades. But there were moderate and sometimes severe sleep disturbances in both groups. Also, there was no significant difference between sleep quality and academic achievement [29]. In another hand, in our results significant difference between sleep quality and academic performances between students was existed. A research between 341 selected students in Iran showed that 59.1% of them had poor sleep quality. Also, there was a significant negative relationship between sleep quality with academic interaction and academic vitality. There was a significant positive relationship between academic motivations with academic vitality. Also, a significant difference between male and female students in academic vitality was observed [30]. Different reasons such as level of family income, family size, intake of supplements and vitamins, social media dependency, addiction to social networks and social issues can affect the academic success in the different students [29, 31]. Improve adolescent sleep including delaying school start times, providing sleep education, and utilizing light therapy to improve the health, wellbeing and academic performance of sleepy teenagers are in the programs of researchers in this field in different countries [30]. The results from this study indicated that sleeping time in Iranian students is more than students in other countries. In this study, sleep duration, sleep onset delay, sleep insufficient and rate of oversleeping was associated with better academic performance in students. Also, difficulty in falling asleep was associated with weaker academic performance. The overall result was that in students without difficulty in falling asleep, a positive influence on the academic performance was observed.", "The result of this study showed that some of sleep characteristics such as sleep duration, sleep onset delay, sleep insufficient and rate of oversleeping of students could be a significant influence on adolescents’ academic performance. Further studies are needed to objectively determine the effect of sleep variables on adolescents’ performance.\nLIMITATIONS First, the design is cross sectional. Therefore, it may be difficult to confirm a cause-effect relationship. Moreover, the selected students were found randomly. Also, findings may not be applicable to all students in other geographical locations.\nFirst, the design is cross sectional. Therefore, it may be difficult to confirm a cause-effect relationship. Moreover, the selected students were found randomly. Also, findings may not be applicable to all students in other geographical locations." ]

[ null, "methods", null, null, null, null, null, null ]

[ "Sleep duration", "School performance", "Adolescents" ]

[ 575, 144, 209, 64, 495, 1084, 151 ]

[ "sleep", "students", "study", "academic", "performance", "hours", "academic performance", "duration", "boys", "girls" ]

[ "sleep insufficiency", "inadequate poor sleep", "reduce sleep body", "diseases effect sleep", "appetite chronic sleep" ]

[ "Adolescent", "Child", "Cross-Sectional Studies", "Dental Caries", "Dental Plaque", "Ethiopia", "Feeding Behavior", "Female", "Humans", "Male", "Oral Hygiene", "Prevalence", "Risk Factors", "Schools", "Surveys and Questionnaires", "Urban Health", "Urban Population" ]

[ "Background", "Study design and area", "Study participants", "Sample size and sampling", "Data collection", "Data analysis", "Ethical considerations", "Sociodemographic characteristics", "Food consumption pattern, dietary habits and practices related to oral hygiene", "Dental caries", "Risk factors associated with dental caries", "Authors’ information" ]

[ "Dental caries is the most prevalent and chronic oral disease particularly in childhood age [1, 2]. Dental caries is a progressive infectious process with a multifactorial etiology [3, 4]. Dietary habits, oral microorganisms that ferment sugars, and host susceptibility have to coexist for dental caries to initiate and develop [3–5]. Dental caries has high morbidity potential. Thus, it has been the main focus of dental health professionals [6].\nDental caries is caused by dental plaque deposits on the tooth surface [7, 8]. After intake of fermentable carbohydrates, Streptococcus mutans undergo fermentation and produce copious amount of acid and lowers the local pH to a level where the minerals of enamel and dentine dissolve [3, 5, 7–11]. The frequent intake of sweets, dry mouth, and poor oral hygiene increase the chances for cavities [8, 12].\nDental caries causes teeth pain, discomfort, eating impairment, loss of tooth and delay language development. Furthermore, dental caries has effects on childern’s concentration in school and a financial burden on the families [6, 13]. Risk factors such as sex, age, dietary habits, socioeconomic and oral hygiene status are associated with increased prevalence and incidence of dental caries in a population [14].\nAlthough, the trend is not clear in developing countries, the burden of dental caries has been increasing among children due to the unlimited consumption of sugary substances, poor oral care practices and inadequate health service utilization [15]. Studies revealed that the prevalence of dental caries was higher among urban children [14, 16]. Similarly, a study conducted in Ethiopia was reported 36.5% prevalence of dental caries among urban children in school [8]. Although, dental caries is more prevalent in school children, there was no documented data on prevailing prevalence and associated factors in primary school children in Bahir Dar city. Therefore, the present study was carried out to determine the prevalence and associated factors of dental caries among primary school children.", "A school based cross-sectional study was conducted from October 2013 to January 2014 among primary school children at Bahir Dar city. Bahir Dar city is 560 kilometers away from the capital city, Addis Ababa. According to 2011 Central Statistical Agency of Ethiopia (CSA) estimates, the city has a total population of 220,344. Urban primary school children under 16 years accounted for 27,511. Of these, 20,340 were from public schools. The town has a total of 35 urban primary schools. Of these, 22 were private and the rest 13 were public schools [17].", "Children from primary schools between 6 to 15 years of age and living at Bahir Dar city were included in the study. Children with 16 years of age and above, and those from private schools were excluded from the study.", "Sample size was calculated using single population proportion formula with an assumption of 95% confidency level, 7% degree of precision and proportion of dental caries, 36.5% [8] to make the final sample size of 180. However, only 147 students provided a complete response. Systematic random sampling technique was employed to select the study participants. Among thirteen government schools, three were selected using systematic random sampling technique. The sample size was allocated proportionally based on the number of children in each selected school. Children were selected randomly based on their name lists taken from their rosters in respective class.", "A structured questionnaire was used to collect socio-demographic characteristics, dietary habits, oral health problems and oral care practices. Dental examination was carried out for all selected children by one trained dental doctor using World Health Organization (WHO) dental caries diagnosis guide line under natural day light [15]. Disposable wooden spatulas were used for intraoral examination. Prior to the study, data collectors were given for two days intensive training on dental caries assessment based on WHO guide line and on how to interview children and fill the questionnaire. Incomplete questionnaires were rejected during data analysis. Dental caries was recorded as being present when a lesion in a pit or fissure or on smooth tooth surface had a detectable softened floor, undermined enamel or softened wall. When any doubt existed, dental caries was not recorded as present. Tooth was considered missing because of caries if a person gave a history of pain and/or presence of cavity prior to extraction.\nThe presence of dental plaque was assessed by direct visual inspection and palpation of the buccal and lingual surfaces of all teeth with clean glove and spatula [18]. Plaque was recorded as being present when visible deposits were detected and then removed following palpation of the teeth by clean gloved hand. Moreover, the presence of both hypo calcification and incipient caries type of white spot lesions were examined by conventional diagnostic technique [18]. First, the wet teeth were inspected for the presence of hypo calcification type of white spot lesion then the teeth were allowed to wipe cleaned and dried with gauze and compressed air to inspect incipient caries type of white spot lesion. White spot lesion was recorded as being present when a white chalky appearance spot were revealed either in dehydrated or desiccated or both type of the upper and lower anterior of enamel.", "Data was entered and analyzed using statistical package for social science (SPSS) version 20. Frequency and percentage were computed from univariate analysis to get summary values. Odds ratio with 95% confidence interval (CI) was computed using logistic regression analysis to assess the presence and degree of association between dental caries and independent variables. Significance was set at p < 0.05 (significance level 95%). For those variables that had a p-value < 0.05 on binary logistic regression, binary multiple logistic regression analysis was computed.", "Ethical clearance was obtained from ethical review committee of College of Medicine and Health Sciences, Bahir Dar University. A written consent was obtained from children’s parents before interview and dental examination. Cases of dental caries were advised to attend the nearby dental clinic.", "A total of 147 children were participated in the study. Of these, 82 (55.4%) were girls. The majority of children (69.6%) were from 11 to 15 years of age. Nearly half of the study participants were grade 1–4. Twenty one (14%) of the students’ parent had an education above grade 12. Eighty (54.1%) of the participants family earned below 1000 Ethiopian birr per month (Table 1).Table 1\nPrevalence of dental caries and socio-demographic characteristics amon\ng primary school children at Bahir Dar city, 2014 (n = 147)\nSocio-demographic variablesDental carriesCOR (95% CI)P- valuePositive N (%)Negative N (%)Total N (%)\nAge in years\n6 -1015 (33.3)30 (66.7)45 (30.4)0.4 (0.18 - 1.0)0.0311 -1517 (16.7)85 (83.3)102 (69.6\n1\n\nSex\nBoys12 (18.5)53 (81.5)65 (44.6)\n1\nGirls20 (24.4)62 (75.6)82 (55.4)0.7 (0.31 - 1.57)0.87\nGrade\n1- 423 (31.9)49 (68.1)72 (49)0.26 (0.07 - 0.92)0.035- 89 (12.2)65 (87.8)75 (51)\n1\n\nFamily income\n<100021 (26.3)59 (73.8)80 (54.1)0.23 (0.07- 0.72)0.21001-20005 (11.9)37 (88.1)42 (28.4)0.24 (0.63 - 8.66)0.82>20006 (24)19 (76)25 (17.6)\n1\n\nFamily educational status\nIlliterate10 (18.5)44 (81.5)54 (36.7)Can read and write7 (38.9)11 (61.1)18 (12.2)0.031-8 grade level7 (31.8)15 (68.2)22 (15)9-12 grade level8 (25)24 (75)32 (21.8)>12 grade level021 (100)21 (14.3)\nTotal\n\n32 (21.8)\n\n115 (78.2)\n\n147 (100)\nKey: COR (crude odds ratio), CI (Confidence interval), 1(Reference category).\n\nPrevalence of dental caries and socio-demographic characteristics amon\ng primary school children at Bahir Dar city, 2014 (n = 147)\n\nKey: COR (crude odds ratio), CI (Confidence interval), 1(Reference category).", "One hundred four (70.7%) of the children had breakfast bread with tea. Most of the children (85%) usually drank tea with sugar. Thirty one (21.1%) of the participants drank coffee with sugar. Fifty five (37.4%) of the children drank soft drinks. Seventy one, (48.3%) of the children used to eat sweet foods. One hundred five (71.4%) of the children were used to clean their teeth. Of whom, 16 (15.2%) cleaned their teeth before and after meal. However, nearly half of the children cleaned their teeth only after meal intake. Majority of the children (67.6%) used a traditional small stick of wood (termed as Mafaqiya) made of a special type of plant to clean their teeth. However, 18 (9.5%) and 5 (4.8%) were used teeth brush with and without paste, respectively to clean their teeth (Table 2).Table 2\nFood consumption pattern, dietary habits and practices of oral hygiene among primary school children at Bahir Dar city, 2014\nVariablesDental cariesP-valuePositive N (%)Negative N (%)Total\nConsumption of sugared tea (n = 147)\nYes26 (20.8)99 (79.2)125 (85)0.33No6 (27.3)16 (72.7)22 (15)\nConsumption of sugared coffee (n = 147)\nYes6 (19.4)25 (80.6)31 (21.1)0.81No26 (22.6)90 (73.4)116 (78.9)\nType of food for breakfast (n = 147)\nBread with tea24 (23.1)80 (76.9)104 (70.7)0.47Pasta7 (21.1)28 (78.8)35 (23.8)Other (Makorony, Rice)088 (5.4)\nConsumption of sweet foods (n = 147)\nYes21 (29.6)50 (70.4)71 (48.3)0.03No11 (14.7)64 (85.3)76 (51.7)\nConsumption of soft drinks (n = 147)\nYes15 (27.3)40 (72.7)55 (37.4)0.21No17 (18.5)75 (81.5)92 (62.6)\nFrequency of taking soft drinks (n = 55)\n<4/day11 (22)39 (76.5)50 (91.1)0.02>4/day3 (75)1 (25)5 (7.3)\nCleaning teeth ( n = 147)\nYes17 (16)88 (84)105 (71.4)0.016No15 (36.6)26(63.4)41 (27.9)\nTime of tooth cleaning (n = 105)\nBefore meal5 (16.7)25 (83.3)30 (28.6)0.63After meal7 (13.2)46 (86.8)53 (50.5)Before and after meal3 (18.8)13 (81.2)16 (15.2)No fixed time2 (33.3)4 (66.7)6 (5.7)\nWay of cleaning teeth (n = 105)\nTooth stick10 (14.1)61 (85.9)71 (67.6)0.52Tooth brush without paste055 (4.8)Tooth brush with paste5 (27.8)13(72.2)18 (9.5)Charcoal011 (0.96)Rinse with water1 (25)3(75)4 (3.8)Other means1 (16.7)5(83.3)6 (5.7)Key: N (number), % (percent).\n\nFood consumption pattern, dietary habits and practices of oral hygiene among primary school children at Bahir Dar city, 2014\n\nKey: N (number), % (percent).", "Of the 147 study participants, 32 (21.8%) had decayed tooth. The proportion of dental caries was 33.3% in children from 6 to 10 years of age. Girls were with a higher proportion of dental caries (24.4%) than boys (18.5%). However, the difference was not statistically significant (P = 0.87). The proportion of dental caries was 23 (31.9%) and 9 (12.2%) among children from grade1- 4 and 5–8, respectively. Children belonging to the lowest income group had the highest proportion of dental caries but the highest income group had higher prevalence than the middle income group (Table 1). Among the total dental caries, the majority 24 (75%) had primary tooth decay. Of the children who had dental caries, 12 (50%) had more than one affected tooth and 7 (21.9%) revealed missed teeth. Toothache and white spot lesions were found in 40 (27.2%) and 12 (8.2%) of children, respectively. However, dental plaque was clinically visible in 99 (67.3%) of the children. To get treatment for dental caries, 9 children (6.1%) had consulted a dentist (Table 3).Table 3\nDental cases of primary school children aged 6–15 years, at Bahir Dar city, 2014\nVariablesFrequencyPercent\nTooth decay/dental caries (any type) (n = 147)\n   Yes3221.8   No11578.2\nType of tooth decayed (n = 32)\n   Decay of primary tooth2475   Decay of permanent tooth825\nNumber of primary teeth decay (n = 32)\n   1 tooth1250   2 teeth625   3 teeth520.8   4 teeth14.2\nType of missed teeth (n = 7)\n   Primary571.4   Permanent228.6\nTooth ache (n = 147)\n   Yes4027.2   No10772.8\nWhite spot lesions (n = 147)\n   Yes128.2   No13591.8\nPlaque accumulation (n = 147)\n   Yes9967.3   No4832.7\nHealth institution / Dental visit (n = 147)\n   Yes96.1   No13893.9\n\nDental cases of primary school children aged 6–15 years, at Bahir Dar city, 2014\n", "Based on bivariate analysis, significant association between dental caries and educational status of children’s parents was found (P = 0.03). Dental caries among children, whose parents’ education are above grade 12 were 100% times at a lower risk compared to those who had non-educated parents (Table 1). There was significant association between dental caries and grade levels of children (AOR = 3.9, 95% CI = 1.49 - 10.4). Children whose grade level 1–4 were more likely to have dental caries compared to grade level of 5 – 8. Children who did not clean their teeth were 2.6 times more likely to have caries than those who cleaned (AOR = 2.6, 95% CI = 1.08 - 6.2). Children who had dental plaque were 5.3 times more likely to have dental caries than those who had not (AOR = 5.3, 95% C I = 1.6 – 17.7). Moreover, the odds of having dental caries was significantly higher among children suffer with tooth ache than those children who had not (AOR = 6.3, 95% CI = 2.4 -15.4) (Table 4).Table 4\nFactors associated with dental caries in primary school children at Bahir Dar city, 2014\nCharacteristicsDental cariesCOR (95% CI)AOR (95% CI)P-valueYesNo\nAge in years\n6-1015300.4 (0.18 - 0.89)**0.62 (0.2 -1.92)0.6211-151785\n1\n\n1\n\nGrade level\n1- 423500.3 (0.13 - 0.71**3.9 (1.49 -10.4)0.0065- 8965\n1\n\n1\n\nCleaning teeth\nYes1789\n1\n\n1\nNo15260.35 (0.15 - 0.79)**2.6 (1.08-6.2)0.033\nUse of sweet foods\nYes21500.4 (0.18 - 0.91)**0.46 (0.18 -1.16)0.09No1165\n1\n\n1\n\nTooth ache\nYes17234.5 (1.97 - 10.41)***6.3 (2.4 - 15.4)<0.001No1592\n1\n\n1\n\nPlague accumulation\nYes28714.3 (1.43 – 13.2)**5.3 (1.6 - 17.7)0.006No4411Key: COR (Crude odds ratio), AOR (adjusted odds ratio), CI (Confidence interval).**:P-value < 0.05, ***:0.05 < P-value < 0.001.\n\nFactors associated with dental caries in primary school children at Bahir Dar city, 2014\n\nKey: COR (Crude odds ratio), AOR (adjusted odds ratio), CI (Confidence interval).\n**:P-value < 0.05, ***:0.05 < P-value < 0.001.", "WM is an assistant professor at College of Medicine and Health Sciences, Bahir Dar University in Medical Microbiology, BA is an associate professor at college of Medicine and Health Sciences, Bahir Dar University in Medical Microbiology and department head of Medical Microbiology, Immunology and Parasitology, MY is an assistant professor at College of Medicine and Health Sciences, Bahir Dar University in Medical Parasitology. KM is lecturer at college of Medicine and Health Sciences, Bahir Dar University in doctor of dentistry." ]

[ null, null, null, null, null, null, null, null, null, null, null, null ]

[ "Background", "Methods", "Study design and area", "Study participants", "Sample size and sampling", "Data collection", "Data analysis", "Ethical considerations", "Results", "Sociodemographic characteristics", "Food consumption pattern, dietary habits and practices related to oral hygiene", "Dental caries", "Risk factors associated with dental caries", "Discussion", "Conclusion", "Authors’ information" ]

[ "Dental caries is the most prevalent and chronic oral disease particularly in childhood age [1, 2]. Dental caries is a progressive infectious process with a multifactorial etiology [3, 4]. Dietary habits, oral microorganisms that ferment sugars, and host susceptibility have to coexist for dental caries to initiate and develop [3–5]. Dental caries has high morbidity potential. Thus, it has been the main focus of dental health professionals [6].\nDental caries is caused by dental plaque deposits on the tooth surface [7, 8]. After intake of fermentable carbohydrates, Streptococcus mutans undergo fermentation and produce copious amount of acid and lowers the local pH to a level where the minerals of enamel and dentine dissolve [3, 5, 7–11]. The frequent intake of sweets, dry mouth, and poor oral hygiene increase the chances for cavities [8, 12].\nDental caries causes teeth pain, discomfort, eating impairment, loss of tooth and delay language development. Furthermore, dental caries has effects on childern’s concentration in school and a financial burden on the families [6, 13]. Risk factors such as sex, age, dietary habits, socioeconomic and oral hygiene status are associated with increased prevalence and incidence of dental caries in a population [14].\nAlthough, the trend is not clear in developing countries, the burden of dental caries has been increasing among children due to the unlimited consumption of sugary substances, poor oral care practices and inadequate health service utilization [15]. Studies revealed that the prevalence of dental caries was higher among urban children [14, 16]. Similarly, a study conducted in Ethiopia was reported 36.5% prevalence of dental caries among urban children in school [8]. Although, dental caries is more prevalent in school children, there was no documented data on prevailing prevalence and associated factors in primary school children in Bahir Dar city. Therefore, the present study was carried out to determine the prevalence and associated factors of dental caries among primary school children.", " Study design and area A school based cross-sectional study was conducted from October 2013 to January 2014 among primary school children at Bahir Dar city. Bahir Dar city is 560 kilometers away from the capital city, Addis Ababa. According to 2011 Central Statistical Agency of Ethiopia (CSA) estimates, the city has a total population of 220,344. Urban primary school children under 16 years accounted for 27,511. Of these, 20,340 were from public schools. The town has a total of 35 urban primary schools. Of these, 22 were private and the rest 13 were public schools [17].\nA school based cross-sectional study was conducted from October 2013 to January 2014 among primary school children at Bahir Dar city. Bahir Dar city is 560 kilometers away from the capital city, Addis Ababa. According to 2011 Central Statistical Agency of Ethiopia (CSA) estimates, the city has a total population of 220,344. Urban primary school children under 16 years accounted for 27,511. Of these, 20,340 were from public schools. The town has a total of 35 urban primary schools. Of these, 22 were private and the rest 13 were public schools [17].\n Study participants Children from primary schools between 6 to 15 years of age and living at Bahir Dar city were included in the study. Children with 16 years of age and above, and those from private schools were excluded from the study.\nChildren from primary schools between 6 to 15 years of age and living at Bahir Dar city were included in the study. Children with 16 years of age and above, and those from private schools were excluded from the study.\n Sample size and sampling Sample size was calculated using single population proportion formula with an assumption of 95% confidency level, 7% degree of precision and proportion of dental caries, 36.5% [8] to make the final sample size of 180. However, only 147 students provided a complete response. Systematic random sampling technique was employed to select the study participants. Among thirteen government schools, three were selected using systematic random sampling technique. The sample size was allocated proportionally based on the number of children in each selected school. Children were selected randomly based on their name lists taken from their rosters in respective class.\nSample size was calculated using single population proportion formula with an assumption of 95% confidency level, 7% degree of precision and proportion of dental caries, 36.5% [8] to make the final sample size of 180. However, only 147 students provided a complete response. Systematic random sampling technique was employed to select the study participants. Among thirteen government schools, three were selected using systematic random sampling technique. The sample size was allocated proportionally based on the number of children in each selected school. Children were selected randomly based on their name lists taken from their rosters in respective class.\n Data collection A structured questionnaire was used to collect socio-demographic characteristics, dietary habits, oral health problems and oral care practices. Dental examination was carried out for all selected children by one trained dental doctor using World Health Organization (WHO) dental caries diagnosis guide line under natural day light [15]. Disposable wooden spatulas were used for intraoral examination. Prior to the study, data collectors were given for two days intensive training on dental caries assessment based on WHO guide line and on how to interview children and fill the questionnaire. Incomplete questionnaires were rejected during data analysis. Dental caries was recorded as being present when a lesion in a pit or fissure or on smooth tooth surface had a detectable softened floor, undermined enamel or softened wall. When any doubt existed, dental caries was not recorded as present. Tooth was considered missing because of caries if a person gave a history of pain and/or presence of cavity prior to extraction.\nThe presence of dental plaque was assessed by direct visual inspection and palpation of the buccal and lingual surfaces of all teeth with clean glove and spatula [18]. Plaque was recorded as being present when visible deposits were detected and then removed following palpation of the teeth by clean gloved hand. Moreover, the presence of both hypo calcification and incipient caries type of white spot lesions were examined by conventional diagnostic technique [18]. First, the wet teeth were inspected for the presence of hypo calcification type of white spot lesion then the teeth were allowed to wipe cleaned and dried with gauze and compressed air to inspect incipient caries type of white spot lesion. White spot lesion was recorded as being present when a white chalky appearance spot were revealed either in dehydrated or desiccated or both type of the upper and lower anterior of enamel.\nA structured questionnaire was used to collect socio-demographic characteristics, dietary habits, oral health problems and oral care practices. Dental examination was carried out for all selected children by one trained dental doctor using World Health Organization (WHO) dental caries diagnosis guide line under natural day light [15]. Disposable wooden spatulas were used for intraoral examination. Prior to the study, data collectors were given for two days intensive training on dental caries assessment based on WHO guide line and on how to interview children and fill the questionnaire. Incomplete questionnaires were rejected during data analysis. Dental caries was recorded as being present when a lesion in a pit or fissure or on smooth tooth surface had a detectable softened floor, undermined enamel or softened wall. When any doubt existed, dental caries was not recorded as present. Tooth was considered missing because of caries if a person gave a history of pain and/or presence of cavity prior to extraction.\nThe presence of dental plaque was assessed by direct visual inspection and palpation of the buccal and lingual surfaces of all teeth with clean glove and spatula [18]. Plaque was recorded as being present when visible deposits were detected and then removed following palpation of the teeth by clean gloved hand. Moreover, the presence of both hypo calcification and incipient caries type of white spot lesions were examined by conventional diagnostic technique [18]. First, the wet teeth were inspected for the presence of hypo calcification type of white spot lesion then the teeth were allowed to wipe cleaned and dried with gauze and compressed air to inspect incipient caries type of white spot lesion. White spot lesion was recorded as being present when a white chalky appearance spot were revealed either in dehydrated or desiccated or both type of the upper and lower anterior of enamel.\n Data analysis Data was entered and analyzed using statistical package for social science (SPSS) version 20. Frequency and percentage were computed from univariate analysis to get summary values. Odds ratio with 95% confidence interval (CI) was computed using logistic regression analysis to assess the presence and degree of association between dental caries and independent variables. Significance was set at p < 0.05 (significance level 95%). For those variables that had a p-value < 0.05 on binary logistic regression, binary multiple logistic regression analysis was computed.\nData was entered and analyzed using statistical package for social science (SPSS) version 20. Frequency and percentage were computed from univariate analysis to get summary values. Odds ratio with 95% confidence interval (CI) was computed using logistic regression analysis to assess the presence and degree of association between dental caries and independent variables. Significance was set at p < 0.05 (significance level 95%). For those variables that had a p-value < 0.05 on binary logistic regression, binary multiple logistic regression analysis was computed.\n Ethical considerations Ethical clearance was obtained from ethical review committee of College of Medicine and Health Sciences, Bahir Dar University. A written consent was obtained from children’s parents before interview and dental examination. Cases of dental caries were advised to attend the nearby dental clinic.\nEthical clearance was obtained from ethical review committee of College of Medicine and Health Sciences, Bahir Dar University. A written consent was obtained from children’s parents before interview and dental examination. Cases of dental caries were advised to attend the nearby dental clinic.", "A school based cross-sectional study was conducted from October 2013 to January 2014 among primary school children at Bahir Dar city. Bahir Dar city is 560 kilometers away from the capital city, Addis Ababa. According to 2011 Central Statistical Agency of Ethiopia (CSA) estimates, the city has a total population of 220,344. Urban primary school children under 16 years accounted for 27,511. Of these, 20,340 were from public schools. The town has a total of 35 urban primary schools. Of these, 22 were private and the rest 13 were public schools [17].", "Children from primary schools between 6 to 15 years of age and living at Bahir Dar city were included in the study. Children with 16 years of age and above, and those from private schools were excluded from the study.", "Sample size was calculated using single population proportion formula with an assumption of 95% confidency level, 7% degree of precision and proportion of dental caries, 36.5% [8] to make the final sample size of 180. However, only 147 students provided a complete response. Systematic random sampling technique was employed to select the study participants. Among thirteen government schools, three were selected using systematic random sampling technique. The sample size was allocated proportionally based on the number of children in each selected school. Children were selected randomly based on their name lists taken from their rosters in respective class.", "A structured questionnaire was used to collect socio-demographic characteristics, dietary habits, oral health problems and oral care practices. Dental examination was carried out for all selected children by one trained dental doctor using World Health Organization (WHO) dental caries diagnosis guide line under natural day light [15]. Disposable wooden spatulas were used for intraoral examination. Prior to the study, data collectors were given for two days intensive training on dental caries assessment based on WHO guide line and on how to interview children and fill the questionnaire. Incomplete questionnaires were rejected during data analysis. Dental caries was recorded as being present when a lesion in a pit or fissure or on smooth tooth surface had a detectable softened floor, undermined enamel or softened wall. When any doubt existed, dental caries was not recorded as present. Tooth was considered missing because of caries if a person gave a history of pain and/or presence of cavity prior to extraction.\nThe presence of dental plaque was assessed by direct visual inspection and palpation of the buccal and lingual surfaces of all teeth with clean glove and spatula [18]. Plaque was recorded as being present when visible deposits were detected and then removed following palpation of the teeth by clean gloved hand. Moreover, the presence of both hypo calcification and incipient caries type of white spot lesions were examined by conventional diagnostic technique [18]. First, the wet teeth were inspected for the presence of hypo calcification type of white spot lesion then the teeth were allowed to wipe cleaned and dried with gauze and compressed air to inspect incipient caries type of white spot lesion. White spot lesion was recorded as being present when a white chalky appearance spot were revealed either in dehydrated or desiccated or both type of the upper and lower anterior of enamel.", "Data was entered and analyzed using statistical package for social science (SPSS) version 20. Frequency and percentage were computed from univariate analysis to get summary values. Odds ratio with 95% confidence interval (CI) was computed using logistic regression analysis to assess the presence and degree of association between dental caries and independent variables. Significance was set at p < 0.05 (significance level 95%). For those variables that had a p-value < 0.05 on binary logistic regression, binary multiple logistic regression analysis was computed.", "Ethical clearance was obtained from ethical review committee of College of Medicine and Health Sciences, Bahir Dar University. A written consent was obtained from children’s parents before interview and dental examination. Cases of dental caries were advised to attend the nearby dental clinic.", " Sociodemographic characteristics A total of 147 children were participated in the study. Of these, 82 (55.4%) were girls. The majority of children (69.6%) were from 11 to 15 years of age. Nearly half of the study participants were grade 1–4. Twenty one (14%) of the students’ parent had an education above grade 12. Eighty (54.1%) of the participants family earned below 1000 Ethiopian birr per month (Table 1).Table 1\nPrevalence of dental caries and socio-demographic characteristics amon\ng primary school children at Bahir Dar city, 2014 (n = 147)\nSocio-demographic variablesDental carriesCOR (95% CI)P- valuePositive N (%)Negative N (%)Total N (%)\nAge in years\n6 -1015 (33.3)30 (66.7)45 (30.4)0.4 (0.18 - 1.0)0.0311 -1517 (16.7)85 (83.3)102 (69.6\n1\n\nSex\nBoys12 (18.5)53 (81.5)65 (44.6)\n1\nGirls20 (24.4)62 (75.6)82 (55.4)0.7 (0.31 - 1.57)0.87\nGrade\n1- 423 (31.9)49 (68.1)72 (49)0.26 (0.07 - 0.92)0.035- 89 (12.2)65 (87.8)75 (51)\n1\n\nFamily income\n<100021 (26.3)59 (73.8)80 (54.1)0.23 (0.07- 0.72)0.21001-20005 (11.9)37 (88.1)42 (28.4)0.24 (0.63 - 8.66)0.82>20006 (24)19 (76)25 (17.6)\n1\n\nFamily educational status\nIlliterate10 (18.5)44 (81.5)54 (36.7)Can read and write7 (38.9)11 (61.1)18 (12.2)0.031-8 grade level7 (31.8)15 (68.2)22 (15)9-12 grade level8 (25)24 (75)32 (21.8)>12 grade level021 (100)21 (14.3)\nTotal\n\n32 (21.8)\n\n115 (78.2)\n\n147 (100)\nKey: COR (crude odds ratio), CI (Confidence interval), 1(Reference category).\n\nPrevalence of dental caries and socio-demographic characteristics amon\ng primary school children at Bahir Dar city, 2014 (n = 147)\n\nKey: COR (crude odds ratio), CI (Confidence interval), 1(Reference category).\nA total of 147 children were participated in the study. Of these, 82 (55.4%) were girls. The majority of children (69.6%) were from 11 to 15 years of age. Nearly half of the study participants were grade 1–4. Twenty one (14%) of the students’ parent had an education above grade 12. Eighty (54.1%) of the participants family earned below 1000 Ethiopian birr per month (Table 1).Table 1\nPrevalence of dental caries and socio-demographic characteristics amon\ng primary school children at Bahir Dar city, 2014 (n = 147)\nSocio-demographic variablesDental carriesCOR (95% CI)P- valuePositive N (%)Negative N (%)Total N (%)\nAge in years\n6 -1015 (33.3)30 (66.7)45 (30.4)0.4 (0.18 - 1.0)0.0311 -1517 (16.7)85 (83.3)102 (69.6\n1\n\nSex\nBoys12 (18.5)53 (81.5)65 (44.6)\n1\nGirls20 (24.4)62 (75.6)82 (55.4)0.7 (0.31 - 1.57)0.87\nGrade\n1- 423 (31.9)49 (68.1)72 (49)0.26 (0.07 - 0.92)0.035- 89 (12.2)65 (87.8)75 (51)\n1\n\nFamily income\n<100021 (26.3)59 (73.8)80 (54.1)0.23 (0.07- 0.72)0.21001-20005 (11.9)37 (88.1)42 (28.4)0.24 (0.63 - 8.66)0.82>20006 (24)19 (76)25 (17.6)\n1\n\nFamily educational status\nIlliterate10 (18.5)44 (81.5)54 (36.7)Can read and write7 (38.9)11 (61.1)18 (12.2)0.031-8 grade level7 (31.8)15 (68.2)22 (15)9-12 grade level8 (25)24 (75)32 (21.8)>12 grade level021 (100)21 (14.3)\nTotal\n\n32 (21.8)\n\n115 (78.2)\n\n147 (100)\nKey: COR (crude odds ratio), CI (Confidence interval), 1(Reference category).\n\nPrevalence of dental caries and socio-demographic characteristics amon\ng primary school children at Bahir Dar city, 2014 (n = 147)\n\nKey: COR (crude odds ratio), CI (Confidence interval), 1(Reference category).\n Food consumption pattern, dietary habits and practices related to oral hygiene One hundred four (70.7%) of the children had breakfast bread with tea. Most of the children (85%) usually drank tea with sugar. Thirty one (21.1%) of the participants drank coffee with sugar. Fifty five (37.4%) of the children drank soft drinks. Seventy one, (48.3%) of the children used to eat sweet foods. One hundred five (71.4%) of the children were used to clean their teeth. Of whom, 16 (15.2%) cleaned their teeth before and after meal. However, nearly half of the children cleaned their teeth only after meal intake. Majority of the children (67.6%) used a traditional small stick of wood (termed as Mafaqiya) made of a special type of plant to clean their teeth. However, 18 (9.5%) and 5 (4.8%) were used teeth brush with and without paste, respectively to clean their teeth (Table 2).Table 2\nFood consumption pattern, dietary habits and practices of oral hygiene among primary school children at Bahir Dar city, 2014\nVariablesDental cariesP-valuePositive N (%)Negative N (%)Total\nConsumption of sugared tea (n = 147)\nYes26 (20.8)99 (79.2)125 (85)0.33No6 (27.3)16 (72.7)22 (15)\nConsumption of sugared coffee (n = 147)\nYes6 (19.4)25 (80.6)31 (21.1)0.81No26 (22.6)90 (73.4)116 (78.9)\nType of food for breakfast (n = 147)\nBread with tea24 (23.1)80 (76.9)104 (70.7)0.47Pasta7 (21.1)28 (78.8)35 (23.8)Other (Makorony, Rice)088 (5.4)\nConsumption of sweet foods (n = 147)\nYes21 (29.6)50 (70.4)71 (48.3)0.03No11 (14.7)64 (85.3)76 (51.7)\nConsumption of soft drinks (n = 147)\nYes15 (27.3)40 (72.7)55 (37.4)0.21No17 (18.5)75 (81.5)92 (62.6)\nFrequency of taking soft drinks (n = 55)\n<4/day11 (22)39 (76.5)50 (91.1)0.02>4/day3 (75)1 (25)5 (7.3)\nCleaning teeth ( n = 147)\nYes17 (16)88 (84)105 (71.4)0.016No15 (36.6)26(63.4)41 (27.9)\nTime of tooth cleaning (n = 105)\nBefore meal5 (16.7)25 (83.3)30 (28.6)0.63After meal7 (13.2)46 (86.8)53 (50.5)Before and after meal3 (18.8)13 (81.2)16 (15.2)No fixed time2 (33.3)4 (66.7)6 (5.7)\nWay of cleaning teeth (n = 105)\nTooth stick10 (14.1)61 (85.9)71 (67.6)0.52Tooth brush without paste055 (4.8)Tooth brush with paste5 (27.8)13(72.2)18 (9.5)Charcoal011 (0.96)Rinse with water1 (25)3(75)4 (3.8)Other means1 (16.7)5(83.3)6 (5.7)Key: N (number), % (percent).\n\nFood consumption pattern, dietary habits and practices of oral hygiene among primary school children at Bahir Dar city, 2014\n\nKey: N (number), % (percent).\nOne hundred four (70.7%) of the children had breakfast bread with tea. Most of the children (85%) usually drank tea with sugar. Thirty one (21.1%) of the participants drank coffee with sugar. Fifty five (37.4%) of the children drank soft drinks. Seventy one, (48.3%) of the children used to eat sweet foods. One hundred five (71.4%) of the children were used to clean their teeth. Of whom, 16 (15.2%) cleaned their teeth before and after meal. However, nearly half of the children cleaned their teeth only after meal intake. Majority of the children (67.6%) used a traditional small stick of wood (termed as Mafaqiya) made of a special type of plant to clean their teeth. However, 18 (9.5%) and 5 (4.8%) were used teeth brush with and without paste, respectively to clean their teeth (Table 2).Table 2\nFood consumption pattern, dietary habits and practices of oral hygiene among primary school children at Bahir Dar city, 2014\nVariablesDental cariesP-valuePositive N (%)Negative N (%)Total\nConsumption of sugared tea (n = 147)\nYes26 (20.8)99 (79.2)125 (85)0.33No6 (27.3)16 (72.7)22 (15)\nConsumption of sugared coffee (n = 147)\nYes6 (19.4)25 (80.6)31 (21.1)0.81No26 (22.6)90 (73.4)116 (78.9)\nType of food for breakfast (n = 147)\nBread with tea24 (23.1)80 (76.9)104 (70.7)0.47Pasta7 (21.1)28 (78.8)35 (23.8)Other (Makorony, Rice)088 (5.4)\nConsumption of sweet foods (n = 147)\nYes21 (29.6)50 (70.4)71 (48.3)0.03No11 (14.7)64 (85.3)76 (51.7)\nConsumption of soft drinks (n = 147)\nYes15 (27.3)40 (72.7)55 (37.4)0.21No17 (18.5)75 (81.5)92 (62.6)\nFrequency of taking soft drinks (n = 55)\n<4/day11 (22)39 (76.5)50 (91.1)0.02>4/day3 (75)1 (25)5 (7.3)\nCleaning teeth ( n = 147)\nYes17 (16)88 (84)105 (71.4)0.016No15 (36.6)26(63.4)41 (27.9)\nTime of tooth cleaning (n = 105)\nBefore meal5 (16.7)25 (83.3)30 (28.6)0.63After meal7 (13.2)46 (86.8)53 (50.5)Before and after meal3 (18.8)13 (81.2)16 (15.2)No fixed time2 (33.3)4 (66.7)6 (5.7)\nWay of cleaning teeth (n = 105)\nTooth stick10 (14.1)61 (85.9)71 (67.6)0.52Tooth brush without paste055 (4.8)Tooth brush with paste5 (27.8)13(72.2)18 (9.5)Charcoal011 (0.96)Rinse with water1 (25)3(75)4 (3.8)Other means1 (16.7)5(83.3)6 (5.7)Key: N (number), % (percent).\n\nFood consumption pattern, dietary habits and practices of oral hygiene among primary school children at Bahir Dar city, 2014\n\nKey: N (number), % (percent).\n Dental caries Of the 147 study participants, 32 (21.8%) had decayed tooth. The proportion of dental caries was 33.3% in children from 6 to 10 years of age. Girls were with a higher proportion of dental caries (24.4%) than boys (18.5%). However, the difference was not statistically significant (P = 0.87). The proportion of dental caries was 23 (31.9%) and 9 (12.2%) among children from grade1- 4 and 5–8, respectively. Children belonging to the lowest income group had the highest proportion of dental caries but the highest income group had higher prevalence than the middle income group (Table 1). Among the total dental caries, the majority 24 (75%) had primary tooth decay. Of the children who had dental caries, 12 (50%) had more than one affected tooth and 7 (21.9%) revealed missed teeth. Toothache and white spot lesions were found in 40 (27.2%) and 12 (8.2%) of children, respectively. However, dental plaque was clinically visible in 99 (67.3%) of the children. To get treatment for dental caries, 9 children (6.1%) had consulted a dentist (Table 3).Table 3\nDental cases of primary school children aged 6–15 years, at Bahir Dar city, 2014\nVariablesFrequencyPercent\nTooth decay/dental caries (any type) (n = 147)\n   Yes3221.8   No11578.2\nType of tooth decayed (n = 32)\n   Decay of primary tooth2475   Decay of permanent tooth825\nNumber of primary teeth decay (n = 32)\n   1 tooth1250   2 teeth625   3 teeth520.8   4 teeth14.2\nType of missed teeth (n = 7)\n   Primary571.4   Permanent228.6\nTooth ache (n = 147)\n   Yes4027.2   No10772.8\nWhite spot lesions (n = 147)\n   Yes128.2   No13591.8\nPlaque accumulation (n = 147)\n   Yes9967.3   No4832.7\nHealth institution / Dental visit (n = 147)\n   Yes96.1   No13893.9\n\nDental cases of primary school children aged 6–15 years, at Bahir Dar city, 2014\n\nOf the 147 study participants, 32 (21.8%) had decayed tooth. The proportion of dental caries was 33.3% in children from 6 to 10 years of age. Girls were with a higher proportion of dental caries (24.4%) than boys (18.5%). However, the difference was not statistically significant (P = 0.87). The proportion of dental caries was 23 (31.9%) and 9 (12.2%) among children from grade1- 4 and 5–8, respectively. Children belonging to the lowest income group had the highest proportion of dental caries but the highest income group had higher prevalence than the middle income group (Table 1). Among the total dental caries, the majority 24 (75%) had primary tooth decay. Of the children who had dental caries, 12 (50%) had more than one affected tooth and 7 (21.9%) revealed missed teeth. Toothache and white spot lesions were found in 40 (27.2%) and 12 (8.2%) of children, respectively. However, dental plaque was clinically visible in 99 (67.3%) of the children. To get treatment for dental caries, 9 children (6.1%) had consulted a dentist (Table 3).Table 3\nDental cases of primary school children aged 6–15 years, at Bahir Dar city, 2014\nVariablesFrequencyPercent\nTooth decay/dental caries (any type) (n = 147)\n   Yes3221.8   No11578.2\nType of tooth decayed (n = 32)\n   Decay of primary tooth2475   Decay of permanent tooth825\nNumber of primary teeth decay (n = 32)\n   1 tooth1250   2 teeth625   3 teeth520.8   4 teeth14.2\nType of missed teeth (n = 7)\n   Primary571.4   Permanent228.6\nTooth ache (n = 147)\n   Yes4027.2   No10772.8\nWhite spot lesions (n = 147)\n   Yes128.2   No13591.8\nPlaque accumulation (n = 147)\n   Yes9967.3   No4832.7\nHealth institution / Dental visit (n = 147)\n   Yes96.1   No13893.9\n\nDental cases of primary school children aged 6–15 years, at Bahir Dar city, 2014\n\n Risk factors associated with dental caries Based on bivariate analysis, significant association between dental caries and educational status of children’s parents was found (P = 0.03). Dental caries among children, whose parents’ education are above grade 12 were 100% times at a lower risk compared to those who had non-educated parents (Table 1). There was significant association between dental caries and grade levels of children (AOR = 3.9, 95% CI = 1.49 - 10.4). Children whose grade level 1–4 were more likely to have dental caries compared to grade level of 5 – 8. Children who did not clean their teeth were 2.6 times more likely to have caries than those who cleaned (AOR = 2.6, 95% CI = 1.08 - 6.2). Children who had dental plaque were 5.3 times more likely to have dental caries than those who had not (AOR = 5.3, 95% C I = 1.6 – 17.7). Moreover, the odds of having dental caries was significantly higher among children suffer with tooth ache than those children who had not (AOR = 6.3, 95% CI = 2.4 -15.4) (Table 4).Table 4\nFactors associated with dental caries in primary school children at Bahir Dar city, 2014\nCharacteristicsDental cariesCOR (95% CI)AOR (95% CI)P-valueYesNo\nAge in years\n6-1015300.4 (0.18 - 0.89)**0.62 (0.2 -1.92)0.6211-151785\n1\n\n1\n\nGrade level\n1- 423500.3 (0.13 - 0.71**3.9 (1.49 -10.4)0.0065- 8965\n1\n\n1\n\nCleaning teeth\nYes1789\n1\n\n1\nNo15260.35 (0.15 - 0.79)**2.6 (1.08-6.2)0.033\nUse of sweet foods\nYes21500.4 (0.18 - 0.91)**0.46 (0.18 -1.16)0.09No1165\n1\n\n1\n\nTooth ache\nYes17234.5 (1.97 - 10.41)***6.3 (2.4 - 15.4)<0.001No1592\n1\n\n1\n\nPlague accumulation\nYes28714.3 (1.43 – 13.2)**5.3 (1.6 - 17.7)0.006No4411Key: COR (Crude odds ratio), AOR (adjusted odds ratio), CI (Confidence interval).**:P-value < 0.05, ***:0.05 < P-value < 0.001.\n\nFactors associated with dental caries in primary school children at Bahir Dar city, 2014\n\nKey: COR (Crude odds ratio), AOR (adjusted odds ratio), CI (Confidence interval).\n**:P-value < 0.05, ***:0.05 < P-value < 0.001.\nBased on bivariate analysis, significant association between dental caries and educational status of children’s parents was found (P = 0.03). Dental caries among children, whose parents’ education are above grade 12 were 100% times at a lower risk compared to those who had non-educated parents (Table 1). There was significant association between dental caries and grade levels of children (AOR = 3.9, 95% CI = 1.49 - 10.4). Children whose grade level 1–4 were more likely to have dental caries compared to grade level of 5 – 8. Children who did not clean their teeth were 2.6 times more likely to have caries than those who cleaned (AOR = 2.6, 95% CI = 1.08 - 6.2). Children who had dental plaque were 5.3 times more likely to have dental caries than those who had not (AOR = 5.3, 95% C I = 1.6 – 17.7). Moreover, the odds of having dental caries was significantly higher among children suffer with tooth ache than those children who had not (AOR = 6.3, 95% CI = 2.4 -15.4) (Table 4).Table 4\nFactors associated with dental caries in primary school children at Bahir Dar city, 2014\nCharacteristicsDental cariesCOR (95% CI)AOR (95% CI)P-valueYesNo\nAge in years\n6-1015300.4 (0.18 - 0.89)**0.62 (0.2 -1.92)0.6211-151785\n1\n\n1\n\nGrade level\n1- 423500.3 (0.13 - 0.71**3.9 (1.49 -10.4)0.0065- 8965\n1\n\n1\n\nCleaning teeth\nYes1789\n1\n\n1\nNo15260.35 (0.15 - 0.79)**2.6 (1.08-6.2)0.033\nUse of sweet foods\nYes21500.4 (0.18 - 0.91)**0.46 (0.18 -1.16)0.09No1165\n1\n\n1\n\nTooth ache\nYes17234.5 (1.97 - 10.41)***6.3 (2.4 - 15.4)<0.001No1592\n1\n\n1\n\nPlague accumulation\nYes28714.3 (1.43 – 13.2)**5.3 (1.6 - 17.7)0.006No4411Key: COR (Crude odds ratio), AOR (adjusted odds ratio), CI (Confidence interval).**:P-value < 0.05, ***:0.05 < P-value < 0.001.\n\nFactors associated with dental caries in primary school children at Bahir Dar city, 2014\n\nKey: COR (Crude odds ratio), AOR (adjusted odds ratio), CI (Confidence interval).\n**:P-value < 0.05, ***:0.05 < P-value < 0.001.", "A total of 147 children were participated in the study. Of these, 82 (55.4%) were girls. The majority of children (69.6%) were from 11 to 15 years of age. Nearly half of the study participants were grade 1–4. Twenty one (14%) of the students’ parent had an education above grade 12. Eighty (54.1%) of the participants family earned below 1000 Ethiopian birr per month (Table 1).Table 1\nPrevalence of dental caries and socio-demographic characteristics amon\ng primary school children at Bahir Dar city, 2014 (n = 147)\nSocio-demographic variablesDental carriesCOR (95% CI)P- valuePositive N (%)Negative N (%)Total N (%)\nAge in years\n6 -1015 (33.3)30 (66.7)45 (30.4)0.4 (0.18 - 1.0)0.0311 -1517 (16.7)85 (83.3)102 (69.6\n1\n\nSex\nBoys12 (18.5)53 (81.5)65 (44.6)\n1\nGirls20 (24.4)62 (75.6)82 (55.4)0.7 (0.31 - 1.57)0.87\nGrade\n1- 423 (31.9)49 (68.1)72 (49)0.26 (0.07 - 0.92)0.035- 89 (12.2)65 (87.8)75 (51)\n1\n\nFamily income\n<100021 (26.3)59 (73.8)80 (54.1)0.23 (0.07- 0.72)0.21001-20005 (11.9)37 (88.1)42 (28.4)0.24 (0.63 - 8.66)0.82>20006 (24)19 (76)25 (17.6)\n1\n\nFamily educational status\nIlliterate10 (18.5)44 (81.5)54 (36.7)Can read and write7 (38.9)11 (61.1)18 (12.2)0.031-8 grade level7 (31.8)15 (68.2)22 (15)9-12 grade level8 (25)24 (75)32 (21.8)>12 grade level021 (100)21 (14.3)\nTotal\n\n32 (21.8)\n\n115 (78.2)\n\n147 (100)\nKey: COR (crude odds ratio), CI (Confidence interval), 1(Reference category).\n\nPrevalence of dental caries and socio-demographic characteristics amon\ng primary school children at Bahir Dar city, 2014 (n = 147)\n\nKey: COR (crude odds ratio), CI (Confidence interval), 1(Reference category).", "One hundred four (70.7%) of the children had breakfast bread with tea. Most of the children (85%) usually drank tea with sugar. Thirty one (21.1%) of the participants drank coffee with sugar. Fifty five (37.4%) of the children drank soft drinks. Seventy one, (48.3%) of the children used to eat sweet foods. One hundred five (71.4%) of the children were used to clean their teeth. Of whom, 16 (15.2%) cleaned their teeth before and after meal. However, nearly half of the children cleaned their teeth only after meal intake. Majority of the children (67.6%) used a traditional small stick of wood (termed as Mafaqiya) made of a special type of plant to clean their teeth. However, 18 (9.5%) and 5 (4.8%) were used teeth brush with and without paste, respectively to clean their teeth (Table 2).Table 2\nFood consumption pattern, dietary habits and practices of oral hygiene among primary school children at Bahir Dar city, 2014\nVariablesDental cariesP-valuePositive N (%)Negative N (%)Total\nConsumption of sugared tea (n = 147)\nYes26 (20.8)99 (79.2)125 (85)0.33No6 (27.3)16 (72.7)22 (15)\nConsumption of sugared coffee (n = 147)\nYes6 (19.4)25 (80.6)31 (21.1)0.81No26 (22.6)90 (73.4)116 (78.9)\nType of food for breakfast (n = 147)\nBread with tea24 (23.1)80 (76.9)104 (70.7)0.47Pasta7 (21.1)28 (78.8)35 (23.8)Other (Makorony, Rice)088 (5.4)\nConsumption of sweet foods (n = 147)\nYes21 (29.6)50 (70.4)71 (48.3)0.03No11 (14.7)64 (85.3)76 (51.7)\nConsumption of soft drinks (n = 147)\nYes15 (27.3)40 (72.7)55 (37.4)0.21No17 (18.5)75 (81.5)92 (62.6)\nFrequency of taking soft drinks (n = 55)\n<4/day11 (22)39 (76.5)50 (91.1)0.02>4/day3 (75)1 (25)5 (7.3)\nCleaning teeth ( n = 147)\nYes17 (16)88 (84)105 (71.4)0.016No15 (36.6)26(63.4)41 (27.9)\nTime of tooth cleaning (n = 105)\nBefore meal5 (16.7)25 (83.3)30 (28.6)0.63After meal7 (13.2)46 (86.8)53 (50.5)Before and after meal3 (18.8)13 (81.2)16 (15.2)No fixed time2 (33.3)4 (66.7)6 (5.7)\nWay of cleaning teeth (n = 105)\nTooth stick10 (14.1)61 (85.9)71 (67.6)0.52Tooth brush without paste055 (4.8)Tooth brush with paste5 (27.8)13(72.2)18 (9.5)Charcoal011 (0.96)Rinse with water1 (25)3(75)4 (3.8)Other means1 (16.7)5(83.3)6 (5.7)Key: N (number), % (percent).\n\nFood consumption pattern, dietary habits and practices of oral hygiene among primary school children at Bahir Dar city, 2014\n\nKey: N (number), % (percent).", "Of the 147 study participants, 32 (21.8%) had decayed tooth. The proportion of dental caries was 33.3% in children from 6 to 10 years of age. Girls were with a higher proportion of dental caries (24.4%) than boys (18.5%). However, the difference was not statistically significant (P = 0.87). The proportion of dental caries was 23 (31.9%) and 9 (12.2%) among children from grade1- 4 and 5–8, respectively. Children belonging to the lowest income group had the highest proportion of dental caries but the highest income group had higher prevalence than the middle income group (Table 1). Among the total dental caries, the majority 24 (75%) had primary tooth decay. Of the children who had dental caries, 12 (50%) had more than one affected tooth and 7 (21.9%) revealed missed teeth. Toothache and white spot lesions were found in 40 (27.2%) and 12 (8.2%) of children, respectively. However, dental plaque was clinically visible in 99 (67.3%) of the children. To get treatment for dental caries, 9 children (6.1%) had consulted a dentist (Table 3).Table 3\nDental cases of primary school children aged 6–15 years, at Bahir Dar city, 2014\nVariablesFrequencyPercent\nTooth decay/dental caries (any type) (n = 147)\n   Yes3221.8   No11578.2\nType of tooth decayed (n = 32)\n   Decay of primary tooth2475   Decay of permanent tooth825\nNumber of primary teeth decay (n = 32)\n   1 tooth1250   2 teeth625   3 teeth520.8   4 teeth14.2\nType of missed teeth (n = 7)\n   Primary571.4   Permanent228.6\nTooth ache (n = 147)\n   Yes4027.2   No10772.8\nWhite spot lesions (n = 147)\n   Yes128.2   No13591.8\nPlaque accumulation (n = 147)\n   Yes9967.3   No4832.7\nHealth institution / Dental visit (n = 147)\n   Yes96.1   No13893.9\n\nDental cases of primary school children aged 6–15 years, at Bahir Dar city, 2014\n", "Based on bivariate analysis, significant association between dental caries and educational status of children’s parents was found (P = 0.03). Dental caries among children, whose parents’ education are above grade 12 were 100% times at a lower risk compared to those who had non-educated parents (Table 1). There was significant association between dental caries and grade levels of children (AOR = 3.9, 95% CI = 1.49 - 10.4). Children whose grade level 1–4 were more likely to have dental caries compared to grade level of 5 – 8. Children who did not clean their teeth were 2.6 times more likely to have caries than those who cleaned (AOR = 2.6, 95% CI = 1.08 - 6.2). Children who had dental plaque were 5.3 times more likely to have dental caries than those who had not (AOR = 5.3, 95% C I = 1.6 – 17.7). Moreover, the odds of having dental caries was significantly higher among children suffer with tooth ache than those children who had not (AOR = 6.3, 95% CI = 2.4 -15.4) (Table 4).Table 4\nFactors associated with dental caries in primary school children at Bahir Dar city, 2014\nCharacteristicsDental cariesCOR (95% CI)AOR (95% CI)P-valueYesNo\nAge in years\n6-1015300.4 (0.18 - 0.89)**0.62 (0.2 -1.92)0.6211-151785\n1\n\n1\n\nGrade level\n1- 423500.3 (0.13 - 0.71**3.9 (1.49 -10.4)0.0065- 8965\n1\n\n1\n\nCleaning teeth\nYes1789\n1\n\n1\nNo15260.35 (0.15 - 0.79)**2.6 (1.08-6.2)0.033\nUse of sweet foods\nYes21500.4 (0.18 - 0.91)**0.46 (0.18 -1.16)0.09No1165\n1\n\n1\n\nTooth ache\nYes17234.5 (1.97 - 10.41)***6.3 (2.4 - 15.4)<0.001No1592\n1\n\n1\n\nPlague accumulation\nYes28714.3 (1.43 – 13.2)**5.3 (1.6 - 17.7)0.006No4411Key: COR (Crude odds ratio), AOR (adjusted odds ratio), CI (Confidence interval).**:P-value < 0.05, ***:0.05 < P-value < 0.001.\n\nFactors associated with dental caries in primary school children at Bahir Dar city, 2014\n\nKey: COR (Crude odds ratio), AOR (adjusted odds ratio), CI (Confidence interval).\n**:P-value < 0.05, ***:0.05 < P-value < 0.001.", "In Ethiopia, there is scarcity of data on dental caries in primary school children. In this study, dental caries is a common health problem among primary school children. The prevalence of dental caries found in the present study was comparable with a study conducted in Tanzania (17.6%) [19]. However, it was lower than studies carried out in other parts of Ethiopia (36.3- 48.1%) [8, 20], Nigeria (35.5%) [1], Nepal (52%) [21] and India (77%) [22]. The difference could be due to difference in knowledge, attitude and practices (KAPs) on oral hygiene as our study was under taken among urban school children. In addition, the lower sample size in the present study might be accounted for the difference.\nAs a result of shortage of dental insurance and families lacking usual source of dental care, previous studies have shown that children living in poverty have higher prevalence of dental caries [7, 23, 24] but in this study a clear association between family income and dental caries was not observed. This was comparable with a study done in Sirilanka [23]. Though, the lowest income group had highest prevalence of dental caries, the middle income group had lower prevalence than the highest income group. This discrepancy might be due to other confounding factors like dietary habits.\nChildren’s grade level was found to be statistically significant for dental caries, hence as their grade levels increased the chance of dental caries get reduced. This finding was also supported by similar findings [8, 25, 26]. On bivariate analysis, habit of consumption of sweet foods has significantly associated with dental caries. This finding was in agreement with a study done in Saudi Arabia [5]. This might be associated with copious acid production by cariogenic bacteria like Streptococcus mutans that are adherent to teeth as a result of fermentation of the sweet foods. Later the enamel of the tooth went into tooth decay [5, 26]. Moreover, poor habit of tooth cleaning was significantly associated with dental caries. Children who had cleaned their teeth revealed a lower prevalence of dental caries. It is generally true that cleaning teeth will remove away the food debris from the oral. Therefore, Streptococcus mutans cannot get enough nutrient and time for growth and no acid production that causes dental caries development [18, 26].\nIt is known that toothache is one of the major indicators of tooth decay [18, 27]. In this study, children who had toothache were 6.3 times more likely to have dental caries. This finding conforms to a study conducted in Dare Salaam [27]. Moreover, dental plaque was significantly associated with dental caries. Children having visible dental plaque were more likely to have dental caries than their free counter parts. This is also a good indicator of poor oral hygiene practices. Because, dental plaque retention increases Streptococcus mutans colonization and in severe cases, the loss of enamel. In other studies, children with visible dental plaque recognized as a proxy of poor tooth brushing frequency. Similar finding was reported in Tanzania [19] stating that poor oral hygiene practice was the associated factor of dental carries. Moreover, such associations were established in other studies [28, 29].\nThis study reported that 67.6% of children cleaned their teeth using traditional small stick of wood (Mafaqiya) for maintaining oral hygiene. In contrast, other studies showed that tooth brush and tooth paste were the most common means of maintaining oral hygiene [8, 21]. In this study, a large proportion of children had never visit to a dentist. Similar finding was reported in Ethiopia [8], Nepal [21] and Sirilanka [23].\nThis study has the following main limitations: Detection of dental caries using dental mirror and radiology was not possible because of lack of instruments and laboratory set up. Therefore, dental caries was identified only with clinical diagnosis. In addition, the number of study participants was lower compared to other studies. Thus, this might reduce the true magnitude of the problem.", "Dental caries is a common public health problem among primary school children at Bahir Dar city. Low grade level, poor oral hygiene and dietary along with lack of dental visit were the associated factors for dental caries. Therefore, health education on oral hygiene, dietary habits and dental visit should be given for children to prevent and control dental caries. Moreover, further studies including private and rural school children using all methods of diagnosis of dental caries and assessment of knowledge, attitude and practices of children and their parents on oral hygiene should be recommended.", "WM is an assistant professor at College of Medicine and Health Sciences, Bahir Dar University in Medical Microbiology, BA is an associate professor at college of Medicine and Health Sciences, Bahir Dar University in Medical Microbiology and department head of Medical Microbiology, Immunology and Parasitology, MY is an assistant professor at College of Medicine and Health Sciences, Bahir Dar University in Medical Parasitology. KM is lecturer at college of Medicine and Health Sciences, Bahir Dar University in doctor of dentistry." ]

[ null, "methods", null, null, null, null, null, null, "results", null, null, null, null, "discussion", "conclusions", null ]

[ "Dental caries", "Dental plaque", "Children" ]

[ 388, 110, 45, 113, 334, 103, 48, 389, 560, 421, 476, 89 ]

[ "dental", "children", "caries", "dental caries", "teeth", "147", "primary", "tooth", "school", "study" ]

[ "caries caused dental", "develop dental caries", "study dental caries", "factors dental caries", "dental caries effects" ]

[ "Adult", "Aged", "Aged, 80 and over", "Air Pollutants", "Air Pollution", "Asthma", "Cities", "Europe", "Female", "Humans", "Male", "Middle Aged", "Nitrogen Dioxide", "Odds Ratio", "Ozone", "Particulate Matter", "Pulmonary Disease, Chronic Obstructive", "Respiration Disorders", "Walking" ]

[ "Background", "Study design", "Study population", "Symptom diary", "Air pollution exposure", "Confounder data", "Quality assurance/quality control", "Statistical analysis", "Panel characteristics", "Symptoms", "Air pollution concentrations", "Air pollution effects on symptoms-limitation in activities due to breathing problems", "Prevalence analyses", "Incidence analyses", "Abbreviations", "Competing interests", "Authors’ contributions" ]

[ "Over the last decades numerous epidemiological studies have clearly shown that urban air pollution can produce a variety of adverse health effects\n[1,2]. Ambient particulate matter (PM) either characterized as the mass concentration of particles less than 10 μm (PM10) or less than 2.5 μm (PM2.5) are considered to be the major culprit. Therefore, current air quality standards or guidelines refer to PM10 and/or PM2.5[3,4]. However, in reality ambient PM is a mixture of coarse (2.5-10 μm), PM2.5 (named also fine particles) and ultrafine (<0.1 μm) particles generated from different processes, having variable chemical composition and atmospheric behavior. It should also be noted that although the ultrafine fraction accounts for less than 1% of the mass of particulate matter, it represents the greatest proportion in terms of number of particles (typically >80%)\n[5-7]. Furthermore, the mechanism and the fraction of PM that are mainly responsible for the observed health effects is a matter of controversy\n[1]. In 1995 Seaton hypothesized that the number of ultrafine particles may be a more health relevant property than the usually measured mass of inhaled PM10 and PM2.5[8]. This is because of the greater surface area available to react with epithelial and inflammatory cells in the lung and because of the capacity of ultrafine particles to penetrate deeper in the lung parenchyma, potentially reaching the circulation and exerting adverse biological effects by releasing toxic free radicals\n[8-11]. In meantime other studies were published, however, the role of ultrafine particles is still under discussion\n[9,12-14].\nThe only systematic review of studies that have analysed fine and coarse PM jointly demonstrates that the health effects of coarse particles are significant and should not be overlooked\n[15]. Thus, special consideration should be given to each fraction of the particles and their effects on health. Better characterization of the health relevant particle fraction will have major implications for air quality policy since it will determine which sources should be controlled.\nThe RUPIOH (Relationship between Ultrafine and fine Particulate matter in Indoor and Outdoor air and respiratory Health) is an EU-funded multicentre study designed to examine the distribution of various particle metrics both indoors and outdoors in four European cities and assess their health effects in individuals with asthma or chronic obstructive pulmonary disease (COPD), based on a detailed exposure assessment. The study consisted of two parts: i) the diary study in which participants were asked to complete a daily diary for six months while exposure was assessed based on a central site measurements and ii) the intensive week measurements during which, for each subject, more intensive health and exposure measurements were conducted. In this paper, we report the association of ambient PM10, PM2.5, coarse particle mass (PM10-2.5) and particle number concentrations (PNC), measured at the central site, with respiratory symptoms and limitation in activities due to breathing problems in participants having either asthma or COPD who have been followed for six months. Associations of the health outcomes with gaseous air pollutants were also examined based on data collected from existing national monitoring networks in each country. The relationships between central site outdoor, residential outdoor and indoor concentrations, as well as the association between outdoor and indoor exposure to fine and ultrafine particles and lung function in the same participants but based on the intensive week measurements have been published before\n[16-20].", "In the context of RUPIOH, a multicentre study was conducted from October 2002 to March 2004 in four European metropolitan areas, namely, Amsterdam (The Netherlands), Athens (Greece), Birmingham (United Kingdom) and Helsinki (Finland). During the whole study period a central site in each city was used to monitor particle mass and PNC on a daily basis. At various locations covering the entire metropolitan area, homes of participants with either asthma or COPD were selected. The criteria for the central site and homes selection have been described in detail in a previous publication\n[17]. Respiratory health status of each participant was monitored for six months by a daily symptom diary. We used a staged entry of the participants (based on the real date the participants started to fill out the diaries) in order to increase the period of data collection and thus, decrease the likelihood for uncontrolled factors or unexpected events to influence the associations between air pollution and health\n[21]. In all centres, participants were recruited between October 2002 and March 2004.", "Inclusion criteria and recruitment procedures have been described in detail before\n[19]. Briefly, in each city the recruitment criteria for participants were age 35 or more, a doctor diagnosis of either asthma (as defined by Global Initiative for Asthma) or COPD (as defined by Global Initiative for Chronic Obstructive Lung Disease) and having had experienced respiratory symptoms in the past 12 months\n[22,23]. Especially, in the Netherlands some patients who had not received a definite diagnosis of asthma or COPD were classified as chronic non-specific lung disease (CNSLD) as a relic of tradition (term previously used to indicate either asthma or COPD)\n[24]. Severe patients defined as those using relief bronchodilating medications more than three times per day or using nebulised bronchodilators or long-term oxygen therapy as well as participants unable to perform a satisfactory spirometry test were excluded from the study. An attempt was made to select non-working, non-smoking patients living in a non-smoking household to eliminate potential confounding by occupational exposures to airborne particles and by environmental tobacco smoke. The same screening questionnaire was used across the four centres to ascertain eligibility. However, each centre was allowed to choose the optimal subject recruitment method. Specifically, in Amsterdam, the panelists were recruited through distribution of 10,000 information letters accompanied by screening questionnaires. Inclusion criteria were checked using the returned screening questionnaires followed by participants’ homes’ visits. In Athens, subjects recruited through local hospitals and pulmonary chest physicians were visited at home by a pulmonologist (A.K.) and one of the investigators of the exposure assessment team (I.K.) who checked whether inclusion criteria were met. In Finland, subjects were selected from the Helsinki Metropolitan Area (including cities of Helsinki, Espoo and Vantaa) by placing advertisement on two issues of the respiratory patient association magazine (circulation ~3500 households) and notice boards of pulmonary disease clinics of four major hospitals within the study area. Candidate subjects were interviewed and screened by telephone and invited to an information session when they met the criteria. In the United Kingdom, potential study subjects living in the greater area of Birmingham were selected from the Clinic for Respiratory illnesses (CRI) database of respiratory patients at the Heartlands Hospital. Privacy regulations restricted the selections to only those that had given their written consent to be approached for research studies.\nMedical ethical clearance was acquired from the relevant local medical ethics committees in all centres before the start of the recruitment. Written informed consent was obtained from each subject.", "The diary was based upon diaries used in previous studies of acute effects of air pollution such as the PEACE study\n[21]. Although there is no real objective method of validating symptoms, a previous study by Hoek et al. provide evidence that symptoms, assessed with the same diary, are reflected in lung function drops\n[25]. Participants were instructed to complete a daily record about respiratory symptoms and medication taken “as needed” for six months, grading shortness of breath, wheeze, cough, phlegm and woken with breathing problems as absent (0), slight (1), or moderate/severe (2). In addition, they were asked about any limitation in performing daily life activities categorized as vigorous (such as running, lifting heavy objects, participating in strenuous sports), moderate (such as moving a table, pushing a vacuum cleaner, bowling or playing golf), walking one block/climbing one flight of stairs and leaving one’s home, because of breathing problems. This limitation could be reported in three grades: no limitation (0), yes, did activity slowly (1) and yes, avoided activity completely (2). Questions on whether they have been outside the house or town and for how long have also been included.\nDuring the study period there was personal contact with the participants once a month to collect the completed diary forms, discuss potential problems and keep the motivation at a good level.", "Exposure assessment has been described in previous publications\n[16-18,20]. In brief, during the entire study period (October 2002 to March 2004) in each city measurements of PM2.5, PM10 and PNC were performed continuously at a central site representing urban background levels\n[17]. The same type of condensation particle counter (TSI 3022A, TSI Inc., St. Paul, MN, USA) was used in each city to monitor PNC. 24-hour average particle mass concentration was measured with Harvard impactors for PM2.5 and PM10. Coarse particles concentrations were calculated by subtracting PM2.5 from PM10. After weighing, the absorbance of the PM2.5 filters (a good surrogate for elemental carbon/soot) was determined using reflectometry. PNC was transformed to “noon-to-noon” 24-hour means to coincide with the PM2.5 measurements. Data on concentrations of other air pollutants (ozone, nitrogen dioxide) and meteorology (air temperature, relative humidity) were collected from existing national monitoring networks in each country. We did not replace missing values in exposures variables by imputation.", "Time trend in health endpoints (e.g. fatigue in reporting), weather (outdoor temperature, relative humidity), medication use and day of the week were taken into account as potential confounders. Because of the staged entry of participants, we evaluated two time variables: calendar date (proxy for unmeasured confounders) and day of study for a specific subject (possibly related to fatigue).", "Air pollution and health measurements were performed according to standard operating procedures (SOPs). A training workshop was organized before the start of the fieldwork and site visits were implemented during the fieldwork to identify any deviations from SOPs.", "Data analysis was done according to a predefined analysis plan. The symptom variables, initially coded as 0 for no symptoms (absent), 1 for slight symptoms and 2 for moderate/severe symptoms, were dichotomised for the analysis by setting 0 for no symptoms and 1 for slight to moderate/severe symptoms. Each symptom was analysed separately either as prevalent (irrespective of its occurrence on the previous day) or incident (when that symptom was reported to be absent on the previous day). Medication use was coded as 0 (no medication) versus 1 (intake of one or more doses) independently of the initial medication group. Every person was included in the analysis regardless of how many diary entries were made. Moreover, diary entries were excluded when participants had left the study area during the measurement period. For every pollutant the following lags were evaluated: lag 0, 1, 2 and the average of lag 0–6 days. Lag 0 was defined as the 24-hour period starting from noon of the calendar day before the health response.\nA hierarchical modelling approach was used. First, regression models were fitted in each city separately to allow specific control for seasonal effects, weather and other potential confounders. Results of the individual city analysis were used in a second stage analysis (meta-analysis) to provide overall estimates\n[26]. We computed both fixed and random effects combined estimates. Furthermore, a chi-square test of heterogeneity of the four city-specific estimates was computed.\nWe applied logistic regression to obtain centre-specific effect estimates. A smooth function (natural splines with 6 degrees of freedom per year) of time was used to remove the seasonal patterns and long time trends from the data. Afterwards, same-day (lag 0) and previous-day (lag 1) mean daily temperatures were introduced simultaneously into the model. For both lags of temperature, a linear term was compared with a smoothed function (natural splines) with 2, 3 and 4 degrees of freedom and the model with the lowest Akaike’s Information Criterion (AIC) was selected. A linear term of relative humidity (lag 0) was added to the model as another indicator of weather. Finally, indicator variables for day of the week, medication use and individual differences in frequency of symptoms, were added to the model. After setting up the baseline model, the effects of the various lags of the pollutants were evaluated.\nIn the city specific analysis we fitted fixed effects models, described above, as well as random intercept logistic regression models using “glmmPQL” function from MASS library in R software, to take into account the correlation among each subject’s measurements. Results from the random effects analysis were very similar to those derived from fixed effects. In a few cases though, we faced convergence issues. This was even more the case when we tested a first order autoregressive correlation structure. The significance of the associations was similar between random intercept models and the models incorporating an autoregressive term.\nBecause of the heterogeneity of the study population, we repeated the analysis (for all air pollution measures) for the subgroup of asthmatic patients. There were not enough COPD patients to analyse these patients separately. We also fitted two pollutant models by including simultaneously PM2.5 and PM10-2.5 in order to better characterize which of the two components of PM10 (PM10-2.5 or PM2.5) was responsible for the observed health effects.\nEffect estimates are expressed as odds ratios (OR) for an increase of 10 μg/m3 in PM10, 10 μg/m3 in PM2.5, 10 μg/m3 in PM10-2.5, 10,000 particles/cm3 for PNC and 1·10-5 m-1 for absorbance, in order to be comparable with other studies. For gaseous pollutants the effect estimates are expressed as OR for an increase of 10 μg/m3 in ozone and NO2 concentrations.All analyses were performed using R software\n[27].", "A brief description of the study population is presented in Table\n1. Mean age and age range were about the same in all cities. Three participants in Athens were slightly below of the recruitment criterion of ≥35 years. In Amsterdam a large group was reported to have CNSLD. Medication use was high in the panels. Seventy seven per cent of the participants (77%) used reliever medication. Use of “as needed medication” was recorded in 26.5% of total person days in Helsinki, 13.9% in Athens, 37.9% in Amsterdam and 59.7% in Birmingham. Twenty-nine participants (21%) worked outside their home especially from Amsterdam and Birmingham. Those who worked outside their home, worked on average 19 h/week.\nCharacteristics of four European panels of asthmatic/COPD patients\na Total participants in panel.\nb Asthma + COPD or chronic non-specific lung disease.\nc Given as mean and [range].\nd Chronic non-specific lung disease.\ne Environmental tobacco smoke.\nf Includes short acting β2-agonist, long acting β2-agonist, anticholinergic drugs and combination of an anticholinergic drug and a β2-agonist.", "In total between 4,760 and 6,003 person days were available for analysis in the four cities. In Amsterdam, Athens and Birmingham participants filled out the diary from October 2002 to March 2004 whilst in Helsinki between October 2002 and February 2004. Missing values (person days) ranged between 9.4-15.1% in Amsterdam, 4.7-5.5% in Athens, 8.7-8.8% in Birmingham and 8.6-12.1% in Helsinki. Consistent with the composition of the panel, fairly high symptom prevalence occurred during the study period. Person days with severe symptoms were low, except for cough and phlegm. There were small differences between the cities (Table\n2).\nPerson days with symptoms in the diary (n = number of expected person days)\na due to breathing problems.", "Helsinki had the lowest median concentrations for all PM components whilst Athens had the highest. However, maximum concentrations of PM2.5 were observed in Amsterdam (103.4 μg/m3) and of PM10-2.5 (152.6 μg/m3) in Helsinki (Table\n3).\nDaily (24 hours noon-to-noon, central site) median air pollution concentration and meteorology in the four cities", " Prevalence analyses We observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n\nWe observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n", "We observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n", "Patterns similar to those in the combined prevalence analyses were observed for the associations of incident symptoms and particles especially the coarse fraction. Shortness of breath was consistently associated with PM10 and PM10-2.5 in lag 1 with no indication of heterogeneity between the centres (OR = 1.045, 95% CI: 1.008, 1.083 and OR = 1.065, 95% CI: 1.009, 1.124 respectively). There was also a tendency towards positive associations between PM10-2.5 and incidence of wheezing, cough and limitation in walking but none of the associations were statistically significant. Additionally, ozone was positively associated with cough in lags 1 and 2 as well as the average lag 0–6 days but only in lag 2 the association reached the nominal level of significance (Table\n7).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith incidence of symptoms in the four panels (random effects pooled estimates)\n\nBold are significant pooled effects.", "AIC: Akaike’s information criterion; CNSLD: Chronic non-specific lung disease; COPD: Chronic obstructive pulmonary disease; EBC NOx: Total nitrate and nitrite concentrations in exhaled breath condensate; OR: Odds ratio; PM: Particulate matter; PM10-2.5: Coarse particles; PM10: Mass concentration of particles less than 10 μm; PM2.5: Mass concentration of particles less than 2.5 μm; PNC: Particle number concentrations; RUPIOH: Relationship between Ultrafine and fine Particulate matter in Indoor and Outdoor air and respiratory Health; SOPs: Standard operating procedures; 95% CI: 95% confidence interval.", "The authors declare that they have no competing interests.", "All authors of this paper have critically read and approved the final version submitted. They have also made substantive intellectual contributions by directly participating either in the planning, execution, or analysis of the study. AK contributed to the development of the study design, acquisition and interpretation of data and drafted the paper. AA did the analysis, contributed to the interpretation of data and wrote the statistical analysis section of the paper. DP, IGK, JJdeH contributed substantially to acquisition and interpretation of data. JGA, RMH, AK, JP, KH, GPAK, KK contributed to the study design, interpretation of data and have been involved in drafting the manuscript. GH conceived and developed the study design, contributed to the interpretation of data and was involved in drafting the paper. All authors have revised drafts and contributed to the revisions." ]

[ null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null ]

[ "Background", "Methods", "Study design", "Study population", "Symptom diary", "Air pollution exposure", "Confounder data", "Quality assurance/quality control", "Statistical analysis", "Results", "Panel characteristics", "Symptoms", "Air pollution concentrations", "Air pollution effects on symptoms-limitation in activities due to breathing problems", "Prevalence analyses", "Incidence analyses", "Discussion", "Conclusions", "Abbreviations", "Competing interests", "Authors’ contributions" ]

[ "Over the last decades numerous epidemiological studies have clearly shown that urban air pollution can produce a variety of adverse health effects\n[1,2]. Ambient particulate matter (PM) either characterized as the mass concentration of particles less than 10 μm (PM10) or less than 2.5 μm (PM2.5) are considered to be the major culprit. Therefore, current air quality standards or guidelines refer to PM10 and/or PM2.5[3,4]. However, in reality ambient PM is a mixture of coarse (2.5-10 μm), PM2.5 (named also fine particles) and ultrafine (<0.1 μm) particles generated from different processes, having variable chemical composition and atmospheric behavior. It should also be noted that although the ultrafine fraction accounts for less than 1% of the mass of particulate matter, it represents the greatest proportion in terms of number of particles (typically >80%)\n[5-7]. Furthermore, the mechanism and the fraction of PM that are mainly responsible for the observed health effects is a matter of controversy\n[1]. In 1995 Seaton hypothesized that the number of ultrafine particles may be a more health relevant property than the usually measured mass of inhaled PM10 and PM2.5[8]. This is because of the greater surface area available to react with epithelial and inflammatory cells in the lung and because of the capacity of ultrafine particles to penetrate deeper in the lung parenchyma, potentially reaching the circulation and exerting adverse biological effects by releasing toxic free radicals\n[8-11]. In meantime other studies were published, however, the role of ultrafine particles is still under discussion\n[9,12-14].\nThe only systematic review of studies that have analysed fine and coarse PM jointly demonstrates that the health effects of coarse particles are significant and should not be overlooked\n[15]. Thus, special consideration should be given to each fraction of the particles and their effects on health. Better characterization of the health relevant particle fraction will have major implications for air quality policy since it will determine which sources should be controlled.\nThe RUPIOH (Relationship between Ultrafine and fine Particulate matter in Indoor and Outdoor air and respiratory Health) is an EU-funded multicentre study designed to examine the distribution of various particle metrics both indoors and outdoors in four European cities and assess their health effects in individuals with asthma or chronic obstructive pulmonary disease (COPD), based on a detailed exposure assessment. The study consisted of two parts: i) the diary study in which participants were asked to complete a daily diary for six months while exposure was assessed based on a central site measurements and ii) the intensive week measurements during which, for each subject, more intensive health and exposure measurements were conducted. In this paper, we report the association of ambient PM10, PM2.5, coarse particle mass (PM10-2.5) and particle number concentrations (PNC), measured at the central site, with respiratory symptoms and limitation in activities due to breathing problems in participants having either asthma or COPD who have been followed for six months. Associations of the health outcomes with gaseous air pollutants were also examined based on data collected from existing national monitoring networks in each country. The relationships between central site outdoor, residential outdoor and indoor concentrations, as well as the association between outdoor and indoor exposure to fine and ultrafine particles and lung function in the same participants but based on the intensive week measurements have been published before\n[16-20].", " Study design In the context of RUPIOH, a multicentre study was conducted from October 2002 to March 2004 in four European metropolitan areas, namely, Amsterdam (The Netherlands), Athens (Greece), Birmingham (United Kingdom) and Helsinki (Finland). During the whole study period a central site in each city was used to monitor particle mass and PNC on a daily basis. At various locations covering the entire metropolitan area, homes of participants with either asthma or COPD were selected. The criteria for the central site and homes selection have been described in detail in a previous publication\n[17]. Respiratory health status of each participant was monitored for six months by a daily symptom diary. We used a staged entry of the participants (based on the real date the participants started to fill out the diaries) in order to increase the period of data collection and thus, decrease the likelihood for uncontrolled factors or unexpected events to influence the associations between air pollution and health\n[21]. In all centres, participants were recruited between October 2002 and March 2004.\nIn the context of RUPIOH, a multicentre study was conducted from October 2002 to March 2004 in four European metropolitan areas, namely, Amsterdam (The Netherlands), Athens (Greece), Birmingham (United Kingdom) and Helsinki (Finland). During the whole study period a central site in each city was used to monitor particle mass and PNC on a daily basis. At various locations covering the entire metropolitan area, homes of participants with either asthma or COPD were selected. The criteria for the central site and homes selection have been described in detail in a previous publication\n[17]. Respiratory health status of each participant was monitored for six months by a daily symptom diary. We used a staged entry of the participants (based on the real date the participants started to fill out the diaries) in order to increase the period of data collection and thus, decrease the likelihood for uncontrolled factors or unexpected events to influence the associations between air pollution and health\n[21]. In all centres, participants were recruited between October 2002 and March 2004.\n Study population Inclusion criteria and recruitment procedures have been described in detail before\n[19]. Briefly, in each city the recruitment criteria for participants were age 35 or more, a doctor diagnosis of either asthma (as defined by Global Initiative for Asthma) or COPD (as defined by Global Initiative for Chronic Obstructive Lung Disease) and having had experienced respiratory symptoms in the past 12 months\n[22,23]. Especially, in the Netherlands some patients who had not received a definite diagnosis of asthma or COPD were classified as chronic non-specific lung disease (CNSLD) as a relic of tradition (term previously used to indicate either asthma or COPD)\n[24]. Severe patients defined as those using relief bronchodilating medications more than three times per day or using nebulised bronchodilators or long-term oxygen therapy as well as participants unable to perform a satisfactory spirometry test were excluded from the study. An attempt was made to select non-working, non-smoking patients living in a non-smoking household to eliminate potential confounding by occupational exposures to airborne particles and by environmental tobacco smoke. The same screening questionnaire was used across the four centres to ascertain eligibility. However, each centre was allowed to choose the optimal subject recruitment method. Specifically, in Amsterdam, the panelists were recruited through distribution of 10,000 information letters accompanied by screening questionnaires. Inclusion criteria were checked using the returned screening questionnaires followed by participants’ homes’ visits. In Athens, subjects recruited through local hospitals and pulmonary chest physicians were visited at home by a pulmonologist (A.K.) and one of the investigators of the exposure assessment team (I.K.) who checked whether inclusion criteria were met. In Finland, subjects were selected from the Helsinki Metropolitan Area (including cities of Helsinki, Espoo and Vantaa) by placing advertisement on two issues of the respiratory patient association magazine (circulation ~3500 households) and notice boards of pulmonary disease clinics of four major hospitals within the study area. Candidate subjects were interviewed and screened by telephone and invited to an information session when they met the criteria. In the United Kingdom, potential study subjects living in the greater area of Birmingham were selected from the Clinic for Respiratory illnesses (CRI) database of respiratory patients at the Heartlands Hospital. Privacy regulations restricted the selections to only those that had given their written consent to be approached for research studies.\nMedical ethical clearance was acquired from the relevant local medical ethics committees in all centres before the start of the recruitment. Written informed consent was obtained from each subject.\nInclusion criteria and recruitment procedures have been described in detail before\n[19]. Briefly, in each city the recruitment criteria for participants were age 35 or more, a doctor diagnosis of either asthma (as defined by Global Initiative for Asthma) or COPD (as defined by Global Initiative for Chronic Obstructive Lung Disease) and having had experienced respiratory symptoms in the past 12 months\n[22,23]. Especially, in the Netherlands some patients who had not received a definite diagnosis of asthma or COPD were classified as chronic non-specific lung disease (CNSLD) as a relic of tradition (term previously used to indicate either asthma or COPD)\n[24]. Severe patients defined as those using relief bronchodilating medications more than three times per day or using nebulised bronchodilators or long-term oxygen therapy as well as participants unable to perform a satisfactory spirometry test were excluded from the study. An attempt was made to select non-working, non-smoking patients living in a non-smoking household to eliminate potential confounding by occupational exposures to airborne particles and by environmental tobacco smoke. The same screening questionnaire was used across the four centres to ascertain eligibility. However, each centre was allowed to choose the optimal subject recruitment method. Specifically, in Amsterdam, the panelists were recruited through distribution of 10,000 information letters accompanied by screening questionnaires. Inclusion criteria were checked using the returned screening questionnaires followed by participants’ homes’ visits. In Athens, subjects recruited through local hospitals and pulmonary chest physicians were visited at home by a pulmonologist (A.K.) and one of the investigators of the exposure assessment team (I.K.) who checked whether inclusion criteria were met. In Finland, subjects were selected from the Helsinki Metropolitan Area (including cities of Helsinki, Espoo and Vantaa) by placing advertisement on two issues of the respiratory patient association magazine (circulation ~3500 households) and notice boards of pulmonary disease clinics of four major hospitals within the study area. Candidate subjects were interviewed and screened by telephone and invited to an information session when they met the criteria. In the United Kingdom, potential study subjects living in the greater area of Birmingham were selected from the Clinic for Respiratory illnesses (CRI) database of respiratory patients at the Heartlands Hospital. Privacy regulations restricted the selections to only those that had given their written consent to be approached for research studies.\nMedical ethical clearance was acquired from the relevant local medical ethics committees in all centres before the start of the recruitment. Written informed consent was obtained from each subject.\n Symptom diary The diary was based upon diaries used in previous studies of acute effects of air pollution such as the PEACE study\n[21]. Although there is no real objective method of validating symptoms, a previous study by Hoek et al. provide evidence that symptoms, assessed with the same diary, are reflected in lung function drops\n[25]. Participants were instructed to complete a daily record about respiratory symptoms and medication taken “as needed” for six months, grading shortness of breath, wheeze, cough, phlegm and woken with breathing problems as absent (0), slight (1), or moderate/severe (2). In addition, they were asked about any limitation in performing daily life activities categorized as vigorous (such as running, lifting heavy objects, participating in strenuous sports), moderate (such as moving a table, pushing a vacuum cleaner, bowling or playing golf), walking one block/climbing one flight of stairs and leaving one’s home, because of breathing problems. This limitation could be reported in three grades: no limitation (0), yes, did activity slowly (1) and yes, avoided activity completely (2). Questions on whether they have been outside the house or town and for how long have also been included.\nDuring the study period there was personal contact with the participants once a month to collect the completed diary forms, discuss potential problems and keep the motivation at a good level.\nThe diary was based upon diaries used in previous studies of acute effects of air pollution such as the PEACE study\n[21]. Although there is no real objective method of validating symptoms, a previous study by Hoek et al. provide evidence that symptoms, assessed with the same diary, are reflected in lung function drops\n[25]. Participants were instructed to complete a daily record about respiratory symptoms and medication taken “as needed” for six months, grading shortness of breath, wheeze, cough, phlegm and woken with breathing problems as absent (0), slight (1), or moderate/severe (2). In addition, they were asked about any limitation in performing daily life activities categorized as vigorous (such as running, lifting heavy objects, participating in strenuous sports), moderate (such as moving a table, pushing a vacuum cleaner, bowling or playing golf), walking one block/climbing one flight of stairs and leaving one’s home, because of breathing problems. This limitation could be reported in three grades: no limitation (0), yes, did activity slowly (1) and yes, avoided activity completely (2). Questions on whether they have been outside the house or town and for how long have also been included.\nDuring the study period there was personal contact with the participants once a month to collect the completed diary forms, discuss potential problems and keep the motivation at a good level.\n Air pollution exposure Exposure assessment has been described in previous publications\n[16-18,20]. In brief, during the entire study period (October 2002 to March 2004) in each city measurements of PM2.5, PM10 and PNC were performed continuously at a central site representing urban background levels\n[17]. The same type of condensation particle counter (TSI 3022A, TSI Inc., St. Paul, MN, USA) was used in each city to monitor PNC. 24-hour average particle mass concentration was measured with Harvard impactors for PM2.5 and PM10. Coarse particles concentrations were calculated by subtracting PM2.5 from PM10. After weighing, the absorbance of the PM2.5 filters (a good surrogate for elemental carbon/soot) was determined using reflectometry. PNC was transformed to “noon-to-noon” 24-hour means to coincide with the PM2.5 measurements. Data on concentrations of other air pollutants (ozone, nitrogen dioxide) and meteorology (air temperature, relative humidity) were collected from existing national monitoring networks in each country. We did not replace missing values in exposures variables by imputation.\nExposure assessment has been described in previous publications\n[16-18,20]. In brief, during the entire study period (October 2002 to March 2004) in each city measurements of PM2.5, PM10 and PNC were performed continuously at a central site representing urban background levels\n[17]. The same type of condensation particle counter (TSI 3022A, TSI Inc., St. Paul, MN, USA) was used in each city to monitor PNC. 24-hour average particle mass concentration was measured with Harvard impactors for PM2.5 and PM10. Coarse particles concentrations were calculated by subtracting PM2.5 from PM10. After weighing, the absorbance of the PM2.5 filters (a good surrogate for elemental carbon/soot) was determined using reflectometry. PNC was transformed to “noon-to-noon” 24-hour means to coincide with the PM2.5 measurements. Data on concentrations of other air pollutants (ozone, nitrogen dioxide) and meteorology (air temperature, relative humidity) were collected from existing national monitoring networks in each country. We did not replace missing values in exposures variables by imputation.\n Confounder data Time trend in health endpoints (e.g. fatigue in reporting), weather (outdoor temperature, relative humidity), medication use and day of the week were taken into account as potential confounders. Because of the staged entry of participants, we evaluated two time variables: calendar date (proxy for unmeasured confounders) and day of study for a specific subject (possibly related to fatigue).\nTime trend in health endpoints (e.g. fatigue in reporting), weather (outdoor temperature, relative humidity), medication use and day of the week were taken into account as potential confounders. Because of the staged entry of participants, we evaluated two time variables: calendar date (proxy for unmeasured confounders) and day of study for a specific subject (possibly related to fatigue).\n Quality assurance/quality control Air pollution and health measurements were performed according to standard operating procedures (SOPs). A training workshop was organized before the start of the fieldwork and site visits were implemented during the fieldwork to identify any deviations from SOPs.\nAir pollution and health measurements were performed according to standard operating procedures (SOPs). A training workshop was organized before the start of the fieldwork and site visits were implemented during the fieldwork to identify any deviations from SOPs.\n Statistical analysis Data analysis was done according to a predefined analysis plan. The symptom variables, initially coded as 0 for no symptoms (absent), 1 for slight symptoms and 2 for moderate/severe symptoms, were dichotomised for the analysis by setting 0 for no symptoms and 1 for slight to moderate/severe symptoms. Each symptom was analysed separately either as prevalent (irrespective of its occurrence on the previous day) or incident (when that symptom was reported to be absent on the previous day). Medication use was coded as 0 (no medication) versus 1 (intake of one or more doses) independently of the initial medication group. Every person was included in the analysis regardless of how many diary entries were made. Moreover, diary entries were excluded when participants had left the study area during the measurement period. For every pollutant the following lags were evaluated: lag 0, 1, 2 and the average of lag 0–6 days. Lag 0 was defined as the 24-hour period starting from noon of the calendar day before the health response.\nA hierarchical modelling approach was used. First, regression models were fitted in each city separately to allow specific control for seasonal effects, weather and other potential confounders. Results of the individual city analysis were used in a second stage analysis (meta-analysis) to provide overall estimates\n[26]. We computed both fixed and random effects combined estimates. Furthermore, a chi-square test of heterogeneity of the four city-specific estimates was computed.\nWe applied logistic regression to obtain centre-specific effect estimates. A smooth function (natural splines with 6 degrees of freedom per year) of time was used to remove the seasonal patterns and long time trends from the data. Afterwards, same-day (lag 0) and previous-day (lag 1) mean daily temperatures were introduced simultaneously into the model. For both lags of temperature, a linear term was compared with a smoothed function (natural splines) with 2, 3 and 4 degrees of freedom and the model with the lowest Akaike’s Information Criterion (AIC) was selected. A linear term of relative humidity (lag 0) was added to the model as another indicator of weather. Finally, indicator variables for day of the week, medication use and individual differences in frequency of symptoms, were added to the model. After setting up the baseline model, the effects of the various lags of the pollutants were evaluated.\nIn the city specific analysis we fitted fixed effects models, described above, as well as random intercept logistic regression models using “glmmPQL” function from MASS library in R software, to take into account the correlation among each subject’s measurements. Results from the random effects analysis were very similar to those derived from fixed effects. In a few cases though, we faced convergence issues. This was even more the case when we tested a first order autoregressive correlation structure. The significance of the associations was similar between random intercept models and the models incorporating an autoregressive term.\nBecause of the heterogeneity of the study population, we repeated the analysis (for all air pollution measures) for the subgroup of asthmatic patients. There were not enough COPD patients to analyse these patients separately. We also fitted two pollutant models by including simultaneously PM2.5 and PM10-2.5 in order to better characterize which of the two components of PM10 (PM10-2.5 or PM2.5) was responsible for the observed health effects.\nEffect estimates are expressed as odds ratios (OR) for an increase of 10 μg/m3 in PM10, 10 μg/m3 in PM2.5, 10 μg/m3 in PM10-2.5, 10,000 particles/cm3 for PNC and 1·10-5 m-1 for absorbance, in order to be comparable with other studies. For gaseous pollutants the effect estimates are expressed as OR for an increase of 10 μg/m3 in ozone and NO2 concentrations.All analyses were performed using R software\n[27].\nData analysis was done according to a predefined analysis plan. The symptom variables, initially coded as 0 for no symptoms (absent), 1 for slight symptoms and 2 for moderate/severe symptoms, were dichotomised for the analysis by setting 0 for no symptoms and 1 for slight to moderate/severe symptoms. Each symptom was analysed separately either as prevalent (irrespective of its occurrence on the previous day) or incident (when that symptom was reported to be absent on the previous day). Medication use was coded as 0 (no medication) versus 1 (intake of one or more doses) independently of the initial medication group. Every person was included in the analysis regardless of how many diary entries were made. Moreover, diary entries were excluded when participants had left the study area during the measurement period. For every pollutant the following lags were evaluated: lag 0, 1, 2 and the average of lag 0–6 days. Lag 0 was defined as the 24-hour period starting from noon of the calendar day before the health response.\nA hierarchical modelling approach was used. First, regression models were fitted in each city separately to allow specific control for seasonal effects, weather and other potential confounders. Results of the individual city analysis were used in a second stage analysis (meta-analysis) to provide overall estimates\n[26]. We computed both fixed and random effects combined estimates. Furthermore, a chi-square test of heterogeneity of the four city-specific estimates was computed.\nWe applied logistic regression to obtain centre-specific effect estimates. A smooth function (natural splines with 6 degrees of freedom per year) of time was used to remove the seasonal patterns and long time trends from the data. Afterwards, same-day (lag 0) and previous-day (lag 1) mean daily temperatures were introduced simultaneously into the model. For both lags of temperature, a linear term was compared with a smoothed function (natural splines) with 2, 3 and 4 degrees of freedom and the model with the lowest Akaike’s Information Criterion (AIC) was selected. A linear term of relative humidity (lag 0) was added to the model as another indicator of weather. Finally, indicator variables for day of the week, medication use and individual differences in frequency of symptoms, were added to the model. After setting up the baseline model, the effects of the various lags of the pollutants were evaluated.\nIn the city specific analysis we fitted fixed effects models, described above, as well as random intercept logistic regression models using “glmmPQL” function from MASS library in R software, to take into account the correlation among each subject’s measurements. Results from the random effects analysis were very similar to those derived from fixed effects. In a few cases though, we faced convergence issues. This was even more the case when we tested a first order autoregressive correlation structure. The significance of the associations was similar between random intercept models and the models incorporating an autoregressive term.\nBecause of the heterogeneity of the study population, we repeated the analysis (for all air pollution measures) for the subgroup of asthmatic patients. There were not enough COPD patients to analyse these patients separately. We also fitted two pollutant models by including simultaneously PM2.5 and PM10-2.5 in order to better characterize which of the two components of PM10 (PM10-2.5 or PM2.5) was responsible for the observed health effects.\nEffect estimates are expressed as odds ratios (OR) for an increase of 10 μg/m3 in PM10, 10 μg/m3 in PM2.5, 10 μg/m3 in PM10-2.5, 10,000 particles/cm3 for PNC and 1·10-5 m-1 for absorbance, in order to be comparable with other studies. For gaseous pollutants the effect estimates are expressed as OR for an increase of 10 μg/m3 in ozone and NO2 concentrations.All analyses were performed using R software\n[27].", "In the context of RUPIOH, a multicentre study was conducted from October 2002 to March 2004 in four European metropolitan areas, namely, Amsterdam (The Netherlands), Athens (Greece), Birmingham (United Kingdom) and Helsinki (Finland). During the whole study period a central site in each city was used to monitor particle mass and PNC on a daily basis. At various locations covering the entire metropolitan area, homes of participants with either asthma or COPD were selected. The criteria for the central site and homes selection have been described in detail in a previous publication\n[17]. Respiratory health status of each participant was monitored for six months by a daily symptom diary. We used a staged entry of the participants (based on the real date the participants started to fill out the diaries) in order to increase the period of data collection and thus, decrease the likelihood for uncontrolled factors or unexpected events to influence the associations between air pollution and health\n[21]. In all centres, participants were recruited between October 2002 and March 2004.", "Inclusion criteria and recruitment procedures have been described in detail before\n[19]. Briefly, in each city the recruitment criteria for participants were age 35 or more, a doctor diagnosis of either asthma (as defined by Global Initiative for Asthma) or COPD (as defined by Global Initiative for Chronic Obstructive Lung Disease) and having had experienced respiratory symptoms in the past 12 months\n[22,23]. Especially, in the Netherlands some patients who had not received a definite diagnosis of asthma or COPD were classified as chronic non-specific lung disease (CNSLD) as a relic of tradition (term previously used to indicate either asthma or COPD)\n[24]. Severe patients defined as those using relief bronchodilating medications more than three times per day or using nebulised bronchodilators or long-term oxygen therapy as well as participants unable to perform a satisfactory spirometry test were excluded from the study. An attempt was made to select non-working, non-smoking patients living in a non-smoking household to eliminate potential confounding by occupational exposures to airborne particles and by environmental tobacco smoke. The same screening questionnaire was used across the four centres to ascertain eligibility. However, each centre was allowed to choose the optimal subject recruitment method. Specifically, in Amsterdam, the panelists were recruited through distribution of 10,000 information letters accompanied by screening questionnaires. Inclusion criteria were checked using the returned screening questionnaires followed by participants’ homes’ visits. In Athens, subjects recruited through local hospitals and pulmonary chest physicians were visited at home by a pulmonologist (A.K.) and one of the investigators of the exposure assessment team (I.K.) who checked whether inclusion criteria were met. In Finland, subjects were selected from the Helsinki Metropolitan Area (including cities of Helsinki, Espoo and Vantaa) by placing advertisement on two issues of the respiratory patient association magazine (circulation ~3500 households) and notice boards of pulmonary disease clinics of four major hospitals within the study area. Candidate subjects were interviewed and screened by telephone and invited to an information session when they met the criteria. In the United Kingdom, potential study subjects living in the greater area of Birmingham were selected from the Clinic for Respiratory illnesses (CRI) database of respiratory patients at the Heartlands Hospital. Privacy regulations restricted the selections to only those that had given their written consent to be approached for research studies.\nMedical ethical clearance was acquired from the relevant local medical ethics committees in all centres before the start of the recruitment. Written informed consent was obtained from each subject.", "The diary was based upon diaries used in previous studies of acute effects of air pollution such as the PEACE study\n[21]. Although there is no real objective method of validating symptoms, a previous study by Hoek et al. provide evidence that symptoms, assessed with the same diary, are reflected in lung function drops\n[25]. Participants were instructed to complete a daily record about respiratory symptoms and medication taken “as needed” for six months, grading shortness of breath, wheeze, cough, phlegm and woken with breathing problems as absent (0), slight (1), or moderate/severe (2). In addition, they were asked about any limitation in performing daily life activities categorized as vigorous (such as running, lifting heavy objects, participating in strenuous sports), moderate (such as moving a table, pushing a vacuum cleaner, bowling or playing golf), walking one block/climbing one flight of stairs and leaving one’s home, because of breathing problems. This limitation could be reported in three grades: no limitation (0), yes, did activity slowly (1) and yes, avoided activity completely (2). Questions on whether they have been outside the house or town and for how long have also been included.\nDuring the study period there was personal contact with the participants once a month to collect the completed diary forms, discuss potential problems and keep the motivation at a good level.", "Exposure assessment has been described in previous publications\n[16-18,20]. In brief, during the entire study period (October 2002 to March 2004) in each city measurements of PM2.5, PM10 and PNC were performed continuously at a central site representing urban background levels\n[17]. The same type of condensation particle counter (TSI 3022A, TSI Inc., St. Paul, MN, USA) was used in each city to monitor PNC. 24-hour average particle mass concentration was measured with Harvard impactors for PM2.5 and PM10. Coarse particles concentrations were calculated by subtracting PM2.5 from PM10. After weighing, the absorbance of the PM2.5 filters (a good surrogate for elemental carbon/soot) was determined using reflectometry. PNC was transformed to “noon-to-noon” 24-hour means to coincide with the PM2.5 measurements. Data on concentrations of other air pollutants (ozone, nitrogen dioxide) and meteorology (air temperature, relative humidity) were collected from existing national monitoring networks in each country. We did not replace missing values in exposures variables by imputation.", "Time trend in health endpoints (e.g. fatigue in reporting), weather (outdoor temperature, relative humidity), medication use and day of the week were taken into account as potential confounders. Because of the staged entry of participants, we evaluated two time variables: calendar date (proxy for unmeasured confounders) and day of study for a specific subject (possibly related to fatigue).", "Air pollution and health measurements were performed according to standard operating procedures (SOPs). A training workshop was organized before the start of the fieldwork and site visits were implemented during the fieldwork to identify any deviations from SOPs.", "Data analysis was done according to a predefined analysis plan. The symptom variables, initially coded as 0 for no symptoms (absent), 1 for slight symptoms and 2 for moderate/severe symptoms, were dichotomised for the analysis by setting 0 for no symptoms and 1 for slight to moderate/severe symptoms. Each symptom was analysed separately either as prevalent (irrespective of its occurrence on the previous day) or incident (when that symptom was reported to be absent on the previous day). Medication use was coded as 0 (no medication) versus 1 (intake of one or more doses) independently of the initial medication group. Every person was included in the analysis regardless of how many diary entries were made. Moreover, diary entries were excluded when participants had left the study area during the measurement period. For every pollutant the following lags were evaluated: lag 0, 1, 2 and the average of lag 0–6 days. Lag 0 was defined as the 24-hour period starting from noon of the calendar day before the health response.\nA hierarchical modelling approach was used. First, regression models were fitted in each city separately to allow specific control for seasonal effects, weather and other potential confounders. Results of the individual city analysis were used in a second stage analysis (meta-analysis) to provide overall estimates\n[26]. We computed both fixed and random effects combined estimates. Furthermore, a chi-square test of heterogeneity of the four city-specific estimates was computed.\nWe applied logistic regression to obtain centre-specific effect estimates. A smooth function (natural splines with 6 degrees of freedom per year) of time was used to remove the seasonal patterns and long time trends from the data. Afterwards, same-day (lag 0) and previous-day (lag 1) mean daily temperatures were introduced simultaneously into the model. For both lags of temperature, a linear term was compared with a smoothed function (natural splines) with 2, 3 and 4 degrees of freedom and the model with the lowest Akaike’s Information Criterion (AIC) was selected. A linear term of relative humidity (lag 0) was added to the model as another indicator of weather. Finally, indicator variables for day of the week, medication use and individual differences in frequency of symptoms, were added to the model. After setting up the baseline model, the effects of the various lags of the pollutants were evaluated.\nIn the city specific analysis we fitted fixed effects models, described above, as well as random intercept logistic regression models using “glmmPQL” function from MASS library in R software, to take into account the correlation among each subject’s measurements. Results from the random effects analysis were very similar to those derived from fixed effects. In a few cases though, we faced convergence issues. This was even more the case when we tested a first order autoregressive correlation structure. The significance of the associations was similar between random intercept models and the models incorporating an autoregressive term.\nBecause of the heterogeneity of the study population, we repeated the analysis (for all air pollution measures) for the subgroup of asthmatic patients. There were not enough COPD patients to analyse these patients separately. We also fitted two pollutant models by including simultaneously PM2.5 and PM10-2.5 in order to better characterize which of the two components of PM10 (PM10-2.5 or PM2.5) was responsible for the observed health effects.\nEffect estimates are expressed as odds ratios (OR) for an increase of 10 μg/m3 in PM10, 10 μg/m3 in PM2.5, 10 μg/m3 in PM10-2.5, 10,000 particles/cm3 for PNC and 1·10-5 m-1 for absorbance, in order to be comparable with other studies. For gaseous pollutants the effect estimates are expressed as OR for an increase of 10 μg/m3 in ozone and NO2 concentrations.All analyses were performed using R software\n[27].", " Panel characteristics A brief description of the study population is presented in Table\n1. Mean age and age range were about the same in all cities. Three participants in Athens were slightly below of the recruitment criterion of ≥35 years. In Amsterdam a large group was reported to have CNSLD. Medication use was high in the panels. Seventy seven per cent of the participants (77%) used reliever medication. Use of “as needed medication” was recorded in 26.5% of total person days in Helsinki, 13.9% in Athens, 37.9% in Amsterdam and 59.7% in Birmingham. Twenty-nine participants (21%) worked outside their home especially from Amsterdam and Birmingham. Those who worked outside their home, worked on average 19 h/week.\nCharacteristics of four European panels of asthmatic/COPD patients\na Total participants in panel.\nb Asthma + COPD or chronic non-specific lung disease.\nc Given as mean and [range].\nd Chronic non-specific lung disease.\ne Environmental tobacco smoke.\nf Includes short acting β2-agonist, long acting β2-agonist, anticholinergic drugs and combination of an anticholinergic drug and a β2-agonist.\nA brief description of the study population is presented in Table\n1. Mean age and age range were about the same in all cities. Three participants in Athens were slightly below of the recruitment criterion of ≥35 years. In Amsterdam a large group was reported to have CNSLD. Medication use was high in the panels. Seventy seven per cent of the participants (77%) used reliever medication. Use of “as needed medication” was recorded in 26.5% of total person days in Helsinki, 13.9% in Athens, 37.9% in Amsterdam and 59.7% in Birmingham. Twenty-nine participants (21%) worked outside their home especially from Amsterdam and Birmingham. Those who worked outside their home, worked on average 19 h/week.\nCharacteristics of four European panels of asthmatic/COPD patients\na Total participants in panel.\nb Asthma + COPD or chronic non-specific lung disease.\nc Given as mean and [range].\nd Chronic non-specific lung disease.\ne Environmental tobacco smoke.\nf Includes short acting β2-agonist, long acting β2-agonist, anticholinergic drugs and combination of an anticholinergic drug and a β2-agonist.\n Symptoms In total between 4,760 and 6,003 person days were available for analysis in the four cities. In Amsterdam, Athens and Birmingham participants filled out the diary from October 2002 to March 2004 whilst in Helsinki between October 2002 and February 2004. Missing values (person days) ranged between 9.4-15.1% in Amsterdam, 4.7-5.5% in Athens, 8.7-8.8% in Birmingham and 8.6-12.1% in Helsinki. Consistent with the composition of the panel, fairly high symptom prevalence occurred during the study period. Person days with severe symptoms were low, except for cough and phlegm. There were small differences between the cities (Table\n2).\nPerson days with symptoms in the diary (n = number of expected person days)\na due to breathing problems.\nIn total between 4,760 and 6,003 person days were available for analysis in the four cities. In Amsterdam, Athens and Birmingham participants filled out the diary from October 2002 to March 2004 whilst in Helsinki between October 2002 and February 2004. Missing values (person days) ranged between 9.4-15.1% in Amsterdam, 4.7-5.5% in Athens, 8.7-8.8% in Birmingham and 8.6-12.1% in Helsinki. Consistent with the composition of the panel, fairly high symptom prevalence occurred during the study period. Person days with severe symptoms were low, except for cough and phlegm. There were small differences between the cities (Table\n2).\nPerson days with symptoms in the diary (n = number of expected person days)\na due to breathing problems.\n Air pollution concentrations Helsinki had the lowest median concentrations for all PM components whilst Athens had the highest. However, maximum concentrations of PM2.5 were observed in Amsterdam (103.4 μg/m3) and of PM10-2.5 (152.6 μg/m3) in Helsinki (Table\n3).\nDaily (24 hours noon-to-noon, central site) median air pollution concentration and meteorology in the four cities\nHelsinki had the lowest median concentrations for all PM components whilst Athens had the highest. However, maximum concentrations of PM2.5 were observed in Amsterdam (103.4 μg/m3) and of PM10-2.5 (152.6 μg/m3) in Helsinki (Table\n3).\nDaily (24 hours noon-to-noon, central site) median air pollution concentration and meteorology in the four cities\n Air pollution effects on symptoms-limitation in activities due to breathing problems Prevalence analyses We observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n\nWe observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n\n Prevalence analyses We observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n\nWe observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n\n Incidence analyses Patterns similar to those in the combined prevalence analyses were observed for the associations of incident symptoms and particles especially the coarse fraction. Shortness of breath was consistently associated with PM10 and PM10-2.5 in lag 1 with no indication of heterogeneity between the centres (OR = 1.045, 95% CI: 1.008, 1.083 and OR = 1.065, 95% CI: 1.009, 1.124 respectively). There was also a tendency towards positive associations between PM10-2.5 and incidence of wheezing, cough and limitation in walking but none of the associations were statistically significant. Additionally, ozone was positively associated with cough in lags 1 and 2 as well as the average lag 0–6 days but only in lag 2 the association reached the nominal level of significance (Table\n7).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith incidence of symptoms in the four panels (random effects pooled estimates)\n\nBold are significant pooled effects.\nPatterns similar to those in the combined prevalence analyses were observed for the associations of incident symptoms and particles especially the coarse fraction. Shortness of breath was consistently associated with PM10 and PM10-2.5 in lag 1 with no indication of heterogeneity between the centres (OR = 1.045, 95% CI: 1.008, 1.083 and OR = 1.065, 95% CI: 1.009, 1.124 respectively). There was also a tendency towards positive associations between PM10-2.5 and incidence of wheezing, cough and limitation in walking but none of the associations were statistically significant. Additionally, ozone was positively associated with cough in lags 1 and 2 as well as the average lag 0–6 days but only in lag 2 the association reached the nominal level of significance (Table\n7).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith incidence of symptoms in the four panels (random effects pooled estimates)\n\nBold are significant pooled effects.", "A brief description of the study population is presented in Table\n1. Mean age and age range were about the same in all cities. Three participants in Athens were slightly below of the recruitment criterion of ≥35 years. In Amsterdam a large group was reported to have CNSLD. Medication use was high in the panels. Seventy seven per cent of the participants (77%) used reliever medication. Use of “as needed medication” was recorded in 26.5% of total person days in Helsinki, 13.9% in Athens, 37.9% in Amsterdam and 59.7% in Birmingham. Twenty-nine participants (21%) worked outside their home especially from Amsterdam and Birmingham. Those who worked outside their home, worked on average 19 h/week.\nCharacteristics of four European panels of asthmatic/COPD patients\na Total participants in panel.\nb Asthma + COPD or chronic non-specific lung disease.\nc Given as mean and [range].\nd Chronic non-specific lung disease.\ne Environmental tobacco smoke.\nf Includes short acting β2-agonist, long acting β2-agonist, anticholinergic drugs and combination of an anticholinergic drug and a β2-agonist.", "In total between 4,760 and 6,003 person days were available for analysis in the four cities. In Amsterdam, Athens and Birmingham participants filled out the diary from October 2002 to March 2004 whilst in Helsinki between October 2002 and February 2004. Missing values (person days) ranged between 9.4-15.1% in Amsterdam, 4.7-5.5% in Athens, 8.7-8.8% in Birmingham and 8.6-12.1% in Helsinki. Consistent with the composition of the panel, fairly high symptom prevalence occurred during the study period. Person days with severe symptoms were low, except for cough and phlegm. There were small differences between the cities (Table\n2).\nPerson days with symptoms in the diary (n = number of expected person days)\na due to breathing problems.", "Helsinki had the lowest median concentrations for all PM components whilst Athens had the highest. However, maximum concentrations of PM2.5 were observed in Amsterdam (103.4 μg/m3) and of PM10-2.5 (152.6 μg/m3) in Helsinki (Table\n3).\nDaily (24 hours noon-to-noon, central site) median air pollution concentration and meteorology in the four cities", " Prevalence analyses We observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n\nWe observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n", "We observed very small differences in fixed and random effects combined estimates. In Tables\n4 and\n5 combined odds ratios for the association of particulate matter indices, NO2, ozone and prevalence of symptoms and limitation in activities are presented, using random effects models adjusting for the above mentioned confounders and “as needed” medication. When all participants were included in the analysis as a total, we found that a 10 μg/m3 increase in PM10 was significantly associated at the nominal level with shortness of breath in the lag 1 whilst the association in the lags 2 and 0 to 6 was of borderline significance. However, none of the associations was significant for the asthma group. Significant association was also observed for wheezing and limitation in walking due to breathing problems (lag 1). The association was driven by the PM10-2.5 component of PM10 and much less by PM2.5. Coarse particles concentrations were positively associated with most symptom and restriction of activities variables in lag1. In addition, the modest correlations between PM10-2.5 and PM2.5 (0.08, 0.40, 0.35 and 0.13 for Amsterdam, Athens, Birmingham and Helsinki respectively) did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5 (Table\n6).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith prevalence of symptoms in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith limitation in activities due to breathing problems in all participants and the subgroup of asthmatics (random effects pooled estimates)\n\nBold are significant pooled effects.\n\nAssociations of PM\n\n10-2.5 \n\nand PM\n\n2.5 \n\nwith prevalence of symptoms and limitation in activities due to breathing problems after applying two pollutant models (random effects pooled estimates)\n\nBold are significant pooled effects.\nThe above-mentioned positive associations with PM10-2.5 (Tables\n4 and\n5) were reduced and no longer significant after restricting the analysis to the asthmatic only participants. A significant association remained with restricting walking activities and wheeze (borderline).\nOzone was significantly associated with cough at lag 0, lag 1, lag 2 and with woken with breathing problems at lag 0. Furthermore, the associations with wheezing, limitation in vigorous activities and walking due to breathing problems remained positive across all examined lags although, non significant. Negative but non significant associations were observed with shortness of breath across all examined lags. However, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2, in the asthma group. Moreover, in the asthmatics, negative associations were also observed for ozone with woken with breathing problems (lag 0), wheezing (lag 0 and lag 1), and with limitation in activities due to breathing problems (most of the lags), although non significant (Tables\n4 and\n5).\nNeither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst the positive associations with woken with breathing problems and cough in lag 1 as well as in limitation of activities due to breathing problems (mainly vigorous and moderate) in lags 0, 1, 2 did not reach the nominal level of significance. Moreover, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant (Tables\n4 and\n5).\nCentre specific and overall effect estimates with 95 percent confidence intervals (95% CI) for the association of each symptom and air pollutant in lag1 are presented in Figure\n1. Odds ratios (OR) for the effect of PM10-2.5 were consistently above one in almost every city as well as in the pooled data using random effects meta-analysis.\n\nOdds ratio (95% CI) for prevalence of symptoms and limitation in activities associated with an increase of 10 μg/m\n\n3 \n\nin previous day (lag1) concentrations of each pollutant (10,000/cm\n\n3 \n\nfor PNC) in each participating city and overall estimate (random effects pooled estimates).\n", "Patterns similar to those in the combined prevalence analyses were observed for the associations of incident symptoms and particles especially the coarse fraction. Shortness of breath was consistently associated with PM10 and PM10-2.5 in lag 1 with no indication of heterogeneity between the centres (OR = 1.045, 95% CI: 1.008, 1.083 and OR = 1.065, 95% CI: 1.009, 1.124 respectively). There was also a tendency towards positive associations between PM10-2.5 and incidence of wheezing, cough and limitation in walking but none of the associations were statistically significant. Additionally, ozone was positively associated with cough in lags 1 and 2 as well as the average lag 0–6 days but only in lag 2 the association reached the nominal level of significance (Table\n7).\n\nAssociations of particulate matter indices, NO\n\n2 \n\nand O\n\n3 \n\nwith incidence of symptoms in the four panels (random effects pooled estimates)\n\nBold are significant pooled effects.", "In this multicentre study we found consistent positive associations between coarse particles central sites concentrations and prevalence of respiratory symptoms, as recorded in a 6-month diary, in four panels of participants with predominantly mild to moderate asthma or COPD in four European cities participating in the RUPIOH study. We also found a significant association of ozone with cough and woken with breathing problems, but not with other symptoms. Neither PM2.5 nor NO2 were consistently associated with any symptom or limitation in activities variable. As for PNC a (mostly non-significant) negative association was observed with most symptoms whilst positive associations with woken with breathing problems and cough as well as with limitation of vigorous and moderate activities due to breathing problems, did not reach the nominal level of significance. Interestingly, for PNC a change of the negative associations with woken with breathing problems towards positive values, across all lags, was observed when the analysis was restricted to the asthmatic participants, although non significant at the nominal level. An analysis of the asthmatic subgroup showed generally lower odds ratios for PM10-2.5.\nOne particularity and strength of the RUPIOH study is the in depth assessment of particulate air pollution by measuring PM10, PM2.5 (then deriving coarse particles), filters absorbance as well as the number of ultrafine particles. Previous work from RUPIOH that included air pollution monitoring for one week inside and directly outside participants’ homes reported no association with lung function\n[19]. As the authors stated a potential explanation could be the high prevalence of medication use, the short period of measurements (one week) that limited the ability to assess lagged effects over several days or absence of an effect. The high prevalence of medication use may also have covered some associations in the present study.\nA limitation of the study is the inclusion of both COPD and asthma patients. COPD and asthma are two diseases with different underlying pathophysiological mechanisms and day to day variability in their symptoms\n[22,23]. Mixing of the two diseases does not create bias in the analysis in the full population as we adjusted for differences in health status between individuals. The generalizability of the size of the effect estimates is more affected by the population. Though asthma and COPD are different diseases, we are not aware of studies that have demonstrated differences in the magnitude of response to air pollution. In our recently accepted paper in the same panel we did not find any difference in the effect of PM10-2.5 on total nitrate and nitrite concentrations in exhaled breath condensate (EBC NOx), a marker of oxidative stress between asthma and COPD patients\n[28]. In that study we could evaluate disease status as the outcome was a continuous variable. Unfortunately, an analysis restricted to COPD patients was not possible due to the small number of COPD patients participating in Helsinki and Birmingham. Hence, we also could not test whether the smaller PM10-2.5 effect in the asthmatic subgroup differed significantly from the COPD subgroup.\nOur coarse particle findings are however consistent with the observation that in the RUPIOH study only the PM10-2.5 concentration at central sites was significantly associated with increased EBC NOx collected during the same week as the spirometry\n[28]. EBC NOx has been suggested as a reliable marker of oxidative stress\n[29-31]. The link between PM10-2.5 with oxidative stress and airway inflammation may explain the increase in respiratory symptoms we found.\nIn this study, we also report significant positive associations of ozone with cough throughout most of the examined lags both in the analysis of total participants and the subgroup of the asthmatics that are consistent with previous epidemiological and toxicological studies\n[32]. In addition, positive associations, but not significant in the nominal level, were observed with most of symptoms when total participants were included in the analysis. However, when we restricted the analysis to the subgroup of asthmatics, a significant preventive effect of ozone for shortness of breath was revealed for lags 1 and 2. Negative associations were also observed with woken with breathing problems, wheezing and with limitation in activities due to breathing problems, although non significant. Factors like high medication use, intrinsic differences in responsiveness to ozone among individuals, adaptation to ozone issues or other spurious effects may have been responsible for these findings\n[32].\nIn the last two decades a substantial body of literature has focused on the harmful health effects of PM10 and PM2.5[15,32]. As a result guideline values have been recommended by the U.S. Environmental Protection Agency and World Health Organization for both indicators of PM pollution to protect public health\n[2,3]. However, from recent studies there is increasing evidence that the health effects of coarse particles should not be underestimated. In a systematic review of epidemiological studies that have analyzed fine and coarse PM jointly, Brunekreef and Forsberg examined the epidemiological evidence for effects of coarse particles on health\n[15]. They concluded that the effects of PM10-2.5 were stronger than or as strong as PM2.5 on short-term respiratory morbidity. Furthermore, in a national multicity study, Zanobetti and Schwartz found a strong association of both fine and coarse particles with daily deaths in 112 U.S. cities\n[33]. A 10 μg/m3 increase in PM10-2.5 was significantly associated with total mortality, stroke, cardiovascular, and respiratory mortality, the latter of which showing the largest effect (a 1.2% increase). Mechanistically, these effects may be due either to biogenic factors or to metals carried by PM10-2.5 by activation of inflammatory and oxidative stress pathways\n[34-36]. The findings of our study support previous epidemiological and toxicological evidence that health effects due to the coarse fraction may be substantial\n[37].\nThe large number of calculations we have done could have given some statistically significant associations by chance. However, multiple testing is an unlikely explanation of the findings in the current study. In the full study population we found 14 significant associations of which 10 were positive; in the asthmatics subgroup we found 12 significant associations of which 8 were positive. The consistency of associations (e.g. for ozone and cough) further argues against chance as the main explanation for our findings. Additionally, in the full study population the significant associations for PM10-2.5 were supported by elevated though not nominally significant ORs for other lags and symptoms. Finally, ORs were mainly homogeneous across centers. Moreover, the modest correlations between PM10-2.5 and PM2.5 did allow us to apply a two-pollutant model in order to separate and further evaluate the effects of the two components of PM10. The magnitude of the associations for PM10-2.5 with prevalence of symptoms and restriction of activities remained approximately the same or increased when we applied a two-pollutant model with PM2.5.\nThe majority of studies that investigated health effects of particulate pollutants have expressed results on a mass basis. It has been suggested that when taking into consideration particle number or surface area, the pulmonary dose of toxic material related to PM2.5 may be much larger than the dose related to PM10-2.5 that for this reason alone, comparison on a mass basis may be less informative\n[15]. In our study we separately investigated the mass and the number effect. Neither central site PM2.5 nor PNC were consistently associated with symptoms. The association we observed with PM10-2.5, if not by chance, may also imply that a central measurement site is more appropriate for measurements of mass concentrations than for PNC. The analysis of RUPIOH data by Puustinen et al. showed generally high correlations between 24 hour average central site and residential outdoor concentrations for PM2.5 and soot with a lesser median correlation for PM10 and a lower correlation for PNC and PM10-2.5[17]. For PM10-2.5 correlations between central site and home outdoor measurements were 0.66, 0.74, 0.89 and 0.64 in Helsinki, Athens, Amsterdam and Birmingham respectively. A central site thus provides a reasonably good estimate of more local exposures even for coarse particles.\nThe relatively high divergence of PM10-2.5 concentrations between proximate sites in the UK has recently been confirmed by Liu and Harrison\n[38]. Consequently, for both PNC and PM10-2.5, there is a higher probability of exposure misclassification than for PM2.5 or soot. The finding of significant associations with respiratory health outcomes for PM10-2.5 but not for PNC is therefore quite striking but consistent with the recent findings of a time series study in London which found significant associations between PNC and cardiovascular health outcomes whilst PM mass metrics were associated with respiratory outcomes\n[39]. A plausible explanation could be the existence of different biological and pathophysiological mechanisms through which PM10-2.5 and PNC exert their adverse effects or different target organs. The results of recent toxicological studies support the theory that PM10-2.5 exert their effects at the site of deposition in the airways whereas PNC, after crossing the alveolar epithelial barrier, enter into the systemic circulation and affect cardiovascular function\n[40,41]. This theory could explain the positive associations we found between PM10-2.5 and respiratory symptoms.\nIn summary, our study contributes to the literature on the health effects of PM in respiratory patients. Moreover, the results of our study are in agreement with the findings of recent epidemiological and toxicological studies and provide enough evidence to conclude that it is prudent to keep PM10-2.5 regulated in addition to fine particles.", "Our study adds to the limited existing evidence of recent epidemiological and toxicological studies that health effects due to the coarse fraction of ambient PM may be substantial. Further studies are needed to clarify possible different effects of PM on COPD and asthmatic patients. The observed associations suggest it is prudent to regulate also coarse particles in addition to fine particles.", "AIC: Akaike’s information criterion; CNSLD: Chronic non-specific lung disease; COPD: Chronic obstructive pulmonary disease; EBC NOx: Total nitrate and nitrite concentrations in exhaled breath condensate; OR: Odds ratio; PM: Particulate matter; PM10-2.5: Coarse particles; PM10: Mass concentration of particles less than 10 μm; PM2.5: Mass concentration of particles less than 2.5 μm; PNC: Particle number concentrations; RUPIOH: Relationship between Ultrafine and fine Particulate matter in Indoor and Outdoor air and respiratory Health; SOPs: Standard operating procedures; 95% CI: 95% confidence interval.", "The authors declare that they have no competing interests.", "All authors of this paper have critically read and approved the final version submitted. They have also made substantive intellectual contributions by directly participating either in the planning, execution, or analysis of the study. AK contributed to the development of the study design, acquisition and interpretation of data and drafted the paper. AA did the analysis, contributed to the interpretation of data and wrote the statistical analysis section of the paper. DP, IGK, JJdeH contributed substantially to acquisition and interpretation of data. JGA, RMH, AK, JP, KH, GPAK, KK contributed to the study design, interpretation of data and have been involved in drafting the manuscript. GH conceived and developed the study design, contributed to the interpretation of data and was involved in drafting the paper. All authors have revised drafts and contributed to the revisions." ]

[ null, "methods", null, null, null, null, null, null, null, "results", null, null, null, null, null, null, "discussion", "conclusions", null, null, null ]

[ "Air pollution", "Asthma", "Chronic obstructive pulmonary disease", "Coarse particles", "Particle number concentration", "Respiratory health" ]

[ 657, 203, 477, 281, 205, 73, 42, 756, 230, 151, 75, 1750, 873, 186, 114, 10, 157 ]

[ "effects", "pm10", "associations", "significant", "symptoms", "lag", "participants", "problems", "breathing", "breathing problems" ]

[ "particulate pollutants expressed", "outdoor concentrations pm2", "pm10 concentrations", "mass particulate matter", "ambient particulate matter" ]

[ "Brazil", "Cryptococcosis", "Cryptococcus gattii", "Cryptococcus neoformans", "Genotype", "Humans", "Polymorphism, Restriction Fragment Length" ]

[]

[]

[]

[ "INTRODUCTION", "METHODS", "RESULTS", "DISCUSSION" ]

[ "In the last two decades, there has been a decline in the occurrence of opportunistic cryptococcosis in AIDS cases with a simultaneous increase in the incidence of cryptococcal disease in non-HIV-infected patients1\n,\n2. HIV seronegative patients infected with Cryptococcus spp. are a heterogeneous population that includes cases of therapeutic immunosuppression, comorbidities, solid organ transplantation, and immunocompetent individuals with no apparent comorbidity. Differences in the clinical characteristics and lethality of cryptococcal disease have been observed between these groups of patients and with cryptococcosis associated with AIDS3\n,\n4.\nCryptococcal disease is caused predominantly by species of C. neoformans (CN) and C. gattii (CG) complexes. The CN complex includes the species C. neoformans (genotype VNI/VNII/VNB), C. deneoformans (genotype VNIV), and a hybrid species (genotype VNIII). CG complex includes C. gattii strictu senso, C. deuterogattii, C. bacillisporus, C. tetragattii, and C. decagattii, respectively, genotypes VGI, VGII, VGIII, VGIV, and VGV/IIIc; hydrids between species of the two Cryptococcus complexes have been reported5. CN is more prevalent and has a wider geographical distribution, while CG is more isolated in tropical and subtropical regions, although species of this complex have also been isolated in temperate climate environments6. CG is often associated with infections in immunocompetent individuals, in addition to more frequent lung and brain parenchyma lesions7\n,\n8. Different geographical areas may show differences in the predominant genotype/species of Cryptococcus spp. and eventually in the clinical presentation of cryptococcal disease9. C. neoformans molecular type VNI is the major agent of cryptococcal disease in Brazil, followed by C. gattii, and the prevalence of this last species increases from the southern to northern region of the country6.\nThis study aimed to assess the characteristics of cryptococcosis in non-HIV-infected patients in southeastern Brazil. The clinical aspects of cryptococcal disease in immunocompetent individuals and the comparison of cases with isolation of CN and CG species complexes were analyzed. This study has clinical relevance because of the scarcity of studies on cryptococcosis comparing CN or CG complex infections in non-HIV-infected patients, including immunocompetent individuals, from Brazil.", "This retrospective study analyzed the clinical and epidemiological data of non-HIV-infected patients with cryptococcal disease. The patients received medical assistance between 2000 and 2016 at the University Hospital of the Ribeirão Preto Medical School, University of São Paulo (SP), and lived in the region of Ribeirão Preto, SP, Brazil. The data were analyzed according to the results of the Cryptococcus spp. genotyping, which were divided into two groups: 1) the CN group with 24 patients infected with species of the C. neoformans complex; and 2) the CG group with 12 patients infected with species of the C. gattii complex. Further analysis compared data from immunocompetent (apparently healthy) patients (10/36) with other patients with comorbidities and/or immunosuppressed patients (26/36). Three other patients were excluded due to a lack of clinical data, or because the isolation of Cryptococcus spp. was considered as only colonization. \nClinical and epidemiological data were collected from patients’ medical records, including age, sex, underlying diseases, and predisposing factors for cryptococcal disease. The involvement of organs and tissues by Cryptococcus spp. was assessed by clinical manifestations, radiographic images, the isolation site of this yeast, cerebrospinal fluid (CSF) analysis, and biopsy of the lung, skin, and lymph nodes. Antifungal treatment for meningitis and bloodstream infection was performed with deoxycholate amphotericin B (CN = 11/17; CG = 4/7) or liposomal amphotericin B (CN = 6/17; CG = 3/7) and was maintained until there was no fungal growth in the CSF, and these drugs were associated or not with fluconazole. The consolidation and maintenance phases of the antifungal therapy were performed using fluconazole. Patients with lung and skin lesions and without meningitis were treated orally with fluconazole or itraconazole. The outcome was determined one year after diagnosis, and cases were classified as Cure-Improvement or Death.\nCryptococcus spp. were isolated from the following clinical samples: CSF (n=23), blood (n=9), skin biopsy (n=4), bronchoalveolar lavage (n=2), and a sample of each of the following materials: sputum, lung biopsy, pleural fluid, lymph node biopsy, and urine. Sabouraud dextrose agar with or without chloramphenicol was used to isolate the fungus, and Bact Alert (Biomérieux Brasil) or BD (Becton Dickinson and Company, USA) flasks were used for blood culture. Identification of the genus Cryptococcus was carried out by conventional laboratory methods of clinical mycology and/or the automated Vitek (BioMérieux Brasil) system. Clinical isolates of Cryptococcus spp. were maintained in the laboratory using periodic subcultures.\nMolecular identification of CN/CG species complexes was carried out by polymerase chain reaction (PCR) using pairs of specific primers that amplify DNA fragments to 695 bp for C. neoformans (CNa-70a/CNa-70s) and 448 bp for C. gattii (CNb-49a/CNb-49s)10. The molecular types of the CN (VNI, VNII, VNIII, and VNIV) and CG complexes (VGI, VGII, VGIII, and VGIV) were assessed by restriction fragment length polymorphism (RFLP) of the URA5 gene. After amplification, the products were subjected to enzymatic restriction with the restriction endonucleases HhaI (Invitrogen, Thermo Fisher) and Cfr13I (Invitrogen,Thermo Fisher)11\n,\n12. PCR-RFLP patterns were assigned visually by comparing them to the standard strains C. neoformans, molecular type VNI (WM148); C. neoformans, molecular type VNII (WM626); C. neoformans × C. deneoformans hybrid, molecular type VNIII (WM628); C. deneoformans, molecular type VNIV (WM629); C. gattii, molecular type VGI (WM179); C. deuterogattii, molecular type VGII (WM178); C. bacillisporus, molecular type VGIII (WM175), and C. tetragattii, molecular type VGIV (WM779) from the Laboratory of Mycology (Pathogenic Fungi Collection) at the Oswaldo Cruz Foundation (FIOCRUZ)-INI/FIOCRUZ in Brazil.\nThe titer of the cryptococcal antigen in the CSF of 18 patients was determined by the latex agglutination method using the CALAS® Kit (Meridian Bioscience, USA). The titer of anti-Cryptococcus antibodies in the serum of 27 patients was measured by counterimmunoelectrophoresis using in-house prepared antigen, which were obtained by sonicating four samples of clinical isolates of C. neoformans (identified by cultivation in L-canavanine-glycine-bromothimol blue medium).\nStatistical analysis was performed using GraphPad Prism v.6 (GraphPad Software, La Jolla, CA, USA). The proportions were compared using the chi-square test or Fisher’s exact test. The Mann-Whitney U test was used to assess CSF parameters and the titer of the serum anti-Cryptococcus spp. antibody. The significance level was set at P < 0.05.\nThe research project was approved by the Research Ethics Committee of University Hospital, Ribeirão Preto Medical School (No. 12247/2010).", "Thirty-six cases of cryptococcosis in non-HIV-infected patients in southeastern Brazil were analyzed in this study. Among the patients, 72.3% (26/36) were individuals with comorbidities and/or immunosuppressed and 27.7% (10/36) were immunocompetent (apparently healthy) patients. Strains isolated from the CN group were 100% (24/24) of the VNI genotype (C. neoformans). The CG group consisted of 17% (2/12) of the VGI genotype (C. gattii sensu stricto) and 83% (10/12) of the VGII genotype (C. deuterogattii).\nAge, sex, and associated conditions of the patients showed no significant differences between the CG and CN groups. The proportion of immunocompetent patients was higher in the CG group (Table 1). Patients from the CG group reported more prolonged exposure in a rural environment and/or working with wood or raising birds (P=0.0002). Immunosuppression by corticosteroids or cytotoxic drugs was observed only in the CN group, while alcoholism and diabetes mellitus were observed in patients from both groups (Table 1).\n\nTABLE 1:Demographic data, associated conditions, and predisposing factors according to the C. neoformans/C. gattii species complex. \nC. gattii*\n\nC. neoformans\nTotal\nP value**\n n/%n/%n/%\nAge - median (range)46.5 (4 - 73)46.0 (2.5 - 80)46.5 (2.5 - 80)0.7355Gender (Male: Female)10:216:826:100.6667Associated Conditions 7/5819/7926/720.2474Diabetes mellitus4/334/178/220.3974Malignancya\n0/05/215/140.1494Chronic visceral diseasesb\n6/5011/4618/501.0Systemic erythematosus lupus0/02/82/60.5429Kidney transplantation0/03/133/80.5361Non-associated conditions5/425/2110/280.2474Predisposing factors \n\n\n\nEnvironmental expositionc\n10/834/1714/390.0002Pharmacologic immunosuppression0/06/256/170.0793 Alcoholism3/253/136/170.3781Malnutrition0/03/133/80.5361Patients - Total12/10024/10036/100\na) Hematologic malignancies or solid organ neoplasm; b) chronic diseases and/or dysfunction in one of the following organs: lungs, liver, kidney, heart, brain, intestine; c) living or working in rural areas and/or regular contact with birds or wood; * C. gattii = C. deuterogattii (n=10), C. gattii s.s. (n=2); **statistical analysis.\n\na) Hematologic malignancies or solid organ neoplasm; b) chronic diseases and/or dysfunction in one of the following organs: lungs, liver, kidney, heart, brain, intestine; c) living or working in rural areas and/or regular contact with birds or wood; * C. gattii = C. deuterogattii (n=10), C. gattii s.s. (n=2); **statistical analysis.\nMeningitis was the most common clinical manifestation in both of the groups. Cryptococcemia occurred only in the CN group (P=0.0163), while a trend towards a higher proportion of lung and skin lesions, in addition to a patient with generalized lymphadenopathy, was observed in the CG group (Table 2). Patients in the two groups presented no differences in cellularity, glucose, protein, and cryptococcal antigen levels in the CSF. Patients in the CG group showed higher serum reactivity and slightly higher antibody titers against Cryptococcus spp. antigens (Table 2).\n\nTABLE 2:Clinical and laboratory manifestations of the cryptococcal disease according to the C. neoformans/C. gattii species complex. \nC. gattii complex\n\nC. neoformans\nTotal\nP value*\n n/%n/%n/%\nClinical Manifestation \n\n\n\nMeningitis7/5815/6322/611.0Brain granuloma or pseudocyst3/251/44/110.0980Cryptococcemia0/09/389/250.0163Pulmonary lesion8/679/3817/470.1582Cutaneous lesion3/253/136/170.3781Lymphadenopathy1/80/01/30.3333\n\n\n\n\nCSF alterations(n=7)(n=15)(n=22)\nCells - no./µL - mean ± SD100.42 ± 11.99115.73 ± 236.80110.8 ± 80.050.6867Protein - mg/dL - mean ± SD161.28 ± 236.80144.42 ± 182.22150± 196.10.8582Glucose - mg/dL - mean ± SD45.85± 45.3830.28 ± 24.0535.4 ± 32,40.3115Cryptococcal antigen titera - median ange)≥ 40963072≥40960.4714\n(64 - ≥ 4096)(NR - ≥ 4096)(NR - ≥ 4096)\nAntibodies anti-Cryptococcus in serumb\n\n\n\n\nReactive patients / total 7;11/644;16/2511;27/410.0608Antibodies titer - median (range)8 (1 - 16)2.5 (1 - 8)4 (1 - 16)0.0264Patients - Total12/10024/10036/100\n a) Latex agglutination test: titer-inverse of CSF dilution; b) counterimmunoelectrophoresis test: titer-inverse of serum dilution; *statistical analysis; SD: standard deviation; CSF: cerebrospinal fluid; NR: non-reactive.\n\n a) Latex agglutination test: titer-inverse of CSF dilution; b) counterimmunoelectrophoresis test: titer-inverse of serum dilution; *statistical analysis; SD: standard deviation; CSF: cerebrospinal fluid; NR: non-reactive.\nPatients in the CN and CG groups received similar antifungal treatments, except for the use of amphotericin B monotherapy in six patients in the CN group. Lethality was higher in the CN group (48%) than in the CG group (18%), without reaching statistical significance. Cure/improvement in the CG group was verified in 2/2 patients with C. gattii s.s. infection and in 5/7 cases of C. deuterogattii infection. The period between the diagnosis of cryptococcosis and death was significantly shorter in the CN group (Table 3). Sequelae occurred in three patients: in the CG group, a child had amaurosis and delay in neuromotor development, and another patient had hypoacusis. In the CN group, amaurosis occurred in one patient.\n\nTABLE 3:Antifungal drug treatment and outcome of patients with cryptococcosis due to the C. neoformans/C. gattii species complex. C. gattii complex\nC. neoformans\nTotalp value* n/%n/%n/%\nAntifungal treatment \n\n\n\nAmphotericin B → Fluconazole4/33a\n5/219/250.4428Amphotericin B + Fluconazole3/256/259/251.0Amphotericin B 0/06/256/170.0793Fluconazole4/333/137/190.1904Itraconazole0/01/41/31.0No antifungal1/83/134/111.0Outcome \n\n\n\nCure/improvementb\n9/8212/5221/620.1398Deathb\n2/1811/4813/38\nDiagnosis - death time (days)\n\n\n\nmedian (range)53 (31-75)7 (2-79)12 (2-79)0.0803Unknown112\nSequels 2/171/43/80.2527Patients - Total 12/10024/10036/100\n a) One patient used ketoconazole instead of fluconazole; b) survival and lethality rates excluded patients whose outcome is unknown; * statistical analysis.\n\n a) One patient used ketoconazole instead of fluconazole; b) survival and lethality rates excluded patients whose outcome is unknown; * statistical analysis.\nTable 4 compares cryptococcal disease in patients with the presence or absence of comorbidities or organ transplants. Immunocompetent patients (absence of comorbidity or organ transplantation) had cryptococcosis caused by both CG and CN species complexes and a higher frequency of meningitis (90%) (P=0.0245). Lethality was higher in patients with comorbidities (46%) than in immunocompetent patients (20%), although the difference was not statistically significant (Table 4).\n\nTABLE 4:Cryptococcal disease and outcome of patients with or without comorbidities and organ transplantation, regardless of the Cryptococcus species. Comorbidity or organ transplantation \nP-value*\n Absent n/%Present n/%\nCryptococcus species complex   \nC. gattii complex5/507/270.2474\nC. neoformans\n5/5019/73\nCryptococcosis site \n\n\nMeningitis9/9012/460.0245Cryptococcemia1/108/310.3921Pulmonary2/2014/540.1326Cutaneous1/105/190.6546Lymphadenopathy0/01/41.0Outcomea\n\n\n\nCure/improvement8/8013/540.2508Death2/2011/46\nUnknown02-Patients - Total10/10026/100\nCure/improvement and death rates excluded patients whose outcome was unknown; *statistical analysis.\n\nCure/improvement and death rates excluded patients whose outcome was unknown; *statistical analysis.", "The cryptococcal disease of non-HIV-infected patients evaluated in this study was mainly related to the genotypes VNI of CN (C. neoformans) and VGII of CG (C. deuterogattii), a finding similar to that observed in clinical isolates in Brazil6\n,\n13. The CG species complex has been associated with immunocompetent individuals14, but our study revealed that C. neoformans also infects such people, although they are more prevalent in comorbid or organ-transplanted patients. Immunocompetent individuals who are apparently healthy may have small defects in their immune capacity, facilitating cryptococcal infection15. In the studied cases, 72% of the patients had previously altered health conditions, and both patients infected with the CN and CG species complex had similar rates of chronic non-infectious diseases. The immunocompetent cases (42%) and comorbidity rates in patients infected by the CG species complex were similar to those found in a large series of cases in Canada16. Only patients infected by C. neoformans had neoplasms and immunosuppression by corticosteroids or cytotoxic drugs, which suggests some specificity in the pathogenesis of the disease caused by different species of Cryptococcus. However, the disease caused by the CG species complex has already been associated with severe immunosuppression17. Diabetes mellitus and alcoholism are likely predisposing factors for the disease caused by both the CN and CG species complexes18\n,\n19. Exposure to environmental sources that contain Cryptococcus spp. was associated with C. gattii infection, similar to that observed in Australia7, in which many patients lived in rural areas. The CG group included a child with cryptococcal meningitis caused by C. deuterogattii (VGII genotype) after returning from a trip to a CG species complex endemic area in northeastern Brazil. C. gattii s.s. was isolated from a patient who caught wild birds and presented with Moyamoya disease and generalized cryptococcal lymphadenopathy. C. gattii s.s. was also isolated from an immunocompetent man who presented with a chronic cutaneous ulcer and had past contact with house birds.\nMeningitis and disseminated disease are the most common clinical manifestations in non-HIV-infected patients, but CG species can lead to a predominance of lung involvement or cause primary skin lesions9\n,\n20. The frequency of meningeal, lung, and skin lesions showed no difference between the CN and CG group. A significant difference was found in bloodstream infection, which occurred only in patients infected with C. neoformans. This is probably a consequence of more severe immunosuppression in patients in the CN group, facilitating fungal dissemination. The frequency of patients with cryptococcal fungemia was higher in this investigation (25%) than in other Brazilian report of cryptococcosis in non-HIV-infected and non-transplanted patients (6.8%)21. This lower percentage is probably due to the high proportion of immunocompetent patients and CG complex infections in the later study.\nThe higher level of serum anti-Cryptococcus antibodies among patients infected by CG could be related to the lower frequency of immunosuppression in the patients in this group. A previous study found a higher humoral response of IgG and IgA antibodies in patients infected with C. gattii than in those infected with C. neoformans\n22.\nThe outcome of antifungal treatment with amphotericin B and/or azole drugs showed a trend towards lower lethality among patients infected by CG species. The overall lethality in patients in this study (38%) was higher than the 21% lethality found in another series of cryptococcosis cases in non-HIV-infected patients in Brazil, although that study had a higher percentage of immunocompetent individuals21. The death of patients at the beginning of antifungal treatment has been observed in other hospitals and has been associated with cryptococcemia and high lactate levels in CSF23\n,\n24. Patients with cryptococcal disease due to CG species showed a lethality after 12 months which reached 18% in this study, 23.3% in a British Colombia-Canada study, and was more elevated in some series of cases that included immunosuppressed individuals and children16\n,\n25\n,\n26. \nThe comparison of clinical manifestations possibly attributable to the type of causative agent, CN or CG complexes, may have been impaired by the higher proportion of immunosuppressed patients in the CN group. Thus, the clinical picture and outcome were compared between immunocompetent patients and those with previous diseases and/or immunosuppression, regardless of the Cryptococcus species. Cryptococcal meningitis was most commonly seen among immunocompetent patients, with a trend for cryptococcemia and pulmonary involvement manifested mainly in cases with comorbidities and/or immunosuppression. The high proportion of clinically expressed meningitis may be a consequence of the high immunological reactivity of immunocompetent patients. Tissue injury and damage in cryptococcosis are more likely to occur when the immune response of the host is very weak or very intense27. Immunocompetent patients also had lower lethality, although the difference was not statistically significant. HIV-infected patients with controlled or active disease (AIDS) showed similar lethality among those infected with C. gattii or non-C. gattii species28. Such data suggest that not only the Cryptococcus species but also the health condition and immunological capacity of the host are important in defining the clinical presentation and outcome of patients14.\nThis study was limited by the small number of cases, making it difficult to differentiate patient groups based on the analyzed parameters. Other cases of cryptococcosis in non-HIV-infected patients were recognized at the institution during the same period, but without the availability of isolated microorganisms for genotyping.\nIn conclusion, cryptococcal disease caused by C. neoformans (VNI genotype) was associated with immunodepressed patients and fungemia, and patients infected with C. deuterogattii and C. gattii s.s. (genotypes VGII and VGI of CG) were exposed to environmental sources of Cryptococcus spp. and showed a higher humoral immune response. Chronic non-infectious diseases, diabetes mellitus, and alcoholism were likely predisposing factors for infection by both CN and CG species. Immunocompetent patients (without comorbidity or solid organ transplantation) showed a high incidence of cryptococcal meningitis, a trend toward less fungal dissemination, and longer patient survival, regardless of the infecting species. The clinical expression and outcome of cryptococcal disease in non-HIV-infected patients are probably more related to the health and immunological conditions of the host than to the Cryptococcus species complexes." ]

[ "intro", "methods", "results", "discussion" ]

[ "Cryptococcal disease", "Cryptococcal meningitis", "Cryptococcus neoformans complex", "Cryptococcus gattii complex" ]

[]

[ "patients", "cg", "species", "cn", "gattii", "infected", "neoformans", "cryptococcal", "group", "cryptococcus" ]

[ "cryptococcal disease9 neoformans", "patients cryptococcosis neoformans", "associated aids3 cryptococcal", "cryptococcosis non hiv", "aids3 cryptococcal disease" ]

[ "Adenocarcinoma", "Adenocarcinoma of Lung", "Antibodies", "Antibody Specificity", "Buffers", "China", "DNA Mutational Analysis", "ErbB Receptors", "Exons", "Humans", "Hydrogen-Ion Concentration", "Immunohistochemistry", "Lung Neoplasms", "Mutation", "Predictive Value of Tests", "Prospective Studies", "Reproducibility of Results", "Retrospective Studies", "Specimen Handling" ]

[ "Background", "Patient samples", "Immunohistochemistry", "IHC scoring", "DNA sequencing", "DNA preparation", "Mutant-enriched PCR for EGFR exon 19 and 21", "DNA sequencing for EGFR exon 19 and 21", "Statistical analysis", "Ethical approval", "DNA sequencing", "Evaluation of antigen retrieval buffer", "IHC results", "Concordance analysis of IHC and DNA sequencing", "Consent", "Abbreviations", "Competing interests", "Authors’ contributions", "Authors’ information" ]

[ "Somatic mutations within the tyrosine kinase (TK) domain of the epidermal growth factor receptor (EGFR) gene are found in approximately 30% of lung adenocarcinomas in Asian populations\n[1]. Studies support that some of these activating mutations are not only reliable predictors of response to the small molecule EGFR tyrosine kinase inhibitors (TKIs) gefitinib and erlotinib but also prognostic factors for survival\n[2-4]. Among numerous TK domain mutations, 85–90% are exon 19 E746_A750 deletions or exon 21 L858R point mutations\n[5]. A variety of DNA-based molecular methods are used to detect EGFR mutations. These methods have respective advantages and disadvantages, with no consensus on which one is the best. For example, direct sequencing of PCR-amplified genomic DNA can detect all mutations in the regions analyzed, but has limited analytical sensitivity when the tumor cells are not a large fraction of the specimen. The amplification refractory mutation system (ARMS) assay is more sensitive, but detects fewer mutations, usually only one per reaction. In general, direct analysis of DNA is expensive because of the cost of the equipment and reagents. In addition it is technically complex, and usually done in laboratories that specialize in molecular pathology\n[6].\nYu et al.\n[7] developed mutation specific rabbit monoclonal antibodies against the two most common EGFR mutations and showed that these antibodies can be applied to the immunohistochemical (IHC) detection of these mutations in formalin-fixed paraffin-embedded (FFPE) tissue. Several independent groups have investigated the sensitivity and specificity of these antibodies in the detection of EGFR mutations in non-small cell lung cancer (NSCLC). Most of them confirmed a high degree of specificity, but the reported sensitivities were quite variable ranging from 24% to 100% (Table\n1)\n[8-14]. This inconsistency may be related to differences in methodology and interpretation\n[10,13,15], as well as population specific differences in gene mutations and differences in the level of protein expression\n[16]. This inconsistency suggests that further study is needed in diverse populations before EGFR mutation-specific IHC can be implemented as a clinical tool.\nLiterature review of sensitivity and specificity of mutation-specific immunohistochemistry\nIn the study reported here we optimized the methodology and interpretive aspects of IHC for detection of EGFR mutations, and evaluated the success of this effort by comparison with DNA sequencing. This study investigated the staining protocol, staining pattern, scoring methods, and cut off value to determine the diagnostic power of EGFR mutation-specific IHC in Chinese lung adenocarcinoma patients.", "Samples for study were selected according to the following criteria: lung adenocarcinoma, surgically resected, primary, solitary and no preoperative therapy. A total of 50 cases were collected retrospectively and prospectively from the Department of Pathology, Peking University First Hospital during January 2010 to January 2012.\nAll specimens were dissected and immersed in 10% neutral buffered formalin, then fixed overnight. The number of sections for histology depended on the greatest dimension of tumors, i.e. one section per centimeter. If a tumor was less than 2 cm in greatest dimension, the tumor was totally sampled for microscopic examination. Sectioned tissues were embedded in paraffin routinely.\nInformed consent for the use of these specimens for medical studies was obtained before surgery.", "50 tissue blocks were cut into 4-μm-thick whole sections. EGFR mutation specific antibodies were Rabbit XP® mAbs obtained from Cell Signaling Technology (Danvers, MA), 6B6 specific for the E746-A750del mutation, and 43B2 for the L858R mutation. The antibodies were diluted 1:100 with antigen retrieval buffer before use. The antigen retrieval buffers tested were sodium citrate (pH 6.0), EDTA (pH 8.0) and EDTA (pH 9.0). Cytokeratin AE1/AE3 IHC was used as a quality control for tissue and protocol. The IHC protocol is described in greater detail in the Additional file\n1.", "Three sets of criteria were used for interpretation of the IHC results, referred to as Score A, B and C, respectively, in this study. A positive result using score A was moderate to strong staining of membrane and/or cytoplasm in >10% tumor cells\n[15]. A positive result using score B was membrane staining in >10% tumor cells with any intensity\n[10]. A positive result using score C was membrane and/or cytoplasmic staining in >50% of the tumor cells with any intensity\n[13]. In this study all 50 specimens were analyzed using Score A, B and C separately, so as to evaluate the validity of these scoring methods by comparing to the results of DNA sequencing. Both the intensity and percentage of stained cells were assessed at low magnification (objective magnification ×10). The distribution of staining, membrane or cytoplasm, was assessed at high magnification (objective magnification ×40). Four experienced pathologists (Yan Xiong, Ying Dong, Lin Nong and Jing Zhao) reviewed all of the slides independently, and then replicated the analysis 16 to 18 weeks later. The intra- and inter-observer reliability was analyzed.", " DNA preparation H&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.\nH&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.\n Mutant-enriched PCR for EGFR exon 19 and 21 All samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.\nAll samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.\n DNA sequencing for EGFR exon 19 and 21 The amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.\nThe amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.", "H&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.", "All samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.", "The amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.", "Statistical analysis was done using the statistics software SPSS V16.0 (SPSS Inc., Chicago, IL). Fleiss’ Kappa was used to determine inter-observer agreement. Cohen’s Kappa was used to determine intra -observer agreement and agreement of IHC and DNA sequencing. A Kappa value between 0.81 and 1.0 was defined as nearly perfect agreement, between 0.61 and 0.8 as substantial agreement, between 0.41 and 0.60 as moderate agreement, between 0.21 and 0.40 as fair agreement, between 0.00 and 0.20 as slight agreement. For each Kappa, the 95% confidence interval (CI) was calculated. Difference was considered significant (P < 0.05), if the lower and upper boundary of the 95% CI showed no overlap.", "All experiments above have been performed with the approval of Peking University First Hospital Ethics Committee.", "In the total cohort of 50 samples L858R was identified in 16 cases, a deletion in exon 19 in 17 cases, and neither of them in 17 cases. Of the 17 cases with exon 19 deletion, 14 had a p.E746_A750del (c.del2235_2249 on the DNA level), one had a p.L747_T751del (c.2240_2254del), one had a p.L747_P753delinsS (c.2240_2257del), and one had a p.L747_T751delinsPT (c.2239_2253delinsCCAACG) that had not been previously reported. In our study, of all 33 cases with EGFR mutations, L858R and E746_A750del together comprised 90% (30/33) and the others, including L747_T751del, L747_P753delinsS and L747_T751delinsPT, comprised 10%, which was concordant with other studies\n[17]. From this point of view L858R and E746_A750del are recognized as the most common mutations and the other mutation types are described as uncommon mutations.", "We evaluated three different antigen retrieval buffers on all 50 specimens to optimize the IHC results: sodium citrate (pH 6.0), EDTA (pH 8.0) and EDTA (pH 9.0). Slides in the EDTA (pH 8.0) group showed the best histological pictures with strongly specific staining and minimal background. The intensity of the positive cells in the sodium citrate (pH 6.0) group was too faint to distinguish from the background. Mesenchymal cells on slides exposed to EDTA (pH 9.0) were stained as strong as tumor cells, which made it impossible to identify the specificity of staining (Figure\n1). Inter-observer agreement was nearly perfect in the EDTA (pH 8.0) group, substantial in the sodium citrate (pH 6.0) group and moderate in the EDTA (pH 9.0) group. The Fleiss’ Kappa (95% confidence interval) was 0.912 (0.862, 0.962), 0.753 (0.677, 0.829), and 0.643(0.558, 0.728) in the three groups, respectively. The difference between EDTA (pH 8.0) and the others was significant (P < 0.05). Intra-observer agreement was highest in EDTA (pH 8.0), moderate in sodium citrate (pH 6.0), and lowest in EDTA (pH 9.0). The Cohen’s Kappa (95% confidence interval) was 0.955 (0.918, 0.992), 0.853 (0.790, 0.916), and 0.801 (0.730, 0.872), respectively. The difference between EDTA (pH 8.0) and the others was statistically significant (P < 0.05) (Table\n3).\nSample case of predominant solid adenocarcinoma immunostained with E746_A750del-specific antibody using different antigen retrieval buffers (original magnification x400). A Sodium citrate (pH 6.0). B EDTA (pH 8.0). C EDTA (pH 9.0).\nIntra- and inter-observer agreement based on slides treated with different antigen retrieval buffers", "The staining distribution included cytoplasm only or cytoplasm together with membrane. Normal tissue adjacent to adenocarcinoma was negative (Figure\n2). In the cases with lepidic pattern staining of lepidic tumor cells was either negative or fainter than the tumor cells of other patterns (Figure\n3). In our study, 80% of the cases were either negative or positive in 100% of the tumor cells, although the intensity was diverse ranging from + to +++. In only 20% of cases the tumor cells were stained in some areas and completely negative in other areas. Overall, the staining pattern showed characteristics of homogeneity more than heterogeneity.\nPredominant acinar adenocarcinoma with adjacent normal alveoli. A H & E (original magnification x100). B The cytoplasm and membrane of the tumor cells were stained strongly with L858-specific antibody. In contrast, the adjacent normal alveolar epithelial cells were completely negative (original magnification x100).\nAdenocarcinoma with acinar and lepidic patterns. A H & E (original magnification x100). B The cytoplasm and membrane of acinar component stained strongly, but lepidic tumor cells stained weakly with L858-specfic antibody (original magnification x100).\nBased on different scoring systems, the percentage of positive cases was different too. For L858R-specific IHC it was 28% (14/50) on Score A, 16% (8/50) on Score B, and 40% (20/50) on Score C; for E746_A750del-specific IHC it was 20% (10/50) on Score A, 14% (7/50) on Score B, and 24% (12/50) on Score C.", "Of the 16 cases with L858R, the L858R-specific IHC was positive in 13 on Score A, 7 on Score B and 11 on Score C (Figure\n4). Of 34 cases without L858R the L858R-specific IHC was negative in 33 on Score A, 33 on Score B and 25 on Score C (Figure\n5). L858R-specific IHC showed a sensitivity of 81%, a specificity of 97%, a positive predictive value (PPV) of 93%, and a negative predictive value (NPV) of 92% on Score A; a sensitivity of 44%, a specificity of 97%, a PPV of 88%, and a NPV of 79% on Score B; and a sensitivity of 69%, a specificity of 74%, a PPV of 55%, and a NPV of 83% on Score C. Reliability analysis for L858R-specific IHC and DNA sequencing was found to be Cohen’s Kappa = 0.810 and 95% CI (0.701, 0.919) on Score A, Cohen’s Kappa = 0.470 and 95% CI (0.332, 0.608) on Score B, and Cohen’s Kappa = 0.397 and 95% CI (0.261, 0.533) on Score C. The agreement between L858R-specific IHC and DNA sequencing was the best using Score A compared to Score B and C. The difference was significant (P < 0.05) (Table\n4).\nA case of predominant solid adenocarcinoma with the L858R mutation. A DNA sequencing of EGFR showing normal (upper panel) and L858R mutant (lower panel). The position of the mutation is boxed. The mutant sequence appears to be homozygous with complete absence of normal sequence. This is because of the use of a “clamp” strategy to suppress amplification of the normal sequence, as described in the methods section. B Immunohistochemical staining with L858R-specific antibody showing strong positivity (original magnification x200). C Immunohistochemical staining with E746-A750 del-specific antibody showing complete negativity (original magnification x200).\nA case of predominant acinar adenocarcinoma with normal EGFR. A and B DNA sequencing shown normal EGFR exon 21 and 19, respectively, in the region that is frequently subject to mutation. C and D Tumor cells were not stained with either L858R-specfic or E746_A750del-specific antibodies, respectively (original magnification x200).\nDiagnostic power of L858R-specific IHC on score A, B and C\nIHC indicates immunohistochemistry; PPV, positive predictive value; NPV, negative predictive value.\nOf 14 cases with the E746_A750del by DNA sequencing E746_A750del-specific IHC was positive in 10 on Score A, 7 on Score B, and 9 on Score C (Figure\n6). All of the 3 cases with uncommon types of exon 19 deletion, includingL747_T751del, L747_P753delinsS and L747_T751delinsPT, were negative by E746_A750del-specific IHC regardless of the scoring method.\nA case of predominant acinar adenocarcinoma with the exon 19 deletion mutation (E746-A750 del). A DNA sequencing of EGFR showing normal (upper panel) and the E746-A750 del mutant (lower panel). The position of the mutation is indicated. The mutant sequence appears to be homozygous with complete absence of normal sequence. This is because of the use of a “clamp” strategy to suppress amplification of the normal sequence, as described in the methods section. B Immunohistochemical staining with E746-A750 del-specific antibody showing strong positivity (original magnification x200). C Immunohistochemical staining with L858R-specific antibody showing complete negativity (original magnification x200).\nIncluding all 17 specimens with an exon 19 deletion detected by DNA sequencing, the E746_A750del-specific IHC found 10 (59%) were positive and 7 (41%) negative on Score A, 7 (41%) were positive and 10 (59%) negative on Score B, 9 (53%) were positive and 8 (47%) negative on Score C. Of the 33 cases without an exon 19 deletion detected by DNA sequencing all were negative by E746_A750del-specific IHC both on Score A and B (Figure\n5), while 3 were positive on Score C. As a method to detect deletions in exon 19 despite the exact structure of the deletion, E746_A750del-specific IHC showed a sensitivity of 59%, a specificity of 100%, a PPV of 100%, and a NPV of 82% on Score A; a sensitivity of 41%, a specificity of 100%, a PPV of 100%, and a NPV of 77% on Score B; and a sensitivity of 53%, a specificity of 91%, a PPV of 75%, and a NPV of 79% on Score C. Reliability analysis for E746_A750del-specifiac IHC and DNA sequencing was found to be Cohen’s Kappa = 0.653 and 95% CI (0.521, 0.785) on Score A, Cohen’s Kappa = 0.480 and 95% CI (0.342, 0.618) on Score B, Cohen’s Kappa = 0.472 and 95% CI (0.334, 0.610) on Score C. Similar to L858R-specific IHC the agreement between E746_A750del-specific IHC and DNA sequencing was the best using Score A compared to Score B and C, but the difference was not significant (P > 0.05) (Table\n5).\nDiagnostic power of E746_A750del-specific IHC on score A, B and C\nIHC indicates immunohistochemistry; PPV, positive predictive value; NPV, negative predictive value.", "Written informed consent was obtained from the patient for publication of this report and any accompanying images.", "ARMS: Amplification refractory mutation system; CI: Confidence interval; CLMP1: Clamps for exons 19; CLMP2: Clamps for exons 21; EGFR: Epidermal growth factor; FFPE: Formalin-fixed paraffin- embedded; IHC: Immunohistochemistry; LNA: Locked nucleic acid; NSCL: Non-small cell lung cancer; NPV: Negative predictive value; PPV: Positive predictive value; TK: Tyrosine kinase; TKI: Tyrosine kinase inhibitor.", "The authors declare that we do not have any financial competing interests.", "XY participated in the design of the study and drafted the manuscript. BY carried out the immunoassays and collected patient’s clinic data. LN and LTS carried out the molecular genetic studies. RPG carried out the molecular genetic studies, and helped to draft the manuscript. XH contributed to the design of the study, helped to draft the manuscript, and participated in coordination. NL, ZJ and DY participated in the review of the histologic slides. NL carried out the molecular genetic studies as well. LT conceived of the study, participated in its design and coordination, and helped to draft the manuscript. All authors read and approved the final manuscript.", "Yan Xiong, MD, Associated Professor in Department of Pathology, Peking University First Hospital, Beijing, China. Ting Li, MD, Full professor and Chair in Department of Pathology, Peking University First Hospital, Beijing, China." ]

[ null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null ]

[ "Background", "Methods", "Patient samples", "Immunohistochemistry", "IHC scoring", "DNA sequencing", "DNA preparation", "Mutant-enriched PCR for EGFR exon 19 and 21", "DNA sequencing for EGFR exon 19 and 21", "Statistical analysis", "Ethical approval", "Results", "DNA sequencing", "Evaluation of antigen retrieval buffer", "IHC results", "Concordance analysis of IHC and DNA sequencing", "Discussion", "Conclusions", "Consent", "Abbreviations", "Competing interests", "Authors’ contributions", "Authors’ information", "Supplementary Material" ]

[ "Somatic mutations within the tyrosine kinase (TK) domain of the epidermal growth factor receptor (EGFR) gene are found in approximately 30% of lung adenocarcinomas in Asian populations\n[1]. Studies support that some of these activating mutations are not only reliable predictors of response to the small molecule EGFR tyrosine kinase inhibitors (TKIs) gefitinib and erlotinib but also prognostic factors for survival\n[2-4]. Among numerous TK domain mutations, 85–90% are exon 19 E746_A750 deletions or exon 21 L858R point mutations\n[5]. A variety of DNA-based molecular methods are used to detect EGFR mutations. These methods have respective advantages and disadvantages, with no consensus on which one is the best. For example, direct sequencing of PCR-amplified genomic DNA can detect all mutations in the regions analyzed, but has limited analytical sensitivity when the tumor cells are not a large fraction of the specimen. The amplification refractory mutation system (ARMS) assay is more sensitive, but detects fewer mutations, usually only one per reaction. In general, direct analysis of DNA is expensive because of the cost of the equipment and reagents. In addition it is technically complex, and usually done in laboratories that specialize in molecular pathology\n[6].\nYu et al.\n[7] developed mutation specific rabbit monoclonal antibodies against the two most common EGFR mutations and showed that these antibodies can be applied to the immunohistochemical (IHC) detection of these mutations in formalin-fixed paraffin-embedded (FFPE) tissue. Several independent groups have investigated the sensitivity and specificity of these antibodies in the detection of EGFR mutations in non-small cell lung cancer (NSCLC). Most of them confirmed a high degree of specificity, but the reported sensitivities were quite variable ranging from 24% to 100% (Table\n1)\n[8-14]. This inconsistency may be related to differences in methodology and interpretation\n[10,13,15], as well as population specific differences in gene mutations and differences in the level of protein expression\n[16]. This inconsistency suggests that further study is needed in diverse populations before EGFR mutation-specific IHC can be implemented as a clinical tool.\nLiterature review of sensitivity and specificity of mutation-specific immunohistochemistry\nIn the study reported here we optimized the methodology and interpretive aspects of IHC for detection of EGFR mutations, and evaluated the success of this effort by comparison with DNA sequencing. This study investigated the staining protocol, staining pattern, scoring methods, and cut off value to determine the diagnostic power of EGFR mutation-specific IHC in Chinese lung adenocarcinoma patients.", " Patient samples Samples for study were selected according to the following criteria: lung adenocarcinoma, surgically resected, primary, solitary and no preoperative therapy. A total of 50 cases were collected retrospectively and prospectively from the Department of Pathology, Peking University First Hospital during January 2010 to January 2012.\nAll specimens were dissected and immersed in 10% neutral buffered formalin, then fixed overnight. The number of sections for histology depended on the greatest dimension of tumors, i.e. one section per centimeter. If a tumor was less than 2 cm in greatest dimension, the tumor was totally sampled for microscopic examination. Sectioned tissues were embedded in paraffin routinely.\nInformed consent for the use of these specimens for medical studies was obtained before surgery.\nSamples for study were selected according to the following criteria: lung adenocarcinoma, surgically resected, primary, solitary and no preoperative therapy. A total of 50 cases were collected retrospectively and prospectively from the Department of Pathology, Peking University First Hospital during January 2010 to January 2012.\nAll specimens were dissected and immersed in 10% neutral buffered formalin, then fixed overnight. The number of sections for histology depended on the greatest dimension of tumors, i.e. one section per centimeter. If a tumor was less than 2 cm in greatest dimension, the tumor was totally sampled for microscopic examination. Sectioned tissues were embedded in paraffin routinely.\nInformed consent for the use of these specimens for medical studies was obtained before surgery.\n Immunohistochemistry 50 tissue blocks were cut into 4-μm-thick whole sections. EGFR mutation specific antibodies were Rabbit XP® mAbs obtained from Cell Signaling Technology (Danvers, MA), 6B6 specific for the E746-A750del mutation, and 43B2 for the L858R mutation. The antibodies were diluted 1:100 with antigen retrieval buffer before use. The antigen retrieval buffers tested were sodium citrate (pH 6.0), EDTA (pH 8.0) and EDTA (pH 9.0). Cytokeratin AE1/AE3 IHC was used as a quality control for tissue and protocol. The IHC protocol is described in greater detail in the Additional file\n1.\n50 tissue blocks were cut into 4-μm-thick whole sections. EGFR mutation specific antibodies were Rabbit XP® mAbs obtained from Cell Signaling Technology (Danvers, MA), 6B6 specific for the E746-A750del mutation, and 43B2 for the L858R mutation. The antibodies were diluted 1:100 with antigen retrieval buffer before use. The antigen retrieval buffers tested were sodium citrate (pH 6.0), EDTA (pH 8.0) and EDTA (pH 9.0). Cytokeratin AE1/AE3 IHC was used as a quality control for tissue and protocol. The IHC protocol is described in greater detail in the Additional file\n1.\n IHC scoring Three sets of criteria were used for interpretation of the IHC results, referred to as Score A, B and C, respectively, in this study. A positive result using score A was moderate to strong staining of membrane and/or cytoplasm in >10% tumor cells\n[15]. A positive result using score B was membrane staining in >10% tumor cells with any intensity\n[10]. A positive result using score C was membrane and/or cytoplasmic staining in >50% of the tumor cells with any intensity\n[13]. In this study all 50 specimens were analyzed using Score A, B and C separately, so as to evaluate the validity of these scoring methods by comparing to the results of DNA sequencing. Both the intensity and percentage of stained cells were assessed at low magnification (objective magnification ×10). The distribution of staining, membrane or cytoplasm, was assessed at high magnification (objective magnification ×40). Four experienced pathologists (Yan Xiong, Ying Dong, Lin Nong and Jing Zhao) reviewed all of the slides independently, and then replicated the analysis 16 to 18 weeks later. The intra- and inter-observer reliability was analyzed.\nThree sets of criteria were used for interpretation of the IHC results, referred to as Score A, B and C, respectively, in this study. A positive result using score A was moderate to strong staining of membrane and/or cytoplasm in >10% tumor cells\n[15]. A positive result using score B was membrane staining in >10% tumor cells with any intensity\n[10]. A positive result using score C was membrane and/or cytoplasmic staining in >50% of the tumor cells with any intensity\n[13]. In this study all 50 specimens were analyzed using Score A, B and C separately, so as to evaluate the validity of these scoring methods by comparing to the results of DNA sequencing. Both the intensity and percentage of stained cells were assessed at low magnification (objective magnification ×10). The distribution of staining, membrane or cytoplasm, was assessed at high magnification (objective magnification ×40). Four experienced pathologists (Yan Xiong, Ying Dong, Lin Nong and Jing Zhao) reviewed all of the slides independently, and then replicated the analysis 16 to 18 weeks later. The intra- and inter-observer reliability was analyzed.\n DNA sequencing DNA preparation H&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.\nH&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.\n Mutant-enriched PCR for EGFR exon 19 and 21 All samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.\nAll samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.\n DNA sequencing for EGFR exon 19 and 21 The amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.\nThe amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.\n DNA preparation H&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.\nH&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.\n Mutant-enriched PCR for EGFR exon 19 and 21 All samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.\nAll samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.\n DNA sequencing for EGFR exon 19 and 21 The amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.\nThe amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.\n Statistical analysis Statistical analysis was done using the statistics software SPSS V16.0 (SPSS Inc., Chicago, IL). Fleiss’ Kappa was used to determine inter-observer agreement. Cohen’s Kappa was used to determine intra -observer agreement and agreement of IHC and DNA sequencing. A Kappa value between 0.81 and 1.0 was defined as nearly perfect agreement, between 0.61 and 0.8 as substantial agreement, between 0.41 and 0.60 as moderate agreement, between 0.21 and 0.40 as fair agreement, between 0.00 and 0.20 as slight agreement. For each Kappa, the 95% confidence interval (CI) was calculated. Difference was considered significant (P < 0.05), if the lower and upper boundary of the 95% CI showed no overlap.\nStatistical analysis was done using the statistics software SPSS V16.0 (SPSS Inc., Chicago, IL). Fleiss’ Kappa was used to determine inter-observer agreement. Cohen’s Kappa was used to determine intra -observer agreement and agreement of IHC and DNA sequencing. A Kappa value between 0.81 and 1.0 was defined as nearly perfect agreement, between 0.61 and 0.8 as substantial agreement, between 0.41 and 0.60 as moderate agreement, between 0.21 and 0.40 as fair agreement, between 0.00 and 0.20 as slight agreement. For each Kappa, the 95% confidence interval (CI) was calculated. Difference was considered significant (P < 0.05), if the lower and upper boundary of the 95% CI showed no overlap.\n Ethical approval All experiments above have been performed with the approval of Peking University First Hospital Ethics Committee.\nAll experiments above have been performed with the approval of Peking University First Hospital Ethics Committee.", "Samples for study were selected according to the following criteria: lung adenocarcinoma, surgically resected, primary, solitary and no preoperative therapy. A total of 50 cases were collected retrospectively and prospectively from the Department of Pathology, Peking University First Hospital during January 2010 to January 2012.\nAll specimens were dissected and immersed in 10% neutral buffered formalin, then fixed overnight. The number of sections for histology depended on the greatest dimension of tumors, i.e. one section per centimeter. If a tumor was less than 2 cm in greatest dimension, the tumor was totally sampled for microscopic examination. Sectioned tissues were embedded in paraffin routinely.\nInformed consent for the use of these specimens for medical studies was obtained before surgery.", "50 tissue blocks were cut into 4-μm-thick whole sections. EGFR mutation specific antibodies were Rabbit XP® mAbs obtained from Cell Signaling Technology (Danvers, MA), 6B6 specific for the E746-A750del mutation, and 43B2 for the L858R mutation. The antibodies were diluted 1:100 with antigen retrieval buffer before use. The antigen retrieval buffers tested were sodium citrate (pH 6.0), EDTA (pH 8.0) and EDTA (pH 9.0). Cytokeratin AE1/AE3 IHC was used as a quality control for tissue and protocol. The IHC protocol is described in greater detail in the Additional file\n1.", "Three sets of criteria were used for interpretation of the IHC results, referred to as Score A, B and C, respectively, in this study. A positive result using score A was moderate to strong staining of membrane and/or cytoplasm in >10% tumor cells\n[15]. A positive result using score B was membrane staining in >10% tumor cells with any intensity\n[10]. A positive result using score C was membrane and/or cytoplasmic staining in >50% of the tumor cells with any intensity\n[13]. In this study all 50 specimens were analyzed using Score A, B and C separately, so as to evaluate the validity of these scoring methods by comparing to the results of DNA sequencing. Both the intensity and percentage of stained cells were assessed at low magnification (objective magnification ×10). The distribution of staining, membrane or cytoplasm, was assessed at high magnification (objective magnification ×40). Four experienced pathologists (Yan Xiong, Ying Dong, Lin Nong and Jing Zhao) reviewed all of the slides independently, and then replicated the analysis 16 to 18 weeks later. The intra- and inter-observer reliability was analyzed.", " DNA preparation H&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.\nH&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.\n Mutant-enriched PCR for EGFR exon 19 and 21 All samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.\nAll samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.\n DNA sequencing for EGFR exon 19 and 21 The amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.\nThe amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.", "H&E stained sections of FFPE tissue were reviewed for each sample and those with greater than 50% tumor volume were selected for molecular testing. Genomic DNA was extracted using the QIAamp DNA FFPE Tissue Kit (Qiagen, Valencia, CA) according to the manufacturer’s protocol.", "All samples were studied by DNA sequencing after mutation enriched PCR of exons 19 and 21. The PCR was done in a total volume of 12 μL with primers at a final concentration of 1 μM each, 50 μM of each dNTP, 0.75 units of HotStarTaq DNA polymerase, and 1.2 μL of the 10X buffer provided by the enzyme manufacturer (Qiagen). Template was added in a volume of 1 μL containing approximately 50 ng of DNA. An oligonucleotide clamp was added to suppress amplification of the normal sequence and enhance amplification of the L858R mutation and the exon 19 deletions. The clamp was synthesized with several locked nucleic acid (LNA) positions to increase its avidity. The clamps for exons 19 (CLMP1) and 21 (CLMP2) were used at final concentrations of 25 nM, and 5 nM, respectively. The sequences of the primers and clamps are described in Table\n2. The PCR primers were synthesized with M13 tail sequences appended to the 5′-end to facilitate sequencing. The reactions were cycled 40 times between 95°C for 10 seconds, 68°C for 15 seconds and 72°C for 20 seconds, preceded by 10 minutes at 95°C, and followed by 5 minutes at 72°C. The primers were purchased from Integrated DNA Technologies (Coralville, IA) and the LNA clamp was purchased from Biosynthesis, Inc. (Lewisville, TX).\nSequences of oligodeoxyribonucleotides\n1Other than the bases A, T, C and G, the oligos are described by a Y for mixed T and C; the positions of the LNA nucleotides (+ before the LNA) in the clamps (CLMP1 and CLMP2); and the C3 spacer (-C3) on the 3′ terminus of the clamps.", "The amplicons were treated with ExoSap (Amersham Biosciences, Piscataway, NJ) to remove the primers and dNTPs; then 1 μL was sequenced using the M13 tail primers as sequencing primers and Applied Biosystems (ABI, Foster City, CA) BigDye Terminator v.3.1 chemistry. The sequencing reactions were purified using the CleanSeq system (Agencourt Bioscience, Beverly, MA) and then resolved by capillary electrophoresis on the ABI 3100 Prism Genetic Analyzer.", "Statistical analysis was done using the statistics software SPSS V16.0 (SPSS Inc., Chicago, IL). Fleiss’ Kappa was used to determine inter-observer agreement. Cohen’s Kappa was used to determine intra -observer agreement and agreement of IHC and DNA sequencing. A Kappa value between 0.81 and 1.0 was defined as nearly perfect agreement, between 0.61 and 0.8 as substantial agreement, between 0.41 and 0.60 as moderate agreement, between 0.21 and 0.40 as fair agreement, between 0.00 and 0.20 as slight agreement. For each Kappa, the 95% confidence interval (CI) was calculated. Difference was considered significant (P < 0.05), if the lower and upper boundary of the 95% CI showed no overlap.", "All experiments above have been performed with the approval of Peking University First Hospital Ethics Committee.", " DNA sequencing In the total cohort of 50 samples L858R was identified in 16 cases, a deletion in exon 19 in 17 cases, and neither of them in 17 cases. Of the 17 cases with exon 19 deletion, 14 had a p.E746_A750del (c.del2235_2249 on the DNA level), one had a p.L747_T751del (c.2240_2254del), one had a p.L747_P753delinsS (c.2240_2257del), and one had a p.L747_T751delinsPT (c.2239_2253delinsCCAACG) that had not been previously reported. In our study, of all 33 cases with EGFR mutations, L858R and E746_A750del together comprised 90% (30/33) and the others, including L747_T751del, L747_P753delinsS and L747_T751delinsPT, comprised 10%, which was concordant with other studies\n[17]. From this point of view L858R and E746_A750del are recognized as the most common mutations and the other mutation types are described as uncommon mutations.\nIn the total cohort of 50 samples L858R was identified in 16 cases, a deletion in exon 19 in 17 cases, and neither of them in 17 cases. Of the 17 cases with exon 19 deletion, 14 had a p.E746_A750del (c.del2235_2249 on the DNA level), one had a p.L747_T751del (c.2240_2254del), one had a p.L747_P753delinsS (c.2240_2257del), and one had a p.L747_T751delinsPT (c.2239_2253delinsCCAACG) that had not been previously reported. In our study, of all 33 cases with EGFR mutations, L858R and E746_A750del together comprised 90% (30/33) and the others, including L747_T751del, L747_P753delinsS and L747_T751delinsPT, comprised 10%, which was concordant with other studies\n[17]. From this point of view L858R and E746_A750del are recognized as the most common mutations and the other mutation types are described as uncommon mutations.\n Evaluation of antigen retrieval buffer We evaluated three different antigen retrieval buffers on all 50 specimens to optimize the IHC results: sodium citrate (pH 6.0), EDTA (pH 8.0) and EDTA (pH 9.0). Slides in the EDTA (pH 8.0) group showed the best histological pictures with strongly specific staining and minimal background. The intensity of the positive cells in the sodium citrate (pH 6.0) group was too faint to distinguish from the background. Mesenchymal cells on slides exposed to EDTA (pH 9.0) were stained as strong as tumor cells, which made it impossible to identify the specificity of staining (Figure\n1). Inter-observer agreement was nearly perfect in the EDTA (pH 8.0) group, substantial in the sodium citrate (pH 6.0) group and moderate in the EDTA (pH 9.0) group. The Fleiss’ Kappa (95% confidence interval) was 0.912 (0.862, 0.962), 0.753 (0.677, 0.829), and 0.643(0.558, 0.728) in the three groups, respectively. The difference between EDTA (pH 8.0) and the others was significant (P < 0.05). Intra-observer agreement was highest in EDTA (pH 8.0), moderate in sodium citrate (pH 6.0), and lowest in EDTA (pH 9.0). The Cohen’s Kappa (95% confidence interval) was 0.955 (0.918, 0.992), 0.853 (0.790, 0.916), and 0.801 (0.730, 0.872), respectively. The difference between EDTA (pH 8.0) and the others was statistically significant (P < 0.05) (Table\n3).\nSample case of predominant solid adenocarcinoma immunostained with E746_A750del-specific antibody using different antigen retrieval buffers (original magnification x400). A Sodium citrate (pH 6.0). B EDTA (pH 8.0). C EDTA (pH 9.0).\nIntra- and inter-observer agreement based on slides treated with different antigen retrieval buffers\nWe evaluated three different antigen retrieval buffers on all 50 specimens to optimize the IHC results: sodium citrate (pH 6.0), EDTA (pH 8.0) and EDTA (pH 9.0). Slides in the EDTA (pH 8.0) group showed the best histological pictures with strongly specific staining and minimal background. The intensity of the positive cells in the sodium citrate (pH 6.0) group was too faint to distinguish from the background. Mesenchymal cells on slides exposed to EDTA (pH 9.0) were stained as strong as tumor cells, which made it impossible to identify the specificity of staining (Figure\n1). Inter-observer agreement was nearly perfect in the EDTA (pH 8.0) group, substantial in the sodium citrate (pH 6.0) group and moderate in the EDTA (pH 9.0) group. The Fleiss’ Kappa (95% confidence interval) was 0.912 (0.862, 0.962), 0.753 (0.677, 0.829), and 0.643(0.558, 0.728) in the three groups, respectively. The difference between EDTA (pH 8.0) and the others was significant (P < 0.05). Intra-observer agreement was highest in EDTA (pH 8.0), moderate in sodium citrate (pH 6.0), and lowest in EDTA (pH 9.0). The Cohen’s Kappa (95% confidence interval) was 0.955 (0.918, 0.992), 0.853 (0.790, 0.916), and 0.801 (0.730, 0.872), respectively. The difference between EDTA (pH 8.0) and the others was statistically significant (P < 0.05) (Table\n3).\nSample case of predominant solid adenocarcinoma immunostained with E746_A750del-specific antibody using different antigen retrieval buffers (original magnification x400). A Sodium citrate (pH 6.0). B EDTA (pH 8.0). C EDTA (pH 9.0).\nIntra- and inter-observer agreement based on slides treated with different antigen retrieval buffers\n IHC results The staining distribution included cytoplasm only or cytoplasm together with membrane. Normal tissue adjacent to adenocarcinoma was negative (Figure\n2). In the cases with lepidic pattern staining of lepidic tumor cells was either negative or fainter than the tumor cells of other patterns (Figure\n3). In our study, 80% of the cases were either negative or positive in 100% of the tumor cells, although the intensity was diverse ranging from + to +++. In only 20% of cases the tumor cells were stained in some areas and completely negative in other areas. Overall, the staining pattern showed characteristics of homogeneity more than heterogeneity.\nPredominant acinar adenocarcinoma with adjacent normal alveoli. A H & E (original magnification x100). B The cytoplasm and membrane of the tumor cells were stained strongly with L858-specific antibody. In contrast, the adjacent normal alveolar epithelial cells were completely negative (original magnification x100).\nAdenocarcinoma with acinar and lepidic patterns. A H & E (original magnification x100). B The cytoplasm and membrane of acinar component stained strongly, but lepidic tumor cells stained weakly with L858-specfic antibody (original magnification x100).\nBased on different scoring systems, the percentage of positive cases was different too. For L858R-specific IHC it was 28% (14/50) on Score A, 16% (8/50) on Score B, and 40% (20/50) on Score C; for E746_A750del-specific IHC it was 20% (10/50) on Score A, 14% (7/50) on Score B, and 24% (12/50) on Score C.\nThe staining distribution included cytoplasm only or cytoplasm together with membrane. Normal tissue adjacent to adenocarcinoma was negative (Figure\n2). In the cases with lepidic pattern staining of lepidic tumor cells was either negative or fainter than the tumor cells of other patterns (Figure\n3). In our study, 80% of the cases were either negative or positive in 100% of the tumor cells, although the intensity was diverse ranging from + to +++. In only 20% of cases the tumor cells were stained in some areas and completely negative in other areas. Overall, the staining pattern showed characteristics of homogeneity more than heterogeneity.\nPredominant acinar adenocarcinoma with adjacent normal alveoli. A H & E (original magnification x100). B The cytoplasm and membrane of the tumor cells were stained strongly with L858-specific antibody. In contrast, the adjacent normal alveolar epithelial cells were completely negative (original magnification x100).\nAdenocarcinoma with acinar and lepidic patterns. A H & E (original magnification x100). B The cytoplasm and membrane of acinar component stained strongly, but lepidic tumor cells stained weakly with L858-specfic antibody (original magnification x100).\nBased on different scoring systems, the percentage of positive cases was different too. For L858R-specific IHC it was 28% (14/50) on Score A, 16% (8/50) on Score B, and 40% (20/50) on Score C; for E746_A750del-specific IHC it was 20% (10/50) on Score A, 14% (7/50) on Score B, and 24% (12/50) on Score C.\n Concordance analysis of IHC and DNA sequencing Of the 16 cases with L858R, the L858R-specific IHC was positive in 13 on Score A, 7 on Score B and 11 on Score C (Figure\n4). Of 34 cases without L858R the L858R-specific IHC was negative in 33 on Score A, 33 on Score B and 25 on Score C (Figure\n5). L858R-specific IHC showed a sensitivity of 81%, a specificity of 97%, a positive predictive value (PPV) of 93%, and a negative predictive value (NPV) of 92% on Score A; a sensitivity of 44%, a specificity of 97%, a PPV of 88%, and a NPV of 79% on Score B; and a sensitivity of 69%, a specificity of 74%, a PPV of 55%, and a NPV of 83% on Score C. Reliability analysis for L858R-specific IHC and DNA sequencing was found to be Cohen’s Kappa = 0.810 and 95% CI (0.701, 0.919) on Score A, Cohen’s Kappa = 0.470 and 95% CI (0.332, 0.608) on Score B, and Cohen’s Kappa = 0.397 and 95% CI (0.261, 0.533) on Score C. The agreement between L858R-specific IHC and DNA sequencing was the best using Score A compared to Score B and C. The difference was significant (P < 0.05) (Table\n4).\nA case of predominant solid adenocarcinoma with the L858R mutation. A DNA sequencing of EGFR showing normal (upper panel) and L858R mutant (lower panel). The position of the mutation is boxed. The mutant sequence appears to be homozygous with complete absence of normal sequence. This is because of the use of a “clamp” strategy to suppress amplification of the normal sequence, as described in the methods section. B Immunohistochemical staining with L858R-specific antibody showing strong positivity (original magnification x200). C Immunohistochemical staining with E746-A750 del-specific antibody showing complete negativity (original magnification x200).\nA case of predominant acinar adenocarcinoma with normal EGFR. A and B DNA sequencing shown normal EGFR exon 21 and 19, respectively, in the region that is frequently subject to mutation. C and D Tumor cells were not stained with either L858R-specfic or E746_A750del-specific antibodies, respectively (original magnification x200).\nDiagnostic power of L858R-specific IHC on score A, B and C\nIHC indicates immunohistochemistry; PPV, positive predictive value; NPV, negative predictive value.\nOf 14 cases with the E746_A750del by DNA sequencing E746_A750del-specific IHC was positive in 10 on Score A, 7 on Score B, and 9 on Score C (Figure\n6). All of the 3 cases with uncommon types of exon 19 deletion, includingL747_T751del, L747_P753delinsS and L747_T751delinsPT, were negative by E746_A750del-specific IHC regardless of the scoring method.\nA case of predominant acinar adenocarcinoma with the exon 19 deletion mutation (E746-A750 del). A DNA sequencing of EGFR showing normal (upper panel) and the E746-A750 del mutant (lower panel). The position of the mutation is indicated. The mutant sequence appears to be homozygous with complete absence of normal sequence. This is because of the use of a “clamp” strategy to suppress amplification of the normal sequence, as described in the methods section. B Immunohistochemical staining with E746-A750 del-specific antibody showing strong positivity (original magnification x200). C Immunohistochemical staining with L858R-specific antibody showing complete negativity (original magnification x200).\nIncluding all 17 specimens with an exon 19 deletion detected by DNA sequencing, the E746_A750del-specific IHC found 10 (59%) were positive and 7 (41%) negative on Score A, 7 (41%) were positive and 10 (59%) negative on Score B, 9 (53%) were positive and 8 (47%) negative on Score C. Of the 33 cases without an exon 19 deletion detected by DNA sequencing all were negative by E746_A750del-specific IHC both on Score A and B (Figure\n5), while 3 were positive on Score C. As a method to detect deletions in exon 19 despite the exact structure of the deletion, E746_A750del-specific IHC showed a sensitivity of 59%, a specificity of 100%, a PPV of 100%, and a NPV of 82% on Score A; a sensitivity of 41%, a specificity of 100%, a PPV of 100%, and a NPV of 77% on Score B; and a sensitivity of 53%, a specificity of 91%, a PPV of 75%, and a NPV of 79% on Score C. Reliability analysis for E746_A750del-specifiac IHC and DNA sequencing was found to be Cohen’s Kappa = 0.653 and 95% CI (0.521, 0.785) on Score A, Cohen’s Kappa = 0.480 and 95% CI (0.342, 0.618) on Score B, Cohen’s Kappa = 0.472 and 95% CI (0.334, 0.610) on Score C. Similar to L858R-specific IHC the agreement between E746_A750del-specific IHC and DNA sequencing was the best using Score A compared to Score B and C, but the difference was not significant (P > 0.05) (Table\n5).\nDiagnostic power of E746_A750del-specific IHC on score A, B and C\nIHC indicates immunohistochemistry; PPV, positive predictive value; NPV, negative predictive value.\nOf the 16 cases with L858R, the L858R-specific IHC was positive in 13 on Score A, 7 on Score B and 11 on Score C (Figure\n4). Of 34 cases without L858R the L858R-specific IHC was negative in 33 on Score A, 33 on Score B and 25 on Score C (Figure\n5). L858R-specific IHC showed a sensitivity of 81%, a specificity of 97%, a positive predictive value (PPV) of 93%, and a negative predictive value (NPV) of 92% on Score A; a sensitivity of 44%, a specificity of 97%, a PPV of 88%, and a NPV of 79% on Score B; and a sensitivity of 69%, a specificity of 74%, a PPV of 55%, and a NPV of 83% on Score C. Reliability analysis for L858R-specific IHC and DNA sequencing was found to be Cohen’s Kappa = 0.810 and 95% CI (0.701, 0.919) on Score A, Cohen’s Kappa = 0.470 and 95% CI (0.332, 0.608) on Score B, and Cohen’s Kappa = 0.397 and 95% CI (0.261, 0.533) on Score C. The agreement between L858R-specific IHC and DNA sequencing was the best using Score A compared to Score B and C. The difference was significant (P < 0.05) (Table\n4).\nA case of predominant solid adenocarcinoma with the L858R mutation. A DNA sequencing of EGFR showing normal (upper panel) and L858R mutant (lower panel). The position of the mutation is boxed. The mutant sequence appears to be homozygous with complete absence of normal sequence. This is because of the use of a “clamp” strategy to suppress amplification of the normal sequence, as described in the methods section. B Immunohistochemical staining with L858R-specific antibody showing strong positivity (original magnification x200). C Immunohistochemical staining with E746-A750 del-specific antibody showing complete negativity (original magnification x200).\nA case of predominant acinar adenocarcinoma with normal EGFR. A and B DNA sequencing shown normal EGFR exon 21 and 19, respectively, in the region that is frequently subject to mutation. C and D Tumor cells were not stained with either L858R-specfic or E746_A750del-specific antibodies, respectively (original magnification x200).\nDiagnostic power of L858R-specific IHC on score A, B and C\nIHC indicates immunohistochemistry; PPV, positive predictive value; NPV, negative predictive value.\nOf 14 cases with the E746_A750del by DNA sequencing E746_A750del-specific IHC was positive in 10 on Score A, 7 on Score B, and 9 on Score C (Figure\n6). All of the 3 cases with uncommon types of exon 19 deletion, includingL747_T751del, L747_P753delinsS and L747_T751delinsPT, were negative by E746_A750del-specific IHC regardless of the scoring method.\nA case of predominant acinar adenocarcinoma with the exon 19 deletion mutation (E746-A750 del). A DNA sequencing of EGFR showing normal (upper panel) and the E746-A750 del mutant (lower panel). The position of the mutation is indicated. The mutant sequence appears to be homozygous with complete absence of normal sequence. This is because of the use of a “clamp” strategy to suppress amplification of the normal sequence, as described in the methods section. B Immunohistochemical staining with E746-A750 del-specific antibody showing strong positivity (original magnification x200). C Immunohistochemical staining with L858R-specific antibody showing complete negativity (original magnification x200).\nIncluding all 17 specimens with an exon 19 deletion detected by DNA sequencing, the E746_A750del-specific IHC found 10 (59%) were positive and 7 (41%) negative on Score A, 7 (41%) were positive and 10 (59%) negative on Score B, 9 (53%) were positive and 8 (47%) negative on Score C. Of the 33 cases without an exon 19 deletion detected by DNA sequencing all were negative by E746_A750del-specific IHC both on Score A and B (Figure\n5), while 3 were positive on Score C. As a method to detect deletions in exon 19 despite the exact structure of the deletion, E746_A750del-specific IHC showed a sensitivity of 59%, a specificity of 100%, a PPV of 100%, and a NPV of 82% on Score A; a sensitivity of 41%, a specificity of 100%, a PPV of 100%, and a NPV of 77% on Score B; and a sensitivity of 53%, a specificity of 91%, a PPV of 75%, and a NPV of 79% on Score C. Reliability analysis for E746_A750del-specifiac IHC and DNA sequencing was found to be Cohen’s Kappa = 0.653 and 95% CI (0.521, 0.785) on Score A, Cohen’s Kappa = 0.480 and 95% CI (0.342, 0.618) on Score B, Cohen’s Kappa = 0.472 and 95% CI (0.334, 0.610) on Score C. Similar to L858R-specific IHC the agreement between E746_A750del-specific IHC and DNA sequencing was the best using Score A compared to Score B and C, but the difference was not significant (P > 0.05) (Table\n5).\nDiagnostic power of E746_A750del-specific IHC on score A, B and C\nIHC indicates immunohistochemistry; PPV, positive predictive value; NPV, negative predictive value.", "In the total cohort of 50 samples L858R was identified in 16 cases, a deletion in exon 19 in 17 cases, and neither of them in 17 cases. Of the 17 cases with exon 19 deletion, 14 had a p.E746_A750del (c.del2235_2249 on the DNA level), one had a p.L747_T751del (c.2240_2254del), one had a p.L747_P753delinsS (c.2240_2257del), and one had a p.L747_T751delinsPT (c.2239_2253delinsCCAACG) that had not been previously reported. In our study, of all 33 cases with EGFR mutations, L858R and E746_A750del together comprised 90% (30/33) and the others, including L747_T751del, L747_P753delinsS and L747_T751delinsPT, comprised 10%, which was concordant with other studies\n[17]. From this point of view L858R and E746_A750del are recognized as the most common mutations and the other mutation types are described as uncommon mutations.", "We evaluated three different antigen retrieval buffers on all 50 specimens to optimize the IHC results: sodium citrate (pH 6.0), EDTA (pH 8.0) and EDTA (pH 9.0). Slides in the EDTA (pH 8.0) group showed the best histological pictures with strongly specific staining and minimal background. The intensity of the positive cells in the sodium citrate (pH 6.0) group was too faint to distinguish from the background. Mesenchymal cells on slides exposed to EDTA (pH 9.0) were stained as strong as tumor cells, which made it impossible to identify the specificity of staining (Figure\n1). Inter-observer agreement was nearly perfect in the EDTA (pH 8.0) group, substantial in the sodium citrate (pH 6.0) group and moderate in the EDTA (pH 9.0) group. The Fleiss’ Kappa (95% confidence interval) was 0.912 (0.862, 0.962), 0.753 (0.677, 0.829), and 0.643(0.558, 0.728) in the three groups, respectively. The difference between EDTA (pH 8.0) and the others was significant (P < 0.05). Intra-observer agreement was highest in EDTA (pH 8.0), moderate in sodium citrate (pH 6.0), and lowest in EDTA (pH 9.0). The Cohen’s Kappa (95% confidence interval) was 0.955 (0.918, 0.992), 0.853 (0.790, 0.916), and 0.801 (0.730, 0.872), respectively. The difference between EDTA (pH 8.0) and the others was statistically significant (P < 0.05) (Table\n3).\nSample case of predominant solid adenocarcinoma immunostained with E746_A750del-specific antibody using different antigen retrieval buffers (original magnification x400). A Sodium citrate (pH 6.0). B EDTA (pH 8.0). C EDTA (pH 9.0).\nIntra- and inter-observer agreement based on slides treated with different antigen retrieval buffers", "The staining distribution included cytoplasm only or cytoplasm together with membrane. Normal tissue adjacent to adenocarcinoma was negative (Figure\n2). In the cases with lepidic pattern staining of lepidic tumor cells was either negative or fainter than the tumor cells of other patterns (Figure\n3). In our study, 80% of the cases were either negative or positive in 100% of the tumor cells, although the intensity was diverse ranging from + to +++. In only 20% of cases the tumor cells were stained in some areas and completely negative in other areas. Overall, the staining pattern showed characteristics of homogeneity more than heterogeneity.\nPredominant acinar adenocarcinoma with adjacent normal alveoli. A H & E (original magnification x100). B The cytoplasm and membrane of the tumor cells were stained strongly with L858-specific antibody. In contrast, the adjacent normal alveolar epithelial cells were completely negative (original magnification x100).\nAdenocarcinoma with acinar and lepidic patterns. A H & E (original magnification x100). B The cytoplasm and membrane of acinar component stained strongly, but lepidic tumor cells stained weakly with L858-specfic antibody (original magnification x100).\nBased on different scoring systems, the percentage of positive cases was different too. For L858R-specific IHC it was 28% (14/50) on Score A, 16% (8/50) on Score B, and 40% (20/50) on Score C; for E746_A750del-specific IHC it was 20% (10/50) on Score A, 14% (7/50) on Score B, and 24% (12/50) on Score C.", "Of the 16 cases with L858R, the L858R-specific IHC was positive in 13 on Score A, 7 on Score B and 11 on Score C (Figure\n4). Of 34 cases without L858R the L858R-specific IHC was negative in 33 on Score A, 33 on Score B and 25 on Score C (Figure\n5). L858R-specific IHC showed a sensitivity of 81%, a specificity of 97%, a positive predictive value (PPV) of 93%, and a negative predictive value (NPV) of 92% on Score A; a sensitivity of 44%, a specificity of 97%, a PPV of 88%, and a NPV of 79% on Score B; and a sensitivity of 69%, a specificity of 74%, a PPV of 55%, and a NPV of 83% on Score C. Reliability analysis for L858R-specific IHC and DNA sequencing was found to be Cohen’s Kappa = 0.810 and 95% CI (0.701, 0.919) on Score A, Cohen’s Kappa = 0.470 and 95% CI (0.332, 0.608) on Score B, and Cohen’s Kappa = 0.397 and 95% CI (0.261, 0.533) on Score C. The agreement between L858R-specific IHC and DNA sequencing was the best using Score A compared to Score B and C. The difference was significant (P < 0.05) (Table\n4).\nA case of predominant solid adenocarcinoma with the L858R mutation. A DNA sequencing of EGFR showing normal (upper panel) and L858R mutant (lower panel). The position of the mutation is boxed. The mutant sequence appears to be homozygous with complete absence of normal sequence. This is because of the use of a “clamp” strategy to suppress amplification of the normal sequence, as described in the methods section. B Immunohistochemical staining with L858R-specific antibody showing strong positivity (original magnification x200). C Immunohistochemical staining with E746-A750 del-specific antibody showing complete negativity (original magnification x200).\nA case of predominant acinar adenocarcinoma with normal EGFR. A and B DNA sequencing shown normal EGFR exon 21 and 19, respectively, in the region that is frequently subject to mutation. C and D Tumor cells were not stained with either L858R-specfic or E746_A750del-specific antibodies, respectively (original magnification x200).\nDiagnostic power of L858R-specific IHC on score A, B and C\nIHC indicates immunohistochemistry; PPV, positive predictive value; NPV, negative predictive value.\nOf 14 cases with the E746_A750del by DNA sequencing E746_A750del-specific IHC was positive in 10 on Score A, 7 on Score B, and 9 on Score C (Figure\n6). All of the 3 cases with uncommon types of exon 19 deletion, includingL747_T751del, L747_P753delinsS and L747_T751delinsPT, were negative by E746_A750del-specific IHC regardless of the scoring method.\nA case of predominant acinar adenocarcinoma with the exon 19 deletion mutation (E746-A750 del). A DNA sequencing of EGFR showing normal (upper panel) and the E746-A750 del mutant (lower panel). The position of the mutation is indicated. The mutant sequence appears to be homozygous with complete absence of normal sequence. This is because of the use of a “clamp” strategy to suppress amplification of the normal sequence, as described in the methods section. B Immunohistochemical staining with E746-A750 del-specific antibody showing strong positivity (original magnification x200). C Immunohistochemical staining with L858R-specific antibody showing complete negativity (original magnification x200).\nIncluding all 17 specimens with an exon 19 deletion detected by DNA sequencing, the E746_A750del-specific IHC found 10 (59%) were positive and 7 (41%) negative on Score A, 7 (41%) were positive and 10 (59%) negative on Score B, 9 (53%) were positive and 8 (47%) negative on Score C. Of the 33 cases without an exon 19 deletion detected by DNA sequencing all were negative by E746_A750del-specific IHC both on Score A and B (Figure\n5), while 3 were positive on Score C. As a method to detect deletions in exon 19 despite the exact structure of the deletion, E746_A750del-specific IHC showed a sensitivity of 59%, a specificity of 100%, a PPV of 100%, and a NPV of 82% on Score A; a sensitivity of 41%, a specificity of 100%, a PPV of 100%, and a NPV of 77% on Score B; and a sensitivity of 53%, a specificity of 91%, a PPV of 75%, and a NPV of 79% on Score C. Reliability analysis for E746_A750del-specifiac IHC and DNA sequencing was found to be Cohen’s Kappa = 0.653 and 95% CI (0.521, 0.785) on Score A, Cohen’s Kappa = 0.480 and 95% CI (0.342, 0.618) on Score B, Cohen’s Kappa = 0.472 and 95% CI (0.334, 0.610) on Score C. Similar to L858R-specific IHC the agreement between E746_A750del-specific IHC and DNA sequencing was the best using Score A compared to Score B and C, but the difference was not significant (P > 0.05) (Table\n5).\nDiagnostic power of E746_A750del-specific IHC on score A, B and C\nIHC indicates immunohistochemistry; PPV, positive predictive value; NPV, negative predictive value.", "The studies that established the relationship between mutations in the EGFR gene and response to the small molecule EGFR TKIs gefitinib and erlotinib were done using analysis of DNA extracted from the tumor\n[18]. The recent availability of antibodies that are specific for the mutations most clearly associated with response to EGFR TKIs, L858R and E746_A750del, create the opportunity to exploit an alternative method to evaluate NSCLC for EGFR mutations to aid decisions with regard to EGFR TKI therapy\n[11].\nIHC has the advantage of being a method that is routinely applied in solid tumor diagnosis in pathology. Also, it can be used on specimens that are not optimal for DNA analysis such as small tissue samples or individual cells obtained from body fluids, bronchial washings, and fine needle aspirates. Although some studies have shown that EGFR gene sequencing could be successfully applied to cytological specimens\n[19], it is still a problem to get enough DNA for sequencing from such samples in routine practice. Thus, the development of antibodies that specifically detect mutant EGFR protein by IHC could be a valuable addition to the current methods used to diagnose and predict response to treatment of lung cancer.\nIn 2009, Yu et al.\n[7] reported generating two mAbs from New Zealand rabbits, one against the E746_A750del and the other against the L858R point mutation, and evaluated them by Western blotting, immunofluorescence and IHC. They tested these antibodies in a series of cell lines and in tumor tissues from patients with primary NSCLC, with known and unknown EGFR mutations, comparing the IHC results with DNA sequencing. They found that IHC with these mutation-specific antibodies showed a sensitivity of 92% and a specificity of 99%. Recently, several studies examined the presence of EGFR mutations in NSCLC by IHC using the same two antibodies and the reported sensitivity ranged from 24% to 100% and specificity ranged from 77% to 100%\n[8-15]. IHC is known to sometimes suffer from high inter-laboratory variability in assay performance, and high inter-observer variability in assay interpretation. These drawbacks may explain the variability in results of the studies described above. There is still much work to be done before IHC can be considered an adequate substitute for direct analysis of mutations in the EGFR gene in NSCLC.\nIn our study we found that slides treated by EDTA (pH 8.0) showed the best histological pictures with strongly specific staining and minimal background. As a result when pathologists reviewed these slides the intra- and inter-observer agreement was better than those treated by sodium citrate (pH 6.0) and EDTA (pH 9.0). The difference was statistically significant (P < 0.05). In conclusion, EDTA (pH 8.0) is the preferred buffer for antigen retrieval for IHC using EGFR mutation specific antibodies.\nScoring is the final step involved in the IHC protocol, but is not the least one, because the scoring system plays a critical role in obtaining a reliable result. In our study, we compared three scoring systems that have been used by other investigators, using DNA sequencing as the gold standard\n[10,13,15]. For L858R-specific IHC the agreement with DNA sequencing using Score A was superior to Score B and C. The difference was statistically significant. For E746_A750del-specific IHC the agreement with DNA sequencing was good for Score A, which was superior to Score B and C, but the difference was not statistically significant. In conclusion our study showed that Score A is the most appropriate way to interpret the EGFR mutation-specific IHC.\nBased on Score A the specificity of EGFR mutation-specific IHC was very high, 100% for exon 19 deletions and 97% for L858R, while sensitivity was lower, 81% for L858R and 59% for exon 19 deletions. In another words, mutation-specific IHC demonstrated extremely high specificities, but much lower sensitivity. The low sensitivity of the exon 19 del IHC is mostly due to the presence of several exon 19 deletions other than the most common E746_A750del, which is the target of the exon 19 del antibodies. We conclude, based on our work, that NSLC cases positive by IHC could be selected as candidates for EGFR-TKI, while negative cases should be referred for further testing by DNA analysis.\nIn our study the majority of cases were either negative or positive in 100% of the tumor cells. The staining pattern showed characteristics of homogeneity more than heterogeneity. Consequently, we expect that evaluation of the mutation status by IHC should be reliable using small biopsy specimens or tissue microarray.\nOur study also showed that all of the normal alveolar epithelial cells were completely negative and the intensity of mutation-specific immunostaining was much fainter in tumor cells with a lepidic pattern comparing to other patterns. This demonstrates that the EGFR mutations are tumor-specific, and likely an initiating event in lung cancer tumorigenesis.", "Immunohistochemistry using mutation-specific mAbs is demonstrated to be a reliable test for detecting EGFR mutations in adenocarcinoma of the lung in our study. The diagnostic power of EGFR mutation-specific IHC is influenced by the antigen retrieval and scoring methods. Based on our study EDTA (pH 8.0) is better than sodium citrate (pH 6.0) and EDTA (pH 9.0) as the antigen retrieval buffer. A practical and reliable scoring method, i.e. positive is interpreted as moderate to strong staining of membrane and/or cytoplasm in >10% tumor cells, is recommended. However its final validation depends on strict quality control of the whole protocol, including antibody manufacture, IHC method, scoring system, criteria for interpretation, and the proper way to integrate with molecular methods, etc.\nThe specificity of EGFR mutation-specific IHC was very high, 100% for exon 19 deletions and 97% for L858R, while sensitivity was relatively lower, 81% for L858R and 59% for Exon 19 deletions. Considering the use of IHC has the advantage of being a method routinely applied in solid tumor diagnosis in pathology, EGFR mutation-specific IHC could be used as a prescreening method for selecting EGFR-TKI candidates. The positive cases by IHC could be selected as candidates for EGFR-TKI, while negative cases should be referred for DNA analysis. Additionally, as the staining pattern showed characteristics of homogeneity more than heterogeneity, it should be reliable to evaluate the mutation status of biopsy specimens or tissue microarray using IHC. Furthermore, it may be possible to use IHC as a substitute when the quantity of the sample DNA is not sufficient for molecular methods, e.g., small tissue samples or individual cells obtained from body fluids, bronchial washings, and fine needle aspirates etc.", "Written informed consent was obtained from the patient for publication of this report and any accompanying images.", "ARMS: Amplification refractory mutation system; CI: Confidence interval; CLMP1: Clamps for exons 19; CLMP2: Clamps for exons 21; EGFR: Epidermal growth factor; FFPE: Formalin-fixed paraffin- embedded; IHC: Immunohistochemistry; LNA: Locked nucleic acid; NSCL: Non-small cell lung cancer; NPV: Negative predictive value; PPV: Positive predictive value; TK: Tyrosine kinase; TKI: Tyrosine kinase inhibitor.", "The authors declare that we do not have any financial competing interests.", "XY participated in the design of the study and drafted the manuscript. BY carried out the immunoassays and collected patient’s clinic data. LN and LTS carried out the molecular genetic studies. RPG carried out the molecular genetic studies, and helped to draft the manuscript. XH contributed to the design of the study, helped to draft the manuscript, and participated in coordination. NL, ZJ and DY participated in the review of the histologic slides. NL carried out the molecular genetic studies as well. LT conceived of the study, participated in its design and coordination, and helped to draft the manuscript. All authors read and approved the final manuscript.", "Yan Xiong, MD, Associated Professor in Department of Pathology, Peking University First Hospital, Beijing, China. Ting Li, MD, Full professor and Chair in Department of Pathology, Peking University First Hospital, Beijing, China.", "Protocol for EGFR mutation-specific immunohistochemistry.\nClick here for file" ]

[ null, "methods", null, null, null, null, null, null, null, null, null, "results", null, null, null, null, "discussion", "conclusions", null, null, null, null, null, "supplementary-material" ]

[ "Lung adenocarcinoma", "Epidermal growth factor receptor", "Mutation", "Immunohistochemistry" ]

[ 499, 137, 121, 225, 961, 51, 334, 81, 137, 17, 161, 384, 311, 1063, 18, 82, 13, 123, 44 ]

[ "score", "ihc", "dna", "specific", "sequencing", "l858r", "ph", "19", "mutation", "dna sequencing" ]

[ "common egfr mutations", "egfr tyrosine kinase", "egfr mutations methods", "egfr mutations comparing", "detect egfr mutations" ]

[ "COVID-19", "First Aid", "Germany", "Hotlines", "Humans", "Mental Health", "Pandemics", "Psychological First Aid", "SARS-CoV-2" ]

[ "Hintergrund", "Ziel des Projekts", "Methodik", "Rekrutierung der Berater*innen", "Implementierung der Webseite", "Datenerhebung", "Dokumentation der Beratungen", "Abschließende Berater*innenbefragung", "Statistische Analysen", "Ergebnisse", "Dokumentation der Beratungsgespräche", "Inanspruchnahme der Webseite", "Abschließende Berater*innenbefragung", "Diskussion", "Limitationen", "Schlussfolgerung" ]

[ "Die Covid-19-Pandemie stellt Deutschland und die gesamte Welt vor große medizinische, wirtschaftliche, soziale und auch psychologische Herausforderungen. Erste Längsschnittstudien deuten auf eine Zunahme der psychischen Belastung in der Bevölkerung hin [1, 2]. Unter einer Vielzahl pandemiebedingter Belastungsfaktoren stehen vor allem die Angst vor Infektionen, die staatlich verordneten Maßnahmen zur Eindämmung der Pandemie, wirtschaftliche Existenzängste und arbeitsbezogene Belastungen im Vordergrund.\nDie Angst vor einer Infektion mit dem Coronavirus ist weit verbreitet. Demnach äußerten im März 2020, zum Höhepunkt der ersten Infektionswelle, 49 % aller Deutschen eine große oder sehr große Angst vor einer möglichen Infektion [3]. Eine deutsche Studie zeigte, dass die Corona-Angst zu Beginn der Pandemie schnell zunahm und mit einer späteren generalisierten Angst und Schlafstörungen einherging [4]. Die Angst zeigte sich besonders ausgeprägt bei vulnerablen Gruppen (z. B. Personen mit körperlichen Vorerkrankungen) oder aber bei Beschäftigten im Gesundheitswesen [5–7]. Dabei scheint es einen Zusammenhang zwischen der Angst vor einer Corona-Infektion mit psychischen Symptomen wie depressiven Beschwerden und suizidalen Gedanken sowie mit deutlichen psychosozialen Funktionseinschränkungen zu geben [8]. Neben der Angst, selbst durch das Coronavirus infiziert zu werden, bestehen ebenso Ängste, andere Personen anzustecken oder dass sich Personen im Freundes- und Familienkreis mit dem Virus infizieren könnten.\nDie Eindämmung der Pandemie erforderte gerade während der ersten Welle zudem eine Serie von gänzlich neuen, staatlich verordneten Verhaltensempfehlungen und Präventionsmaßnahmen. Durch Isolations- und Quarantänemaßnahmen waren viele Menschen gezwungen, über lange Zeit alleine oder auf enge Räume beschränkt zu leben. Außerdem waren soziale Kontakte erschwert, wodurch wichtige Unterstützung entfiel. Die Auswirkungen dieser einschränkenden Maßnahmen auf die psychische Gesundheit von Menschen werden seit Beginn der Pandemie kritisch diskutiert. Schon im März 2020 erschien eine Übersichtsarbeit, in der traumaähnliche Symptome, Verwirrung und Ärger als mögliche Folgen von Quarantänemaßnahmen beschrieben wurden [9]. Die Belastungssymptome können auch noch Monate nach den Quarantänemaßnahmen bestehen bleiben [10]. Doch nicht nur die verordnete Quarantäne führt zu psychischen Belastungen, auch schon weniger extreme Formen der sozialen Distanzierung wie die Reduktion von Kontakten oder die Empfehlung, das häusliche Umfeld nicht zu verlassen, zeigte sich mit vermehrten Depressionssymptomen, generalisierten Angstsymptomen, intrusiven Gedanken, Insomnie und akutem Stress assoziiert [11]. Gerade ältere Menschen, aber auch Jugendliche, leiden demnach an den Folgen der Einschränkungen und berichten eine Zunahme von Einsamkeit und psychischer Belastung während der Pandemie [1, 12].\nNicht zuletzt stellen die volkswirtschaftlichen Folgen der Pandemie eine besondere Herausforderung dar. So führen Arbeitsplatzunsicherheit und finanzielle Sorgen im Rahmen der Corona-Pandemie zu Beeinträchtigungen der psychischen Gesundheit [13]. Homeoffice-Regelungen und Schulschließungen führen zu einer radikalen Veränderung des Lebensalltags vieler Menschen und stellen bisher ungekannte Herausforderungen dar, Arbeit, Kinderbetreuung und Haushalt zu vereinen. Eine Studie über die Auswirkungen von Homeschooling in sieben europäischen Ländern (u. a. Deutschland) zeigte negative Auswirkungen von Homeschooling sowohl für Eltern als auch für Kinder in Form von vermehrtem Stress, Sorgen, sozialer Isolation, häuslicher Konflikte und teilweise auch erhöhtem Alkoholkonsum [14]. Nicht zuletzt lösen die wirtschaftlichen Restriktionen für viele Menschen existenzielle Ängste aus [15], insbesondere bei Beschäftigten in besonders betroffenen Branchen wie der Gastronomie‑, Freizeit- oder Kulturbranche.\nIn vielen Fällen können diese Belastungen zu dysfunktionalen Bewältigungsversuchen führen wie erhöhtem Alkohol- und Substanzkonsum [16], Aggressivität und Wutausbrüchen oder ein Anstieg von Gewalt im häuslichen Umfeld [17–19]. In den europäischen Ländern wurde demnach im April 2020 ein Anstieg von Notrufen aufgrund häuslicher Gewalt gegen Frauen um 60 % im Vergleich zum Vorjahr verzeichnet [20]. Eine repräsentative Studie in Deutschland zeigte, dass die häusliche Gewalt gegen Frauen und Kinder unter Quarantänebedingungen um das zwei‑ bis dreifache zunahm [21].\nVor diesem Hintergrund besteht eine ernste Gefahr, dass psychische Erkrankungen durch die Pandemie zunehmen werden [22]. Die Bundespsychotherapeutenkammer warnt, dass die Corona-Pandemie manifeste psychische Erkrankungen auslösen oder verstärken werde, und benennt dabei v. a. Depressionen und Angststörungen, akute und posttraumatische Belastungsstörungen (PTBS), Alkohol- und Medikamentenabhängigkeit, Zwangsstörungen und Psychosen [23].\nEs scheint daher von zentraler Bedeutung zu sein, Menschen bei der Bewältigung der psychischen Belastungen möglichst schnell, unkompliziert und früh zu unterstützen. Schon zu Beginn der Pandemie wurde daher auf die Wichtigkeit von Frühintervention hingewiesen [24]. Im Rahmen der Pandemie spielen dabei vor allem flexible und schnelle Hilfsangebote wie Online- oder Telefonhilfen eine wichtige Rolle, um die Menschen zu erreichen [25, 26]. Eine große Herausforderung besteht dabei, die Hotlinedienste mit ausreichend qualifizierten Berater*innen auszustatten [26].\nIn Deutschland konnten die gut etablierten Versorgungsstrukturen gerade zu Beginn der Pandemie nur bedingt greifen und pandemietaugliche Alternativkonzepte fehlten [27]. Daher gab es trotz der hohen Unsicherheit und subjektiv erlebten Bedrohung kaum Hilfsangebote, die eine unkomplizierte, schnelle und gleichzeitig therapeutisch professionelle Hilfe bei Corona-bedingten psychischen Belastungen sicherstellten. Aus Fachkreisen wurde daher darauf hingewiesen, dass eine Überlastung des bestehenden psychotherapeutischen und psychiatrischen Versorgungssystems, einschließlich der psychosozialen Dienste, wahrscheinlich sei und alternative Hilfsangebote wie Telefonhotlines dringend ausgebaut werden sollten [27, 28]. Aus dieser Situation heraus entstand das Vorhaben, die gegebenen psychotherapeutischen Versorgungsstrukturen zu nutzen, um ein professionelles, flexibles und niederschwelliges Angebot zur psychologischen Ersthilfe für die Allgemeinbevölkerung zu schaffen.\nZiel des Projekts Ziel des Projekts war es, Menschen mit psychischen Belastungen aufgrund der Corona-Pandemie mit einer schnellen, professionell ausgeübten Ersthilfe zu unterstützen. Dazu wurde sowohl eine Hotline eingerichtet, über die persönliche Beratungsgespräche durch Fachpersonen ermöglicht wurden, als auch eine Webseite, auf der Unterstützungsangebote zur psychologischen Selbsthilfe zusammengetragen wurden. Die Inanspruchnahme und Nutzung der Hilfsangebote wurde wissenschaftlich begleitet.\nZiel des Projekts war es, Menschen mit psychischen Belastungen aufgrund der Corona-Pandemie mit einer schnellen, professionell ausgeübten Ersthilfe zu unterstützen. Dazu wurde sowohl eine Hotline eingerichtet, über die persönliche Beratungsgespräche durch Fachpersonen ermöglicht wurden, als auch eine Webseite, auf der Unterstützungsangebote zur psychologischen Selbsthilfe zusammengetragen wurden. Die Inanspruchnahme und Nutzung der Hilfsangebote wurde wissenschaftlich begleitet.", "Ziel des Projekts war es, Menschen mit psychischen Belastungen aufgrund der Corona-Pandemie mit einer schnellen, professionell ausgeübten Ersthilfe zu unterstützen. Dazu wurde sowohl eine Hotline eingerichtet, über die persönliche Beratungsgespräche durch Fachpersonen ermöglicht wurden, als auch eine Webseite, auf der Unterstützungsangebote zur psychologischen Selbsthilfe zusammengetragen wurden. Die Inanspruchnahme und Nutzung der Hilfsangebote wurde wissenschaftlich begleitet.", "Das Hotlineprojekt wurde auf Initiative des Zentralinstituts für Seelische Gesundheit (ZI) und des Ministeriums für Soziales und Integration Baden-Württemberg in Zusammenarbeit mit der Landespsychotherapeutenkammer, der Landesärztekammer sowie der Kassenärztlichen Vereinigung Baden-Württemberg eingerichtet. Für die Hotline wurde eine zentrale Rufnummer des Landes Baden-Württemberg bereitgestellt. Die Hotline war zwischen dem 22.04. 2020 und dem 24.07.2020 täglich von 08:00 bis 20:00 Uhr erreichbar. Im Juli 2020 wurde die Hotline aufgrund der zwischenzeitlichen Entspannung der Corona-Situation und der nur noch marginalen Beteiligung der Berater*innen eingestellt.\nRekrutierung der Berater*innen Als potenzielle Berater*innen wurden durch die Landespsychotherapeutenkammer und die Landesärztekammer approbierte ärztliche und psychologische Psychotherapeut*innen sowie Kinder- und Jugendpsychotherapeut*innen per E‑Mail kontaktiert und um Beteiligung gebeten. Über staatlich anerkannte Ausbildungsinstitute wurden psychologische und Kinder- und Jugendpsychologische Psychotherapeut*innen in Ausbildung kontaktiert. Über das Sozialministerium wurden zudem erfahrene Sozialarbeiter*innen aus der gemeindepsychiatrischen Versorgung kontaktiert. Alle potenziellen Berater*innen erhielten ein Informationsschreiben, in dem sie über Ziel und Zweck der Hotline, den Ablauf der Registrierung und der Beratungen, der Freiwilligkeit ihrer Teilnahme sowie die bestehenden Datenschutzbestimmungen aufgeklärt wurden. Die Berater*innen konnten sich anschließend freiwillig über einen URL-Link registrieren. Nachdem die personale und berufliche Identität der Berater*innen geprüft wurde, wurden ihnen die Einwahldaten für die Hotline übersendet und ihre Telefonnummern zur Einwahl in die Hotline zugelassen. Die Berater*innen konnten sich anschließend zeitlich und örtlich flexibel mit ihrem eigenen Endgerät in die Hotline einwählen und so eingehende Anrufe überstellt bekommen. Über eine telefonische Direktschaltung mit der Kassenärztlichen Vereinigung war es den Berater*innen im Bedarfsfall möglich, den Anrufenden einen raschen Zugang zu psychotherapeutischer Behandlung zu ermöglichen. Insgesamt registrierten sich 753 Berater*innen, unter ihnen 469 ärztliche und psychologische sowie Kinder- und Jugendpsychotherapeut*innen, 107 Psychotherapeut*innen in Ausbildung, 160 Sozialarbeiter*innen im ambulant betreuten Wohnen und 17 Personen mit anderen psychosozialen Berufen.\nAls potenzielle Berater*innen wurden durch die Landespsychotherapeutenkammer und die Landesärztekammer approbierte ärztliche und psychologische Psychotherapeut*innen sowie Kinder- und Jugendpsychotherapeut*innen per E‑Mail kontaktiert und um Beteiligung gebeten. Über staatlich anerkannte Ausbildungsinstitute wurden psychologische und Kinder- und Jugendpsychologische Psychotherapeut*innen in Ausbildung kontaktiert. Über das Sozialministerium wurden zudem erfahrene Sozialarbeiter*innen aus der gemeindepsychiatrischen Versorgung kontaktiert. Alle potenziellen Berater*innen erhielten ein Informationsschreiben, in dem sie über Ziel und Zweck der Hotline, den Ablauf der Registrierung und der Beratungen, der Freiwilligkeit ihrer Teilnahme sowie die bestehenden Datenschutzbestimmungen aufgeklärt wurden. Die Berater*innen konnten sich anschließend freiwillig über einen URL-Link registrieren. Nachdem die personale und berufliche Identität der Berater*innen geprüft wurde, wurden ihnen die Einwahldaten für die Hotline übersendet und ihre Telefonnummern zur Einwahl in die Hotline zugelassen. Die Berater*innen konnten sich anschließend zeitlich und örtlich flexibel mit ihrem eigenen Endgerät in die Hotline einwählen und so eingehende Anrufe überstellt bekommen. Über eine telefonische Direktschaltung mit der Kassenärztlichen Vereinigung war es den Berater*innen im Bedarfsfall möglich, den Anrufenden einen raschen Zugang zu psychotherapeutischer Behandlung zu ermöglichen. Insgesamt registrierten sich 753 Berater*innen, unter ihnen 469 ärztliche und psychologische sowie Kinder- und Jugendpsychotherapeut*innen, 107 Psychotherapeut*innen in Ausbildung, 160 Sozialarbeiter*innen im ambulant betreuten Wohnen und 17 Personen mit anderen psychosozialen Berufen.\nImplementierung der Webseite Zusätzlich wurde eine Webseite mit Inhalten zur psychologischen Selbsthilfe eingerichtet. Betroffene konnten dort Tipps zu den folgenden Themen erhalten: „Psychische Krise & Gewalt“, „Tipps für den Alltag“, „Umgang mit Belastung und Stress“, „Sorgen und Ängste“, „Schlafprobleme“, „Mit sich alleine sein“, „Ärger & Konflikte“, „Familienleben“. Die Webseite ist unter der Adresse www.psyhotline-corona-bw.de weiterhin erreichbar.\nZusätzlich wurde eine Webseite mit Inhalten zur psychologischen Selbsthilfe eingerichtet. Betroffene konnten dort Tipps zu den folgenden Themen erhalten: „Psychische Krise & Gewalt“, „Tipps für den Alltag“, „Umgang mit Belastung und Stress“, „Sorgen und Ängste“, „Schlafprobleme“, „Mit sich alleine sein“, „Ärger & Konflikte“, „Familienleben“. Die Webseite ist unter der Adresse www.psyhotline-corona-bw.de weiterhin erreichbar.\nDatenerhebung Das Projekt wurde durch Mitarbeiter*innen des ZI wissenschaftlich begleitet. Hierzu wurden Daten zur Inanspruchnahme der Hotline und zu den Beratungsgesprächen erhoben. Alle Datenerhebungen erfolgten vollständig anonym. Ein positives Ethikvotum der Ethikkommission II der Universität Heidelberg liegt vor (2020559N).\nDokumentation der Beratungen Die Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nDie Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nAbschließende Berater*innenbefragung In einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nIn einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nDas Projekt wurde durch Mitarbeiter*innen des ZI wissenschaftlich begleitet. Hierzu wurden Daten zur Inanspruchnahme der Hotline und zu den Beratungsgesprächen erhoben. Alle Datenerhebungen erfolgten vollständig anonym. Ein positives Ethikvotum der Ethikkommission II der Universität Heidelberg liegt vor (2020559N).\nDokumentation der Beratungen Die Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nDie Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nAbschließende Berater*innenbefragung In einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nIn einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nStatistische Analysen Die Dokumentationen der Beratungen wurden deskriptiv ausgewertet. Die Einschätzungen der Berater*innen, den Anrufenden helfen zu können, wurde in Abhängigkeit der vorliegenden Symptomatik mithilfe von t‑Tests für unabhängige Stichproben inferenzstatistisch analysiert. Hierbei wurden die Angaben der Beratungsgespräche, in denen die jeweilige Symptomatik berichtet wurde, mit den Angaben der übrigen Beratungsgespräche verglichen, in denen die jeweilige Symptomatik nicht berichtet wurde (1 = liegt vor, 0 = liegt nicht vor). Um einen möglichen Zusammenhang zwischen der Inanspruchnahme der Hotline und dem Infektionsgeschehen in Deutschland zu analysieren, wurde die vom Robert-Koch-Institut veröffentlichte Anzahl der Neuinfektionen pro Tag in die Analysen mit einbezogen.\nDie Dokumentationen der Beratungen wurden deskriptiv ausgewertet. Die Einschätzungen der Berater*innen, den Anrufenden helfen zu können, wurde in Abhängigkeit der vorliegenden Symptomatik mithilfe von t‑Tests für unabhängige Stichproben inferenzstatistisch analysiert. Hierbei wurden die Angaben der Beratungsgespräche, in denen die jeweilige Symptomatik berichtet wurde, mit den Angaben der übrigen Beratungsgespräche verglichen, in denen die jeweilige Symptomatik nicht berichtet wurde (1 = liegt vor, 0 = liegt nicht vor). Um einen möglichen Zusammenhang zwischen der Inanspruchnahme der Hotline und dem Infektionsgeschehen in Deutschland zu analysieren, wurde die vom Robert-Koch-Institut veröffentlichte Anzahl der Neuinfektionen pro Tag in die Analysen mit einbezogen.", "Als potenzielle Berater*innen wurden durch die Landespsychotherapeutenkammer und die Landesärztekammer approbierte ärztliche und psychologische Psychotherapeut*innen sowie Kinder- und Jugendpsychotherapeut*innen per E‑Mail kontaktiert und um Beteiligung gebeten. Über staatlich anerkannte Ausbildungsinstitute wurden psychologische und Kinder- und Jugendpsychologische Psychotherapeut*innen in Ausbildung kontaktiert. Über das Sozialministerium wurden zudem erfahrene Sozialarbeiter*innen aus der gemeindepsychiatrischen Versorgung kontaktiert. Alle potenziellen Berater*innen erhielten ein Informationsschreiben, in dem sie über Ziel und Zweck der Hotline, den Ablauf der Registrierung und der Beratungen, der Freiwilligkeit ihrer Teilnahme sowie die bestehenden Datenschutzbestimmungen aufgeklärt wurden. Die Berater*innen konnten sich anschließend freiwillig über einen URL-Link registrieren. Nachdem die personale und berufliche Identität der Berater*innen geprüft wurde, wurden ihnen die Einwahldaten für die Hotline übersendet und ihre Telefonnummern zur Einwahl in die Hotline zugelassen. Die Berater*innen konnten sich anschließend zeitlich und örtlich flexibel mit ihrem eigenen Endgerät in die Hotline einwählen und so eingehende Anrufe überstellt bekommen. Über eine telefonische Direktschaltung mit der Kassenärztlichen Vereinigung war es den Berater*innen im Bedarfsfall möglich, den Anrufenden einen raschen Zugang zu psychotherapeutischer Behandlung zu ermöglichen. Insgesamt registrierten sich 753 Berater*innen, unter ihnen 469 ärztliche und psychologische sowie Kinder- und Jugendpsychotherapeut*innen, 107 Psychotherapeut*innen in Ausbildung, 160 Sozialarbeiter*innen im ambulant betreuten Wohnen und 17 Personen mit anderen psychosozialen Berufen.", "Zusätzlich wurde eine Webseite mit Inhalten zur psychologischen Selbsthilfe eingerichtet. Betroffene konnten dort Tipps zu den folgenden Themen erhalten: „Psychische Krise & Gewalt“, „Tipps für den Alltag“, „Umgang mit Belastung und Stress“, „Sorgen und Ängste“, „Schlafprobleme“, „Mit sich alleine sein“, „Ärger & Konflikte“, „Familienleben“. Die Webseite ist unter der Adresse www.psyhotline-corona-bw.de weiterhin erreichbar.", "Das Projekt wurde durch Mitarbeiter*innen des ZI wissenschaftlich begleitet. Hierzu wurden Daten zur Inanspruchnahme der Hotline und zu den Beratungsgesprächen erhoben. Alle Datenerhebungen erfolgten vollständig anonym. Ein positives Ethikvotum der Ethikkommission II der Universität Heidelberg liegt vor (2020559N).\nDokumentation der Beratungen Die Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nDie Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nAbschließende Berater*innenbefragung In einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nIn einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.", "Die Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).", "In einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.", "Die Dokumentationen der Beratungen wurden deskriptiv ausgewertet. Die Einschätzungen der Berater*innen, den Anrufenden helfen zu können, wurde in Abhängigkeit der vorliegenden Symptomatik mithilfe von t‑Tests für unabhängige Stichproben inferenzstatistisch analysiert. Hierbei wurden die Angaben der Beratungsgespräche, in denen die jeweilige Symptomatik berichtet wurde, mit den Angaben der übrigen Beratungsgespräche verglichen, in denen die jeweilige Symptomatik nicht berichtet wurde (1 = liegt vor, 0 = liegt nicht vor). Um einen möglichen Zusammenhang zwischen der Inanspruchnahme der Hotline und dem Infektionsgeschehen in Deutschland zu analysieren, wurde die vom Robert-Koch-Institut veröffentlichte Anzahl der Neuinfektionen pro Tag in die Analysen mit einbezogen.", "Insgesamt war die Hotline über 13 Wochen vom 22.04. bis zum 24.07.2020 aktiv geschaltet. In diesem Zeitraum wurden insgesamt 8577 Anrufe registriert (Abb. 1). Die meisten Anrufe (27 %) gingen abends zwischen 18:00 und 20:00 Uhr ein. Es zeigte sich kein signifikanter Zusammenhang zwischen der Anzahl der eingehenden Anrufe und der Anzahl der Neuinfektionen pro Tag (r = −0,02, p = 0,983). Die Dauer eines Beratungsgesprächs betrug durchschnittlich 23,7 min (SD = 17,1). Insgesamt konnten 481 Anrufe (5,6 %) aufgrund eines vorübergehenden Beratermangels nicht durchgestellt werden, davon lassen sich 121 Anrufe auf technische Probleme an einem singulären Tag zurückführen. Die Anzahl der nicht durchgestellten Anrufe zeigte sich besonders hoch in den Morgenstunden zwischen 08:00 bis 09:00 Uhr (79 Anrufe), in den Mittagsstunden zwischen 12:00 bis 13:00 Uhr (77 Anrufe) und in den Abendstunden zwischen 18:00 bis 19:00 Uhr (95 Anrufe). Somit konnten insgesamt 8096 eingehende Anrufe durchgestellt werden.\nDokumentation der Beratungsgespräche Von den 8096 durchgestellten Anrufen wurden insgesamt 1292 Telefonberatungen dokumentiert (16 %). Demnach, waren 64 % der Anrufenden weiblich und die große Mehrheit der Anrufenden erwachsen (79 %), gefolgt von Anrufenden höheren Alters (17 %). Nur ein kleiner Teil der Anrufenden waren Jugendliche (3,8 %) oder Kinder (0,2 %). Mehr als jeder zweite dokumentierte Anruf stand in Zusammenhang mit einer vorbestehenden psychischen Erkrankung (55 %), gefolgt von Anrufen aufgrund von Quarantänemaßnahmen oder sozialer Isolation (42 %), Problemen im Alltag (30 %) oder Corona-spezifischen Ängsten (25 %). Nur ein kleiner Teil der Anrufe bezog sich auf häusliche Gewalt (3 %; Abb. 2).\nDie mit Abstand am häufigsten dokumentierte Symptomatik waren depressive Symptome (36 %), gefolgt von psychotischen Symptomen (19 %) und Angstsymptomen (18 %). 15 % der dokumentierten Anrufe berichteten eine unklare Symptomatik. Nur ein kleiner Teil (4 %) berichtete Suizidgedanken oder -absichten (Abb. 3). Die durchschnittliche Belastung der Anrufenden lag bei M = 3,3 (SD = 0,8) auf einer Skala von 1 bis 5.\nIn 42 % aller Beratungen gaben die Berater*innen an, kurze therapeutische Interventionen durchgeführt zu haben. Diese umfassten vor allem unterstützende Gesprächstechniken (z. B. Validierung, supportive Techniken), Anleitung zu Entspannung (z. B. Atemübungen), emotionsfokussierte Interventionen (z. B. Benennen und Sortieren von Gefühlen, entgegengesetztes Handeln), kognitive Interventionen (z. B. hilfreiche Sätze, Perspektivwechsel, Grübelmanagement), verhaltensbezogene Interventionen (z. B. Tagesstruktur, positive Aktivitäten), psychoedukative Interventionen (z. B. Schlafhygiene), Unterstützung bei Problemlösung (z. B. Konfliktberatung) sowie Ressourcenaktivierung. In 26 % aller dokumentierten Beratungen wurde eine psychotherapeutische Behandlung empfohlen, in 11 % aller Beratungen auf weitere, spezialisierte Telefonangebote verwiesen. In 16 % der Beratungen wurden Informationen zu Corona vermittelt. Nur in wenigen Beratungen (2 %) war ein akutes Krisencoaching notwendig (z. B. aufgrund akuter Suizidalität).\nInsgesamt gaben die Berater*innen an, dass sie in 27 % aller Gespräche „(sehr) viel“, in 40 % „etwas“ und in 33 % „nur wenig“ oder „gar nicht“ haben helfen können. Dabei zeigte sich, dass die Berater*innen in Beratungsgesprächen, bei denen ein klarer Anrufgrund vorlag (z. B. Corona-Angst, Isolation und Quarantäne, Schwierigkeit im Alltag oder Gewalt), signifikant mehr helfen konnten, im Vergleich zu den übrigen Beratungsgesprächen, in denen diese Anrufgründe nicht vorlagen (Tab. 1). In Beratungsgesprächen, in denen der Anrufgrund unklar war, konnten die Berater*innen im Vergleich zu den übrigen Beratungsgesprächen signifikant weniger helfen (M = 1,84). Ein ähnliches Bild zeigte sich auf der Symptomebene. Beim Vorliegen einer klaren Symptomatik wie Depressions‑, Angst‑, Zwangssymptomen, Gewalterleben oder PTBS konnten die Berater*innen signifikant mehr helfen (M = 2,98–3,34) als bei den übrigen Beratungen, in denen die jeweilige Symptomatik nicht vorlag. Signifikant weniger helfen konnten die Berater*innen hingegen im Vergleich zu den übrigen Beratungsgesprächen beim Vorliegen von psychotischen Symptomen, Störungen auf Persönlichkeitsebene oder bei unklarer Symptomatik (M = 2,31–2,50). Keine Unterschiede zeigten sich zwischen Beratungsgesprächen, in denen akute Suizidalität, Essstörungen oder Gewaltausübungen vorlagen (M = 2,71–3,27) im Vergleich zu den übrigen Beratungsgesprächen.Grund/Symptom liegt vorGrund/Symptom liegt nicht vort-TestEffektMSDnMSDnt-WertDfp-WertdGrund für AnrufCorona-Angst3,300,873152,671,0690010,32664<0,0010,61Quarantäne3,060,945352,651,106806,931205<0,0010,39Psych. Erkrankung2,791,026922,891,10523−1,7110770,087−0,10Schwierigkeit Alltag2,970,973692,771,088463,087740,0020,18Häusliche Gewalt3,240,86372,821,0511782,92390,0060,40Unklar1,841,02942,921,011121−9,83109<0,001−1,06SymptomeSuizidalität2,710,94492,841,061166−0,90530,372−0,12Depression2,980,944562,741,107594,021078<0,0010,23Angst3,320,792212,721,079949,49423<0,0010,58Zwang3,180,90342,821,0511812,24360,0320,34Essverhalten3,270,90112,831,0512041,62100,1360,42Psychose/Wahn2,421,032432,941,03972−6,93374<0,001−0,50Gewaltausübung2,930,83142,831,0512010,43140,6720,09Gewalterleben3,330,89402,821,0511753,5443<0,0010,49Posttraumatische Belastungsstörungen3,340,89382,821,0511773,6141<0,0010,5Persönlichkeit2,501,001812,891,051034−4,76254<0,001−0,37Unklar2,311,061752,921,021040−7,11232<0,001−0,60Restkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)M Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nRestkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)\nM Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nMit einem Mittelwert von 1,9 (SD = 0,9) auf einer Skala von 1 (gar nicht) bis 5 (sehr), empfanden die Berater*innen die Gespräche im Durchschnitt nur wenig belastend. Nur 5 % der Gespräche wurden als (sehr) stark belastend empfunden.\nVon den 8096 durchgestellten Anrufen wurden insgesamt 1292 Telefonberatungen dokumentiert (16 %). Demnach, waren 64 % der Anrufenden weiblich und die große Mehrheit der Anrufenden erwachsen (79 %), gefolgt von Anrufenden höheren Alters (17 %). Nur ein kleiner Teil der Anrufenden waren Jugendliche (3,8 %) oder Kinder (0,2 %). Mehr als jeder zweite dokumentierte Anruf stand in Zusammenhang mit einer vorbestehenden psychischen Erkrankung (55 %), gefolgt von Anrufen aufgrund von Quarantänemaßnahmen oder sozialer Isolation (42 %), Problemen im Alltag (30 %) oder Corona-spezifischen Ängsten (25 %). Nur ein kleiner Teil der Anrufe bezog sich auf häusliche Gewalt (3 %; Abb. 2).\nDie mit Abstand am häufigsten dokumentierte Symptomatik waren depressive Symptome (36 %), gefolgt von psychotischen Symptomen (19 %) und Angstsymptomen (18 %). 15 % der dokumentierten Anrufe berichteten eine unklare Symptomatik. Nur ein kleiner Teil (4 %) berichtete Suizidgedanken oder -absichten (Abb. 3). Die durchschnittliche Belastung der Anrufenden lag bei M = 3,3 (SD = 0,8) auf einer Skala von 1 bis 5.\nIn 42 % aller Beratungen gaben die Berater*innen an, kurze therapeutische Interventionen durchgeführt zu haben. Diese umfassten vor allem unterstützende Gesprächstechniken (z. B. Validierung, supportive Techniken), Anleitung zu Entspannung (z. B. Atemübungen), emotionsfokussierte Interventionen (z. B. Benennen und Sortieren von Gefühlen, entgegengesetztes Handeln), kognitive Interventionen (z. B. hilfreiche Sätze, Perspektivwechsel, Grübelmanagement), verhaltensbezogene Interventionen (z. B. Tagesstruktur, positive Aktivitäten), psychoedukative Interventionen (z. B. Schlafhygiene), Unterstützung bei Problemlösung (z. B. Konfliktberatung) sowie Ressourcenaktivierung. In 26 % aller dokumentierten Beratungen wurde eine psychotherapeutische Behandlung empfohlen, in 11 % aller Beratungen auf weitere, spezialisierte Telefonangebote verwiesen. In 16 % der Beratungen wurden Informationen zu Corona vermittelt. Nur in wenigen Beratungen (2 %) war ein akutes Krisencoaching notwendig (z. B. aufgrund akuter Suizidalität).\nInsgesamt gaben die Berater*innen an, dass sie in 27 % aller Gespräche „(sehr) viel“, in 40 % „etwas“ und in 33 % „nur wenig“ oder „gar nicht“ haben helfen können. Dabei zeigte sich, dass die Berater*innen in Beratungsgesprächen, bei denen ein klarer Anrufgrund vorlag (z. B. Corona-Angst, Isolation und Quarantäne, Schwierigkeit im Alltag oder Gewalt), signifikant mehr helfen konnten, im Vergleich zu den übrigen Beratungsgesprächen, in denen diese Anrufgründe nicht vorlagen (Tab. 1). In Beratungsgesprächen, in denen der Anrufgrund unklar war, konnten die Berater*innen im Vergleich zu den übrigen Beratungsgesprächen signifikant weniger helfen (M = 1,84). Ein ähnliches Bild zeigte sich auf der Symptomebene. Beim Vorliegen einer klaren Symptomatik wie Depressions‑, Angst‑, Zwangssymptomen, Gewalterleben oder PTBS konnten die Berater*innen signifikant mehr helfen (M = 2,98–3,34) als bei den übrigen Beratungen, in denen die jeweilige Symptomatik nicht vorlag. Signifikant weniger helfen konnten die Berater*innen hingegen im Vergleich zu den übrigen Beratungsgesprächen beim Vorliegen von psychotischen Symptomen, Störungen auf Persönlichkeitsebene oder bei unklarer Symptomatik (M = 2,31–2,50). Keine Unterschiede zeigten sich zwischen Beratungsgesprächen, in denen akute Suizidalität, Essstörungen oder Gewaltausübungen vorlagen (M = 2,71–3,27) im Vergleich zu den übrigen Beratungsgesprächen.Grund/Symptom liegt vorGrund/Symptom liegt nicht vort-TestEffektMSDnMSDnt-WertDfp-WertdGrund für AnrufCorona-Angst3,300,873152,671,0690010,32664<0,0010,61Quarantäne3,060,945352,651,106806,931205<0,0010,39Psych. Erkrankung2,791,026922,891,10523−1,7110770,087−0,10Schwierigkeit Alltag2,970,973692,771,088463,087740,0020,18Häusliche Gewalt3,240,86372,821,0511782,92390,0060,40Unklar1,841,02942,921,011121−9,83109<0,001−1,06SymptomeSuizidalität2,710,94492,841,061166−0,90530,372−0,12Depression2,980,944562,741,107594,021078<0,0010,23Angst3,320,792212,721,079949,49423<0,0010,58Zwang3,180,90342,821,0511812,24360,0320,34Essverhalten3,270,90112,831,0512041,62100,1360,42Psychose/Wahn2,421,032432,941,03972−6,93374<0,001−0,50Gewaltausübung2,930,83142,831,0512010,43140,6720,09Gewalterleben3,330,89402,821,0511753,5443<0,0010,49Posttraumatische Belastungsstörungen3,340,89382,821,0511773,6141<0,0010,5Persönlichkeit2,501,001812,891,051034−4,76254<0,001−0,37Unklar2,311,061752,921,021040−7,11232<0,001−0,60Restkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)M Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nRestkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)\nM Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nMit einem Mittelwert von 1,9 (SD = 0,9) auf einer Skala von 1 (gar nicht) bis 5 (sehr), empfanden die Berater*innen die Gespräche im Durchschnitt nur wenig belastend. Nur 5 % der Gespräche wurden als (sehr) stark belastend empfunden.\nInanspruchnahme der Webseite Die begleitende Webseite verzeichnete im selben Zeitraum, zwischen März und Juli 2020, insgesamt 5544 Aufrufe, also durchschnittlich 656 Aufrufe pro Tag (Abb. 4). Die Aufrufe auf der Webseite verteilten sich gleichmäßig über alle Rubriken hinweg.\nDie begleitende Webseite verzeichnete im selben Zeitraum, zwischen März und Juli 2020, insgesamt 5544 Aufrufe, also durchschnittlich 656 Aufrufe pro Tag (Abb. 4). Die Aufrufe auf der Webseite verteilten sich gleichmäßig über alle Rubriken hinweg.\nAbschließende Berater*innenbefragung In der abschließenden Berater*innenbefragung zeigte sich eine hohe Zufriedenheit der Berater*innen mit dem Projekt. Auf einer 5‑stufigen Skala (1 = überhaupt nicht; 5 = sehr) bewerteten die Berater*innen das Hotlineprojekt durchschnittlich als sinnvoll (M = 4,03, SD = 1,06), die zugehörige Webseite als hilfreich (M = 4,02, SD = 1,04) und die eigene Beteiligung als teilweise befriedigend (M = 3,34, SD = 1,04).\nIn der abschließenden Berater*innenbefragung zeigte sich eine hohe Zufriedenheit der Berater*innen mit dem Projekt. Auf einer 5‑stufigen Skala (1 = überhaupt nicht; 5 = sehr) bewerteten die Berater*innen das Hotlineprojekt durchschnittlich als sinnvoll (M = 4,03, SD = 1,06), die zugehörige Webseite als hilfreich (M = 4,02, SD = 1,04) und die eigene Beteiligung als teilweise befriedigend (M = 3,34, SD = 1,04).", "Von den 8096 durchgestellten Anrufen wurden insgesamt 1292 Telefonberatungen dokumentiert (16 %). Demnach, waren 64 % der Anrufenden weiblich und die große Mehrheit der Anrufenden erwachsen (79 %), gefolgt von Anrufenden höheren Alters (17 %). Nur ein kleiner Teil der Anrufenden waren Jugendliche (3,8 %) oder Kinder (0,2 %). Mehr als jeder zweite dokumentierte Anruf stand in Zusammenhang mit einer vorbestehenden psychischen Erkrankung (55 %), gefolgt von Anrufen aufgrund von Quarantänemaßnahmen oder sozialer Isolation (42 %), Problemen im Alltag (30 %) oder Corona-spezifischen Ängsten (25 %). Nur ein kleiner Teil der Anrufe bezog sich auf häusliche Gewalt (3 %; Abb. 2).\nDie mit Abstand am häufigsten dokumentierte Symptomatik waren depressive Symptome (36 %), gefolgt von psychotischen Symptomen (19 %) und Angstsymptomen (18 %). 15 % der dokumentierten Anrufe berichteten eine unklare Symptomatik. Nur ein kleiner Teil (4 %) berichtete Suizidgedanken oder -absichten (Abb. 3). Die durchschnittliche Belastung der Anrufenden lag bei M = 3,3 (SD = 0,8) auf einer Skala von 1 bis 5.\nIn 42 % aller Beratungen gaben die Berater*innen an, kurze therapeutische Interventionen durchgeführt zu haben. Diese umfassten vor allem unterstützende Gesprächstechniken (z. B. Validierung, supportive Techniken), Anleitung zu Entspannung (z. B. Atemübungen), emotionsfokussierte Interventionen (z. B. Benennen und Sortieren von Gefühlen, entgegengesetztes Handeln), kognitive Interventionen (z. B. hilfreiche Sätze, Perspektivwechsel, Grübelmanagement), verhaltensbezogene Interventionen (z. B. Tagesstruktur, positive Aktivitäten), psychoedukative Interventionen (z. B. Schlafhygiene), Unterstützung bei Problemlösung (z. B. Konfliktberatung) sowie Ressourcenaktivierung. In 26 % aller dokumentierten Beratungen wurde eine psychotherapeutische Behandlung empfohlen, in 11 % aller Beratungen auf weitere, spezialisierte Telefonangebote verwiesen. In 16 % der Beratungen wurden Informationen zu Corona vermittelt. Nur in wenigen Beratungen (2 %) war ein akutes Krisencoaching notwendig (z. B. aufgrund akuter Suizidalität).\nInsgesamt gaben die Berater*innen an, dass sie in 27 % aller Gespräche „(sehr) viel“, in 40 % „etwas“ und in 33 % „nur wenig“ oder „gar nicht“ haben helfen können. Dabei zeigte sich, dass die Berater*innen in Beratungsgesprächen, bei denen ein klarer Anrufgrund vorlag (z. B. Corona-Angst, Isolation und Quarantäne, Schwierigkeit im Alltag oder Gewalt), signifikant mehr helfen konnten, im Vergleich zu den übrigen Beratungsgesprächen, in denen diese Anrufgründe nicht vorlagen (Tab. 1). In Beratungsgesprächen, in denen der Anrufgrund unklar war, konnten die Berater*innen im Vergleich zu den übrigen Beratungsgesprächen signifikant weniger helfen (M = 1,84). Ein ähnliches Bild zeigte sich auf der Symptomebene. Beim Vorliegen einer klaren Symptomatik wie Depressions‑, Angst‑, Zwangssymptomen, Gewalterleben oder PTBS konnten die Berater*innen signifikant mehr helfen (M = 2,98–3,34) als bei den übrigen Beratungen, in denen die jeweilige Symptomatik nicht vorlag. Signifikant weniger helfen konnten die Berater*innen hingegen im Vergleich zu den übrigen Beratungsgesprächen beim Vorliegen von psychotischen Symptomen, Störungen auf Persönlichkeitsebene oder bei unklarer Symptomatik (M = 2,31–2,50). Keine Unterschiede zeigten sich zwischen Beratungsgesprächen, in denen akute Suizidalität, Essstörungen oder Gewaltausübungen vorlagen (M = 2,71–3,27) im Vergleich zu den übrigen Beratungsgesprächen.Grund/Symptom liegt vorGrund/Symptom liegt nicht vort-TestEffektMSDnMSDnt-WertDfp-WertdGrund für AnrufCorona-Angst3,300,873152,671,0690010,32664<0,0010,61Quarantäne3,060,945352,651,106806,931205<0,0010,39Psych. Erkrankung2,791,026922,891,10523−1,7110770,087−0,10Schwierigkeit Alltag2,970,973692,771,088463,087740,0020,18Häusliche Gewalt3,240,86372,821,0511782,92390,0060,40Unklar1,841,02942,921,011121−9,83109<0,001−1,06SymptomeSuizidalität2,710,94492,841,061166−0,90530,372−0,12Depression2,980,944562,741,107594,021078<0,0010,23Angst3,320,792212,721,079949,49423<0,0010,58Zwang3,180,90342,821,0511812,24360,0320,34Essverhalten3,270,90112,831,0512041,62100,1360,42Psychose/Wahn2,421,032432,941,03972−6,93374<0,001−0,50Gewaltausübung2,930,83142,831,0512010,43140,6720,09Gewalterleben3,330,89402,821,0511753,5443<0,0010,49Posttraumatische Belastungsstörungen3,340,89382,821,0511773,6141<0,0010,5Persönlichkeit2,501,001812,891,051034−4,76254<0,001−0,37Unklar2,311,061752,921,021040−7,11232<0,001−0,60Restkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)M Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nRestkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)\nM Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nMit einem Mittelwert von 1,9 (SD = 0,9) auf einer Skala von 1 (gar nicht) bis 5 (sehr), empfanden die Berater*innen die Gespräche im Durchschnitt nur wenig belastend. Nur 5 % der Gespräche wurden als (sehr) stark belastend empfunden.", "Die begleitende Webseite verzeichnete im selben Zeitraum, zwischen März und Juli 2020, insgesamt 5544 Aufrufe, also durchschnittlich 656 Aufrufe pro Tag (Abb. 4). Die Aufrufe auf der Webseite verteilten sich gleichmäßig über alle Rubriken hinweg.", "In der abschließenden Berater*innenbefragung zeigte sich eine hohe Zufriedenheit der Berater*innen mit dem Projekt. Auf einer 5‑stufigen Skala (1 = überhaupt nicht; 5 = sehr) bewerteten die Berater*innen das Hotlineprojekt durchschnittlich als sinnvoll (M = 4,03, SD = 1,06), die zugehörige Webseite als hilfreich (M = 4,02, SD = 1,04) und die eigene Beteiligung als teilweise befriedigend (M = 3,34, SD = 1,04).", "Ziel des Projekts war es, eine schnelle und professionelle psychologische Ersthilfe für Menschen mit psychischen Belastungen in der ersten Welle der Corona-Pandemie zu ermöglichen. Dazu wurden eine Hotline zur persönlichen Beratung und eine Webseite für Selbsthilfezwecke eingerichtet.\nMit 753 registrierten Berater*innen stieß das Hotlineprojekt auf eine große berufsgruppenübergreifende Zustimmung. Durch die hohe Anzahl professioneller Berater*innen konnte eine hohe Beratungsqualität sichergestellt werden, was in bisherigen Hotlineimplementierungen eine große Herausforderung darstellte [26].\nMit insgesamt 8577 Anrufen in 13 Wochen (ca. 700 Anrufe pro Woche) stieß die Hotline auf eine große Nachfrage in der Landesbevölkerung. Die Häufung der Anrufe in den Abendstunden zeigt die Wichtigkeit, diese Hilfsangebote auch außerhalb der regelhaften Arbeitszeit anzubieten [26]. Die Anzahl der eingehenden Anrufe zeigte keinen Zusammenhang mit der Anzahl der Neuinfektionen pro Tag. Dies könnte darauf zurückzuführen sein, dass die Hotline erst Ende April, kurz nach dem Höhepunkt der ersten Welle, in Betrieb genommen wurde. Die Anzahl der Neuinfektionen nahm zu diesem Zeitpunkt bereits kontinuierlich ab, während die Beschränkungsmaßnahmen nur langsam gelockert wurden. Wahrscheinlich ist auch, dass weitere Faktoren, wie die Bekanntheit der Hotline, einen wichtigen Einfluss auf die Anrufzahlen hatten.\nDie Dokumentationen der Berater*innen weisen darauf hin, dass jeder zweite Anruf von Personen getätigt wurde, die eine bereits erkannte psychische Störung angaben. Dies könnte mit besonderen Risiken dieser Personengruppe zusammenhängen, was sich mit Erkenntnissen aus weiteren Studien deckt. So zeigte sich, dass COVID-19-erkrankte Personen mit komorbiden psychischen Störungen im Vergleich zu COVID-19-Erkrankten ohne psychische Störung eine um 48 % erhöhte Mortalitätsrate aufweisen [29]. Diese Befunde unterstreichen die Wichtigkeit, dieser gefährdeten Gruppe besondere Aufmerksamkeit zu schenken [30]. Wie erwartet betrafen die weiteren Gründe für einen Anruf bei der Hotline vor allem Corona-spezifische Ängste, Belastungen aufgrund der Isolations- und Quarantänemaßnahmen oder Schwierigkeiten im Alltag (25–42 %). Der Anteil der Anrufe aufgrund häuslicher Gewalt lag mit 2,9 % vergleichsweise niedrig, obwohl diese auch in Deutschland unter Quarantänemaßnahmen signifikant zugenommen hat [21]. Dies kann auf mehrere Gründe zurückzuführen sein: Zum einen gibt es weitere etablierte und gut ausgebaute telefonische Hilfsangebote für diese spezifische Zielgruppe, zum anderen jedoch sind Hilfsangebote bei häuslicher Gewalt den Betroffenen in vielen Fällen nicht bekannt oder es ist ihnen nur eingeschränkt möglich, bei weitreichender Überwachung und Kontrolle durch eine Partner*in, diese telefonischen Hilfsangebote zu nutzen [21]. Dies legt nahe, dass Hilfsangebote in der Öffentlichkeit besser kommuniziert und unterschiedliche Zugänge zu diesen Angeboten geschaffen werden sollten (z. B. Online und Telefon).\nAuf Symptomebene zeigten sich hohe Anteile von Depressions- und Angstsymptomen sowie ein überraschend hoher Anteil psychotischer und wahnhafter Symptome. Möglicherweise trägt das niedrigschwellig und anonym erreichbare Angebot dazu bei, dass Psychosebetroffene ein psychotherapeutisches Beratungsangebot annehmen. Der hohe Anteil dieser Symptomgruppen deckt sich mit der Einschätzung der Bundespsychotherapeutenkammer, dass Depressionen, Angststörungen und Psychosen zu den psychischen Erkrankungen gehören, die besonders durch die Pandemie ausgelöst oder verstärkt werden können [23]. Der hohe Anteil von Depressions- und Angstsymptomen zeigt sich in Übereinstimmung mit der vergleichsweise hohen Prävalenz dieser Erkrankungen in der deutschen Allgemeinbevölkerung mit 9,3 % für eine affektive Störung und 15,3 % für eine Angststörung [31], während der hohe Anteil psychotischer Symptome den hohen Bedarf an psychotherapeutischen Hilfen für diese Personengruppe anzeigt. Symptome von Zwangsstörungen, posttraumatischer Belastungsstörung oder Substanzkonsum wurden hingegen mit 3–4 % vergleichsweise seltener berichtet.\nBemerkenswert erscheint die Vielzahl an Kurzinterventionen, die trotz der anonymen Anrufe akut telefonisch durchgeführt werden konnten. Diese umfassten unterstützende Gesprächstechniken, Anleitung zu Entspannung, emotionsfokussierte, kognitive und verhaltensbezogene Interventionen, psychoedukative Interventionen, Unterstützung bei Problemlösung sowie Ressourcenaktivierung. An dieser Stelle sei noch einmal die Relevanz psychotherapeutisch geschulter Berater*innen herausgestellt, die auch in kurzer Zeit im telefonischen Kontakt gezielt helfen können.\nDie Einschätzung der Berater*innen, wie sehr sie den Anrufenden haben helfen können, zeigt, dass diese bei etwa zwei Drittel aller Beratungsgespräche den Eindruck hatten, zumindest „etwas“, häufig aber auch „viel“ oder „sehr viel“ helfen zu können. Dabei konnten die Berater*innen vor allem dann helfen, wenn ein klarer Anrufgrund oder eine klare Symptomatik vorlag. Bei unklaren Anrufgründen und unklarer Symptomatik konnten die Berater*innen signifikant weniger helfen; insbesondere bei psychotischen Symptomen und Symptomen von Persönlichkeitsstörungen gaben die Berater*innen an, dass sie weniger helfen konnten. Dennoch ist denkbar, dass auch dieser Personengruppe ein Anstoß gegeben wurde, sich weiterhin um psychotherapeutische Hilfen zu bemühen.\nLimitationen Zur Gewährleistung der Anonymität und des Datenschutzes wurden die Anrufenden nicht direkt um Angaben zur Person oder zu den Beratungsgesprächen gebeten. Bei der Datenerhebung handelt es sich lediglich um freiwillige Angaben der Berater*innen. Aufgrund der Freiwilligkeit der Dokumentation wurden zudem lediglich 16 % der Beratungsgespräche dokumentiert (1292 von 8096 durchgestellten Anrufen). Die Ergebnisse sollten daher mit Vorsicht und die vorliegenden Symptome nicht als Feststellung einer gesicherten Diagnose interpretiert werden. Obwohl die Dokumentationen von Fachpersonal durchgeführt wurden, die langjährige Erfahrungen bei der Einschätzung psychischer Symptome und deren Behandlung aufweisen, stellen die Ergebnisse nur einen ersten Hinweis auf den Nutzen einer entsprechenden Hotline dar. Diesem sollte durch zusätzliche Befragungen der Anrufenden in Zukunft weiter nachgegangen werden.\nZur Gewährleistung der Anonymität und des Datenschutzes wurden die Anrufenden nicht direkt um Angaben zur Person oder zu den Beratungsgesprächen gebeten. Bei der Datenerhebung handelt es sich lediglich um freiwillige Angaben der Berater*innen. Aufgrund der Freiwilligkeit der Dokumentation wurden zudem lediglich 16 % der Beratungsgespräche dokumentiert (1292 von 8096 durchgestellten Anrufen). Die Ergebnisse sollten daher mit Vorsicht und die vorliegenden Symptome nicht als Feststellung einer gesicherten Diagnose interpretiert werden. Obwohl die Dokumentationen von Fachpersonal durchgeführt wurden, die langjährige Erfahrungen bei der Einschätzung psychischer Symptome und deren Behandlung aufweisen, stellen die Ergebnisse nur einen ersten Hinweis auf den Nutzen einer entsprechenden Hotline dar. Diesem sollte durch zusätzliche Befragungen der Anrufenden in Zukunft weiter nachgegangen werden.\nSchlussfolgerung Insgesamt leistete die psychologische Hotline einen wichtigen Beitrag zur Bewältigung psychischer Belastungen während der ersten Infektionswelle der Corona-Pandemie in Baden-Württemberg. Für viele Menschen, die in diesem Zusammenhang unter psychischen Belastungen litten, wurde so eine Möglichkeit zur schnellen und professionellen psychologischen Ersthilfe geschaffen. Ein besonderer Dank gebührt der Vielzahl ehrenamtlicher Berater*innen, die sich über verschiedene Berufsgruppen und Therapieschulen hinweg engagierten, um einen außerordentlich wichtigen gesellschaftlichen Beitrag zur Bewältigung der Pandemie zu leisten.\nInsgesamt leistete die psychologische Hotline einen wichtigen Beitrag zur Bewältigung psychischer Belastungen während der ersten Infektionswelle der Corona-Pandemie in Baden-Württemberg. Für viele Menschen, die in diesem Zusammenhang unter psychischen Belastungen litten, wurde so eine Möglichkeit zur schnellen und professionellen psychologischen Ersthilfe geschaffen. Ein besonderer Dank gebührt der Vielzahl ehrenamtlicher Berater*innen, die sich über verschiedene Berufsgruppen und Therapieschulen hinweg engagierten, um einen außerordentlich wichtigen gesellschaftlichen Beitrag zur Bewältigung der Pandemie zu leisten.", "Zur Gewährleistung der Anonymität und des Datenschutzes wurden die Anrufenden nicht direkt um Angaben zur Person oder zu den Beratungsgesprächen gebeten. Bei der Datenerhebung handelt es sich lediglich um freiwillige Angaben der Berater*innen. Aufgrund der Freiwilligkeit der Dokumentation wurden zudem lediglich 16 % der Beratungsgespräche dokumentiert (1292 von 8096 durchgestellten Anrufen). Die Ergebnisse sollten daher mit Vorsicht und die vorliegenden Symptome nicht als Feststellung einer gesicherten Diagnose interpretiert werden. Obwohl die Dokumentationen von Fachpersonal durchgeführt wurden, die langjährige Erfahrungen bei der Einschätzung psychischer Symptome und deren Behandlung aufweisen, stellen die Ergebnisse nur einen ersten Hinweis auf den Nutzen einer entsprechenden Hotline dar. Diesem sollte durch zusätzliche Befragungen der Anrufenden in Zukunft weiter nachgegangen werden.", "Insgesamt leistete die psychologische Hotline einen wichtigen Beitrag zur Bewältigung psychischer Belastungen während der ersten Infektionswelle der Corona-Pandemie in Baden-Württemberg. Für viele Menschen, die in diesem Zusammenhang unter psychischen Belastungen litten, wurde so eine Möglichkeit zur schnellen und professionellen psychologischen Ersthilfe geschaffen. Ein besonderer Dank gebührt der Vielzahl ehrenamtlicher Berater*innen, die sich über verschiedene Berufsgruppen und Therapieschulen hinweg engagierten, um einen außerordentlich wichtigen gesellschaftlichen Beitrag zur Bewältigung der Pandemie zu leisten." ]

[ null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null ]

[ "Hintergrund", "Ziel des Projekts", "Methodik", "Rekrutierung der Berater*innen", "Implementierung der Webseite", "Datenerhebung", "Dokumentation der Beratungen", "Abschließende Berater*innenbefragung", "Statistische Analysen", "Ergebnisse", "Dokumentation der Beratungsgespräche", "Inanspruchnahme der Webseite", "Abschließende Berater*innenbefragung", "Diskussion", "Limitationen", "Schlussfolgerung", "Fazit für die Praxis" ]

[ "Die Covid-19-Pandemie stellt Deutschland und die gesamte Welt vor große medizinische, wirtschaftliche, soziale und auch psychologische Herausforderungen. Erste Längsschnittstudien deuten auf eine Zunahme der psychischen Belastung in der Bevölkerung hin [1, 2]. Unter einer Vielzahl pandemiebedingter Belastungsfaktoren stehen vor allem die Angst vor Infektionen, die staatlich verordneten Maßnahmen zur Eindämmung der Pandemie, wirtschaftliche Existenzängste und arbeitsbezogene Belastungen im Vordergrund.\nDie Angst vor einer Infektion mit dem Coronavirus ist weit verbreitet. Demnach äußerten im März 2020, zum Höhepunkt der ersten Infektionswelle, 49 % aller Deutschen eine große oder sehr große Angst vor einer möglichen Infektion [3]. Eine deutsche Studie zeigte, dass die Corona-Angst zu Beginn der Pandemie schnell zunahm und mit einer späteren generalisierten Angst und Schlafstörungen einherging [4]. Die Angst zeigte sich besonders ausgeprägt bei vulnerablen Gruppen (z. B. Personen mit körperlichen Vorerkrankungen) oder aber bei Beschäftigten im Gesundheitswesen [5–7]. Dabei scheint es einen Zusammenhang zwischen der Angst vor einer Corona-Infektion mit psychischen Symptomen wie depressiven Beschwerden und suizidalen Gedanken sowie mit deutlichen psychosozialen Funktionseinschränkungen zu geben [8]. Neben der Angst, selbst durch das Coronavirus infiziert zu werden, bestehen ebenso Ängste, andere Personen anzustecken oder dass sich Personen im Freundes- und Familienkreis mit dem Virus infizieren könnten.\nDie Eindämmung der Pandemie erforderte gerade während der ersten Welle zudem eine Serie von gänzlich neuen, staatlich verordneten Verhaltensempfehlungen und Präventionsmaßnahmen. Durch Isolations- und Quarantänemaßnahmen waren viele Menschen gezwungen, über lange Zeit alleine oder auf enge Räume beschränkt zu leben. Außerdem waren soziale Kontakte erschwert, wodurch wichtige Unterstützung entfiel. Die Auswirkungen dieser einschränkenden Maßnahmen auf die psychische Gesundheit von Menschen werden seit Beginn der Pandemie kritisch diskutiert. Schon im März 2020 erschien eine Übersichtsarbeit, in der traumaähnliche Symptome, Verwirrung und Ärger als mögliche Folgen von Quarantänemaßnahmen beschrieben wurden [9]. Die Belastungssymptome können auch noch Monate nach den Quarantänemaßnahmen bestehen bleiben [10]. Doch nicht nur die verordnete Quarantäne führt zu psychischen Belastungen, auch schon weniger extreme Formen der sozialen Distanzierung wie die Reduktion von Kontakten oder die Empfehlung, das häusliche Umfeld nicht zu verlassen, zeigte sich mit vermehrten Depressionssymptomen, generalisierten Angstsymptomen, intrusiven Gedanken, Insomnie und akutem Stress assoziiert [11]. Gerade ältere Menschen, aber auch Jugendliche, leiden demnach an den Folgen der Einschränkungen und berichten eine Zunahme von Einsamkeit und psychischer Belastung während der Pandemie [1, 12].\nNicht zuletzt stellen die volkswirtschaftlichen Folgen der Pandemie eine besondere Herausforderung dar. So führen Arbeitsplatzunsicherheit und finanzielle Sorgen im Rahmen der Corona-Pandemie zu Beeinträchtigungen der psychischen Gesundheit [13]. Homeoffice-Regelungen und Schulschließungen führen zu einer radikalen Veränderung des Lebensalltags vieler Menschen und stellen bisher ungekannte Herausforderungen dar, Arbeit, Kinderbetreuung und Haushalt zu vereinen. Eine Studie über die Auswirkungen von Homeschooling in sieben europäischen Ländern (u. a. Deutschland) zeigte negative Auswirkungen von Homeschooling sowohl für Eltern als auch für Kinder in Form von vermehrtem Stress, Sorgen, sozialer Isolation, häuslicher Konflikte und teilweise auch erhöhtem Alkoholkonsum [14]. Nicht zuletzt lösen die wirtschaftlichen Restriktionen für viele Menschen existenzielle Ängste aus [15], insbesondere bei Beschäftigten in besonders betroffenen Branchen wie der Gastronomie‑, Freizeit- oder Kulturbranche.\nIn vielen Fällen können diese Belastungen zu dysfunktionalen Bewältigungsversuchen führen wie erhöhtem Alkohol- und Substanzkonsum [16], Aggressivität und Wutausbrüchen oder ein Anstieg von Gewalt im häuslichen Umfeld [17–19]. In den europäischen Ländern wurde demnach im April 2020 ein Anstieg von Notrufen aufgrund häuslicher Gewalt gegen Frauen um 60 % im Vergleich zum Vorjahr verzeichnet [20]. Eine repräsentative Studie in Deutschland zeigte, dass die häusliche Gewalt gegen Frauen und Kinder unter Quarantänebedingungen um das zwei‑ bis dreifache zunahm [21].\nVor diesem Hintergrund besteht eine ernste Gefahr, dass psychische Erkrankungen durch die Pandemie zunehmen werden [22]. Die Bundespsychotherapeutenkammer warnt, dass die Corona-Pandemie manifeste psychische Erkrankungen auslösen oder verstärken werde, und benennt dabei v. a. Depressionen und Angststörungen, akute und posttraumatische Belastungsstörungen (PTBS), Alkohol- und Medikamentenabhängigkeit, Zwangsstörungen und Psychosen [23].\nEs scheint daher von zentraler Bedeutung zu sein, Menschen bei der Bewältigung der psychischen Belastungen möglichst schnell, unkompliziert und früh zu unterstützen. Schon zu Beginn der Pandemie wurde daher auf die Wichtigkeit von Frühintervention hingewiesen [24]. Im Rahmen der Pandemie spielen dabei vor allem flexible und schnelle Hilfsangebote wie Online- oder Telefonhilfen eine wichtige Rolle, um die Menschen zu erreichen [25, 26]. Eine große Herausforderung besteht dabei, die Hotlinedienste mit ausreichend qualifizierten Berater*innen auszustatten [26].\nIn Deutschland konnten die gut etablierten Versorgungsstrukturen gerade zu Beginn der Pandemie nur bedingt greifen und pandemietaugliche Alternativkonzepte fehlten [27]. Daher gab es trotz der hohen Unsicherheit und subjektiv erlebten Bedrohung kaum Hilfsangebote, die eine unkomplizierte, schnelle und gleichzeitig therapeutisch professionelle Hilfe bei Corona-bedingten psychischen Belastungen sicherstellten. Aus Fachkreisen wurde daher darauf hingewiesen, dass eine Überlastung des bestehenden psychotherapeutischen und psychiatrischen Versorgungssystems, einschließlich der psychosozialen Dienste, wahrscheinlich sei und alternative Hilfsangebote wie Telefonhotlines dringend ausgebaut werden sollten [27, 28]. Aus dieser Situation heraus entstand das Vorhaben, die gegebenen psychotherapeutischen Versorgungsstrukturen zu nutzen, um ein professionelles, flexibles und niederschwelliges Angebot zur psychologischen Ersthilfe für die Allgemeinbevölkerung zu schaffen.\nZiel des Projekts Ziel des Projekts war es, Menschen mit psychischen Belastungen aufgrund der Corona-Pandemie mit einer schnellen, professionell ausgeübten Ersthilfe zu unterstützen. Dazu wurde sowohl eine Hotline eingerichtet, über die persönliche Beratungsgespräche durch Fachpersonen ermöglicht wurden, als auch eine Webseite, auf der Unterstützungsangebote zur psychologischen Selbsthilfe zusammengetragen wurden. Die Inanspruchnahme und Nutzung der Hilfsangebote wurde wissenschaftlich begleitet.\nZiel des Projekts war es, Menschen mit psychischen Belastungen aufgrund der Corona-Pandemie mit einer schnellen, professionell ausgeübten Ersthilfe zu unterstützen. Dazu wurde sowohl eine Hotline eingerichtet, über die persönliche Beratungsgespräche durch Fachpersonen ermöglicht wurden, als auch eine Webseite, auf der Unterstützungsangebote zur psychologischen Selbsthilfe zusammengetragen wurden. Die Inanspruchnahme und Nutzung der Hilfsangebote wurde wissenschaftlich begleitet.", "Ziel des Projekts war es, Menschen mit psychischen Belastungen aufgrund der Corona-Pandemie mit einer schnellen, professionell ausgeübten Ersthilfe zu unterstützen. Dazu wurde sowohl eine Hotline eingerichtet, über die persönliche Beratungsgespräche durch Fachpersonen ermöglicht wurden, als auch eine Webseite, auf der Unterstützungsangebote zur psychologischen Selbsthilfe zusammengetragen wurden. Die Inanspruchnahme und Nutzung der Hilfsangebote wurde wissenschaftlich begleitet.", "Das Hotlineprojekt wurde auf Initiative des Zentralinstituts für Seelische Gesundheit (ZI) und des Ministeriums für Soziales und Integration Baden-Württemberg in Zusammenarbeit mit der Landespsychotherapeutenkammer, der Landesärztekammer sowie der Kassenärztlichen Vereinigung Baden-Württemberg eingerichtet. Für die Hotline wurde eine zentrale Rufnummer des Landes Baden-Württemberg bereitgestellt. Die Hotline war zwischen dem 22.04. 2020 und dem 24.07.2020 täglich von 08:00 bis 20:00 Uhr erreichbar. Im Juli 2020 wurde die Hotline aufgrund der zwischenzeitlichen Entspannung der Corona-Situation und der nur noch marginalen Beteiligung der Berater*innen eingestellt.\nRekrutierung der Berater*innen Als potenzielle Berater*innen wurden durch die Landespsychotherapeutenkammer und die Landesärztekammer approbierte ärztliche und psychologische Psychotherapeut*innen sowie Kinder- und Jugendpsychotherapeut*innen per E‑Mail kontaktiert und um Beteiligung gebeten. Über staatlich anerkannte Ausbildungsinstitute wurden psychologische und Kinder- und Jugendpsychologische Psychotherapeut*innen in Ausbildung kontaktiert. Über das Sozialministerium wurden zudem erfahrene Sozialarbeiter*innen aus der gemeindepsychiatrischen Versorgung kontaktiert. Alle potenziellen Berater*innen erhielten ein Informationsschreiben, in dem sie über Ziel und Zweck der Hotline, den Ablauf der Registrierung und der Beratungen, der Freiwilligkeit ihrer Teilnahme sowie die bestehenden Datenschutzbestimmungen aufgeklärt wurden. Die Berater*innen konnten sich anschließend freiwillig über einen URL-Link registrieren. Nachdem die personale und berufliche Identität der Berater*innen geprüft wurde, wurden ihnen die Einwahldaten für die Hotline übersendet und ihre Telefonnummern zur Einwahl in die Hotline zugelassen. Die Berater*innen konnten sich anschließend zeitlich und örtlich flexibel mit ihrem eigenen Endgerät in die Hotline einwählen und so eingehende Anrufe überstellt bekommen. Über eine telefonische Direktschaltung mit der Kassenärztlichen Vereinigung war es den Berater*innen im Bedarfsfall möglich, den Anrufenden einen raschen Zugang zu psychotherapeutischer Behandlung zu ermöglichen. Insgesamt registrierten sich 753 Berater*innen, unter ihnen 469 ärztliche und psychologische sowie Kinder- und Jugendpsychotherapeut*innen, 107 Psychotherapeut*innen in Ausbildung, 160 Sozialarbeiter*innen im ambulant betreuten Wohnen und 17 Personen mit anderen psychosozialen Berufen.\nAls potenzielle Berater*innen wurden durch die Landespsychotherapeutenkammer und die Landesärztekammer approbierte ärztliche und psychologische Psychotherapeut*innen sowie Kinder- und Jugendpsychotherapeut*innen per E‑Mail kontaktiert und um Beteiligung gebeten. Über staatlich anerkannte Ausbildungsinstitute wurden psychologische und Kinder- und Jugendpsychologische Psychotherapeut*innen in Ausbildung kontaktiert. Über das Sozialministerium wurden zudem erfahrene Sozialarbeiter*innen aus der gemeindepsychiatrischen Versorgung kontaktiert. Alle potenziellen Berater*innen erhielten ein Informationsschreiben, in dem sie über Ziel und Zweck der Hotline, den Ablauf der Registrierung und der Beratungen, der Freiwilligkeit ihrer Teilnahme sowie die bestehenden Datenschutzbestimmungen aufgeklärt wurden. Die Berater*innen konnten sich anschließend freiwillig über einen URL-Link registrieren. Nachdem die personale und berufliche Identität der Berater*innen geprüft wurde, wurden ihnen die Einwahldaten für die Hotline übersendet und ihre Telefonnummern zur Einwahl in die Hotline zugelassen. Die Berater*innen konnten sich anschließend zeitlich und örtlich flexibel mit ihrem eigenen Endgerät in die Hotline einwählen und so eingehende Anrufe überstellt bekommen. Über eine telefonische Direktschaltung mit der Kassenärztlichen Vereinigung war es den Berater*innen im Bedarfsfall möglich, den Anrufenden einen raschen Zugang zu psychotherapeutischer Behandlung zu ermöglichen. Insgesamt registrierten sich 753 Berater*innen, unter ihnen 469 ärztliche und psychologische sowie Kinder- und Jugendpsychotherapeut*innen, 107 Psychotherapeut*innen in Ausbildung, 160 Sozialarbeiter*innen im ambulant betreuten Wohnen und 17 Personen mit anderen psychosozialen Berufen.\nImplementierung der Webseite Zusätzlich wurde eine Webseite mit Inhalten zur psychologischen Selbsthilfe eingerichtet. Betroffene konnten dort Tipps zu den folgenden Themen erhalten: „Psychische Krise & Gewalt“, „Tipps für den Alltag“, „Umgang mit Belastung und Stress“, „Sorgen und Ängste“, „Schlafprobleme“, „Mit sich alleine sein“, „Ärger & Konflikte“, „Familienleben“. Die Webseite ist unter der Adresse www.psyhotline-corona-bw.de weiterhin erreichbar.\nZusätzlich wurde eine Webseite mit Inhalten zur psychologischen Selbsthilfe eingerichtet. Betroffene konnten dort Tipps zu den folgenden Themen erhalten: „Psychische Krise & Gewalt“, „Tipps für den Alltag“, „Umgang mit Belastung und Stress“, „Sorgen und Ängste“, „Schlafprobleme“, „Mit sich alleine sein“, „Ärger & Konflikte“, „Familienleben“. Die Webseite ist unter der Adresse www.psyhotline-corona-bw.de weiterhin erreichbar.\nDatenerhebung Das Projekt wurde durch Mitarbeiter*innen des ZI wissenschaftlich begleitet. Hierzu wurden Daten zur Inanspruchnahme der Hotline und zu den Beratungsgesprächen erhoben. Alle Datenerhebungen erfolgten vollständig anonym. Ein positives Ethikvotum der Ethikkommission II der Universität Heidelberg liegt vor (2020559N).\nDokumentation der Beratungen Die Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nDie Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nAbschließende Berater*innenbefragung In einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nIn einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nDas Projekt wurde durch Mitarbeiter*innen des ZI wissenschaftlich begleitet. Hierzu wurden Daten zur Inanspruchnahme der Hotline und zu den Beratungsgesprächen erhoben. Alle Datenerhebungen erfolgten vollständig anonym. Ein positives Ethikvotum der Ethikkommission II der Universität Heidelberg liegt vor (2020559N).\nDokumentation der Beratungen Die Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nDie Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nAbschließende Berater*innenbefragung In einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nIn einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nStatistische Analysen Die Dokumentationen der Beratungen wurden deskriptiv ausgewertet. Die Einschätzungen der Berater*innen, den Anrufenden helfen zu können, wurde in Abhängigkeit der vorliegenden Symptomatik mithilfe von t‑Tests für unabhängige Stichproben inferenzstatistisch analysiert. Hierbei wurden die Angaben der Beratungsgespräche, in denen die jeweilige Symptomatik berichtet wurde, mit den Angaben der übrigen Beratungsgespräche verglichen, in denen die jeweilige Symptomatik nicht berichtet wurde (1 = liegt vor, 0 = liegt nicht vor). Um einen möglichen Zusammenhang zwischen der Inanspruchnahme der Hotline und dem Infektionsgeschehen in Deutschland zu analysieren, wurde die vom Robert-Koch-Institut veröffentlichte Anzahl der Neuinfektionen pro Tag in die Analysen mit einbezogen.\nDie Dokumentationen der Beratungen wurden deskriptiv ausgewertet. Die Einschätzungen der Berater*innen, den Anrufenden helfen zu können, wurde in Abhängigkeit der vorliegenden Symptomatik mithilfe von t‑Tests für unabhängige Stichproben inferenzstatistisch analysiert. Hierbei wurden die Angaben der Beratungsgespräche, in denen die jeweilige Symptomatik berichtet wurde, mit den Angaben der übrigen Beratungsgespräche verglichen, in denen die jeweilige Symptomatik nicht berichtet wurde (1 = liegt vor, 0 = liegt nicht vor). Um einen möglichen Zusammenhang zwischen der Inanspruchnahme der Hotline und dem Infektionsgeschehen in Deutschland zu analysieren, wurde die vom Robert-Koch-Institut veröffentlichte Anzahl der Neuinfektionen pro Tag in die Analysen mit einbezogen.", "Als potenzielle Berater*innen wurden durch die Landespsychotherapeutenkammer und die Landesärztekammer approbierte ärztliche und psychologische Psychotherapeut*innen sowie Kinder- und Jugendpsychotherapeut*innen per E‑Mail kontaktiert und um Beteiligung gebeten. Über staatlich anerkannte Ausbildungsinstitute wurden psychologische und Kinder- und Jugendpsychologische Psychotherapeut*innen in Ausbildung kontaktiert. Über das Sozialministerium wurden zudem erfahrene Sozialarbeiter*innen aus der gemeindepsychiatrischen Versorgung kontaktiert. Alle potenziellen Berater*innen erhielten ein Informationsschreiben, in dem sie über Ziel und Zweck der Hotline, den Ablauf der Registrierung und der Beratungen, der Freiwilligkeit ihrer Teilnahme sowie die bestehenden Datenschutzbestimmungen aufgeklärt wurden. Die Berater*innen konnten sich anschließend freiwillig über einen URL-Link registrieren. Nachdem die personale und berufliche Identität der Berater*innen geprüft wurde, wurden ihnen die Einwahldaten für die Hotline übersendet und ihre Telefonnummern zur Einwahl in die Hotline zugelassen. Die Berater*innen konnten sich anschließend zeitlich und örtlich flexibel mit ihrem eigenen Endgerät in die Hotline einwählen und so eingehende Anrufe überstellt bekommen. Über eine telefonische Direktschaltung mit der Kassenärztlichen Vereinigung war es den Berater*innen im Bedarfsfall möglich, den Anrufenden einen raschen Zugang zu psychotherapeutischer Behandlung zu ermöglichen. Insgesamt registrierten sich 753 Berater*innen, unter ihnen 469 ärztliche und psychologische sowie Kinder- und Jugendpsychotherapeut*innen, 107 Psychotherapeut*innen in Ausbildung, 160 Sozialarbeiter*innen im ambulant betreuten Wohnen und 17 Personen mit anderen psychosozialen Berufen.", "Zusätzlich wurde eine Webseite mit Inhalten zur psychologischen Selbsthilfe eingerichtet. Betroffene konnten dort Tipps zu den folgenden Themen erhalten: „Psychische Krise & Gewalt“, „Tipps für den Alltag“, „Umgang mit Belastung und Stress“, „Sorgen und Ängste“, „Schlafprobleme“, „Mit sich alleine sein“, „Ärger & Konflikte“, „Familienleben“. Die Webseite ist unter der Adresse www.psyhotline-corona-bw.de weiterhin erreichbar.", "Das Projekt wurde durch Mitarbeiter*innen des ZI wissenschaftlich begleitet. Hierzu wurden Daten zur Inanspruchnahme der Hotline und zu den Beratungsgesprächen erhoben. Alle Datenerhebungen erfolgten vollständig anonym. Ein positives Ethikvotum der Ethikkommission II der Universität Heidelberg liegt vor (2020559N).\nDokumentation der Beratungen Die Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nDie Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).\nAbschließende Berater*innenbefragung In einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.\nIn einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.", "Die Anzahl der eingehenden Anrufe wurde vom Betreiber der Hotline zur wissenschaftlichen Auswertung bereitgestellt (Abb. 1). Zusätzlich wurden die Berater*innen gebeten, durchgeführte Beratungsgespräche über einen Onlinefragebogen zu dokumentieren. Im Fragebogen wurden Angaben zu Alter und Geschlecht der Anrufenden, dem Anrufgrund (Abb. 2), der aktuellen Belastungssymptomatik (Abb. 3) und den Inhalten der Beratungen erhoben. Bei allen Angaben waren Mehrfachnennungen möglich. Zusätzlich wurde die Gesprächsdauer sowie eine Einschätzung erhoben, wie sehr die Berater*innen den Anrufenden in den telefonischen Beratungen helfen konnten bzw. wie belastend sie diese empfunden haben (1 = überhaupt nicht, 5 = sehr viel).", "In einer abschließenden Berater*innenbefragung wurde die Zufriedenheit der Berater*innen mit dem Projekt und ihrer Teilnahme erhoben.", "Die Dokumentationen der Beratungen wurden deskriptiv ausgewertet. Die Einschätzungen der Berater*innen, den Anrufenden helfen zu können, wurde in Abhängigkeit der vorliegenden Symptomatik mithilfe von t‑Tests für unabhängige Stichproben inferenzstatistisch analysiert. Hierbei wurden die Angaben der Beratungsgespräche, in denen die jeweilige Symptomatik berichtet wurde, mit den Angaben der übrigen Beratungsgespräche verglichen, in denen die jeweilige Symptomatik nicht berichtet wurde (1 = liegt vor, 0 = liegt nicht vor). Um einen möglichen Zusammenhang zwischen der Inanspruchnahme der Hotline und dem Infektionsgeschehen in Deutschland zu analysieren, wurde die vom Robert-Koch-Institut veröffentlichte Anzahl der Neuinfektionen pro Tag in die Analysen mit einbezogen.", "Insgesamt war die Hotline über 13 Wochen vom 22.04. bis zum 24.07.2020 aktiv geschaltet. In diesem Zeitraum wurden insgesamt 8577 Anrufe registriert (Abb. 1). Die meisten Anrufe (27 %) gingen abends zwischen 18:00 und 20:00 Uhr ein. Es zeigte sich kein signifikanter Zusammenhang zwischen der Anzahl der eingehenden Anrufe und der Anzahl der Neuinfektionen pro Tag (r = −0,02, p = 0,983). Die Dauer eines Beratungsgesprächs betrug durchschnittlich 23,7 min (SD = 17,1). Insgesamt konnten 481 Anrufe (5,6 %) aufgrund eines vorübergehenden Beratermangels nicht durchgestellt werden, davon lassen sich 121 Anrufe auf technische Probleme an einem singulären Tag zurückführen. Die Anzahl der nicht durchgestellten Anrufe zeigte sich besonders hoch in den Morgenstunden zwischen 08:00 bis 09:00 Uhr (79 Anrufe), in den Mittagsstunden zwischen 12:00 bis 13:00 Uhr (77 Anrufe) und in den Abendstunden zwischen 18:00 bis 19:00 Uhr (95 Anrufe). Somit konnten insgesamt 8096 eingehende Anrufe durchgestellt werden.\nDokumentation der Beratungsgespräche Von den 8096 durchgestellten Anrufen wurden insgesamt 1292 Telefonberatungen dokumentiert (16 %). Demnach, waren 64 % der Anrufenden weiblich und die große Mehrheit der Anrufenden erwachsen (79 %), gefolgt von Anrufenden höheren Alters (17 %). Nur ein kleiner Teil der Anrufenden waren Jugendliche (3,8 %) oder Kinder (0,2 %). Mehr als jeder zweite dokumentierte Anruf stand in Zusammenhang mit einer vorbestehenden psychischen Erkrankung (55 %), gefolgt von Anrufen aufgrund von Quarantänemaßnahmen oder sozialer Isolation (42 %), Problemen im Alltag (30 %) oder Corona-spezifischen Ängsten (25 %). Nur ein kleiner Teil der Anrufe bezog sich auf häusliche Gewalt (3 %; Abb. 2).\nDie mit Abstand am häufigsten dokumentierte Symptomatik waren depressive Symptome (36 %), gefolgt von psychotischen Symptomen (19 %) und Angstsymptomen (18 %). 15 % der dokumentierten Anrufe berichteten eine unklare Symptomatik. Nur ein kleiner Teil (4 %) berichtete Suizidgedanken oder -absichten (Abb. 3). Die durchschnittliche Belastung der Anrufenden lag bei M = 3,3 (SD = 0,8) auf einer Skala von 1 bis 5.\nIn 42 % aller Beratungen gaben die Berater*innen an, kurze therapeutische Interventionen durchgeführt zu haben. Diese umfassten vor allem unterstützende Gesprächstechniken (z. B. Validierung, supportive Techniken), Anleitung zu Entspannung (z. B. Atemübungen), emotionsfokussierte Interventionen (z. B. Benennen und Sortieren von Gefühlen, entgegengesetztes Handeln), kognitive Interventionen (z. B. hilfreiche Sätze, Perspektivwechsel, Grübelmanagement), verhaltensbezogene Interventionen (z. B. Tagesstruktur, positive Aktivitäten), psychoedukative Interventionen (z. B. Schlafhygiene), Unterstützung bei Problemlösung (z. B. Konfliktberatung) sowie Ressourcenaktivierung. In 26 % aller dokumentierten Beratungen wurde eine psychotherapeutische Behandlung empfohlen, in 11 % aller Beratungen auf weitere, spezialisierte Telefonangebote verwiesen. In 16 % der Beratungen wurden Informationen zu Corona vermittelt. Nur in wenigen Beratungen (2 %) war ein akutes Krisencoaching notwendig (z. B. aufgrund akuter Suizidalität).\nInsgesamt gaben die Berater*innen an, dass sie in 27 % aller Gespräche „(sehr) viel“, in 40 % „etwas“ und in 33 % „nur wenig“ oder „gar nicht“ haben helfen können. Dabei zeigte sich, dass die Berater*innen in Beratungsgesprächen, bei denen ein klarer Anrufgrund vorlag (z. B. Corona-Angst, Isolation und Quarantäne, Schwierigkeit im Alltag oder Gewalt), signifikant mehr helfen konnten, im Vergleich zu den übrigen Beratungsgesprächen, in denen diese Anrufgründe nicht vorlagen (Tab. 1). In Beratungsgesprächen, in denen der Anrufgrund unklar war, konnten die Berater*innen im Vergleich zu den übrigen Beratungsgesprächen signifikant weniger helfen (M = 1,84). Ein ähnliches Bild zeigte sich auf der Symptomebene. Beim Vorliegen einer klaren Symptomatik wie Depressions‑, Angst‑, Zwangssymptomen, Gewalterleben oder PTBS konnten die Berater*innen signifikant mehr helfen (M = 2,98–3,34) als bei den übrigen Beratungen, in denen die jeweilige Symptomatik nicht vorlag. Signifikant weniger helfen konnten die Berater*innen hingegen im Vergleich zu den übrigen Beratungsgesprächen beim Vorliegen von psychotischen Symptomen, Störungen auf Persönlichkeitsebene oder bei unklarer Symptomatik (M = 2,31–2,50). Keine Unterschiede zeigten sich zwischen Beratungsgesprächen, in denen akute Suizidalität, Essstörungen oder Gewaltausübungen vorlagen (M = 2,71–3,27) im Vergleich zu den übrigen Beratungsgesprächen.Grund/Symptom liegt vorGrund/Symptom liegt nicht vort-TestEffektMSDnMSDnt-WertDfp-WertdGrund für AnrufCorona-Angst3,300,873152,671,0690010,32664<0,0010,61Quarantäne3,060,945352,651,106806,931205<0,0010,39Psych. Erkrankung2,791,026922,891,10523−1,7110770,087−0,10Schwierigkeit Alltag2,970,973692,771,088463,087740,0020,18Häusliche Gewalt3,240,86372,821,0511782,92390,0060,40Unklar1,841,02942,921,011121−9,83109<0,001−1,06SymptomeSuizidalität2,710,94492,841,061166−0,90530,372−0,12Depression2,980,944562,741,107594,021078<0,0010,23Angst3,320,792212,721,079949,49423<0,0010,58Zwang3,180,90342,821,0511812,24360,0320,34Essverhalten3,270,90112,831,0512041,62100,1360,42Psychose/Wahn2,421,032432,941,03972−6,93374<0,001−0,50Gewaltausübung2,930,83142,831,0512010,43140,6720,09Gewalterleben3,330,89402,821,0511753,5443<0,0010,49Posttraumatische Belastungsstörungen3,340,89382,821,0511773,6141<0,0010,5Persönlichkeit2,501,001812,891,051034−4,76254<0,001−0,37Unklar2,311,061752,921,021040−7,11232<0,001−0,60Restkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)M Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nRestkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)\nM Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nMit einem Mittelwert von 1,9 (SD = 0,9) auf einer Skala von 1 (gar nicht) bis 5 (sehr), empfanden die Berater*innen die Gespräche im Durchschnitt nur wenig belastend. Nur 5 % der Gespräche wurden als (sehr) stark belastend empfunden.\nVon den 8096 durchgestellten Anrufen wurden insgesamt 1292 Telefonberatungen dokumentiert (16 %). Demnach, waren 64 % der Anrufenden weiblich und die große Mehrheit der Anrufenden erwachsen (79 %), gefolgt von Anrufenden höheren Alters (17 %). Nur ein kleiner Teil der Anrufenden waren Jugendliche (3,8 %) oder Kinder (0,2 %). Mehr als jeder zweite dokumentierte Anruf stand in Zusammenhang mit einer vorbestehenden psychischen Erkrankung (55 %), gefolgt von Anrufen aufgrund von Quarantänemaßnahmen oder sozialer Isolation (42 %), Problemen im Alltag (30 %) oder Corona-spezifischen Ängsten (25 %). Nur ein kleiner Teil der Anrufe bezog sich auf häusliche Gewalt (3 %; Abb. 2).\nDie mit Abstand am häufigsten dokumentierte Symptomatik waren depressive Symptome (36 %), gefolgt von psychotischen Symptomen (19 %) und Angstsymptomen (18 %). 15 % der dokumentierten Anrufe berichteten eine unklare Symptomatik. Nur ein kleiner Teil (4 %) berichtete Suizidgedanken oder -absichten (Abb. 3). Die durchschnittliche Belastung der Anrufenden lag bei M = 3,3 (SD = 0,8) auf einer Skala von 1 bis 5.\nIn 42 % aller Beratungen gaben die Berater*innen an, kurze therapeutische Interventionen durchgeführt zu haben. Diese umfassten vor allem unterstützende Gesprächstechniken (z. B. Validierung, supportive Techniken), Anleitung zu Entspannung (z. B. Atemübungen), emotionsfokussierte Interventionen (z. B. Benennen und Sortieren von Gefühlen, entgegengesetztes Handeln), kognitive Interventionen (z. B. hilfreiche Sätze, Perspektivwechsel, Grübelmanagement), verhaltensbezogene Interventionen (z. B. Tagesstruktur, positive Aktivitäten), psychoedukative Interventionen (z. B. Schlafhygiene), Unterstützung bei Problemlösung (z. B. Konfliktberatung) sowie Ressourcenaktivierung. In 26 % aller dokumentierten Beratungen wurde eine psychotherapeutische Behandlung empfohlen, in 11 % aller Beratungen auf weitere, spezialisierte Telefonangebote verwiesen. In 16 % der Beratungen wurden Informationen zu Corona vermittelt. Nur in wenigen Beratungen (2 %) war ein akutes Krisencoaching notwendig (z. B. aufgrund akuter Suizidalität).\nInsgesamt gaben die Berater*innen an, dass sie in 27 % aller Gespräche „(sehr) viel“, in 40 % „etwas“ und in 33 % „nur wenig“ oder „gar nicht“ haben helfen können. Dabei zeigte sich, dass die Berater*innen in Beratungsgesprächen, bei denen ein klarer Anrufgrund vorlag (z. B. Corona-Angst, Isolation und Quarantäne, Schwierigkeit im Alltag oder Gewalt), signifikant mehr helfen konnten, im Vergleich zu den übrigen Beratungsgesprächen, in denen diese Anrufgründe nicht vorlagen (Tab. 1). In Beratungsgesprächen, in denen der Anrufgrund unklar war, konnten die Berater*innen im Vergleich zu den übrigen Beratungsgesprächen signifikant weniger helfen (M = 1,84). Ein ähnliches Bild zeigte sich auf der Symptomebene. Beim Vorliegen einer klaren Symptomatik wie Depressions‑, Angst‑, Zwangssymptomen, Gewalterleben oder PTBS konnten die Berater*innen signifikant mehr helfen (M = 2,98–3,34) als bei den übrigen Beratungen, in denen die jeweilige Symptomatik nicht vorlag. Signifikant weniger helfen konnten die Berater*innen hingegen im Vergleich zu den übrigen Beratungsgesprächen beim Vorliegen von psychotischen Symptomen, Störungen auf Persönlichkeitsebene oder bei unklarer Symptomatik (M = 2,31–2,50). Keine Unterschiede zeigten sich zwischen Beratungsgesprächen, in denen akute Suizidalität, Essstörungen oder Gewaltausübungen vorlagen (M = 2,71–3,27) im Vergleich zu den übrigen Beratungsgesprächen.Grund/Symptom liegt vorGrund/Symptom liegt nicht vort-TestEffektMSDnMSDnt-WertDfp-WertdGrund für AnrufCorona-Angst3,300,873152,671,0690010,32664<0,0010,61Quarantäne3,060,945352,651,106806,931205<0,0010,39Psych. Erkrankung2,791,026922,891,10523−1,7110770,087−0,10Schwierigkeit Alltag2,970,973692,771,088463,087740,0020,18Häusliche Gewalt3,240,86372,821,0511782,92390,0060,40Unklar1,841,02942,921,011121−9,83109<0,001−1,06SymptomeSuizidalität2,710,94492,841,061166−0,90530,372−0,12Depression2,980,944562,741,107594,021078<0,0010,23Angst3,320,792212,721,079949,49423<0,0010,58Zwang3,180,90342,821,0511812,24360,0320,34Essverhalten3,270,90112,831,0512041,62100,1360,42Psychose/Wahn2,421,032432,941,03972−6,93374<0,001−0,50Gewaltausübung2,930,83142,831,0512010,43140,6720,09Gewalterleben3,330,89402,821,0511753,5443<0,0010,49Posttraumatische Belastungsstörungen3,340,89382,821,0511773,6141<0,0010,5Persönlichkeit2,501,001812,891,051034−4,76254<0,001−0,37Unklar2,311,061752,921,021040−7,11232<0,001−0,60Restkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)M Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nRestkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)\nM Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nMit einem Mittelwert von 1,9 (SD = 0,9) auf einer Skala von 1 (gar nicht) bis 5 (sehr), empfanden die Berater*innen die Gespräche im Durchschnitt nur wenig belastend. Nur 5 % der Gespräche wurden als (sehr) stark belastend empfunden.\nInanspruchnahme der Webseite Die begleitende Webseite verzeichnete im selben Zeitraum, zwischen März und Juli 2020, insgesamt 5544 Aufrufe, also durchschnittlich 656 Aufrufe pro Tag (Abb. 4). Die Aufrufe auf der Webseite verteilten sich gleichmäßig über alle Rubriken hinweg.\nDie begleitende Webseite verzeichnete im selben Zeitraum, zwischen März und Juli 2020, insgesamt 5544 Aufrufe, also durchschnittlich 656 Aufrufe pro Tag (Abb. 4). Die Aufrufe auf der Webseite verteilten sich gleichmäßig über alle Rubriken hinweg.\nAbschließende Berater*innenbefragung In der abschließenden Berater*innenbefragung zeigte sich eine hohe Zufriedenheit der Berater*innen mit dem Projekt. Auf einer 5‑stufigen Skala (1 = überhaupt nicht; 5 = sehr) bewerteten die Berater*innen das Hotlineprojekt durchschnittlich als sinnvoll (M = 4,03, SD = 1,06), die zugehörige Webseite als hilfreich (M = 4,02, SD = 1,04) und die eigene Beteiligung als teilweise befriedigend (M = 3,34, SD = 1,04).\nIn der abschließenden Berater*innenbefragung zeigte sich eine hohe Zufriedenheit der Berater*innen mit dem Projekt. Auf einer 5‑stufigen Skala (1 = überhaupt nicht; 5 = sehr) bewerteten die Berater*innen das Hotlineprojekt durchschnittlich als sinnvoll (M = 4,03, SD = 1,06), die zugehörige Webseite als hilfreich (M = 4,02, SD = 1,04) und die eigene Beteiligung als teilweise befriedigend (M = 3,34, SD = 1,04).", "Von den 8096 durchgestellten Anrufen wurden insgesamt 1292 Telefonberatungen dokumentiert (16 %). Demnach, waren 64 % der Anrufenden weiblich und die große Mehrheit der Anrufenden erwachsen (79 %), gefolgt von Anrufenden höheren Alters (17 %). Nur ein kleiner Teil der Anrufenden waren Jugendliche (3,8 %) oder Kinder (0,2 %). Mehr als jeder zweite dokumentierte Anruf stand in Zusammenhang mit einer vorbestehenden psychischen Erkrankung (55 %), gefolgt von Anrufen aufgrund von Quarantänemaßnahmen oder sozialer Isolation (42 %), Problemen im Alltag (30 %) oder Corona-spezifischen Ängsten (25 %). Nur ein kleiner Teil der Anrufe bezog sich auf häusliche Gewalt (3 %; Abb. 2).\nDie mit Abstand am häufigsten dokumentierte Symptomatik waren depressive Symptome (36 %), gefolgt von psychotischen Symptomen (19 %) und Angstsymptomen (18 %). 15 % der dokumentierten Anrufe berichteten eine unklare Symptomatik. Nur ein kleiner Teil (4 %) berichtete Suizidgedanken oder -absichten (Abb. 3). Die durchschnittliche Belastung der Anrufenden lag bei M = 3,3 (SD = 0,8) auf einer Skala von 1 bis 5.\nIn 42 % aller Beratungen gaben die Berater*innen an, kurze therapeutische Interventionen durchgeführt zu haben. Diese umfassten vor allem unterstützende Gesprächstechniken (z. B. Validierung, supportive Techniken), Anleitung zu Entspannung (z. B. Atemübungen), emotionsfokussierte Interventionen (z. B. Benennen und Sortieren von Gefühlen, entgegengesetztes Handeln), kognitive Interventionen (z. B. hilfreiche Sätze, Perspektivwechsel, Grübelmanagement), verhaltensbezogene Interventionen (z. B. Tagesstruktur, positive Aktivitäten), psychoedukative Interventionen (z. B. Schlafhygiene), Unterstützung bei Problemlösung (z. B. Konfliktberatung) sowie Ressourcenaktivierung. In 26 % aller dokumentierten Beratungen wurde eine psychotherapeutische Behandlung empfohlen, in 11 % aller Beratungen auf weitere, spezialisierte Telefonangebote verwiesen. In 16 % der Beratungen wurden Informationen zu Corona vermittelt. Nur in wenigen Beratungen (2 %) war ein akutes Krisencoaching notwendig (z. B. aufgrund akuter Suizidalität).\nInsgesamt gaben die Berater*innen an, dass sie in 27 % aller Gespräche „(sehr) viel“, in 40 % „etwas“ und in 33 % „nur wenig“ oder „gar nicht“ haben helfen können. Dabei zeigte sich, dass die Berater*innen in Beratungsgesprächen, bei denen ein klarer Anrufgrund vorlag (z. B. Corona-Angst, Isolation und Quarantäne, Schwierigkeit im Alltag oder Gewalt), signifikant mehr helfen konnten, im Vergleich zu den übrigen Beratungsgesprächen, in denen diese Anrufgründe nicht vorlagen (Tab. 1). In Beratungsgesprächen, in denen der Anrufgrund unklar war, konnten die Berater*innen im Vergleich zu den übrigen Beratungsgesprächen signifikant weniger helfen (M = 1,84). Ein ähnliches Bild zeigte sich auf der Symptomebene. Beim Vorliegen einer klaren Symptomatik wie Depressions‑, Angst‑, Zwangssymptomen, Gewalterleben oder PTBS konnten die Berater*innen signifikant mehr helfen (M = 2,98–3,34) als bei den übrigen Beratungen, in denen die jeweilige Symptomatik nicht vorlag. Signifikant weniger helfen konnten die Berater*innen hingegen im Vergleich zu den übrigen Beratungsgesprächen beim Vorliegen von psychotischen Symptomen, Störungen auf Persönlichkeitsebene oder bei unklarer Symptomatik (M = 2,31–2,50). Keine Unterschiede zeigten sich zwischen Beratungsgesprächen, in denen akute Suizidalität, Essstörungen oder Gewaltausübungen vorlagen (M = 2,71–3,27) im Vergleich zu den übrigen Beratungsgesprächen.Grund/Symptom liegt vorGrund/Symptom liegt nicht vort-TestEffektMSDnMSDnt-WertDfp-WertdGrund für AnrufCorona-Angst3,300,873152,671,0690010,32664<0,0010,61Quarantäne3,060,945352,651,106806,931205<0,0010,39Psych. Erkrankung2,791,026922,891,10523−1,7110770,087−0,10Schwierigkeit Alltag2,970,973692,771,088463,087740,0020,18Häusliche Gewalt3,240,86372,821,0511782,92390,0060,40Unklar1,841,02942,921,011121−9,83109<0,001−1,06SymptomeSuizidalität2,710,94492,841,061166−0,90530,372−0,12Depression2,980,944562,741,107594,021078<0,0010,23Angst3,320,792212,721,079949,49423<0,0010,58Zwang3,180,90342,821,0511812,24360,0320,34Essverhalten3,270,90112,831,0512041,62100,1360,42Psychose/Wahn2,421,032432,941,03972−6,93374<0,001−0,50Gewaltausübung2,930,83142,831,0512010,43140,6720,09Gewalterleben3,330,89402,821,0511753,5443<0,0010,49Posttraumatische Belastungsstörungen3,340,89382,821,0511773,6141<0,0010,5Persönlichkeit2,501,001812,891,051034−4,76254<0,001−0,37Unklar2,311,061752,921,021040−7,11232<0,001−0,60Restkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)M Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nRestkategorie „weiß nicht“ (k = 31); keine Angabe (k = 46)\nM Mittelwert, SD Standardabweichung, n Anzahl, d Standardisierte Mittelwertdifferenz (Cohen’s d), Df Freiheitsgrade\nMit einem Mittelwert von 1,9 (SD = 0,9) auf einer Skala von 1 (gar nicht) bis 5 (sehr), empfanden die Berater*innen die Gespräche im Durchschnitt nur wenig belastend. Nur 5 % der Gespräche wurden als (sehr) stark belastend empfunden.", "Die begleitende Webseite verzeichnete im selben Zeitraum, zwischen März und Juli 2020, insgesamt 5544 Aufrufe, also durchschnittlich 656 Aufrufe pro Tag (Abb. 4). Die Aufrufe auf der Webseite verteilten sich gleichmäßig über alle Rubriken hinweg.", "In der abschließenden Berater*innenbefragung zeigte sich eine hohe Zufriedenheit der Berater*innen mit dem Projekt. Auf einer 5‑stufigen Skala (1 = überhaupt nicht; 5 = sehr) bewerteten die Berater*innen das Hotlineprojekt durchschnittlich als sinnvoll (M = 4,03, SD = 1,06), die zugehörige Webseite als hilfreich (M = 4,02, SD = 1,04) und die eigene Beteiligung als teilweise befriedigend (M = 3,34, SD = 1,04).", "Ziel des Projekts war es, eine schnelle und professionelle psychologische Ersthilfe für Menschen mit psychischen Belastungen in der ersten Welle der Corona-Pandemie zu ermöglichen. Dazu wurden eine Hotline zur persönlichen Beratung und eine Webseite für Selbsthilfezwecke eingerichtet.\nMit 753 registrierten Berater*innen stieß das Hotlineprojekt auf eine große berufsgruppenübergreifende Zustimmung. Durch die hohe Anzahl professioneller Berater*innen konnte eine hohe Beratungsqualität sichergestellt werden, was in bisherigen Hotlineimplementierungen eine große Herausforderung darstellte [26].\nMit insgesamt 8577 Anrufen in 13 Wochen (ca. 700 Anrufe pro Woche) stieß die Hotline auf eine große Nachfrage in der Landesbevölkerung. Die Häufung der Anrufe in den Abendstunden zeigt die Wichtigkeit, diese Hilfsangebote auch außerhalb der regelhaften Arbeitszeit anzubieten [26]. Die Anzahl der eingehenden Anrufe zeigte keinen Zusammenhang mit der Anzahl der Neuinfektionen pro Tag. Dies könnte darauf zurückzuführen sein, dass die Hotline erst Ende April, kurz nach dem Höhepunkt der ersten Welle, in Betrieb genommen wurde. Die Anzahl der Neuinfektionen nahm zu diesem Zeitpunkt bereits kontinuierlich ab, während die Beschränkungsmaßnahmen nur langsam gelockert wurden. Wahrscheinlich ist auch, dass weitere Faktoren, wie die Bekanntheit der Hotline, einen wichtigen Einfluss auf die Anrufzahlen hatten.\nDie Dokumentationen der Berater*innen weisen darauf hin, dass jeder zweite Anruf von Personen getätigt wurde, die eine bereits erkannte psychische Störung angaben. Dies könnte mit besonderen Risiken dieser Personengruppe zusammenhängen, was sich mit Erkenntnissen aus weiteren Studien deckt. So zeigte sich, dass COVID-19-erkrankte Personen mit komorbiden psychischen Störungen im Vergleich zu COVID-19-Erkrankten ohne psychische Störung eine um 48 % erhöhte Mortalitätsrate aufweisen [29]. Diese Befunde unterstreichen die Wichtigkeit, dieser gefährdeten Gruppe besondere Aufmerksamkeit zu schenken [30]. Wie erwartet betrafen die weiteren Gründe für einen Anruf bei der Hotline vor allem Corona-spezifische Ängste, Belastungen aufgrund der Isolations- und Quarantänemaßnahmen oder Schwierigkeiten im Alltag (25–42 %). Der Anteil der Anrufe aufgrund häuslicher Gewalt lag mit 2,9 % vergleichsweise niedrig, obwohl diese auch in Deutschland unter Quarantänemaßnahmen signifikant zugenommen hat [21]. Dies kann auf mehrere Gründe zurückzuführen sein: Zum einen gibt es weitere etablierte und gut ausgebaute telefonische Hilfsangebote für diese spezifische Zielgruppe, zum anderen jedoch sind Hilfsangebote bei häuslicher Gewalt den Betroffenen in vielen Fällen nicht bekannt oder es ist ihnen nur eingeschränkt möglich, bei weitreichender Überwachung und Kontrolle durch eine Partner*in, diese telefonischen Hilfsangebote zu nutzen [21]. Dies legt nahe, dass Hilfsangebote in der Öffentlichkeit besser kommuniziert und unterschiedliche Zugänge zu diesen Angeboten geschaffen werden sollten (z. B. Online und Telefon).\nAuf Symptomebene zeigten sich hohe Anteile von Depressions- und Angstsymptomen sowie ein überraschend hoher Anteil psychotischer und wahnhafter Symptome. Möglicherweise trägt das niedrigschwellig und anonym erreichbare Angebot dazu bei, dass Psychosebetroffene ein psychotherapeutisches Beratungsangebot annehmen. Der hohe Anteil dieser Symptomgruppen deckt sich mit der Einschätzung der Bundespsychotherapeutenkammer, dass Depressionen, Angststörungen und Psychosen zu den psychischen Erkrankungen gehören, die besonders durch die Pandemie ausgelöst oder verstärkt werden können [23]. Der hohe Anteil von Depressions- und Angstsymptomen zeigt sich in Übereinstimmung mit der vergleichsweise hohen Prävalenz dieser Erkrankungen in der deutschen Allgemeinbevölkerung mit 9,3 % für eine affektive Störung und 15,3 % für eine Angststörung [31], während der hohe Anteil psychotischer Symptome den hohen Bedarf an psychotherapeutischen Hilfen für diese Personengruppe anzeigt. Symptome von Zwangsstörungen, posttraumatischer Belastungsstörung oder Substanzkonsum wurden hingegen mit 3–4 % vergleichsweise seltener berichtet.\nBemerkenswert erscheint die Vielzahl an Kurzinterventionen, die trotz der anonymen Anrufe akut telefonisch durchgeführt werden konnten. Diese umfassten unterstützende Gesprächstechniken, Anleitung zu Entspannung, emotionsfokussierte, kognitive und verhaltensbezogene Interventionen, psychoedukative Interventionen, Unterstützung bei Problemlösung sowie Ressourcenaktivierung. An dieser Stelle sei noch einmal die Relevanz psychotherapeutisch geschulter Berater*innen herausgestellt, die auch in kurzer Zeit im telefonischen Kontakt gezielt helfen können.\nDie Einschätzung der Berater*innen, wie sehr sie den Anrufenden haben helfen können, zeigt, dass diese bei etwa zwei Drittel aller Beratungsgespräche den Eindruck hatten, zumindest „etwas“, häufig aber auch „viel“ oder „sehr viel“ helfen zu können. Dabei konnten die Berater*innen vor allem dann helfen, wenn ein klarer Anrufgrund oder eine klare Symptomatik vorlag. Bei unklaren Anrufgründen und unklarer Symptomatik konnten die Berater*innen signifikant weniger helfen; insbesondere bei psychotischen Symptomen und Symptomen von Persönlichkeitsstörungen gaben die Berater*innen an, dass sie weniger helfen konnten. Dennoch ist denkbar, dass auch dieser Personengruppe ein Anstoß gegeben wurde, sich weiterhin um psychotherapeutische Hilfen zu bemühen.\nLimitationen Zur Gewährleistung der Anonymität und des Datenschutzes wurden die Anrufenden nicht direkt um Angaben zur Person oder zu den Beratungsgesprächen gebeten. Bei der Datenerhebung handelt es sich lediglich um freiwillige Angaben der Berater*innen. Aufgrund der Freiwilligkeit der Dokumentation wurden zudem lediglich 16 % der Beratungsgespräche dokumentiert (1292 von 8096 durchgestellten Anrufen). Die Ergebnisse sollten daher mit Vorsicht und die vorliegenden Symptome nicht als Feststellung einer gesicherten Diagnose interpretiert werden. Obwohl die Dokumentationen von Fachpersonal durchgeführt wurden, die langjährige Erfahrungen bei der Einschätzung psychischer Symptome und deren Behandlung aufweisen, stellen die Ergebnisse nur einen ersten Hinweis auf den Nutzen einer entsprechenden Hotline dar. Diesem sollte durch zusätzliche Befragungen der Anrufenden in Zukunft weiter nachgegangen werden.\nZur Gewährleistung der Anonymität und des Datenschutzes wurden die Anrufenden nicht direkt um Angaben zur Person oder zu den Beratungsgesprächen gebeten. Bei der Datenerhebung handelt es sich lediglich um freiwillige Angaben der Berater*innen. Aufgrund der Freiwilligkeit der Dokumentation wurden zudem lediglich 16 % der Beratungsgespräche dokumentiert (1292 von 8096 durchgestellten Anrufen). Die Ergebnisse sollten daher mit Vorsicht und die vorliegenden Symptome nicht als Feststellung einer gesicherten Diagnose interpretiert werden. Obwohl die Dokumentationen von Fachpersonal durchgeführt wurden, die langjährige Erfahrungen bei der Einschätzung psychischer Symptome und deren Behandlung aufweisen, stellen die Ergebnisse nur einen ersten Hinweis auf den Nutzen einer entsprechenden Hotline dar. Diesem sollte durch zusätzliche Befragungen der Anrufenden in Zukunft weiter nachgegangen werden.\nSchlussfolgerung Insgesamt leistete die psychologische Hotline einen wichtigen Beitrag zur Bewältigung psychischer Belastungen während der ersten Infektionswelle der Corona-Pandemie in Baden-Württemberg. Für viele Menschen, die in diesem Zusammenhang unter psychischen Belastungen litten, wurde so eine Möglichkeit zur schnellen und professionellen psychologischen Ersthilfe geschaffen. Ein besonderer Dank gebührt der Vielzahl ehrenamtlicher Berater*innen, die sich über verschiedene Berufsgruppen und Therapieschulen hinweg engagierten, um einen außerordentlich wichtigen gesellschaftlichen Beitrag zur Bewältigung der Pandemie zu leisten.\nInsgesamt leistete die psychologische Hotline einen wichtigen Beitrag zur Bewältigung psychischer Belastungen während der ersten Infektionswelle der Corona-Pandemie in Baden-Württemberg. Für viele Menschen, die in diesem Zusammenhang unter psychischen Belastungen litten, wurde so eine Möglichkeit zur schnellen und professionellen psychologischen Ersthilfe geschaffen. Ein besonderer Dank gebührt der Vielzahl ehrenamtlicher Berater*innen, die sich über verschiedene Berufsgruppen und Therapieschulen hinweg engagierten, um einen außerordentlich wichtigen gesellschaftlichen Beitrag zur Bewältigung der Pandemie zu leisten.", "Zur Gewährleistung der Anonymität und des Datenschutzes wurden die Anrufenden nicht direkt um Angaben zur Person oder zu den Beratungsgesprächen gebeten. Bei der Datenerhebung handelt es sich lediglich um freiwillige Angaben der Berater*innen. Aufgrund der Freiwilligkeit der Dokumentation wurden zudem lediglich 16 % der Beratungsgespräche dokumentiert (1292 von 8096 durchgestellten Anrufen). Die Ergebnisse sollten daher mit Vorsicht und die vorliegenden Symptome nicht als Feststellung einer gesicherten Diagnose interpretiert werden. Obwohl die Dokumentationen von Fachpersonal durchgeführt wurden, die langjährige Erfahrungen bei der Einschätzung psychischer Symptome und deren Behandlung aufweisen, stellen die Ergebnisse nur einen ersten Hinweis auf den Nutzen einer entsprechenden Hotline dar. Diesem sollte durch zusätzliche Befragungen der Anrufenden in Zukunft weiter nachgegangen werden.", "Insgesamt leistete die psychologische Hotline einen wichtigen Beitrag zur Bewältigung psychischer Belastungen während der ersten Infektionswelle der Corona-Pandemie in Baden-Württemberg. Für viele Menschen, die in diesem Zusammenhang unter psychischen Belastungen litten, wurde so eine Möglichkeit zur schnellen und professionellen psychologischen Ersthilfe geschaffen. Ein besonderer Dank gebührt der Vielzahl ehrenamtlicher Berater*innen, die sich über verschiedene Berufsgruppen und Therapieschulen hinweg engagierten, um einen außerordentlich wichtigen gesellschaftlichen Beitrag zur Bewältigung der Pandemie zu leisten.", "Die Ergebnisse dieser Arbeit zeigen sowohl den Nutzen als auch die Grenzen von Hotlineangeboten auf. Der Nutzen besteht in erster Linie in der schnellen und einfachen Verfügbarkeit einer psychologischen Ersthilfemaßnahme. Bei unklarer oder komplexer psychischer Symptomatik scheint eine direkte telefonische Hilfe zwar nur eingeschränkt möglich zu sein, sie kann den Zugang zu einem fachärztlichen oder fachpsychotherapeutischen Kontakt zur Bewältigung der Belastungen jedoch erleichtern. Einen Ersatz für persönliche Kontakte sollte ein solches Hotlineangebot aber unter keinen Umständen darstellen. Insgesamt geben die hohe Nachfrage in der Bevölkerung sowie die Angaben zu Inhalt und Nutzen der Beratungsgespräche durch die Berater*innen einen ersten Hinweis darauf, dass Hotlineangebote eine praktikable Möglichkeit zur psychologischen Ersthilfe unter Pandemiebedingungen darstellen." ]

[ null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, "conclusion" ]

[ "Coronavirus", "COVID-19 Pandemie", "Hotline", "Psychologische Hilfe", "Psychische Gesundheit", "Coronavirus", "COVID-19 pandemic", "Hotline", "Psychological counseling", "Mental health" ]

[ 1113, 65, 1632, 223, 80, 336, 124, 17, 119, 2177, 845, 45, 94, 1237, 125, 82 ]

[ "der", "die", "und", "innen", "berater", "zu", "berater innen", "den", "mit", "wurden" ]

[ "der pandemie wirtschaftliche", "vergleich zu covid", "coronavirus infiziert", "hintergrund die covid", "covid 19 pandemie" ]

[ "Animals", "Anopheles", "Blood Chemical Analysis", "DNA", "Electron Transport Complex IV", "Entomology", "Feeding Behavior", "Female", "Meals", "Spain", "Specimen Handling" ]

[ "Background", "Study area", "Blood meal identification", "Statistical analysis", "Competing interests", "Authors’ contributions" ]

[ "The identification of vertebrate feeding sources and host preferences is essential for studies of the dynamics of transmission of vector-borne pathogens. Traditional serological techniques, including precipitin tests and enzyme-linked immunosorbent assays (ELISA), have been used to identify hosts from a diversity of insect vectors [1-3]. Although the use of these methods has provided valuable information, they have several limitations, including the difficulties of obtaining specific antisera against a broad diversity of host species and thereby failing to detect any host being investigated. To solve these limitations, researchers have progressively incorporated molecular approaches based on the amplification of DNA to identify hosts to species level [4].\nSuccessful identification of hosts by PCR-based methods may be limited by the quality and quantity of the host´s DNA contained in the abdomen of mosquitoes [5]. After feeding, the digestion of blood meal in the insect gut favours a quick degradation of host DNA. Therefore, as the stage of blood digestion increases the success of identification of blood meal sources may decrease [6]. Although variable, according to the method employed and the insect species tested, studies in the laboratory have shown that amplification of host DNA fails a few days after feeding [7]. An advance digestion status of blood-fed mosquitoes may be a potential reason explaining the proportion of unidentified blood meals in molecular studies [8-10]. To reduce the potential failure of host identification and the cost derived from these analyses, some studies on field-caught insects, from which the period between feeding and capture is unknown, only include fully engorged females [11,12] or females containing a recent blood meal [13]. However, capturing blood-fed females is a difficult task because they are not attracted to CO2 or other commonly used attractants for mosquitoes [14]. Although other more specific techniques for blood-fed female capture, such as resting boxes or aspirations at resting areas may be used, the number of blood-fed females available for analysis is usually limited. For example, in a recent study on mosquitoes, of the total of 212,987 specimens captured, only 911 (0.43%) engorged females produced a successful amplification [9]. Similar results have been also reported on studies on other haematophagous insects, such as Culicoides[15,16]. Consequently, it is important to describe protocols for blood meal analysis that maximize amplification success.\nHere, the impact of blood meal digestion status and two commonly used DNA extraction protocols on mosquito blood meal identification using DNA sequencing were analysed. Mosquito species studied here have sanitary and ecological importance as potential vectors of pathogens to humans, livestock and wildlife. This is the case for Anopheles atroparvus, the primary vector of human malaria in Spain in the past, which has recently been incriminated in a case of autochthonous malaria transmission [17], and different Culex species involved in the transmission of avian malaria [18,19] and West Nile and Usutu virus [20].", "Mosquito captures were done at Cañada de los Pájaros (Seville, Spain; 6°14’W, 36°57’N), a private natural reserve with a small freshwater pond of about five hectares resulting from the restoration of an abandoned gravel pit and surrounded by ricefields. Cañada de los Pájaros concentrates a large diversity of free-living native birds and captive exotic and native birds and some mammals, including domestic animals and humans. As a part of an extensive study on the transmission of vector-borne diseases, from September to November 2012, 319 blood-fed mosquitoes were captured by direct aspiration while resting in the main building at Cañada de los Pájaros. Furthermore, 28 blood-fed mosquitoes were captured resting outside the building in the same locality and in other surrounding localities (i.e., Doñana National Park) using an aspirator, CDC-type downdraft miniature suction traps (model 1212; J. W. Hock, Gainesville, FL, USA) and BG traps (Biogents, Regensburg, Germany) supplemented with CO2. Subsequently, mosquitoes were enumerated on a chill table under a stereomicroscope to species levels using available morphological keys [21,22]. Culex mosquitoes belonging to the univittatus complex were identified as Culex perexiguus based on the criteria described by Harbach [23]. The digestion status of mosquito blood meals was scored visually according to the Sella score from zero (unfed mosquitoes) to seven (female without visible blood and eggs fully developed in their abdomen), following Detinova [24], see Figure 1. Mosquitoes were stored at −80°C until molecular analyses of blood meal origin were carry out.\nCulex pipiens females with different stages of blood meal digestion. Numbers indicate the stage of blood meal digestion according to the Sella´s score.", "Mosquitoes were randomly assigned to one of two different DNA extraction protocols. The abdomens of 142 blood-fed mosquitoes were treated following the HotSHOT procedure: each abdomen was cut off using sterile tips and subsequently introduced into 75 μl of lysis solution (25 mM NaOH, 0.2 mM EDTA), crushed and incubated at 95°C for 30 minutes. After incubation, the solution was cooled on ice for five minutes and then 75 μl of neutralization solution (40 mM Tris–HCl) was added. At least two negative DNA extraction controls (i e, absence of blood) were included per plate. Abdomens were simultaneously processed using 96-thermowell plates and DNA extracts were stored at −20°C until PCR amplification. In addition, DNA from the abdomen of 205 blood-fed mosquitoes was isolated using the DNeasy Blood and Tissue® kit (QIAGEN, Hilden, Germany) following company specifications.\nVertebrate hosts were identified using a nested-PCR approach [25], which is effective to identify the feeding source of haematophagous insects. A fragment of 758 base pairs (bp) of the mitochondrial cytochrome oxidase 1 (COI) gen was amplified with the primary pair of primers M13BCV-FW and BCV-RV1 and the nested primer pair M13 and BCV-RV2. Positive amplifications were sequenced in one direction according to BigDye 1.1 technology (Applied Biosystems, Carlsbad, CA, USA). Labelled DNA fragments of PCR-positive products were resolved through an ABI 3130xl automated sequencer (Applied Biosystems, Carlsbad, CA, USA). Sequences were edited using the software Sequencher™ v4.9 (Gene Codes Corp, © 1991–2009, Ann Arbor, MI, USA) and assigned to particular vertebrate species when agreement was ≥98% to sequences of known species in GenBank DNA sequence database (National Center for Biotechnology Information Blast) or the Barcode of Life Data Systems (BOLD).", "A generalized linear mixed model (GLMM) with binomial distributed error and logit link function was used to test for the effect of the blood meal digestion status and the DNA extraction protocol on the success of host identification of blood-fed females. Identification success (0 or 1) was included as the dependent variable and DNA extraction protocol, Sella score and the interaction between both factors were included as independent variables. Mosquito species was included as a random factor and the model was fitted using Laplace approximation [26]. The analyses were repeated using only data from An. atroparvus (the species most extensively sampled) but given that the results were qualitatively the same, only the model including data from all species is shown. Analyses were done using SAS 9.2 (SAS Institute Inc 2010).", "The authors declare that they have no competing interests.", "JMP SR RS JF conceived and designed the experiments, and contributed reagents/materials/analysis tools. All authors have read and approved the final manuscript." ]

[ null, null, null, null, null, null ]

[ "Background", "Methods", "Study area", "Blood meal identification", "Statistical analysis", "Results", "Discussion", "Competing interests", "Authors’ contributions" ]

[ "The identification of vertebrate feeding sources and host preferences is essential for studies of the dynamics of transmission of vector-borne pathogens. Traditional serological techniques, including precipitin tests and enzyme-linked immunosorbent assays (ELISA), have been used to identify hosts from a diversity of insect vectors [1-3]. Although the use of these methods has provided valuable information, they have several limitations, including the difficulties of obtaining specific antisera against a broad diversity of host species and thereby failing to detect any host being investigated. To solve these limitations, researchers have progressively incorporated molecular approaches based on the amplification of DNA to identify hosts to species level [4].\nSuccessful identification of hosts by PCR-based methods may be limited by the quality and quantity of the host´s DNA contained in the abdomen of mosquitoes [5]. After feeding, the digestion of blood meal in the insect gut favours a quick degradation of host DNA. Therefore, as the stage of blood digestion increases the success of identification of blood meal sources may decrease [6]. Although variable, according to the method employed and the insect species tested, studies in the laboratory have shown that amplification of host DNA fails a few days after feeding [7]. An advance digestion status of blood-fed mosquitoes may be a potential reason explaining the proportion of unidentified blood meals in molecular studies [8-10]. To reduce the potential failure of host identification and the cost derived from these analyses, some studies on field-caught insects, from which the period between feeding and capture is unknown, only include fully engorged females [11,12] or females containing a recent blood meal [13]. However, capturing blood-fed females is a difficult task because they are not attracted to CO2 or other commonly used attractants for mosquitoes [14]. Although other more specific techniques for blood-fed female capture, such as resting boxes or aspirations at resting areas may be used, the number of blood-fed females available for analysis is usually limited. For example, in a recent study on mosquitoes, of the total of 212,987 specimens captured, only 911 (0.43%) engorged females produced a successful amplification [9]. Similar results have been also reported on studies on other haematophagous insects, such as Culicoides[15,16]. Consequently, it is important to describe protocols for blood meal analysis that maximize amplification success.\nHere, the impact of blood meal digestion status and two commonly used DNA extraction protocols on mosquito blood meal identification using DNA sequencing were analysed. Mosquito species studied here have sanitary and ecological importance as potential vectors of pathogens to humans, livestock and wildlife. This is the case for Anopheles atroparvus, the primary vector of human malaria in Spain in the past, which has recently been incriminated in a case of autochthonous malaria transmission [17], and different Culex species involved in the transmission of avian malaria [18,19] and West Nile and Usutu virus [20].", " Study area Mosquito captures were done at Cañada de los Pájaros (Seville, Spain; 6°14’W, 36°57’N), a private natural reserve with a small freshwater pond of about five hectares resulting from the restoration of an abandoned gravel pit and surrounded by ricefields. Cañada de los Pájaros concentrates a large diversity of free-living native birds and captive exotic and native birds and some mammals, including domestic animals and humans. As a part of an extensive study on the transmission of vector-borne diseases, from September to November 2012, 319 blood-fed mosquitoes were captured by direct aspiration while resting in the main building at Cañada de los Pájaros. Furthermore, 28 blood-fed mosquitoes were captured resting outside the building in the same locality and in other surrounding localities (i.e., Doñana National Park) using an aspirator, CDC-type downdraft miniature suction traps (model 1212; J. W. Hock, Gainesville, FL, USA) and BG traps (Biogents, Regensburg, Germany) supplemented with CO2. Subsequently, mosquitoes were enumerated on a chill table under a stereomicroscope to species levels using available morphological keys [21,22]. Culex mosquitoes belonging to the univittatus complex were identified as Culex perexiguus based on the criteria described by Harbach [23]. The digestion status of mosquito blood meals was scored visually according to the Sella score from zero (unfed mosquitoes) to seven (female without visible blood and eggs fully developed in their abdomen), following Detinova [24], see Figure 1. Mosquitoes were stored at −80°C until molecular analyses of blood meal origin were carry out.\nCulex pipiens females with different stages of blood meal digestion. Numbers indicate the stage of blood meal digestion according to the Sella´s score.\nMosquito captures were done at Cañada de los Pájaros (Seville, Spain; 6°14’W, 36°57’N), a private natural reserve with a small freshwater pond of about five hectares resulting from the restoration of an abandoned gravel pit and surrounded by ricefields. Cañada de los Pájaros concentrates a large diversity of free-living native birds and captive exotic and native birds and some mammals, including domestic animals and humans. As a part of an extensive study on the transmission of vector-borne diseases, from September to November 2012, 319 blood-fed mosquitoes were captured by direct aspiration while resting in the main building at Cañada de los Pájaros. Furthermore, 28 blood-fed mosquitoes were captured resting outside the building in the same locality and in other surrounding localities (i.e., Doñana National Park) using an aspirator, CDC-type downdraft miniature suction traps (model 1212; J. W. Hock, Gainesville, FL, USA) and BG traps (Biogents, Regensburg, Germany) supplemented with CO2. Subsequently, mosquitoes were enumerated on a chill table under a stereomicroscope to species levels using available morphological keys [21,22]. Culex mosquitoes belonging to the univittatus complex were identified as Culex perexiguus based on the criteria described by Harbach [23]. The digestion status of mosquito blood meals was scored visually according to the Sella score from zero (unfed mosquitoes) to seven (female without visible blood and eggs fully developed in their abdomen), following Detinova [24], see Figure 1. Mosquitoes were stored at −80°C until molecular analyses of blood meal origin were carry out.\nCulex pipiens females with different stages of blood meal digestion. Numbers indicate the stage of blood meal digestion according to the Sella´s score.\n Blood meal identification Mosquitoes were randomly assigned to one of two different DNA extraction protocols. The abdomens of 142 blood-fed mosquitoes were treated following the HotSHOT procedure: each abdomen was cut off using sterile tips and subsequently introduced into 75 μl of lysis solution (25 mM NaOH, 0.2 mM EDTA), crushed and incubated at 95°C for 30 minutes. After incubation, the solution was cooled on ice for five minutes and then 75 μl of neutralization solution (40 mM Tris–HCl) was added. At least two negative DNA extraction controls (i e, absence of blood) were included per plate. Abdomens were simultaneously processed using 96-thermowell plates and DNA extracts were stored at −20°C until PCR amplification. In addition, DNA from the abdomen of 205 blood-fed mosquitoes was isolated using the DNeasy Blood and Tissue® kit (QIAGEN, Hilden, Germany) following company specifications.\nVertebrate hosts were identified using a nested-PCR approach [25], which is effective to identify the feeding source of haematophagous insects. A fragment of 758 base pairs (bp) of the mitochondrial cytochrome oxidase 1 (COI) gen was amplified with the primary pair of primers M13BCV-FW and BCV-RV1 and the nested primer pair M13 and BCV-RV2. Positive amplifications were sequenced in one direction according to BigDye 1.1 technology (Applied Biosystems, Carlsbad, CA, USA). Labelled DNA fragments of PCR-positive products were resolved through an ABI 3130xl automated sequencer (Applied Biosystems, Carlsbad, CA, USA). Sequences were edited using the software Sequencher™ v4.9 (Gene Codes Corp, © 1991–2009, Ann Arbor, MI, USA) and assigned to particular vertebrate species when agreement was ≥98% to sequences of known species in GenBank DNA sequence database (National Center for Biotechnology Information Blast) or the Barcode of Life Data Systems (BOLD).\nMosquitoes were randomly assigned to one of two different DNA extraction protocols. The abdomens of 142 blood-fed mosquitoes were treated following the HotSHOT procedure: each abdomen was cut off using sterile tips and subsequently introduced into 75 μl of lysis solution (25 mM NaOH, 0.2 mM EDTA), crushed and incubated at 95°C for 30 minutes. After incubation, the solution was cooled on ice for five minutes and then 75 μl of neutralization solution (40 mM Tris–HCl) was added. At least two negative DNA extraction controls (i e, absence of blood) were included per plate. Abdomens were simultaneously processed using 96-thermowell plates and DNA extracts were stored at −20°C until PCR amplification. In addition, DNA from the abdomen of 205 blood-fed mosquitoes was isolated using the DNeasy Blood and Tissue® kit (QIAGEN, Hilden, Germany) following company specifications.\nVertebrate hosts were identified using a nested-PCR approach [25], which is effective to identify the feeding source of haematophagous insects. A fragment of 758 base pairs (bp) of the mitochondrial cytochrome oxidase 1 (COI) gen was amplified with the primary pair of primers M13BCV-FW and BCV-RV1 and the nested primer pair M13 and BCV-RV2. Positive amplifications were sequenced in one direction according to BigDye 1.1 technology (Applied Biosystems, Carlsbad, CA, USA). Labelled DNA fragments of PCR-positive products were resolved through an ABI 3130xl automated sequencer (Applied Biosystems, Carlsbad, CA, USA). Sequences were edited using the software Sequencher™ v4.9 (Gene Codes Corp, © 1991–2009, Ann Arbor, MI, USA) and assigned to particular vertebrate species when agreement was ≥98% to sequences of known species in GenBank DNA sequence database (National Center for Biotechnology Information Blast) or the Barcode of Life Data Systems (BOLD).\n Statistical analysis A generalized linear mixed model (GLMM) with binomial distributed error and logit link function was used to test for the effect of the blood meal digestion status and the DNA extraction protocol on the success of host identification of blood-fed females. Identification success (0 or 1) was included as the dependent variable and DNA extraction protocol, Sella score and the interaction between both factors were included as independent variables. Mosquito species was included as a random factor and the model was fitted using Laplace approximation [26]. The analyses were repeated using only data from An. atroparvus (the species most extensively sampled) but given that the results were qualitatively the same, only the model including data from all species is shown. Analyses were done using SAS 9.2 (SAS Institute Inc 2010).\nA generalized linear mixed model (GLMM) with binomial distributed error and logit link function was used to test for the effect of the blood meal digestion status and the DNA extraction protocol on the success of host identification of blood-fed females. Identification success (0 or 1) was included as the dependent variable and DNA extraction protocol, Sella score and the interaction between both factors were included as independent variables. Mosquito species was included as a random factor and the model was fitted using Laplace approximation [26]. The analyses were repeated using only data from An. atroparvus (the species most extensively sampled) but given that the results were qualitatively the same, only the model including data from all species is shown. Analyses were done using SAS 9.2 (SAS Institute Inc 2010).", "Mosquito captures were done at Cañada de los Pájaros (Seville, Spain; 6°14’W, 36°57’N), a private natural reserve with a small freshwater pond of about five hectares resulting from the restoration of an abandoned gravel pit and surrounded by ricefields. Cañada de los Pájaros concentrates a large diversity of free-living native birds and captive exotic and native birds and some mammals, including domestic animals and humans. As a part of an extensive study on the transmission of vector-borne diseases, from September to November 2012, 319 blood-fed mosquitoes were captured by direct aspiration while resting in the main building at Cañada de los Pájaros. Furthermore, 28 blood-fed mosquitoes were captured resting outside the building in the same locality and in other surrounding localities (i.e., Doñana National Park) using an aspirator, CDC-type downdraft miniature suction traps (model 1212; J. W. Hock, Gainesville, FL, USA) and BG traps (Biogents, Regensburg, Germany) supplemented with CO2. Subsequently, mosquitoes were enumerated on a chill table under a stereomicroscope to species levels using available morphological keys [21,22]. Culex mosquitoes belonging to the univittatus complex were identified as Culex perexiguus based on the criteria described by Harbach [23]. The digestion status of mosquito blood meals was scored visually according to the Sella score from zero (unfed mosquitoes) to seven (female without visible blood and eggs fully developed in their abdomen), following Detinova [24], see Figure 1. Mosquitoes were stored at −80°C until molecular analyses of blood meal origin were carry out.\nCulex pipiens females with different stages of blood meal digestion. Numbers indicate the stage of blood meal digestion according to the Sella´s score.", "Mosquitoes were randomly assigned to one of two different DNA extraction protocols. The abdomens of 142 blood-fed mosquitoes were treated following the HotSHOT procedure: each abdomen was cut off using sterile tips and subsequently introduced into 75 μl of lysis solution (25 mM NaOH, 0.2 mM EDTA), crushed and incubated at 95°C for 30 minutes. After incubation, the solution was cooled on ice for five minutes and then 75 μl of neutralization solution (40 mM Tris–HCl) was added. At least two negative DNA extraction controls (i e, absence of blood) were included per plate. Abdomens were simultaneously processed using 96-thermowell plates and DNA extracts were stored at −20°C until PCR amplification. In addition, DNA from the abdomen of 205 blood-fed mosquitoes was isolated using the DNeasy Blood and Tissue® kit (QIAGEN, Hilden, Germany) following company specifications.\nVertebrate hosts were identified using a nested-PCR approach [25], which is effective to identify the feeding source of haematophagous insects. A fragment of 758 base pairs (bp) of the mitochondrial cytochrome oxidase 1 (COI) gen was amplified with the primary pair of primers M13BCV-FW and BCV-RV1 and the nested primer pair M13 and BCV-RV2. Positive amplifications were sequenced in one direction according to BigDye 1.1 technology (Applied Biosystems, Carlsbad, CA, USA). Labelled DNA fragments of PCR-positive products were resolved through an ABI 3130xl automated sequencer (Applied Biosystems, Carlsbad, CA, USA). Sequences were edited using the software Sequencher™ v4.9 (Gene Codes Corp, © 1991–2009, Ann Arbor, MI, USA) and assigned to particular vertebrate species when agreement was ≥98% to sequences of known species in GenBank DNA sequence database (National Center for Biotechnology Information Blast) or the Barcode of Life Data Systems (BOLD).", "A generalized linear mixed model (GLMM) with binomial distributed error and logit link function was used to test for the effect of the blood meal digestion status and the DNA extraction protocol on the success of host identification of blood-fed females. Identification success (0 or 1) was included as the dependent variable and DNA extraction protocol, Sella score and the interaction between both factors were included as independent variables. Mosquito species was included as a random factor and the model was fitted using Laplace approximation [26]. The analyses were repeated using only data from An. atroparvus (the species most extensively sampled) but given that the results were qualitatively the same, only the model including data from all species is shown. Analyses were done using SAS 9.2 (SAS Institute Inc 2010).", "Overall, 347 blood-fed mosquitoes were included in this study. The most abundant species sampled was An. atroparvus followed by Cx. perexiguus, Culex pipiens, Culex theileri, Culex modestus, Ochlerotatus caspius and Culiseta sp. (Table 1). The blood meal source was successfully identified for 234 females (Table 1). The success of host identification decreased as the digestion status of the blood meal increased (F5,330=11.08, p<0.0001, Table 1). A higher success was obtained using the QIAGEN kit (F1,330=25.24, p<0.0001, 153/205, 74.6%) than the HotSHOT procedure (81/142, 57.0%). No interaction occurred between extraction method and digestion status (F4,330=1.43, p=0.23), although the differences in success with extraction method were not significant for females with very fresh blood meals (Sella score 2, F1,330=1.13, p=0.29).\nSella score of blood meal digestion, DNA extraction protocol used, and blood meal origin identification success for the 347 female mosquitoes analysed\nTwenty-four host species were identified including four mammals, 14 birds and a single reptile (Table 2). DNA from an additional unidentified bird species was isolated. This unknown species could not be identified by direct comparison with those sequences deposited in Genbank. Rabbit sequences were confirmed by comparison with sequences isolated from fresh muscle tissue. In addition, three samples from Anas sp. and one sample from Grus sp. were identified to the genus level. Dog was the most common feeding source of mosquitoes compromising 71.4% of the identified blood meals. A single human derived blood meal from An. atroparvus was isolated. Evidence of mixed blood meals was not observed.\nVertebrate host species identified for each mosquito species\nNumber of mosquitoes is recorded between brackets.", "The efficiency of the analyses of host blood meal source differs widely between studies (i.e., 17.5%-92%, see [19,27]). In this study, the importance of two sources of variation was quantified. Digestion status of blood meals, visually estimated according to the Sella score, strongly affects the success of host identification using DNA sequencing with efficiencies ranging between 84.5% and 25.0% depending of digestion status. Similar results were obtained by [6] and [28] based on mosquitoes kept at the laboratory, with a significant decrease in the identification success 30 to 36 hours after feeding [6]. Although the time interval between insect feeding and collection was unknown in this study, for a recently fed individual (Sella stage 2) it may take about one day to reach the Sella stages 3 and 4, and 1 or 2 additional days to reach the Sella stages 5 and 6, respectively [29]. Results from the present study support those from previous studies where authors reported a reduction of the proportion of reactions yielding sequences as the Sella score on field-caught mosquitoes increased [30,31]. In this study, a significant drop in success of host identification was found for mosquitoes containing a blood meal in an advanced stage of digestion (Sella stages >5), a similar pattern found in mosquitoes from South Carolina [32]. Obviously, including mosquitoes with blood meals in the highest stages of digestion (scored as 5 and 7 according to the Sella´s method), the overall success of host identification may be reduced, and this may partially explain discrepancies between studies in the rate of host identification success. In addition, in this study, using QIAGEN kit for DNA extraction, the success of host identification significantly increased by 10.2-35.3% depending of the blood meal digestion status. The increase in performance was especially important for the mosquitoes with more digested blood meal (scored from 5 to 7 according to the Sella´s method). Using the QIAGEN kit 47% of blood meal sources was identified while only 12% of those extracted using the HotSHOT procedure was identified. This is also a higher percentage of success than those reported by Tuten et al. [32] where authors, using the DNAzol BD Direct Extraction Kit (Molecular Research Center, Cincinnati, OH, USA), identified 27% (6/22) blood meals from mosquitoes with 5 to 6 Sella´s scores. Consequently, large improvements in blood source determination may be obtained by using more efficient DNA extraction methods. This increase in efficiency is not obtained free as the economic cost of extraction per sample is much higher when using commercial kits, but the extra cost may be worth investing when the number of blood-fed females to analyse is limiting, as used to be the case in most vector ecology studies.\nAt least 25 vertebrate host species of mosquitoes potentially involved in the transmission of pathogens by mosquitoes have been identified. Anopheles atroparvus showed a clear preference to feed on mammals of different sizes, from rats to horses, than on avian species in spite of the presence of a high diversity and abundance of birds in the studied area, supporting results from previous studies [33,34]. Curiously, as recently reported, there is no information on the feeding preference of this species to bite indoors or outdoors [34]. Results obtained in this study clearly indicate that this species feed on surrounding animals located outdoors but use human-made shelters for resting after feeding, adding valuable information to current knowledge on the biology of this species [34]. On the other hand, Cx. perexiguus, the second more extensively sampled species in this study, fed on different bird species in addition to mammals and turtles. Its role as bird feeders, as is the case of other Culex species in this study, supports their importance in the transmission of wildlife diseases in Europe, i.e., West Nile and Usutu virus [9,18-20].", "The authors declare that they have no competing interests.", "JMP SR RS JF conceived and designed the experiments, and contributed reagents/materials/analysis tools. All authors have read and approved the final manuscript." ]

[ null, "methods", null, null, null, "results", "discussion", null, null ]

[ "\nAnopheles atroparvus\n", "COI", "\nCulex\n", "Malaria", "Mosquitoes", "PCR", "Transmission network" ]

[ 566, 330, 357, 150, 10, 29 ]

[ "blood", "mosquitoes", "species", "dna", "blood meal", "meal", "digestion", "fed", "blood fed", "identification" ]

[ "mosquitoes analysed", "blood meal insect", "sequencing analysed mosquito", "protocols mosquito blood", "digestion status mosquito" ]

[ "Carcinoma", "Cell Line, Tumor", "Gene Expression Regulation, Neoplastic", "Gene Regulatory Networks", "Humans", "MicroRNAs", "Oncogenes", "Prognosis", "RNA, Long Noncoding", "RNA, Messenger", "Stomach Neoplasms" ]

[ "INTRODUCTION", "Data resources and differential expression analysis", "ceRNA network construction", "Prognosis‐related lncRNA identification", "Gene sets enrichment analysis", "Cell line culture, RNA extraction, and real‐time PCR", "siRNA knockdown in cell line", "Proliferation assay", "Transwell migration and invasion assay", "Statistical analysis", "Identification of differently expressed lncRNA, miRNA, and mRNA", "ceRNA network construction and function analysis", "Survival‐associated lncRNA and mRNA identification", "Reconstruction and function analysis of hub lncRNA‐associated ceRNA network", "INHBA‐AS1 and CCDC144NL‐AS1 are potential oncogenes in GC", "ETHICAL APPROVAL" ]

[ "Gastric cancer (GC) is one of the most frequent digestive system cancers and is the second cause of cancer mortality worldwide by 2018.\n1\n The cases in China account for more than 40% of the total number of GC worldwide due to a high incidence rate and a large population.\n2\n Moreover, the GC patients were more likely to be in the advanced stage when diagnosed because of non‐early specific symptoms. Unfortunately, the late diagnosis can significantly affect the 5‐year survival rate. In the past decade, due to the advancement of treatment and medicine, the GC prognosis has improved, but it is still not satisfied due to the relatively short disease‐free survival duration.\n3\n Thus, it is challenging and necessary to explore the underlying mechanisms of GC and identify novel biomarkers or treatment targets.\nOn the contrary, the long non‐coding RNA (lncRNA) is a well‐known member of the non‐coding RNA family in the past decade, which is RNA with a length of over 200 nt.\n4\n, \n5\n In the past decade, the accumulating knowledge indicates the important role of the aberrant expression of lncRNA in GC.\n6\n, \n7\n LncRNA can function as sequence‐specific recruitment of proteins, competing for endogenous RNA (ceRNA) regulation, and molecular scaffolding of protein complexes, in which the ceRNA regulation is widely investigated nowadays. It is hypothesized that lncRNA can modulate the miRNA‐regulated mRNA expression by competitively binding miRNAs through endogenous molecular sponges.\n8\n This regulatory mechanism interprets the roles of lncRNA in various cancers, including GC.\n9\n, \n10\n, \n11\n, \n12\n However, the lncRNA‐associated ceRNA networks are far from clear.\nTo further explore the role of specific lncRNA‐miRNA‐mRNA axis in GC, we first construct the ceRNA network via the online database. In addition, the hub lncRNAs with sub‐ceRNA networks related to prognosis were identified. To confirm the reliability and validity of the results, hub lncRNAs were validated in vitro. Overall, the present study aimed to establish a critical ceRNA network and identified novel diagnostic/therapeutic targets.", "We downloaded the RNA sequence data with log2 (fpkm + 1) transformed (lncRNA and mRNA, level 3; Illumina HiSeq RNA‐Seq platform), miRNA sequence data (Illumina HiSeq miRNA‐Seq platform), and clinical information from the Xena dataset (https://xenabrowser.net/datapages/?dataset=TCGA‐STAD.htseq_fpkm‐uq.tsv&host=https%3A%2F%2Fgdc.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443; https://xenabrowser.net/datapages/?dataset=TCGA‐BRCA.htseq_fpkm‐uq.tsv&host=https%3A%2F%2Fgdc.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443). Annotation information for the RNA data was provided by the Ensemble database derived from “biomaRt” package.\n13\n The differentially expressed (DE) lncRNAs, miRNAs, and mRNAs between 32 normal samples and 372 cancer samples were identified by the “limma” package.\n14\n We set the |log2(Fold change)|>1 and p‐value <0.05 as the thresholds to identify DE genes. DE lncRNAs, miRNAs, and mRNAs were presented as a volcano plot. The study was approved by the Ethics Committee of The First Affiliated Hospital of Wenzhou Medical University. Informed patient consent was not required as the results showed are based upon the data generated by the TCGA database.", "To construct a ceRNA network, we extracted the interactions between mRNA and miRNA/lncRNA via multiple online datasets. For the interaction between miRNA and mRNA, we used the “miRNAtap” package (https://bioconductor.org/packages/release/bioc/html/miRNAtap.html), which consisted of commonly used five reliable and online datasets, including Pictar (https://pictar.mdc‐berlin.de/), DIANA,\n15\n Targetscan,\n16\n miranda,\n17\n and mirdb.\n18\n Only when the interaction was identified in at least three datasets, we considered it to be potential interactions. In contrast, the lncRNA‐miRNA interactions were predicted through the miRcode.\n19\n\n\nFor ceRNA construction, we submitted the above interactions and expression dataset to the “GDCRNATools” package (http://bioconductor.org/packages/devel/bioc/vignettes/GDCRNATools/inst/doc/GDCRNATools.html), which can further filter the interaction based on those two criteria, (1) expression of lncRNA and mRNA must be positively correlated, and (2) those common miRNAs should play similar roles in regulating the expression of lncRNA and mRNA. Those two factors were indicated via the Pearson correlation and regulation similarity.\n20\n\n", "To identify the potential prognosis‐related lncRNA, we grouped the samples into high‐ and low‐expression subgroups based on the mean expression. Then, univariate Cox regression analysis was used to find the prognosis‐related lncRNA. Then, the lncRNA achieving statistical significance (p < 0.05) in the univariate analysis was submitted into multivariate Cox analysis adjusting with age, gender, TNM stage, and histological grade. In addition, the KM plot with log‐rank test was carried out to further validate the prognosis‐related lncRNA.", "To reveal the function of the lncRNA‐associated ceRNA network, the DE mRNAs derived from ceRNA were subjected to gene sets enrichment analysis based on gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and REACTOME annotation with the clusterProfiler package.\n21\n GO is a structured standard biological model established by the GO consortium, including biological processes (BP), molecular functions (MF), and cellular components (CC).\n22\n KEGG is widely used as a reference for integrating large‐scale molecular datasets generated by sequencing and high‐throughput experimental technologies.\n23\n RACTOME is an open‐source, open access, manually curated, and peer‐reviewed pathway database (https://reactome.org/). Gene sets with a p‐value <0.05 were considered significant.", "GC cell line, MKN45, was purchased from the Cell Bank of the Chinese Academy of Sciences. Cells were incubated at 37℃ in a humidified atmosphere containing 5% CO2 and cultured in Dulbecco's Modified Eagle's Medium (DMEM) (ThermoFisher Scientific) combined with 10% fetal bovine serum (FBS).\nThe primers were designed and purchased from GenePharma (Table 1). The total RNA was extracted using the ReliaPrep RNA Cell Miniprep System (Promega). It was first reversely transcribed into complementary DNA through the iScript Reverse Transcription Kit (Bio‐Rad), then, SYBR Green Supermix (Bio‐Rad) was employed to perform quantitative‐PCR (qPCR), which was run in Bio‐Rad CFX 96 PCR instrument (Bio‐Rad). The 2−ΔΔCT method was used to evaluate the relative gene expression. The GAPDH and U6 were used as the internal control for lncRNA/mRNA and miRNA, respectively.\nForward and reverse primer sequences used for quantitative PCR\nAbbreviations: F, forward; R, reverse.", "siRNA designed and chemically synthesized by GenePharma. The sequences of siRNA against INHBA‐AS1 were: 5′‐GUCUCAUGACCACAGCUAAtt‐3′ (Sense) and 5′‐UUAGCUGUGGUCAUGAGACct‐3′ (Anti‐Sense), and siRNA against CCDC144NL‐AS1 was 5′‐UAGGUAGAUGGUGGAAUGAtt‐3′ (Sense) and 5′‐UCAUUCCACCAUCUACCUAtg‐3′ (Anti‐Sense). Before the transfection, MKN45 cells were cultured with antibiotic‐free DMEM for 24 h in advance. Transfections of the siRNAs were manipulated via the lipofectamine RNAiMAX (Invitrogen) and the same protocols were carried out following the manufacturer's instruction in all those two lncRNAs.", "To evaluate the proliferation speed, 3 × 103 cells were seeded into a 96‐well plate per well (Greiner bio‐one), and then a time‐series assay every 2 days was carried out in triplicate. We used the MTS CellTiter 96 One Solution Cell Proliferation Assay (Promega) to measure proliferation. With 1, 3, 5, and 7 days, the absorbance at 490 nm was measured using the microtiter plate spectrophotometer (Benchmark Plus, Bio‐Rad) according to the manufacturer's protocol. Subsequently, proliferation was normalized based on the absorbance of the first day and calculated by the changes between the readings.", "For cell migration and invasion assay, 24‐well transwell inserts with a pore size of 0.8 mm (Corning) were used. After siRNA transfection, 1 × 105 cells were seeded in the upper chamber, and 650 µl of complete medium was added to the lower chamber as a chemo‐attractant. The 100‐µl Matrigel (Corning) with a concentration of 20 mg/ml was pre‐coated above the insert for invasion assay. The cells were allowed to migrate or invade toward the chamber for 12 and 16 h, respectively. The migrated and invaded cells below the membrane were fixed with 4% paraformaldehyde, stained with DAPI (Beyotime Biotechnology), and quantified from microscopic fields.", "The student's t test was used to compare the continuous variables. The chi‐squared test was employed for the categorical variables. p values have two tails and only when it is less than 0.05 was considered significant. All the figures and statistical analysis were carried out via R software (version 3.6.1).", "There were 1485 DE lncRNAs (1260 up‐regulation and 225 down‐regulation), 312 DE miRNAs (290 up‐regulation 12 down‐regulation), and 4260 DE mRNAs (2347 up‐regulation and 1913 down‐regulation) identified (Figure 1).\nVolcano plots showing up and downregulated (A) lncRNA, (B) miRNA, and (C) mRNA. The red dots represent high expression lncRNA/miRNA/mRNA with LogFC ≥1 and FDR <0.05, and the blue dots represent low expression of lncRNA/miRNA/mRNA with LogFC ≤ −1 and FDR <0.05", "To construct the ceRNA, we first extracted the interaction between DE miRNAs and DE lncRNAs, and DE miRNAs and DE mRNAs as well via multiple online datasets. Finally, there were 6082 pairs of lncRNA‐miRNA interactions predicted, and 938 interactions between miRNA and mRNA were identified. Based on those interactions, we established 24847 potential lncRNA‐miRNA‐mRNA axes consisting of 892 lncRNAs, 18 miRNAs, and 278 mRNAs preliminarily.\nThen, the ceRNA network was conducted via the “GDCRNATools” package with “gdcCEAnalysis” function, which utilizes the Pearson correlation and regulation pattern to further determine the promising ceRNA. Finally, a ceRNA network (909 edges and 253 nodes), including 76 lncRNA, 18 miRNA, and 159 mRNA, was constructed with Pearson correlation coefficient ≥0.5, Pearson correlation p‐value >0.05, and regulation similarity ≠ 0 (Figure 2).\nThe ceRNA network including lncRNAs, miRNAs, and mRNAs. Red and green for all nodes represent up and downregulated directions between normal and cancer tissues, respectively. The node of V shape: miRNA; the node of triangle shape: lncRNA; the node of the cycle: protein coding gene; connecting line of red: lncRNA‐miRNA; connecting line of blue: miRNA‐mRNA\nNext, we performed gene sets enrichment analysis to understand the potential biological effect of this ceRNA network (159 DE mRNA). We first divided the DE mRNA into two groups, including up‐ (n = 33) and downregulated (n = 126) genes. Then, we employed GO, KEGG, and REACTOME datasets and identified 20 significant GO terms, 1 KEGG, and 22 REACTOME for upregulated genes. In meantime, there were 220 significant GO terms and 5 REACTOME for downregulated genes. Then, we displayed the top 20 significant gene sets in Figure 3.\nGO terms, REACTOME, and KEGG interpretation for functions of (A) up and (B) downregulated mRNAs derived from ceRNA network in GC. BP, biological pathway; CC, cellular component; MF, molecular function", "To identify the potential prognosis‐related lncRNAs, we utilized the univariate Cox analysis to filter 76 lncRNAs derived from the above ceRNA network. Then, the lncRNAs and mRNAs with p < 0.05 were further subjected to multivariable Cox analysis with the adjustment of age, gender, histological grade, and TNM stage. Then, there were 11 lncRNAs associated with the overall survival (OS) (Table 2). Intriguingly, they were all negatively related to OS. The mean expression of lncRNA was utilized as a cut‐off value to determine the high‐ and low‐expression groups. Then, the log‐rank test was applied to validate the relationship between the OS and the lncRNA (Figure 4).\nUnivariate and multivariate Cox analysis for lncRNA and mRNA for overall survival of GC\nKaplan‐Meier survival analysis for the correlation of DE lncRNAs with overall survival of the GC patients. Patients with expression ≥ mean expression were considered as high expression and otherwise as low expression", "We assumed those 11 lncRNAs play critical roles in the GC‐related ceRNA network. Thus, we extracted the corresponding 59 mRNAs that interacted with these 11 lncRNAs derived from the ceRNA network. Then, the univariate and multivariate Cox analyses were employed to filter the prognosis‐related mRNAs with the same procedures as above. There were 13 out of 59 mRNAs showing the independent relationship with OS, which was also validated via KM plot (Table 2 and Figure 5).\nKaplan‐Meier survival analysis for the correlation of DE mRNAs with overall survival of the GC patients. Patients with expression ≥ mean expression were considered as high expression and otherwise as low expression\nNext, a sub ceRNA network (24 edges and 27 nodes), including 9 lncRNAs, 5 miRNAs, and 13 mRNAs, was reconstructed (Figure 6A). A total of 9 lncRNA were considered as hub lncRNA, including LINC02731, MIR99AHG, INHBA‐AS1, CCDC144NL‐AS1, VLDLR‐AS1, LIFR‐AS1, A2M‐AS1, LINC01537, and LINC00702. To reveal the potential biological function of this sub‐network, the functional enrichment analysis was carried out, and it found 8 significant GO terms, 8 KEGG, and 30 REACTOME sets, such as MAPK signaling pathway (Figure 6B).\nThe nine‐hub lncRNAs‐associated sub ceRNA network. (A) The sub‐network including lncRNAs, miRNAs, and mRNAs. Blue circle, red triangle, and yellow V shape represent miRNA, mRNA, and lncRNA, respectively. (B) Functional gene sets derived from GO terms, REACTOME, and KEGG", "To validate the above result, two lncRNAs, INHBA‐AS1 and CCDC144NL‐AS1, were selected for further investigation. Subsequently, to validate the findings, siRNA‐mediated silencing of lncRNA was measured by RT‐qPCR, and the knockdown efficiencies of INHBA‐AS1 and CCDC144NL‐AS1 were significant in the MKN45 cell line (Figure 7A).\nThe expression of INHBA‐AS1/hsa‐miR‐98/COL5A2 and CCDC144NL‐AS1/hsa‐miR‐128‐1/MATN3 axis and in vitro functional assay. (A) The expression of INHBA‐AS1 and CCDC144NL‐AS1 with siRNA knockdown. (B) Proliferation assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (C) Migration assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (D) Invasion assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (E) The expression of hsa‐miR‐98 and hsa‐miR‐128‐1 with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (F) The expression of COL5A2 and MATN3 with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. *** Indicated p < 0.001; ** indicated p < 0.01; * indicated p < 0.05\nNext, the result indicated that the knockdown of the two lncRNAs both suppressed cell proliferation as determined by MTS assays (Figure 7B). The results of migration and invasion assay indicated that MKN45 cell line with INHBA‐AS1 and CCDC144NL‐AS1 knockdown significantly less migrated and invaded than their counterpart (Figure 7C,D). Based on the ceRNA axis, we selected hsa‐miR‐98 and hsa‐miR‐128‐1 to verify the relationships between lncRNA and miRNA. Using RT‐qPCR, we observed that the knockdown of INHBA‐AS1 and CCDC144NL‐AS1 significantly increased the hsa‐miR‐98 and hsa‐miR‐128‐1 expression level, respectively (Figure 7E). Accordingly, the expression of COL5A2 and MATN3, which are corresponding lncRNA‐related mRNAs, showed a significant decrease compared to the controls (Figure 7F). These results indicated that INHBA‐AS1 and CCDC144NL‐AS1 might have an oncogenic function and act as ceRNA to sponge miRNAs in GC.", "The study was approved by the Ethics Committee of The First Affiliated Hospital of Wenzhou Medical University. Informed patient consent was not required as the results shown are based upon the data generated by the TCGA database." ]

[ null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null ]

[ "INTRODUCTION", "MATERIAL AND METHODS", "Data resources and differential expression analysis", "ceRNA network construction", "Prognosis‐related lncRNA identification", "Gene sets enrichment analysis", "Cell line culture, RNA extraction, and real‐time PCR", "siRNA knockdown in cell line", "Proliferation assay", "Transwell migration and invasion assay", "Statistical analysis", "RESULTS", "Identification of differently expressed lncRNA, miRNA, and mRNA", "ceRNA network construction and function analysis", "Survival‐associated lncRNA and mRNA identification", "Reconstruction and function analysis of hub lncRNA‐associated ceRNA network", "INHBA‐AS1 and CCDC144NL‐AS1 are potential oncogenes in GC", "DISCUSSION", "CONCLUSIONS", "CONFLICT OF INTEREST", "ETHICAL APPROVAL" ]

[ "Gastric cancer (GC) is one of the most frequent digestive system cancers and is the second cause of cancer mortality worldwide by 2018.\n1\n The cases in China account for more than 40% of the total number of GC worldwide due to a high incidence rate and a large population.\n2\n Moreover, the GC patients were more likely to be in the advanced stage when diagnosed because of non‐early specific symptoms. Unfortunately, the late diagnosis can significantly affect the 5‐year survival rate. In the past decade, due to the advancement of treatment and medicine, the GC prognosis has improved, but it is still not satisfied due to the relatively short disease‐free survival duration.\n3\n Thus, it is challenging and necessary to explore the underlying mechanisms of GC and identify novel biomarkers or treatment targets.\nOn the contrary, the long non‐coding RNA (lncRNA) is a well‐known member of the non‐coding RNA family in the past decade, which is RNA with a length of over 200 nt.\n4\n, \n5\n In the past decade, the accumulating knowledge indicates the important role of the aberrant expression of lncRNA in GC.\n6\n, \n7\n LncRNA can function as sequence‐specific recruitment of proteins, competing for endogenous RNA (ceRNA) regulation, and molecular scaffolding of protein complexes, in which the ceRNA regulation is widely investigated nowadays. It is hypothesized that lncRNA can modulate the miRNA‐regulated mRNA expression by competitively binding miRNAs through endogenous molecular sponges.\n8\n This regulatory mechanism interprets the roles of lncRNA in various cancers, including GC.\n9\n, \n10\n, \n11\n, \n12\n However, the lncRNA‐associated ceRNA networks are far from clear.\nTo further explore the role of specific lncRNA‐miRNA‐mRNA axis in GC, we first construct the ceRNA network via the online database. In addition, the hub lncRNAs with sub‐ceRNA networks related to prognosis were identified. To confirm the reliability and validity of the results, hub lncRNAs were validated in vitro. Overall, the present study aimed to establish a critical ceRNA network and identified novel diagnostic/therapeutic targets.", "Data resources and differential expression analysis We downloaded the RNA sequence data with log2 (fpkm + 1) transformed (lncRNA and mRNA, level 3; Illumina HiSeq RNA‐Seq platform), miRNA sequence data (Illumina HiSeq miRNA‐Seq platform), and clinical information from the Xena dataset (https://xenabrowser.net/datapages/?dataset=TCGA‐STAD.htseq_fpkm‐uq.tsv&host=https%3A%2F%2Fgdc.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443; https://xenabrowser.net/datapages/?dataset=TCGA‐BRCA.htseq_fpkm‐uq.tsv&host=https%3A%2F%2Fgdc.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443). Annotation information for the RNA data was provided by the Ensemble database derived from “biomaRt” package.\n13\n The differentially expressed (DE) lncRNAs, miRNAs, and mRNAs between 32 normal samples and 372 cancer samples were identified by the “limma” package.\n14\n We set the |log2(Fold change)|>1 and p‐value <0.05 as the thresholds to identify DE genes. DE lncRNAs, miRNAs, and mRNAs were presented as a volcano plot. The study was approved by the Ethics Committee of The First Affiliated Hospital of Wenzhou Medical University. Informed patient consent was not required as the results showed are based upon the data generated by the TCGA database.\nWe downloaded the RNA sequence data with log2 (fpkm + 1) transformed (lncRNA and mRNA, level 3; Illumina HiSeq RNA‐Seq platform), miRNA sequence data (Illumina HiSeq miRNA‐Seq platform), and clinical information from the Xena dataset (https://xenabrowser.net/datapages/?dataset=TCGA‐STAD.htseq_fpkm‐uq.tsv&host=https%3A%2F%2Fgdc.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443; https://xenabrowser.net/datapages/?dataset=TCGA‐BRCA.htseq_fpkm‐uq.tsv&host=https%3A%2F%2Fgdc.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443). Annotation information for the RNA data was provided by the Ensemble database derived from “biomaRt” package.\n13\n The differentially expressed (DE) lncRNAs, miRNAs, and mRNAs between 32 normal samples and 372 cancer samples were identified by the “limma” package.\n14\n We set the |log2(Fold change)|>1 and p‐value <0.05 as the thresholds to identify DE genes. DE lncRNAs, miRNAs, and mRNAs were presented as a volcano plot. The study was approved by the Ethics Committee of The First Affiliated Hospital of Wenzhou Medical University. Informed patient consent was not required as the results showed are based upon the data generated by the TCGA database.\nceRNA network construction To construct a ceRNA network, we extracted the interactions between mRNA and miRNA/lncRNA via multiple online datasets. For the interaction between miRNA and mRNA, we used the “miRNAtap” package (https://bioconductor.org/packages/release/bioc/html/miRNAtap.html), which consisted of commonly used five reliable and online datasets, including Pictar (https://pictar.mdc‐berlin.de/), DIANA,\n15\n Targetscan,\n16\n miranda,\n17\n and mirdb.\n18\n Only when the interaction was identified in at least three datasets, we considered it to be potential interactions. In contrast, the lncRNA‐miRNA interactions were predicted through the miRcode.\n19\n\n\nFor ceRNA construction, we submitted the above interactions and expression dataset to the “GDCRNATools” package (http://bioconductor.org/packages/devel/bioc/vignettes/GDCRNATools/inst/doc/GDCRNATools.html), which can further filter the interaction based on those two criteria, (1) expression of lncRNA and mRNA must be positively correlated, and (2) those common miRNAs should play similar roles in regulating the expression of lncRNA and mRNA. Those two factors were indicated via the Pearson correlation and regulation similarity.\n20\n\n\nTo construct a ceRNA network, we extracted the interactions between mRNA and miRNA/lncRNA via multiple online datasets. For the interaction between miRNA and mRNA, we used the “miRNAtap” package (https://bioconductor.org/packages/release/bioc/html/miRNAtap.html), which consisted of commonly used five reliable and online datasets, including Pictar (https://pictar.mdc‐berlin.de/), DIANA,\n15\n Targetscan,\n16\n miranda,\n17\n and mirdb.\n18\n Only when the interaction was identified in at least three datasets, we considered it to be potential interactions. In contrast, the lncRNA‐miRNA interactions were predicted through the miRcode.\n19\n\n\nFor ceRNA construction, we submitted the above interactions and expression dataset to the “GDCRNATools” package (http://bioconductor.org/packages/devel/bioc/vignettes/GDCRNATools/inst/doc/GDCRNATools.html), which can further filter the interaction based on those two criteria, (1) expression of lncRNA and mRNA must be positively correlated, and (2) those common miRNAs should play similar roles in regulating the expression of lncRNA and mRNA. Those two factors were indicated via the Pearson correlation and regulation similarity.\n20\n\n\nPrognosis‐related lncRNA identification To identify the potential prognosis‐related lncRNA, we grouped the samples into high‐ and low‐expression subgroups based on the mean expression. Then, univariate Cox regression analysis was used to find the prognosis‐related lncRNA. Then, the lncRNA achieving statistical significance (p < 0.05) in the univariate analysis was submitted into multivariate Cox analysis adjusting with age, gender, TNM stage, and histological grade. In addition, the KM plot with log‐rank test was carried out to further validate the prognosis‐related lncRNA.\nTo identify the potential prognosis‐related lncRNA, we grouped the samples into high‐ and low‐expression subgroups based on the mean expression. Then, univariate Cox regression analysis was used to find the prognosis‐related lncRNA. Then, the lncRNA achieving statistical significance (p < 0.05) in the univariate analysis was submitted into multivariate Cox analysis adjusting with age, gender, TNM stage, and histological grade. In addition, the KM plot with log‐rank test was carried out to further validate the prognosis‐related lncRNA.\nGene sets enrichment analysis To reveal the function of the lncRNA‐associated ceRNA network, the DE mRNAs derived from ceRNA were subjected to gene sets enrichment analysis based on gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and REACTOME annotation with the clusterProfiler package.\n21\n GO is a structured standard biological model established by the GO consortium, including biological processes (BP), molecular functions (MF), and cellular components (CC).\n22\n KEGG is widely used as a reference for integrating large‐scale molecular datasets generated by sequencing and high‐throughput experimental technologies.\n23\n RACTOME is an open‐source, open access, manually curated, and peer‐reviewed pathway database (https://reactome.org/). Gene sets with a p‐value <0.05 were considered significant.\nTo reveal the function of the lncRNA‐associated ceRNA network, the DE mRNAs derived from ceRNA were subjected to gene sets enrichment analysis based on gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and REACTOME annotation with the clusterProfiler package.\n21\n GO is a structured standard biological model established by the GO consortium, including biological processes (BP), molecular functions (MF), and cellular components (CC).\n22\n KEGG is widely used as a reference for integrating large‐scale molecular datasets generated by sequencing and high‐throughput experimental technologies.\n23\n RACTOME is an open‐source, open access, manually curated, and peer‐reviewed pathway database (https://reactome.org/). Gene sets with a p‐value <0.05 were considered significant.\nCell line culture, RNA extraction, and real‐time PCR GC cell line, MKN45, was purchased from the Cell Bank of the Chinese Academy of Sciences. Cells were incubated at 37℃ in a humidified atmosphere containing 5% CO2 and cultured in Dulbecco's Modified Eagle's Medium (DMEM) (ThermoFisher Scientific) combined with 10% fetal bovine serum (FBS).\nThe primers were designed and purchased from GenePharma (Table 1). The total RNA was extracted using the ReliaPrep RNA Cell Miniprep System (Promega). It was first reversely transcribed into complementary DNA through the iScript Reverse Transcription Kit (Bio‐Rad), then, SYBR Green Supermix (Bio‐Rad) was employed to perform quantitative‐PCR (qPCR), which was run in Bio‐Rad CFX 96 PCR instrument (Bio‐Rad). The 2−ΔΔCT method was used to evaluate the relative gene expression. The GAPDH and U6 were used as the internal control for lncRNA/mRNA and miRNA, respectively.\nForward and reverse primer sequences used for quantitative PCR\nAbbreviations: F, forward; R, reverse.\nGC cell line, MKN45, was purchased from the Cell Bank of the Chinese Academy of Sciences. Cells were incubated at 37℃ in a humidified atmosphere containing 5% CO2 and cultured in Dulbecco's Modified Eagle's Medium (DMEM) (ThermoFisher Scientific) combined with 10% fetal bovine serum (FBS).\nThe primers were designed and purchased from GenePharma (Table 1). The total RNA was extracted using the ReliaPrep RNA Cell Miniprep System (Promega). It was first reversely transcribed into complementary DNA through the iScript Reverse Transcription Kit (Bio‐Rad), then, SYBR Green Supermix (Bio‐Rad) was employed to perform quantitative‐PCR (qPCR), which was run in Bio‐Rad CFX 96 PCR instrument (Bio‐Rad). The 2−ΔΔCT method was used to evaluate the relative gene expression. The GAPDH and U6 were used as the internal control for lncRNA/mRNA and miRNA, respectively.\nForward and reverse primer sequences used for quantitative PCR\nAbbreviations: F, forward; R, reverse.\nsiRNA knockdown in cell line siRNA designed and chemically synthesized by GenePharma. The sequences of siRNA against INHBA‐AS1 were: 5′‐GUCUCAUGACCACAGCUAAtt‐3′ (Sense) and 5′‐UUAGCUGUGGUCAUGAGACct‐3′ (Anti‐Sense), and siRNA against CCDC144NL‐AS1 was 5′‐UAGGUAGAUGGUGGAAUGAtt‐3′ (Sense) and 5′‐UCAUUCCACCAUCUACCUAtg‐3′ (Anti‐Sense). Before the transfection, MKN45 cells were cultured with antibiotic‐free DMEM for 24 h in advance. Transfections of the siRNAs were manipulated via the lipofectamine RNAiMAX (Invitrogen) and the same protocols were carried out following the manufacturer's instruction in all those two lncRNAs.\nsiRNA designed and chemically synthesized by GenePharma. The sequences of siRNA against INHBA‐AS1 were: 5′‐GUCUCAUGACCACAGCUAAtt‐3′ (Sense) and 5′‐UUAGCUGUGGUCAUGAGACct‐3′ (Anti‐Sense), and siRNA against CCDC144NL‐AS1 was 5′‐UAGGUAGAUGGUGGAAUGAtt‐3′ (Sense) and 5′‐UCAUUCCACCAUCUACCUAtg‐3′ (Anti‐Sense). Before the transfection, MKN45 cells were cultured with antibiotic‐free DMEM for 24 h in advance. Transfections of the siRNAs were manipulated via the lipofectamine RNAiMAX (Invitrogen) and the same protocols were carried out following the manufacturer's instruction in all those two lncRNAs.\nProliferation assay To evaluate the proliferation speed, 3 × 103 cells were seeded into a 96‐well plate per well (Greiner bio‐one), and then a time‐series assay every 2 days was carried out in triplicate. We used the MTS CellTiter 96 One Solution Cell Proliferation Assay (Promega) to measure proliferation. With 1, 3, 5, and 7 days, the absorbance at 490 nm was measured using the microtiter plate spectrophotometer (Benchmark Plus, Bio‐Rad) according to the manufacturer's protocol. Subsequently, proliferation was normalized based on the absorbance of the first day and calculated by the changes between the readings.\nTo evaluate the proliferation speed, 3 × 103 cells were seeded into a 96‐well plate per well (Greiner bio‐one), and then a time‐series assay every 2 days was carried out in triplicate. We used the MTS CellTiter 96 One Solution Cell Proliferation Assay (Promega) to measure proliferation. With 1, 3, 5, and 7 days, the absorbance at 490 nm was measured using the microtiter plate spectrophotometer (Benchmark Plus, Bio‐Rad) according to the manufacturer's protocol. Subsequently, proliferation was normalized based on the absorbance of the first day and calculated by the changes between the readings.\nTranswell migration and invasion assay For cell migration and invasion assay, 24‐well transwell inserts with a pore size of 0.8 mm (Corning) were used. After siRNA transfection, 1 × 105 cells were seeded in the upper chamber, and 650 µl of complete medium was added to the lower chamber as a chemo‐attractant. The 100‐µl Matrigel (Corning) with a concentration of 20 mg/ml was pre‐coated above the insert for invasion assay. The cells were allowed to migrate or invade toward the chamber for 12 and 16 h, respectively. The migrated and invaded cells below the membrane were fixed with 4% paraformaldehyde, stained with DAPI (Beyotime Biotechnology), and quantified from microscopic fields.\nFor cell migration and invasion assay, 24‐well transwell inserts with a pore size of 0.8 mm (Corning) were used. After siRNA transfection, 1 × 105 cells were seeded in the upper chamber, and 650 µl of complete medium was added to the lower chamber as a chemo‐attractant. The 100‐µl Matrigel (Corning) with a concentration of 20 mg/ml was pre‐coated above the insert for invasion assay. The cells were allowed to migrate or invade toward the chamber for 12 and 16 h, respectively. The migrated and invaded cells below the membrane were fixed with 4% paraformaldehyde, stained with DAPI (Beyotime Biotechnology), and quantified from microscopic fields.\nStatistical analysis The student's t test was used to compare the continuous variables. The chi‐squared test was employed for the categorical variables. p values have two tails and only when it is less than 0.05 was considered significant. All the figures and statistical analysis were carried out via R software (version 3.6.1).\nThe student's t test was used to compare the continuous variables. The chi‐squared test was employed for the categorical variables. p values have two tails and only when it is less than 0.05 was considered significant. All the figures and statistical analysis were carried out via R software (version 3.6.1).", "We downloaded the RNA sequence data with log2 (fpkm + 1) transformed (lncRNA and mRNA, level 3; Illumina HiSeq RNA‐Seq platform), miRNA sequence data (Illumina HiSeq miRNA‐Seq platform), and clinical information from the Xena dataset (https://xenabrowser.net/datapages/?dataset=TCGA‐STAD.htseq_fpkm‐uq.tsv&host=https%3A%2F%2Fgdc.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443; https://xenabrowser.net/datapages/?dataset=TCGA‐BRCA.htseq_fpkm‐uq.tsv&host=https%3A%2F%2Fgdc.xenahubs.net&removeHub=https%3A%2F%2Fxena.treehouse.gi.ucsc.edu%3A443). Annotation information for the RNA data was provided by the Ensemble database derived from “biomaRt” package.\n13\n The differentially expressed (DE) lncRNAs, miRNAs, and mRNAs between 32 normal samples and 372 cancer samples were identified by the “limma” package.\n14\n We set the |log2(Fold change)|>1 and p‐value <0.05 as the thresholds to identify DE genes. DE lncRNAs, miRNAs, and mRNAs were presented as a volcano plot. The study was approved by the Ethics Committee of The First Affiliated Hospital of Wenzhou Medical University. Informed patient consent was not required as the results showed are based upon the data generated by the TCGA database.", "To construct a ceRNA network, we extracted the interactions between mRNA and miRNA/lncRNA via multiple online datasets. For the interaction between miRNA and mRNA, we used the “miRNAtap” package (https://bioconductor.org/packages/release/bioc/html/miRNAtap.html), which consisted of commonly used five reliable and online datasets, including Pictar (https://pictar.mdc‐berlin.de/), DIANA,\n15\n Targetscan,\n16\n miranda,\n17\n and mirdb.\n18\n Only when the interaction was identified in at least three datasets, we considered it to be potential interactions. In contrast, the lncRNA‐miRNA interactions were predicted through the miRcode.\n19\n\n\nFor ceRNA construction, we submitted the above interactions and expression dataset to the “GDCRNATools” package (http://bioconductor.org/packages/devel/bioc/vignettes/GDCRNATools/inst/doc/GDCRNATools.html), which can further filter the interaction based on those two criteria, (1) expression of lncRNA and mRNA must be positively correlated, and (2) those common miRNAs should play similar roles in regulating the expression of lncRNA and mRNA. Those two factors were indicated via the Pearson correlation and regulation similarity.\n20\n\n", "To identify the potential prognosis‐related lncRNA, we grouped the samples into high‐ and low‐expression subgroups based on the mean expression. Then, univariate Cox regression analysis was used to find the prognosis‐related lncRNA. Then, the lncRNA achieving statistical significance (p < 0.05) in the univariate analysis was submitted into multivariate Cox analysis adjusting with age, gender, TNM stage, and histological grade. In addition, the KM plot with log‐rank test was carried out to further validate the prognosis‐related lncRNA.", "To reveal the function of the lncRNA‐associated ceRNA network, the DE mRNAs derived from ceRNA were subjected to gene sets enrichment analysis based on gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and REACTOME annotation with the clusterProfiler package.\n21\n GO is a structured standard biological model established by the GO consortium, including biological processes (BP), molecular functions (MF), and cellular components (CC).\n22\n KEGG is widely used as a reference for integrating large‐scale molecular datasets generated by sequencing and high‐throughput experimental technologies.\n23\n RACTOME is an open‐source, open access, manually curated, and peer‐reviewed pathway database (https://reactome.org/). Gene sets with a p‐value <0.05 were considered significant.", "GC cell line, MKN45, was purchased from the Cell Bank of the Chinese Academy of Sciences. Cells were incubated at 37℃ in a humidified atmosphere containing 5% CO2 and cultured in Dulbecco's Modified Eagle's Medium (DMEM) (ThermoFisher Scientific) combined with 10% fetal bovine serum (FBS).\nThe primers were designed and purchased from GenePharma (Table 1). The total RNA was extracted using the ReliaPrep RNA Cell Miniprep System (Promega). It was first reversely transcribed into complementary DNA through the iScript Reverse Transcription Kit (Bio‐Rad), then, SYBR Green Supermix (Bio‐Rad) was employed to perform quantitative‐PCR (qPCR), which was run in Bio‐Rad CFX 96 PCR instrument (Bio‐Rad). The 2−ΔΔCT method was used to evaluate the relative gene expression. The GAPDH and U6 were used as the internal control for lncRNA/mRNA and miRNA, respectively.\nForward and reverse primer sequences used for quantitative PCR\nAbbreviations: F, forward; R, reverse.", "siRNA designed and chemically synthesized by GenePharma. The sequences of siRNA against INHBA‐AS1 were: 5′‐GUCUCAUGACCACAGCUAAtt‐3′ (Sense) and 5′‐UUAGCUGUGGUCAUGAGACct‐3′ (Anti‐Sense), and siRNA against CCDC144NL‐AS1 was 5′‐UAGGUAGAUGGUGGAAUGAtt‐3′ (Sense) and 5′‐UCAUUCCACCAUCUACCUAtg‐3′ (Anti‐Sense). Before the transfection, MKN45 cells were cultured with antibiotic‐free DMEM for 24 h in advance. Transfections of the siRNAs were manipulated via the lipofectamine RNAiMAX (Invitrogen) and the same protocols were carried out following the manufacturer's instruction in all those two lncRNAs.", "To evaluate the proliferation speed, 3 × 103 cells were seeded into a 96‐well plate per well (Greiner bio‐one), and then a time‐series assay every 2 days was carried out in triplicate. We used the MTS CellTiter 96 One Solution Cell Proliferation Assay (Promega) to measure proliferation. With 1, 3, 5, and 7 days, the absorbance at 490 nm was measured using the microtiter plate spectrophotometer (Benchmark Plus, Bio‐Rad) according to the manufacturer's protocol. Subsequently, proliferation was normalized based on the absorbance of the first day and calculated by the changes between the readings.", "For cell migration and invasion assay, 24‐well transwell inserts with a pore size of 0.8 mm (Corning) were used. After siRNA transfection, 1 × 105 cells were seeded in the upper chamber, and 650 µl of complete medium was added to the lower chamber as a chemo‐attractant. The 100‐µl Matrigel (Corning) with a concentration of 20 mg/ml was pre‐coated above the insert for invasion assay. The cells were allowed to migrate or invade toward the chamber for 12 and 16 h, respectively. The migrated and invaded cells below the membrane were fixed with 4% paraformaldehyde, stained with DAPI (Beyotime Biotechnology), and quantified from microscopic fields.", "The student's t test was used to compare the continuous variables. The chi‐squared test was employed for the categorical variables. p values have two tails and only when it is less than 0.05 was considered significant. All the figures and statistical analysis were carried out via R software (version 3.6.1).", "Identification of differently expressed lncRNA, miRNA, and mRNA There were 1485 DE lncRNAs (1260 up‐regulation and 225 down‐regulation), 312 DE miRNAs (290 up‐regulation 12 down‐regulation), and 4260 DE mRNAs (2347 up‐regulation and 1913 down‐regulation) identified (Figure 1).\nVolcano plots showing up and downregulated (A) lncRNA, (B) miRNA, and (C) mRNA. The red dots represent high expression lncRNA/miRNA/mRNA with LogFC ≥1 and FDR <0.05, and the blue dots represent low expression of lncRNA/miRNA/mRNA with LogFC ≤ −1 and FDR <0.05\nThere were 1485 DE lncRNAs (1260 up‐regulation and 225 down‐regulation), 312 DE miRNAs (290 up‐regulation 12 down‐regulation), and 4260 DE mRNAs (2347 up‐regulation and 1913 down‐regulation) identified (Figure 1).\nVolcano plots showing up and downregulated (A) lncRNA, (B) miRNA, and (C) mRNA. The red dots represent high expression lncRNA/miRNA/mRNA with LogFC ≥1 and FDR <0.05, and the blue dots represent low expression of lncRNA/miRNA/mRNA with LogFC ≤ −1 and FDR <0.05\nceRNA network construction and function analysis To construct the ceRNA, we first extracted the interaction between DE miRNAs and DE lncRNAs, and DE miRNAs and DE mRNAs as well via multiple online datasets. Finally, there were 6082 pairs of lncRNA‐miRNA interactions predicted, and 938 interactions between miRNA and mRNA were identified. Based on those interactions, we established 24847 potential lncRNA‐miRNA‐mRNA axes consisting of 892 lncRNAs, 18 miRNAs, and 278 mRNAs preliminarily.\nThen, the ceRNA network was conducted via the “GDCRNATools” package with “gdcCEAnalysis” function, which utilizes the Pearson correlation and regulation pattern to further determine the promising ceRNA. Finally, a ceRNA network (909 edges and 253 nodes), including 76 lncRNA, 18 miRNA, and 159 mRNA, was constructed with Pearson correlation coefficient ≥0.5, Pearson correlation p‐value >0.05, and regulation similarity ≠ 0 (Figure 2).\nThe ceRNA network including lncRNAs, miRNAs, and mRNAs. Red and green for all nodes represent up and downregulated directions between normal and cancer tissues, respectively. The node of V shape: miRNA; the node of triangle shape: lncRNA; the node of the cycle: protein coding gene; connecting line of red: lncRNA‐miRNA; connecting line of blue: miRNA‐mRNA\nNext, we performed gene sets enrichment analysis to understand the potential biological effect of this ceRNA network (159 DE mRNA). We first divided the DE mRNA into two groups, including up‐ (n = 33) and downregulated (n = 126) genes. Then, we employed GO, KEGG, and REACTOME datasets and identified 20 significant GO terms, 1 KEGG, and 22 REACTOME for upregulated genes. In meantime, there were 220 significant GO terms and 5 REACTOME for downregulated genes. Then, we displayed the top 20 significant gene sets in Figure 3.\nGO terms, REACTOME, and KEGG interpretation for functions of (A) up and (B) downregulated mRNAs derived from ceRNA network in GC. BP, biological pathway; CC, cellular component; MF, molecular function\nTo construct the ceRNA, we first extracted the interaction between DE miRNAs and DE lncRNAs, and DE miRNAs and DE mRNAs as well via multiple online datasets. Finally, there were 6082 pairs of lncRNA‐miRNA interactions predicted, and 938 interactions between miRNA and mRNA were identified. Based on those interactions, we established 24847 potential lncRNA‐miRNA‐mRNA axes consisting of 892 lncRNAs, 18 miRNAs, and 278 mRNAs preliminarily.\nThen, the ceRNA network was conducted via the “GDCRNATools” package with “gdcCEAnalysis” function, which utilizes the Pearson correlation and regulation pattern to further determine the promising ceRNA. Finally, a ceRNA network (909 edges and 253 nodes), including 76 lncRNA, 18 miRNA, and 159 mRNA, was constructed with Pearson correlation coefficient ≥0.5, Pearson correlation p‐value >0.05, and regulation similarity ≠ 0 (Figure 2).\nThe ceRNA network including lncRNAs, miRNAs, and mRNAs. Red and green for all nodes represent up and downregulated directions between normal and cancer tissues, respectively. The node of V shape: miRNA; the node of triangle shape: lncRNA; the node of the cycle: protein coding gene; connecting line of red: lncRNA‐miRNA; connecting line of blue: miRNA‐mRNA\nNext, we performed gene sets enrichment analysis to understand the potential biological effect of this ceRNA network (159 DE mRNA). We first divided the DE mRNA into two groups, including up‐ (n = 33) and downregulated (n = 126) genes. Then, we employed GO, KEGG, and REACTOME datasets and identified 20 significant GO terms, 1 KEGG, and 22 REACTOME for upregulated genes. In meantime, there were 220 significant GO terms and 5 REACTOME for downregulated genes. Then, we displayed the top 20 significant gene sets in Figure 3.\nGO terms, REACTOME, and KEGG interpretation for functions of (A) up and (B) downregulated mRNAs derived from ceRNA network in GC. BP, biological pathway; CC, cellular component; MF, molecular function\nSurvival‐associated lncRNA and mRNA identification To identify the potential prognosis‐related lncRNAs, we utilized the univariate Cox analysis to filter 76 lncRNAs derived from the above ceRNA network. Then, the lncRNAs and mRNAs with p < 0.05 were further subjected to multivariable Cox analysis with the adjustment of age, gender, histological grade, and TNM stage. Then, there were 11 lncRNAs associated with the overall survival (OS) (Table 2). Intriguingly, they were all negatively related to OS. The mean expression of lncRNA was utilized as a cut‐off value to determine the high‐ and low‐expression groups. Then, the log‐rank test was applied to validate the relationship between the OS and the lncRNA (Figure 4).\nUnivariate and multivariate Cox analysis for lncRNA and mRNA for overall survival of GC\nKaplan‐Meier survival analysis for the correlation of DE lncRNAs with overall survival of the GC patients. Patients with expression ≥ mean expression were considered as high expression and otherwise as low expression\nTo identify the potential prognosis‐related lncRNAs, we utilized the univariate Cox analysis to filter 76 lncRNAs derived from the above ceRNA network. Then, the lncRNAs and mRNAs with p < 0.05 were further subjected to multivariable Cox analysis with the adjustment of age, gender, histological grade, and TNM stage. Then, there were 11 lncRNAs associated with the overall survival (OS) (Table 2). Intriguingly, they were all negatively related to OS. The mean expression of lncRNA was utilized as a cut‐off value to determine the high‐ and low‐expression groups. Then, the log‐rank test was applied to validate the relationship between the OS and the lncRNA (Figure 4).\nUnivariate and multivariate Cox analysis for lncRNA and mRNA for overall survival of GC\nKaplan‐Meier survival analysis for the correlation of DE lncRNAs with overall survival of the GC patients. Patients with expression ≥ mean expression were considered as high expression and otherwise as low expression\nReconstruction and function analysis of hub lncRNA‐associated ceRNA network We assumed those 11 lncRNAs play critical roles in the GC‐related ceRNA network. Thus, we extracted the corresponding 59 mRNAs that interacted with these 11 lncRNAs derived from the ceRNA network. Then, the univariate and multivariate Cox analyses were employed to filter the prognosis‐related mRNAs with the same procedures as above. There were 13 out of 59 mRNAs showing the independent relationship with OS, which was also validated via KM plot (Table 2 and Figure 5).\nKaplan‐Meier survival analysis for the correlation of DE mRNAs with overall survival of the GC patients. Patients with expression ≥ mean expression were considered as high expression and otherwise as low expression\nNext, a sub ceRNA network (24 edges and 27 nodes), including 9 lncRNAs, 5 miRNAs, and 13 mRNAs, was reconstructed (Figure 6A). A total of 9 lncRNA were considered as hub lncRNA, including LINC02731, MIR99AHG, INHBA‐AS1, CCDC144NL‐AS1, VLDLR‐AS1, LIFR‐AS1, A2M‐AS1, LINC01537, and LINC00702. To reveal the potential biological function of this sub‐network, the functional enrichment analysis was carried out, and it found 8 significant GO terms, 8 KEGG, and 30 REACTOME sets, such as MAPK signaling pathway (Figure 6B).\nThe nine‐hub lncRNAs‐associated sub ceRNA network. (A) The sub‐network including lncRNAs, miRNAs, and mRNAs. Blue circle, red triangle, and yellow V shape represent miRNA, mRNA, and lncRNA, respectively. (B) Functional gene sets derived from GO terms, REACTOME, and KEGG\nWe assumed those 11 lncRNAs play critical roles in the GC‐related ceRNA network. Thus, we extracted the corresponding 59 mRNAs that interacted with these 11 lncRNAs derived from the ceRNA network. Then, the univariate and multivariate Cox analyses were employed to filter the prognosis‐related mRNAs with the same procedures as above. There were 13 out of 59 mRNAs showing the independent relationship with OS, which was also validated via KM plot (Table 2 and Figure 5).\nKaplan‐Meier survival analysis for the correlation of DE mRNAs with overall survival of the GC patients. Patients with expression ≥ mean expression were considered as high expression and otherwise as low expression\nNext, a sub ceRNA network (24 edges and 27 nodes), including 9 lncRNAs, 5 miRNAs, and 13 mRNAs, was reconstructed (Figure 6A). A total of 9 lncRNA were considered as hub lncRNA, including LINC02731, MIR99AHG, INHBA‐AS1, CCDC144NL‐AS1, VLDLR‐AS1, LIFR‐AS1, A2M‐AS1, LINC01537, and LINC00702. To reveal the potential biological function of this sub‐network, the functional enrichment analysis was carried out, and it found 8 significant GO terms, 8 KEGG, and 30 REACTOME sets, such as MAPK signaling pathway (Figure 6B).\nThe nine‐hub lncRNAs‐associated sub ceRNA network. (A) The sub‐network including lncRNAs, miRNAs, and mRNAs. Blue circle, red triangle, and yellow V shape represent miRNA, mRNA, and lncRNA, respectively. (B) Functional gene sets derived from GO terms, REACTOME, and KEGG\nINHBA‐AS1 and CCDC144NL‐AS1 are potential oncogenes in GC To validate the above result, two lncRNAs, INHBA‐AS1 and CCDC144NL‐AS1, were selected for further investigation. Subsequently, to validate the findings, siRNA‐mediated silencing of lncRNA was measured by RT‐qPCR, and the knockdown efficiencies of INHBA‐AS1 and CCDC144NL‐AS1 were significant in the MKN45 cell line (Figure 7A).\nThe expression of INHBA‐AS1/hsa‐miR‐98/COL5A2 and CCDC144NL‐AS1/hsa‐miR‐128‐1/MATN3 axis and in vitro functional assay. (A) The expression of INHBA‐AS1 and CCDC144NL‐AS1 with siRNA knockdown. (B) Proliferation assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (C) Migration assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (D) Invasion assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (E) The expression of hsa‐miR‐98 and hsa‐miR‐128‐1 with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (F) The expression of COL5A2 and MATN3 with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. *** Indicated p < 0.001; ** indicated p < 0.01; * indicated p < 0.05\nNext, the result indicated that the knockdown of the two lncRNAs both suppressed cell proliferation as determined by MTS assays (Figure 7B). The results of migration and invasion assay indicated that MKN45 cell line with INHBA‐AS1 and CCDC144NL‐AS1 knockdown significantly less migrated and invaded than their counterpart (Figure 7C,D). Based on the ceRNA axis, we selected hsa‐miR‐98 and hsa‐miR‐128‐1 to verify the relationships between lncRNA and miRNA. Using RT‐qPCR, we observed that the knockdown of INHBA‐AS1 and CCDC144NL‐AS1 significantly increased the hsa‐miR‐98 and hsa‐miR‐128‐1 expression level, respectively (Figure 7E). Accordingly, the expression of COL5A2 and MATN3, which are corresponding lncRNA‐related mRNAs, showed a significant decrease compared to the controls (Figure 7F). These results indicated that INHBA‐AS1 and CCDC144NL‐AS1 might have an oncogenic function and act as ceRNA to sponge miRNAs in GC.\nTo validate the above result, two lncRNAs, INHBA‐AS1 and CCDC144NL‐AS1, were selected for further investigation. Subsequently, to validate the findings, siRNA‐mediated silencing of lncRNA was measured by RT‐qPCR, and the knockdown efficiencies of INHBA‐AS1 and CCDC144NL‐AS1 were significant in the MKN45 cell line (Figure 7A).\nThe expression of INHBA‐AS1/hsa‐miR‐98/COL5A2 and CCDC144NL‐AS1/hsa‐miR‐128‐1/MATN3 axis and in vitro functional assay. (A) The expression of INHBA‐AS1 and CCDC144NL‐AS1 with siRNA knockdown. (B) Proliferation assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (C) Migration assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (D) Invasion assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (E) The expression of hsa‐miR‐98 and hsa‐miR‐128‐1 with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (F) The expression of COL5A2 and MATN3 with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. *** Indicated p < 0.001; ** indicated p < 0.01; * indicated p < 0.05\nNext, the result indicated that the knockdown of the two lncRNAs both suppressed cell proliferation as determined by MTS assays (Figure 7B). The results of migration and invasion assay indicated that MKN45 cell line with INHBA‐AS1 and CCDC144NL‐AS1 knockdown significantly less migrated and invaded than their counterpart (Figure 7C,D). Based on the ceRNA axis, we selected hsa‐miR‐98 and hsa‐miR‐128‐1 to verify the relationships between lncRNA and miRNA. Using RT‐qPCR, we observed that the knockdown of INHBA‐AS1 and CCDC144NL‐AS1 significantly increased the hsa‐miR‐98 and hsa‐miR‐128‐1 expression level, respectively (Figure 7E). Accordingly, the expression of COL5A2 and MATN3, which are corresponding lncRNA‐related mRNAs, showed a significant decrease compared to the controls (Figure 7F). These results indicated that INHBA‐AS1 and CCDC144NL‐AS1 might have an oncogenic function and act as ceRNA to sponge miRNAs in GC.", "There were 1485 DE lncRNAs (1260 up‐regulation and 225 down‐regulation), 312 DE miRNAs (290 up‐regulation 12 down‐regulation), and 4260 DE mRNAs (2347 up‐regulation and 1913 down‐regulation) identified (Figure 1).\nVolcano plots showing up and downregulated (A) lncRNA, (B) miRNA, and (C) mRNA. The red dots represent high expression lncRNA/miRNA/mRNA with LogFC ≥1 and FDR <0.05, and the blue dots represent low expression of lncRNA/miRNA/mRNA with LogFC ≤ −1 and FDR <0.05", "To construct the ceRNA, we first extracted the interaction between DE miRNAs and DE lncRNAs, and DE miRNAs and DE mRNAs as well via multiple online datasets. Finally, there were 6082 pairs of lncRNA‐miRNA interactions predicted, and 938 interactions between miRNA and mRNA were identified. Based on those interactions, we established 24847 potential lncRNA‐miRNA‐mRNA axes consisting of 892 lncRNAs, 18 miRNAs, and 278 mRNAs preliminarily.\nThen, the ceRNA network was conducted via the “GDCRNATools” package with “gdcCEAnalysis” function, which utilizes the Pearson correlation and regulation pattern to further determine the promising ceRNA. Finally, a ceRNA network (909 edges and 253 nodes), including 76 lncRNA, 18 miRNA, and 159 mRNA, was constructed with Pearson correlation coefficient ≥0.5, Pearson correlation p‐value >0.05, and regulation similarity ≠ 0 (Figure 2).\nThe ceRNA network including lncRNAs, miRNAs, and mRNAs. Red and green for all nodes represent up and downregulated directions between normal and cancer tissues, respectively. The node of V shape: miRNA; the node of triangle shape: lncRNA; the node of the cycle: protein coding gene; connecting line of red: lncRNA‐miRNA; connecting line of blue: miRNA‐mRNA\nNext, we performed gene sets enrichment analysis to understand the potential biological effect of this ceRNA network (159 DE mRNA). We first divided the DE mRNA into two groups, including up‐ (n = 33) and downregulated (n = 126) genes. Then, we employed GO, KEGG, and REACTOME datasets and identified 20 significant GO terms, 1 KEGG, and 22 REACTOME for upregulated genes. In meantime, there were 220 significant GO terms and 5 REACTOME for downregulated genes. Then, we displayed the top 20 significant gene sets in Figure 3.\nGO terms, REACTOME, and KEGG interpretation for functions of (A) up and (B) downregulated mRNAs derived from ceRNA network in GC. BP, biological pathway; CC, cellular component; MF, molecular function", "To identify the potential prognosis‐related lncRNAs, we utilized the univariate Cox analysis to filter 76 lncRNAs derived from the above ceRNA network. Then, the lncRNAs and mRNAs with p < 0.05 were further subjected to multivariable Cox analysis with the adjustment of age, gender, histological grade, and TNM stage. Then, there were 11 lncRNAs associated with the overall survival (OS) (Table 2). Intriguingly, they were all negatively related to OS. The mean expression of lncRNA was utilized as a cut‐off value to determine the high‐ and low‐expression groups. Then, the log‐rank test was applied to validate the relationship between the OS and the lncRNA (Figure 4).\nUnivariate and multivariate Cox analysis for lncRNA and mRNA for overall survival of GC\nKaplan‐Meier survival analysis for the correlation of DE lncRNAs with overall survival of the GC patients. Patients with expression ≥ mean expression were considered as high expression and otherwise as low expression", "We assumed those 11 lncRNAs play critical roles in the GC‐related ceRNA network. Thus, we extracted the corresponding 59 mRNAs that interacted with these 11 lncRNAs derived from the ceRNA network. Then, the univariate and multivariate Cox analyses were employed to filter the prognosis‐related mRNAs with the same procedures as above. There were 13 out of 59 mRNAs showing the independent relationship with OS, which was also validated via KM plot (Table 2 and Figure 5).\nKaplan‐Meier survival analysis for the correlation of DE mRNAs with overall survival of the GC patients. Patients with expression ≥ mean expression were considered as high expression and otherwise as low expression\nNext, a sub ceRNA network (24 edges and 27 nodes), including 9 lncRNAs, 5 miRNAs, and 13 mRNAs, was reconstructed (Figure 6A). A total of 9 lncRNA were considered as hub lncRNA, including LINC02731, MIR99AHG, INHBA‐AS1, CCDC144NL‐AS1, VLDLR‐AS1, LIFR‐AS1, A2M‐AS1, LINC01537, and LINC00702. To reveal the potential biological function of this sub‐network, the functional enrichment analysis was carried out, and it found 8 significant GO terms, 8 KEGG, and 30 REACTOME sets, such as MAPK signaling pathway (Figure 6B).\nThe nine‐hub lncRNAs‐associated sub ceRNA network. (A) The sub‐network including lncRNAs, miRNAs, and mRNAs. Blue circle, red triangle, and yellow V shape represent miRNA, mRNA, and lncRNA, respectively. (B) Functional gene sets derived from GO terms, REACTOME, and KEGG", "To validate the above result, two lncRNAs, INHBA‐AS1 and CCDC144NL‐AS1, were selected for further investigation. Subsequently, to validate the findings, siRNA‐mediated silencing of lncRNA was measured by RT‐qPCR, and the knockdown efficiencies of INHBA‐AS1 and CCDC144NL‐AS1 were significant in the MKN45 cell line (Figure 7A).\nThe expression of INHBA‐AS1/hsa‐miR‐98/COL5A2 and CCDC144NL‐AS1/hsa‐miR‐128‐1/MATN3 axis and in vitro functional assay. (A) The expression of INHBA‐AS1 and CCDC144NL‐AS1 with siRNA knockdown. (B) Proliferation assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (C) Migration assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (D) Invasion assay with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (E) The expression of hsa‐miR‐98 and hsa‐miR‐128‐1 with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. (F) The expression of COL5A2 and MATN3 with INHBA‐AS1 and CCDC144NL‐AS1 knockout, respectively. *** Indicated p < 0.001; ** indicated p < 0.01; * indicated p < 0.05\nNext, the result indicated that the knockdown of the two lncRNAs both suppressed cell proliferation as determined by MTS assays (Figure 7B). The results of migration and invasion assay indicated that MKN45 cell line with INHBA‐AS1 and CCDC144NL‐AS1 knockdown significantly less migrated and invaded than their counterpart (Figure 7C,D). Based on the ceRNA axis, we selected hsa‐miR‐98 and hsa‐miR‐128‐1 to verify the relationships between lncRNA and miRNA. Using RT‐qPCR, we observed that the knockdown of INHBA‐AS1 and CCDC144NL‐AS1 significantly increased the hsa‐miR‐98 and hsa‐miR‐128‐1 expression level, respectively (Figure 7E). Accordingly, the expression of COL5A2 and MATN3, which are corresponding lncRNA‐related mRNAs, showed a significant decrease compared to the controls (Figure 7F). These results indicated that INHBA‐AS1 and CCDC144NL‐AS1 might have an oncogenic function and act as ceRNA to sponge miRNAs in GC.", "So far, the GC is one of the top‐ranking digestive cancers and has become a worldwide public concern. Thus, it is important to investigate the potential biomarkers and therapeutic targets. In our study, we identified an lncRNA‐associated ceRNA network involving GC tumorigenesis, which was based on the analysis of gene expression data obtained from the TCGA databases. Then, we identified nine hub lncRNAs accompanied with the sub ceRNA network related to OS. Among those nine lncRNAs, we validated the two of them, INHBA‐AS1 and CCDC144NL‐AS1, in vitro and found they were promising oncogene in GC.\nAs mentioned above, lncRNA can influence the expression of mRNA via competitively binding to shared miRNA, which is defined as ceRNA and may play a critical role in the regulation of cancer development and progression, including GC.\n8\n For instance, LINC00152 regulated GACAT3 via miR‐103, and both are positively associated with poor clinicopathological characteristics in colorectal cancer.\n24\n For GC, lncRNA LINC01133 can inhibit GC progression by sponging hsa‐miR‐106a‐3p and then influence the APC expression.\n10\n In addition, lncRNA PTENP1 can regulate PTEN expression via binding to miR‐106b and miR‐93 in GC.\n11\n Our study identified a ceRNA network in GC involved in upregulation of MET activates PTK2 signaling, MET promotes cell motility and non‐integrin membrane‐ECM interactions. The MET activating the PTK2 signaling is related to MET receptor activating the focal adhesion kinase FAK1, which plays crucial role in focal adhesions (FAs). Specifically, FAs are large macromolecular complexes of integrins that mediate cell‐ECMs interactions and facilitate the metastatic process in cancer.\n25\n, \n26\n Previous studies identified that FAs is strongly associated with metastasis and lower survival rates.\n25\n, \n27\n, \n28\n, \n29\n Moreover, FAs can impact various tumor behaviors, such as migration, invasion, and proliferation.\n30\n Then, MET promotes cell motility, which may contribute to GC progression.\n31\n Non‐integrin membrane‐ECM interactions, such as dystroglycan and 37/67 laminin receptor, is found to be related to various epithelial cancers.\n32\n\n\nSubsequently, we further identified nine‐hub lncRNAs, including LINC02731, MIR99AHG, INHBA‐AS1, CCDC144NL‐AS1, VLDLR‐AS1, LIFR‐AS1, A2M‐AS1, LINC01537, and LINC00702. Those hub lncRNAs‐associated ceRNA subnetwork is involved in actin filament binding and MAPK signaling pathway. A filament is a form of dense meshwork generated by lamellipodia, which facilitates cellular movement and plays anessential role in tumor cell metastasis.\n33\n MAPK signaling pathway is involved in various promoting‐cancer mechanisms, such as anti‐drug, inflammation, and immune evasion.\n34\n, \n35\n, \n36\n, \n37\n In terms of individual lncRNAs, most of them were related to the development and progression of various cancers in other studies. For instance, MIR100HG has been validated as an oncogene in the development of myeloid leukemia in vitro.\n38\n In addition, it was positively related to worse prognosis in GC via datasets other than TCGA.\n39\n LncRNA INHBA‐AS1 can promote multiple invasion features, including cell growth, migration, and invasion in oral squamous cell carcinoma, which targets on hsa‐miR‐143‐3p.\n40\n The INHBA‐AS1 in GC plasma was overexpressed compared to it in controls without further function assay.\n41\n Knockdown of lncRNA CCDC144NL‐AS1 attenuated migration and invasion in endometrial stromal cells.\n42\n The expression of VLDLR‐AS1 was independently related to the worse prognosis in thymoma.\n43\n The LIFR‐AS1/hsa‐miR‐29a/TNFAIP3 axis played an effect on the resistance of photodynamic therapy in colorectal cancer.\n44\n High expression of LIFR‐AS1 was correlated with poor survival in GC.\n45\n Upregulated A2M‐AS1 was associated with invasion and migration in breast cancer.\n46\n Besides, LINC00702 enhanced the progression of ovarian cancer through increased EZH2 expression.\n47\n Then, LINC00702/has‐miR‐4652‐3p/ZEBI axis can promote the progression of malignant meningioma through activating the Wnt/β‐catenin pathway.\n48\n Taken together, most of those nine‐hub lncRNAs were promising tumor‐promoting genes in diverse cancer and were worthwhile for further investigation in GC.\nThen, to validate our findings, INHBA‐AS1 and CCDC144NL‐AS1 and related axis were further verified in vitro and showed the promoting influence on proliferation, migration, and invasion. This indicated that two lncRNAs were promising oncogenic genes in GC. In terms of the related mRNA, INHBA‐AS1‐regulated COL5A2 and CCDC144NL‐AS1‐regulated MATN3 are related to GC prognosis.\n49\n, \n50\n MATN3 is a member of the Matrilin protein family, a noncollagenous extracellular matrix, which is associated with diverse cancers.\n51\n, \n52\n, \n53\n Specifically, it can induce the expression of MMP1, MMP3, MMP13, pro‐inflammatory cytokines, iNOS, and COX2, indicating MATN3 can regulate extracellular matrix degradation.\n52\n The COL5A2, collagen‐type V alpha 2 chain, encodes an alpha chain for one of the low abundances fibrillar collagens. It plays a critical role in the pathological process in multiple cancers including colorectal cancer, ovarian cancer, and bladder cancer.\n54\n, \n55\n Moreover, COL5A2 was strongly correlated with cell‐extracellular matrix organization, vascularization, and EMTs process function, and those functions were known to be involved in cancer invasion and metastasis.\n54\n Those findings may partially explain the functional effect of INHBA‐AS1 and CCDC144NL‐AS1 in vitro.\nSo far, there are a few studies that are similar to ours. One study applied GEO dataset and paired GC/non‐tumorous tissues to identify DE lncRNAs and miRNAs, respectively. Then, TarBase and miRcode were used to establish a lncRNA‐miRNA‐mRNA network. Subsequently, one pair of ceRNA was validated in vitro without functional assays.\n56\n Another one used a small number of poorly differentiated GC and normal tissues to conducted numerous DE lncRNAs and selected one lncRNA, LINC02535, for further study.\n57\n Specifically, DE genes related to LINC02535 were filtered and used to conduct functional and protein—protein interaction analysis. LINC2535 alone was positively associated with cell proliferation, migration, invasion, and wound healing and negatively related to cell apoptosis via in vitro assays. Besides, one study utilized the TCGA GC dataset to identify DE lncRNAs, and selected LINC01234 for further validation.\n58\n Then, LINC01234 was proved to be positively associated with poor clinical characteristics in GC patients. Besides, they identified potential functions of LINC01234‐ and LINCO1234‐related network, including transcription factor (TF)‐lncRNA regulation, miRNA‐lncRNA relationship, as well as lncRNA‐RNA‐binding proteins interactions, via bioinformatics analysis; however, there was no functional assay. In terms of our study, we also used the TCGA GC data, which is the most well‐known and comprehensive dataset so far. Then, we utilized not only TarBase and miRcode but also other well‐known online tools to predict lncRNA‐miRNA and miRNA‐mRNA relationships and established the ceRNA network. Except for those, we further validated two pairs of ceRNAs in vitro with functional assays. Taken together, our study applied the latest and comprehensive dataset and well design methods to conduct the updated critical ceRNA network in GC. This may compensate for the shortage in this field.\nThere are several limitations to our study. First, we only employed TCGA dataset, a frequently used online comprehensive cancer database. Second, although we combined a well‐designed bioinformatics study and in vitro validation, there was no in‐depth laboratory evidence, for example, a dual‐luciferase reporter assay, and mice model. Third, there is no clinical result in the present study. Taken together, a few vital experiments accompanied by the prospective stududies will be helpful to further validate our findings in the future.", "So far, the role of the ceRNA network in GC is far from understood. In the present study, we established a promising lncRNA‐miRNA‐mRNA triple ceRNA network and identified a ceRNA subnetwork with nine‐hub lncRNAs involved in the prognosis of GC patients. Then, we validated two lncRNAs, INHBA‐AS1, and CCDC144NL‐AS1, accompanied their corresponding miRNAs and mRNAs, as potential oncogenic roles in GC. These findings need to be further confirmed in the future.", "The authors declare no conflict of interest in preparing this article.", "The study was approved by the Ethics Committee of The First Affiliated Hospital of Wenzhou Medical University. Informed patient consent was not required as the results shown are based upon the data generated by the TCGA database." ]

[ null, "materials-and-methods", null, null, null, null, null, null, null, null, null, "results", null, null, null, null, null, "discussion", "conclusions", "COI-statement", null ]

[ "bioinformatics", "competing endogenous RNA", "gastric cancer", "prognosis" ]

[ 402, 179, 196, 92, 141, 191, 90, 120, 131, 57, 106, 388, 181, 289, 350, 41 ]

[ "as1", "lncrna", "expression", "cerna", "lncrnas", "mirna", "mrna", "network", "gc", "de" ]

[ "roles lncrna cancers", "lncrnas promising tumor", "rna lncrna known", "lncrnas overall survival", "prognosis related lncrnas" ]

[ "Adult", "Aged", "Biomarkers, Tumor", "Chemoradiotherapy", "Chromosomes, Human, Pair 1", "Chromosomes, Human, Pair 19", "Female", "Glioma", "Humans", "Isocitrate Dehydrogenase", "Middle Aged", "Mutation", "Neoplasm Grading" ]

[]

[]

[]

[ "Introduction", "Materials and Methods", "Results", "Discussion" ]

[ "Primary malignant brain tumors account for 1.4% of new cancer diagnosis in the United States and 2.7% of the deaths are due to central nervous system (CNS) tumors (Siegel et al., 2016). Incidence of CNS tumors in India ranges from 5 to 10 per 100,000 population with an increasing trend and accounts for 2% of all malignancies (Nair et al.,2015; Yeole et al., 2008). Astrocytomas (38.7%) were the most common primary tumors with the majority being high grade gliomas (59.5%) (Jalali et al., 2008; Nair et al., 2015; Yeole et al., 2008). Glioblastoma multiforme (GBM) accounts for more than half of all primary brain tumours (Jha et al., 2011).The 2007 World Health Organisation (WHO) Classification of tumours of the CNS was based on histopathological analysis (Louis et al., 2007). In 2014, the Haarlem consensus conference under auspices of International society of Neuro-Pathology (ISNP) established guidelines to incorporate molecular findings into histology of brain tumour diagnosis (Louis et al., 2014). Following it, the new CNS WHO classification came into picture in 2016 with major revisions based on molecular parameters to establish brain tumor diagnosis to illustrate prognostic behavior (Louis et al., 2016). However, there is no marked change in current standard of care in high grade gliomas (HGG). Maximal safe surgical resection (MSR) followed by radiotherapy (RT) with or without chemotherapy (CT) remains the standard of care. CT for anaplastic glioma (Anaplastic astrocytoma{AA}, Anaplastic oligodendroglioma {AO} and mixed Anaplastic Oligo-astrocytoma {AOA}) is not standardized even though use of procarbazine, lomustine, vincristine (PCV) improves overall survival (OS) and progression free survival (PFS) (Carincross et al., 2013; Van den bent et al., 2013). However, considering the OS and PFS improvement in GBM with the use of Temozolamide (TMZ) (Stupp et al., 2005), studies are being performed attempting the use of TMZ in Grade 2 and 3 gliomas (van den Bent et al., 2017). In this study we have tried to report the incidence of 1p/19q codeletion, IDH 1 mutation, ATRX deletion and p53 in HGGs in the Indian population and analyse variation of outcomes with respect to individual molecular marker and sub-group thereof. The rationale of the study was to predict response to standard chemo-radiation based on molecular markers in HGGs.", "This is a tertiary-hospital based prospective interventional study in which patients were accrued between January 2016 to January 2018 and followed up till April 2019.The project was approved by institutional ethics committee. The inclusion criteria in the current study were patients with histopathologically proven HGG, age ranging from 18-70 years, normal renal and liver function test and adequate bone marrow reserves. On the other hand, patients with prior history of any malignancy, previous administration of any form of chemotherapy or radiotherapy and patients unfit for chemotherapy and or radiotherapy were excluded from the study. Informed written consent was obtained from all patients. All the patients were assessed for symptoms and the duration of the presenting symptoms were recorded at the time of registration. Detailed neurological examination was performed. The Karnofsky performance score (KPS) were documented for all patients. All the patients were assessed radiologically using Gadolinium enhanced magnetic resonance imaging (MRI) to assess pre-treatment site, size of tumour and other features like edema, necrosis, calcification and vascularity. Histopathological evaluation for confirmation of diagnosis was done in all patients at the central pathology of the institute. The assessment also included molecular studies. Chromosome 1p and 19q deletion status were done by in Fluorescence situ hybridisation studies. IDH 1 mutation was analysed by immunohistochemistry (IHC) using mutation specific antibody. IHC was also used for assessing ATRX. All the patients underwent MSR followed by treatment with concurrent chemo-radiation (CRT) with standard dose and fractionation irrespective of O6-methylguanine-DNA methyl-transferase (MGMT) promotor methylation status followed by adjuvant chemotherapy. \nTotal dose of RT was 60 Gray in 30 fractions (2 Gray per fraction, 5 days per week for total 6 weeks) with 3-dimensional conformal radiotherapy (3D-CRT) (LINAC, Elekta) along with concurrent TMZ followed by adjuvant TMZ for 6 months. The dose of TMZ was 75 mg/ m2 concurrent with RT on all days including public holidays followed by 6 cycles of adjuvant TMZ at 150 -200 mg/ m2 given 5 days every 28 days (Stupp et al., 2005). The patients were followed up monthly for 6 months, then bi-monthly for 6 months followed by three-monthly visits thereafter. They were assessed clinically on monthly basis during each follow-up. The post-treatment gadolinium enhanced MRI was done at 12 weeks after CRT completion. On the basis of response to treatment, cases were divided into complete responders (CR), partial responder (PR), stable disease (SD) and progressive disease (PD) as per the revised response assessment in neuro-oncology (RANO) criteria (Wen et al., 2010). The response assessment was done in the joint clinic with the opinion of neuro-surgeon, oncologist and radiologist. Patient data and treatment files were updated on each subsequent follow-up post-treatment. There were no loss to follow-up. The primary end point was to correlate relationship of various molecular markers 1p/19q codeletion, IDH 1 mutation, ATRX loss and p53 mutation as well different subgroup clusters with similar marker expressions in terms of PFS and OS. The secondary end point was to evaluate frequency of 1p/19q codeletion, IDH 1 mutation, ATRX deletion and p53 in patients with HGGs. The hypothesis was that outcomes of HGGs treated with standard chemo-radiation may be different in terms of different molecular markers assessed and subgroups thereof, i.e., 1p/19q codeletion, IDH 1 mutation, ATRX loss and p53 mutation. \n\nStatistical analysis \n\nOn the basis of response to treatment cases were be divided into CR, PR, SD and PD. The relationship of various molecular markers as well different subgroup clusters with similar marker expression were compared in terms of response to treatment and overall survival. Statistical analysis was performed using statistical package for sciences (SPSS version 23.0). PFS was defined as the time from the day of registration to date of progression or death. OS was measured from the date of registration to the date of death from any cause. Survival analysis was done by Kaplan Meier method with log rank test and all events were calculated from the date of registration.", "Forty-two cases of HGG underwent four molecular studies. All the patients received adjuvant RT dose of 60 Gray with concurrent and adjuvant TMZ for six cycles. The patient characteristics are detailed in Table 1. The median age (range) was 46.5 years (18-67years); 30 (71.4%) were male. The presenting features included headache (95%), vomiting (65%) and seizures (48%). The median duration (mean, SD, range) of symptoms was 6 months (31, 19.8, 1-72). The median (mean, SD, range) diameter of disease was 5.0 cm (3.1, 3.2, 3.4-5.8) on gadolinium enhanced MRI. The median time (mean, SD, range) from the most recent surgery to start of radiotherapy being 1.5 months (2.8, 7.8, 5-6). Following immobilization patients were treated on a 6 /10 MV linear accelerator. The median RT dose (range) was 60Gy (59.4-60 Gray) given in 1.8-2.0 Gray fractions, 5 fractions/week for a duration of 6 weeks. The molecular profiling details are mentioned in Table 2.\nResponse assessment was performed as per RANO criteria (Wen et al., 2010) at a median follow up duration of 22 months was as follows: CR, PR and SD (n=18), PD (n=24). Of the 7 patients with 1p/19q codeletion, one patient had progressive disease at 1 year. All the seven patients with 1p/19q codeletion had IDH1 mutation. Kaplan Meyer curves for the OS and PFS for all patients and 1p/19q codeleted subgroup is as shown in Figure 1 (A and B). Although there was no statistical difference in the OS (p-0.10), there was significant PFS improvement (85.7% vs 62.3%, p-0.01) with 1p/19q codeletion. The details of the variation in OS and PFS with all the molecular analysis are given in Table 3. The supplementary Table shows the variation in OS and PFS with various patient variables: age, gender, KPS, tumour size and tumour location.", "MSR followed by adjuvant chemo-radiation is the most acceptable treatment strategy in the management of HGGs. EORTC 26899/NCIC established the role of concurrent and adjvuant TMZ as standard of care for glioblastoma (Stupp et al., 2005; Stupp et al., 2009). They reported increase in median survival to 14.6 months with CRT as compared to 12.1 months with only RT and 5 year OS increase from 1.9% to 9.8%. At the same time another study (Hegi et al., 2005) reported on subset analysis of 206 glioblastoma patients regarding epigenetic silencing of MGMT by methylation. Regardless of TMZ use, MGMT methylation was associated with improved OS (Median survival 15.3 vs 11.8 months). So, MGMT methylation is both prognostic and predictive for response to TMZ. Use of TMZ in unmethylated patients is controversial while some argued the subset was underpowered and patients may still benefit. Two landmark studies (RTOG 9402, EORTC 26951) were published regarding role of chemotherapy in anaplastic glioma in 2013. Cairncross (Cairncross et al., 2006; Cairncross et al., 2013; Cairncross et al., 2014) reported IDH mutated subgroup with 1p/19q codeleted median survival (MS) was 14.7 years as compared to 1p/19q intact (MS 5.5 years) and IDH wild type (MS-1.8 years) with use of chemotherapy (PCV), while in this trial chemotherapy was used prior to radiotherapy. Simultaneously EORTC 26951 (van den Bent et al., 2006; van den Bent et al., 2013) used adjuvant PCV after radiotherapy and compared with RT alone. They concluded that OS significantly improved with PCV (42.3 months vs 30.6 months). Significant improvement in PFS was noted in both 1p19q co-deleted (157 months vs 50 months) and 1p/19q non-codeleted (15 vs 9 months). IDH mutation and 1p/19q codeletions were independently significant on multivariate prognostic model. In our study, all the 1p/19q codeleted patients had IDH1 mutations. We have observed an improved PFS (85.7% vs 62.3%, p-0.011) and OS (85.7% vs 65.8%, p–0.104) in IDH mutated 1p/19q codeleted subgroup as compared to those with 1p/19q non-codeleted. \nThe role of TMZ in anaplastic gliomas has not been established. Of late, PCV has been replaced by TMZ in many studies performed in diffuse gliomas. A single arm phase II study, RTOG BR0131 (Vogelbaum MA et al., 2015), compared pre-RT and concurrent TMZ with RT with the historical cohort of RTOG 9402. Updates of this trial demonstrated 83% 6-year OS for co-deleted patients (compared with 67% for RTOG 9402, although no statistical difference was found). Similarly, the NOA-O4 study (Wick et al., 2016), which was investigating the optimum treatment sequence for anaplastic gliomas compared TMZ with PCV. In the subgroup analysis, they did not find significant difference in the OS between TMZ and PCV arms in co-deleted patients although there was a trend towards longer time to treatment failure for PCV. Also, the initial interim results of the CATNON study (van den Bent et al., 2017) show adjuvant TMZ was associated with improved survival (5-year OS: 55.9% vs 44.1%) in non-co-deleted anaplastic gliomas. The results have been further rebuffed in the second interim analysis (van den Bent et al., 2019). It shows adjuvant TMZ improves OS and CRT with concurrent TMZ shows trend towards improved OS in IDH mutated tumour. The results from four important trials of radiotherapy and chemotherapy have been discussed by Soltys et al. along with molecular grading of gliomas (Soltys et al., 2018). The future outcomes of studies like CODEL NCCTG N0577 (Jaeckle K et al., 2016) and detailed analysis of CATNON will further clarify the role of TMZ in HGG. RTOG 0131, NOA-04 and early result of CATNON all suggest this is a reasonable substitution to PCV. In the light of the above trials, we have tried to analyse the outcome in HGGs using RT with concurrent TMZ and correlate them with the various molecular markers.\nTill now limited studies have been conducted in Indian patients regarding frequency of 1p/19q codeletion, IDH 1 mutation, ATRX deletion and TP53 and further to correlate individual markers signature with response to chemoradiation and impact on overall survival in high grade gliomas. One study Jha et al., (2011) reported molecular profiling of GBMs in the Indian population. They reported that EGFR (37.3%) and PTEN (54.9%) mutations are relatively common in primary GBM, while TP53 (66.7%) and IDH1 (44.4%) mutations were more common in secondary GBMs. IDH 1 mutation were reported in 68.8%, 85.7% and 12.8% in grade 2, 3 and 4 gliomas, respectively. IDH 1 mutation is present in over 80% of secondary GBMs (Yan et al., 2009), while IDH2 gene have been reported in around 3% of gliomas (Hartmann et al., 2009). Another study (Parsons et al., 2008) has found improved survival with IDH1 mutations. The current study found IDH1 mutations in 42.8% patients. The cases with IDH 1 mutations had better 20-month OS and 20-month PFS, 82.4% vs 30.6% (p-0.014) and 53.6% vs 29.8% (p-0.035), respectively. \nThe incidence of 1p/19q co-deletions among GBMs is low compared to anaplastic oligodendrogliomas. We found 1p/19q codeletion in roughly 17% patients. 1p/19q co-deletions alone did not seem to be associated with good outcomes in one report with GBM (Kaneshiro et al., 2009, Smith et al., 2000), although another study reported GBM long-term survivors with 1p/19q co-deletions (Burton et al., 2002). The current study found non-significant improvement in 20-month OS (85.7% vs 65.8%, p-0.104), but significantly improved 20-month PFS (85.7% vs 62.3%, p-0.011) in 1p/1pq codeleted patients. In light of the fact that 1p/19q codeletion entails improved prognosis and chemosensitivity in anaplastic gliomas, and histo-pathological analysis alone is not sufficient for diagnosis anymore, genetic and molecular testing should be routinely performed for these tumours (Idbaih et al., 2007). The incidence of somatic p53 alterations in GBMs is around 85%, as reported in a study by the Cancer Genome Atlas (Cancer genome atlas, 2008). The present study found p53 mutation in approximately 43% cases.\nSummarising the data in the aforementioned literature, the role of molecular markers in the treatment of HGGs is still an intriguing matter (Clark et al., 2013). With the changing landscape of diagnosis of gliomas, the newer data is largely revolving around next generation sequencing (Parsons et al., 2008) and the Cancer genome atlas (Cancer genome atlas, 2008). Shortcomings of the current study include small sample size which did not allow assessment of various molecular subgroups possible in gliomas (Suzuki et al., 2015). MGMT methylation analysis was not available for all the patients. The optimal duration of treatment with adjuvant temozolamide, 6 months versus 12 months, is still not established. Tumour located in midline were not assessed separately (Weller et al., 2017). \nIn conclusion, the current study has reported the incidence of various molecular markers in HGG and the outcomes after chemo-radiation with regards to these markers. With the integrated classification of diffuse gliomas as advocated by WHO (2016), this study was initiated to standardise treatment related decisions in HGGs. Codeletion of 1p/19q in HGG with IDH1 mutation is associated with a significantly favourable progression free survival (p=0.011). However, longer follow up with a larger cohort is required to assess overall survival (p=0.104). Studies with a larger number of patients and a longer follow up will help to answer the question.\nPatient Demographic Characteristics\nKPS, Karnofsky Performance Score; CR, Complete response; SD, Stable disease; PD, Progressive disease\nFrequency of Molecular Markers\nVariation of OS and PFS with Various Molecular Markers\n*The OS and PFS mentioned here are 20-month OS and 20-month PFS.\nFigure Showing the Kaplan Meyer Curves for OS (A) and PFS (B) Variation with 1p/19q Co-deletion Status (codeleted vs noncodeleted)" ]

[ "intro", "materials|methods", "results", "discussion" ]

[ "High grade glioma", "glioblastoma", "GBM", "anaplastic glioma", "radiotherapy", "1p/19q codeltion", "IDH 1 mutation" ]

[]

[ "patients", "1p", "1p 19q", "19q", "os", "tmz", "study", "molecular", "pfs", "months" ]

[ "diagnosis gliomas newer", "molecular grading gliomas", "malignant brain tumors", "classification tumours cns", "primary brain tumours" ]

[ "Deep Learning", "Genes, p53", "Humans", "Mutation", "Staining and Labeling", "Stomach Neoplasms" ]

[ "INTRODUCTION", "Part I: Tests with The Cancer Genome Atlas whole-slide image datasets", "Part II: Tests on the external cohorts", "Statistical analysis", "RESULTS", "DISCUSSION", "CONCLUSION" ]

[ "Molecular tests to identify specific mutations in solid tumors have improved our ability to stratify cancer patients for more selective treatment regimens[1]. Therefore, molecular tests to detect various mutations are recommended for some tumors, including EGFR mutations in lung cancer, KRAS in colorectal cancer, and BRAF in melanoma. However, it is not routinely applied to cancer patients because molecular tests are not cost- and time-efficient[2]. Furthermore, the clinical significance of many mutations is still not well understood. For example, mutation profiling of gastric cancer (GC) is still proceeding, and the meaning of each mutation is not clearly understood[3]. GC is the fifth most common cancer and the third leading cause of cancer-related deaths worldwide[4]. It is important to evaluate the relationship between the mutational status and clinical characteristics of GC to improve the clinical outcomes of GC patients. Furthermore, many targeted drugs for treating various tumors are not effective in GC therapy because GC is not enriched with known driver mutations[5]. Therefore, research to characterize the roles of GC-related genes on the clinical behavior of tumors and the potential response to targeted therapies will have immense importance for the improvement of treatment response in GC[6]. A cost- and time-effective method to determine the mutational status of GC patients is necessary to promote these studies.\nRecently, deep learning (DL) has been increasingly implemented to predict the mutational status from hematoxylin and eosin (H and E)-stained tissue slides of various cancers[7-11]. The H and E-stained tissue slides were made for almost all cancer patients for basic diagnostic studies by pathologists[12]. Therefore, mutation prediction from the H and E-stained tissue slide based on a computational method can be a cost- and time-effective alternative tool for conventional molecular tests[13-15]. Although it has long been recognized that the morphological features of tissue architecture reflect the underlying molecular alterations[16,17], the features are not easily identifiable by human evaluators[18,19]. DL offers an alternative solution to overcome the limitations of a visual examination of tissue morphology by pathologists. By combining feature learning and model fitting in a unified step, DL can capture the most discriminative features for a given task directly from a large set of tissue images[20]. Digitization of tissue slides has been rapidly increasing after the approval of digitized whole-slide images (WSIs) for diagnostic purposes[21]. Digitized tissue data are rapidly accumulating with their associated mutational profiles. Therefore, the DL-based analysis of tissue slides for the mutational status of cancer tissues has immense potential as an alternative or complementary method for conventional molecular tests. \nBased on the potential of DL for the detection of mutations from digitized tissue slides, in a previous study, we successfully built DL-based classifiers for the prediction of mutational status of APC, KRAS, PIK3CA, SMAD4, and TP53 genes in colorectal cancer tissue slides[11]. This study investigated the feasibility of classifiers for mutations in the CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes in GC tissues. First, the classifiers were trained and tested for GC tissue slides from The Cancer Genome Atlas (TCGA). The generalizability of the classifiers was tested using an external dataset. Then, new classifiers were trained for combined datasets from TCGA and external datasets to investigate the effect of the extended datasets. The results suggest that it is feasible to predict mutational status directly from tissue slides with deep learning-based classifiers. Finally, as the classifiers for KRAS, PIK3CA, and TP53 mutations for both colorectal and GC were available, we also analyzed the generalizability of the DL-based mutation classifiers trained for different cancer types.", "\nPatient cohort: The Cancer Genome Atlas (TCGA) provides extensive archives of digital pathology slides with multi-omics test results to test the possibility of tissue-based mutation detection[22]. After a carefully review of all the WSIs in the TCGA GC dataset (TCGA-STAD), we eliminated WSIs with poor scan quality and very small tumor contents. We selected slides from 25, 19, 34, 64, and 160 patients, which were confirmed to have mutations in CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. There were more than two slides for many patients in the TCGA dataset, with a maximum of four slides for some patients. However, in many cases, one or two slides contained only normal tissues. We excluded normal slides and selected a maximum of two tumor-containing slides per patient. The final number of frozen tissue slides was 34, 26, 50, 94, and 221 and that of formalin-fixed paraffin-embedded (FFPE) tissue slides was 27, 19, 34, 66, and 174 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. We selected 183 patients with wild-type CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes. Therefore, the same patients with wild-type genes for CDH1, ERBB2, KRAS, PIK3CA, and TP53 can be involved in the training of every classifier as a non-mutated group. This may help the comparison of the different classifiers more standardized because they all have the same group of patients as the wild-type group. The TCGA IDs of the patients in each group are listed in Supplementary Table 1. Our previous studies recognized that a DL model cannot perform optimally for both training and testing unless the dataset is forced to have similar amounts of data between classes[23]. Therefore, we limited the difference in patient numbers between the mutation and wild-type groups to less than 1.4 fold by random sampling. For example, only 35 of the 183 wild-type patients were randomly selected as the CDH1 wild-type group because there were only 25 CDH1 mutated patients. Ten-fold cross-validation was performed based on these randomly sampled wild-type patients. However, the classifiers yielded better results when the tumor patches from all wild-type patients other than the test sets were randomly sampled to match the 1.4 fold data ratio of wild-type/mutation groups for training, as this strategy could include a greater variety of tissue images. Therefore, we included all wild-type patients other than the test sets during training and randomly selected patients during testing.\n\nDeep learning model: In general, a WSI is too large to be analyzed simultaneously using a deep neural network. Therefore, the analysis results for small image patches are integrated for conclusion. We divided a WSI into non-overlapping patches of 360 × 360 pixel tissue images at 20× magnification to detect mutational status. To make the classification process fully automated, artifacts in the WSIs such as air bubbles, compression artifacts, out-of-focus blur, pen markings, tissue folding, and white background should be removed automatically. A simple convolutional neural network (CNN), termed as tissue/non-tissue classifier, was trained to discriminate these various artifacts all at once. The structure of the tissue/non-tissue classifier was described in our previous study[11]. The tissue/non-tissue classifier could filter out almost 99.9% of the improper tissue patches. Then, tissue patches classified as “improper” by the tissue/non-tissue classifier were removed, and the remaining “proper” tissue patches were collected. For the tumor or mutation classifiers described below, only proper tissue patches were analyzed (Figure 1).\n\nWorkflow for the fully automated prediction of mutation. Tissue image patches with tumor probability higher than 0.9 were selected by sequential application of the tissue/non-tissue and normal/tumor classifiers. Then the tumor patches were classified into the wild-type or mutated patches. The patch-level probabilities of mutation are averaged to yield the slide-level probability.\nMorphologic features reflecting mutations in specific genes might be expressed mainly in tumor tissues rather than normal tissues[24,25]. Therefore, tumor tissues should be separated from the WSI to predict the mutational status of the WSI. In a previous study, we successfully built normal/tumor classifiers for various tumors, including GC[26]. We concluded that frozen and FFPE slides should be separately analyzed using a deep neural network due to their different morphologic features. Thus, we adopted the normal/tumor classifiers for frozen and FFPE tissue slides from a previous study to delineate the normal/tumor gastric tissues for the frozen and FFPE slides of the TCGA-STAD dataset in the present study. Mutation classifiers were trained separately for the selected tumor patches for frozen and FFPE tissues. We selected tumor patches with a tumor probability higher than 0.9 to collect tissue patches with evident tumor features. We adopted a patient-level ten-fold cross-validation to completely characterize the TCGA-STAD dataset. Therefore, patients in each mutation/wild-type group for the five genes were separated into ten different folds, and one of the ten folds was used to test the classifiers trained with data from the other nine folds. Therefore, ten different classifiers were trained and tested for each group. The same label for all tumor tissue patches in a WSI as either ‘wild-type’ or ‘mutated’ were assigned based on the mutational status of the patient. Thereafter, the Inception-v3 model, a widely used CNN architecture, was trained to classify the tumor patches into ‘wild-type’ or ‘mutated’ tissues, as in our previous study on mutation prediction in colorectal cancer[11]. We fully trained the network from the beginning and did not adopt a transfer-learning scheme. The average probability of all tumor patches in a WSI was calculated to determine the slide-level mutation probability of a WSI. The Inception-v3 model was implemented using the TensorFlow DL library (http://tensorflow.org), and the network was trained with a mini-batch size of 128 and cross-entropy loss function as a loss function. For training, we used the RMSProp optimizer, with an initial learning rate of 0.1, weight decay of 0.9, momentum of 0.9, and epsilon of 1.0. Ten percent of the training slides were used as the validation dataset, and training was stopped when the loss for the validation data started to increase. Data augmentation techniques, including random horizontal/vertical flipping and random rotations by 90°, were applied to the tissue patches during training. Color normalization was applied to the tissue patches to avoid the effect of stain differences[27,28]. At least five classifiers were trained on each fold of mutation for the frozen and FFPE WSIs separately. The classifier with the best area under the curve (AUC) for the receiver operating characteristic (ROC) curves on the test dataset was included in the results. The ROC curves for fold with the lowest AUC, highest AUC, and the concatenated results for data from all ten folds are shown in the figures.\nIn summary, a WSI is analyzed as follows: 1. The whole slide is split into non-overlapping 360 × 360 pixel tissue patches, 2. Proper tissue patches are selected by tissue/non-tissue classifier, 3. Only tumor patches with tumor probability higher than 0.9 are selected, 4. High probability tumor patches are classified by each wild-type/mutation classifier, 5. The probabilities of tumor patches are averaged to obtain the slide-level mutation probability. The number of tissue patches used for the training of all mutation prediction models is summarized in Supplementary Table 2. The average number of training epochs for each classifier is summarized in Supplementary Table 3.", "\nPatient cohort: GC tissue slides were collected from 96 patients who had previously undergone surgical resection at Seoul St. Mary’s Hospital between 2017 and 2020 (SSMH dataset). An Aperio slide scanner (Leica Biosystems) was used to scan the FFPE slides. The Institutional Review Board of the College of Medicine at the Catholic University of Korea approved this study (KC19SESI0787). \n\nMutation prediction on SSMH dataset: For CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, 6, 6, 12, 11, and 39 patients were confirmed to have the mutations, respectively. Thirty-eight patients had wild-type genes for all five genes. For CDH1, ERBB2, KRAS, and PIK3CA genes, we selected the number of wild-type patients to be 1.4 times that of mutated patients. For TP53, all 38 patients with wild-type genes were enrolled. The normal/tumor classifier for TCGA FFPE tissues was also used to discriminate the tumor tissue patches of SSMH WSIs. Our previous study showed that the normal/tumor classifier for TCGA-STAD was valid for SSMH FFPE slides[29]. First, the mutational status of the SSMH slides was analyzed by classifiers trained on TCGA-STAD FFPE WSIs. Subsequently, new classifiers were trained using both TCGA and SSMH FFPE tissues. Patient-level three-fold cross validation was applied to the SSMH datasets because the number of mutated patients was not sufficient for ten-fold cross-validation.", "To demonstrate the performance of each classifier, the ROC curves and their AUCs are presented in the figures. For the concatenated results from all ten folds, 95% confidence intervals (CIs) were also presented using the percentile bootstrap method. In addition, the accuracy, sensitivity, specificity, and F1 score of the classification results of mutation prediction models with cutoff values for maximal Youden index (sensitivity + specificity - 1) were presented. We used a permutation test with 1000 iterations to compare the differences between the two paired or unpaired ROC curves when a comparison was necessary[30]. Statistical significance was set at P < 0.05.", "Tissue patches with high tumor probability were automatically collected from a WSI by sequentially applying the tissue/non-tissue and normal/tumor classifiers to 360 × 360 pixels tissue image patches (Figure 1). Then, classifiers to distinguish the mutational status of CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes in the tumor tissue patches from the TCGA-STAD frozen and FFPE WSI datasets were separately trained with a patient-level ten-fold cross-validation scheme. \nThe classification results of the TCGA-STAD WSIs are presented in Figures 2 to 6 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes. Results for the frozen and FFPE tissues are presented in the upper and lower part of each figure, respectively. Panels A and C demonstrated the representative binary heatmaps of tissue patches classified as wild-type or mutated tissues. The WSIs with gene mutation correctly classified as mutation, with wild-type gene correctly classified as wild-type, with gene mutation falsely classified as wild-type, and with wild-type gene falsely classified as mutation are presented from left to right for panels A and C. The binary heatmaps were drawn with the wild-type/mutation discrimination threshold set to 0.5. We simply set the threshold to 0.5, because every classifier for different folds had different optimal thresholds. Slide-level ROC curves for folds with the lowest and highest AUCs are presented to demonstrate the differences in the performance between folds (left and middle ROC curves in each figure). Finally, the slide-level ROC curves for the concatenated results from all ten folds were used to infer the overall performance (right ROC curves). The results for the CDH1 gene are shown in Figure 2. The AUCs per fold ranged from 0.833 to 1.000 for frozen WSIs and from 0.833 to 1.000 for FFPE WSIs. The AUCs for the concatenated results were 0.842 (95%CI: 0.749-0.936) and 0.781 (95%CI: 0.645-0.917) for frozen and FFPE WSIs, respectively. For ERBB2 (Figure 3), the lowest and highest AUCs per fold were 0.667 and 1.000, respectively, for both frozen and FFPE WSIs. The concatenated AUCs were 0.751 (95%CI: 0.631-0.871) and 0.661 (95%CI: 0.501-0.821), respectively. For the KRAS gene (Figure 4), the AUCs per fold were between 0.775 and 1.000 for frozen WSIs and between 0.750 and 1.000 for FFPE WSIs. The concatenated AUCs were 0.793 (95%CI: 0.706-0.879) and 0.858 (95%CI: 0.738-0.979) for frozen and FFPE WSIs, respectively. For the PIK3CA gene (Figure 5), the concatenated AUC for the frozen WSIs was 0.862 (95%CI: 0.809-0.916), with a range of 0.705 to 0.990. For FFPE WSIs, the lowest and highest AUCs per fold were 0.675 and 1.000, respectively, yielding a concatenated AUC of 0.828 (95%CI: 0.750-0.907). Lastly, the results for the TP53 gene are presented in Figure 6. The AUCs per fold were between 0.666 to 0.810 for frozen WSIs and between 0.702 to 0.847 for FFPE WSIs. The concatenated AUCs were 0.727 (95%CI: 0.683-0.771) and 0.727 (95%CI: 0.671-0.784) for frozen and FFPE WSIs, respectively. For the colorectal cancer dataset from TCGA, mutation classification results for frozen tissues were better than those for FFPE tissues in some genes[11]. However, there were no significant differences between the frozen and FFPE tissues in the TCGA-STAD dataset (P = 0.491, 0.431, 0.187, 0.321, and 0.613 between the concatenated AUCs for the frozen and FFPE tissues by Venkatraman’s permutation test for unpaired ROC curves for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. For a clearer assessment of the performance of each model, the accuracy, sensitivity, specificity, and F1 score of the classification results are presented in Table 1.\n\nClassification results of CDH1 gene in the The Cancer Genome Atlas gastric cancer dataset. A: Representative whole slide images (WSIs) of the frozen slides with CDH1 gene mutation correctly classified as mutation, with wild-type gene correctly classified as wild-type, with gene mutation falsely classified as wild-type, and with wild-type gene falsely classified as mutation, from left to right; B: Receiver operating characteristic (ROC) curves for the fold with lowest area under the curve (AUC), for the fold with highest AUC, and for the concatenated results of all ten folds, from left to right, obtained with the classifiers trained with the frozen tissues; C and D: Same as A and B but the results were for the formalin-fixed paraffin-embedded WSIs. CDH1-M: CDH1 mutated, CDH1-W: CDH1 wild-type.\n\nClassification results of ERBB2 gene in the The Cancer Genome Atlas gastric cancer dataset. The configuration of the figure is the same as in Figure 2. A and B: Upper panels are results for the frozen tissue and lower panels; C and D: Results for the formalin-fixed paraffin-embedded tissues. ERBB2-M: ERBB2 mutated, ERBB2-W: ERBB2 wild-type.\n\nClassification results of KRAS gene in the The Cancer Genome Atlas gastric cancer dataset. The configuration of the figure is the same as in Figure 2. A and B: Upper panels are results for the frozen tissue and lower panels; C and D: Results for the formalin-fixed paraffin-embedded tissues. KRAS-M: KRAS mutated, KRAS-W: KRAS wild-type.\n\nClassification results of PIK3CA gene in the The Cancer Genome Atlas gastric cancer dataset. The configuration of the figure is the same as in Figure 2. A and B: Upper panels are results for the frozen tissue and lower panels; C and D: Results for the formalin-fixed paraffin-embedded tissues. PIK3CA-M: PIK3CA mutated, PIK3CA-W: PIK3CA wild-type.\n\nClassification results of TP53 gene in the The Cancer Genome Atlas gastric cancer dataset. The configuration of the figure is the same as in Figure 2. A and B: Upper panels are results for the frozen tissue and lower panels; C and D: Results for the formalin-fixed paraffin-embedded tissues. TP53-M: TP53 mutated, TP53-W: TP53 wild-type.\nAccuracy, sensitivity, specificity, and F1 score of the classification results of mutation prediction models for the The Cancer Genome Atlas datasets\nTCGA: The Cancer Genome Atlas; FFPE: Formalin-fixed paraffin-embedded.\nThe performance of a DL model on an external dataset should be tested to validate the generalizability of the trained model. Therefore, we collected GC FFPE WSIs with matching mutation data from Seoul St. Mary’s Hospital (SSMH dataset). The normal/tumor classifier for TCGA-STAD FFPE tissues was also applied to select tissue patches with high tumor probabilities. Thereafter, the mutation classifier for each gene trained on the TCGA-STAD FFPE tissues was tested on the SSMH dataset. The slide-level ROC curves for the CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes are presented in Supplementary Figure 1. The AUCs for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes were 0.667, 0.630, 0.657, 0.688, and 0.572, respectively. For the KRAS, PIK3CA, and TP53 genes, the performance of the TCGA-trained mutation classifiers on the SSMH dataset were worse than that of the TCGA dataset (P = 0.389, P = 0.849, P < 0.05, P < 0.05, and P < 0.05 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively, by Venkatraman’s permutation test for unpaired ROC curves). These results demonstrate that the mutation classifiers trained with TCGA-STAD WSI datasets had limited generalizability. It is of interest if the performance can be enhanced by training the classifiers with expanded datasets, including both TCGA and SSMH datasets. Cancer tissues from different ethnic groups can show different features[16,19]; therefore, the performance of the classifiers can be improved by mixing the datasets. When the classifiers trained with the mixed datasets were used, the performance on the SSMH dataset was generally improved because the SSMH data were included in the training data in this setting (Figures 7 and 8). The AUCs became 0.778, 0.833, 0.838, 0.761, and 0.775 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively (P = 0.234, P < 0.05, P < 0.05, P = 0.217, and P < 0.05 between the ROCs of classification results by classifiers trained on the TCGA-STAD dataset and mixed dataset for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively, by Venkatraman’s permutation test for paired ROC curves). Furthermore, the performance on the TCGA-STAD FFPE dataset was also generally improved by the new classifiers trained on both datasets, except for the PIK3CA gene, which showed worse results (Supplementary Figure 2). The AUCs were 0.918, 0.872, 0.885, 0.766, and 0.820 for the CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively (P < 0.05, P < 0.05, P = 0.216, P < 0.05, and P < 0.05 compared with the TCGA-trained classifiers by Venkatraman’s permutation test for paired ROC curves). The accuracy, sensitivity, specificity, and F1 score of the classification results of mutation prediction models trained with both SSMH and TCGA datasets are presented in Supplementary Table 4.\n\nThe classifiers trained with both The Cancer Genome Atlas and SSMH data were used to predict the mutation of CDH1 (A and B), ERBB2 (C and D), and KRAS (E and F) genes. Representative binary heatmaps of the whole slide images (WSIs) correctly classified as mutation, correctly classified as wild-type, falsely classified as wild-type, and falsely classified as mutation were presented. Receiver operating characteristic curves for the folds with the lowest and highest area under the curve and the concatenated ten folds were also presented for each gene. CDH1-M: CDH1 mutated, CDH1-W: CDH1 wild-type, ERBB2-M: ERBB2 mutated, ERBB2-W: ERBB2 wild-type, KRAS-M: KRAS mutated, KRAS-W: KRAS wild-type.\n\nMutation prediction of PIK3CA (A and B) and TP53 (C and D) genes for the SSMH gastric cancer tissue slides by the classifiers trained with both The Cancer Genome Atlas and SSMH data. The configuration of the figure is the same as in figure 7. PIK3CA-M: PIK3CA mutated, PIK3CA-W: PIK3CA wild-type, TP53-M: TP53 mutated, TP53-W: TP53 wild-type.\nAnother interesting question is whether the DL-based classifiers for mutational status can be compatible with other types of cancers. We already built the mutation classifiers for KRAS, PIK3CA, and TP53 genes in the colorectal cancer dataset of TCGA in a previous study[11]. Therefore, we tested whether the classifiers trained on colorectal cancer can distinguish the mutational status in GC. As shown in Figure 9, the classifiers trained to discriminate the mutational status of KRAS, PIK3CA, and TP53 genes in the FFPE tissues of colorectal cancer approximately failed to distinguish the mutational status in the FFPE tissues of the TCGA-STAD dataset, with AUCs of 0.458, 0.550, and 0.538 for the KRAS, PIK3CA, and TP53 genes, respectively. The results indicate that the tissue morphologic features reflecting the wild-type and mutated genes are relatively different between cancers originating from different organs. \n\nMutation prediction of KRAS, PIK3CA, and TP53 genes for the The Cancer Genome Atlas gastric cancer tissue slides by the classifiers trained with The Cancer Genome Atlas colorectal cancer tissues. Receiver operating characteristic curves of the classification results for each gene were presented.", "Recently, many drugs targeting specific biological molecules have been introduced to improve the survival of patients with advanced GC[31]. However, patient stratification strategies to maximize the treatment response of these new drugs are not yet well established. Targeted therapies can yield different responses depending on the mutational status of genes in patients with cancer[32]. To overcome this complexity, clinical trials for new drugs have begun to adopt the umbrella platform strategy, which assigns treatment arms based on the mutational status of cancer patients[6,33]. Therefore, data regarding the mutational status of cancer patients is essential for patient stratification in modern-day medicine. However, molecular tests to detect gene mutations are still not affordable for all cancer patients. If cost- and time-effective alternative methods for mutation detection can be introduced, it will promote prospective clinical trials and retrospective studies to correlate the treatment response with the mutational profiles of cancer patients, which can be retrospectively obtained from clinical data and stored tissue samples. Therefore, the new cost- and time-effective methods will help to establish molecular stratification of cancer patients that can be used to determine effective treatment and improve clinical outcomes[34].\nCancer tissue slides are made and stored for most cancer patients. As a result, DL-based mutation prediction from the tissue slides can be a good candidate for alternative methods. It has been well recognized that the molecular alterations are manifested as morphologic changes in tissue architecture[35]. For example, some morphological features in GC tissues have been associated with specific mutations, including CDH1 and KRAS genes[36,37]. Although it is impractical to quantitatively assess these features for the detection of mutations by visual inspection, DL can learn and distinguish subtle discriminative features for mutation detection in various cancer tissues[7-11]. This study demonstrated the feasibility of DL-based prediction of mutations in CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, which are prevalent in both the TCGA and SSMH GC datasets, from tissue slide images of GC. Other studies have also shown that mutations in these genes are frequently observed in GC[3,5]. Furthermore, many studies have attempted to evaluate the prognostic value of these mutations[3,5,38]. However, the clinical relevance of these mutations for prognosis and treatment response has not been completely determined because the studies often presented discordant results. Various factors, including the relatively low incidence of mutation, small study size, and ethnicity of the studied groups, may have contributed to the inconsistent study results. Although it is still unclear how specific mutations are involved in the prognosis and treatment response in GC patients, further studies for the fine molecular stratification of patients based on mutational status are ongoing[6]. DL-based mutation prediction from the tissue slides could provide valuable tools to support these efforts because the mutational status can be promptly obtained with minimal cost from the existing H and E-stained tissue slides. \nFurthermore, DL-based classifiers can provide important information for the study of tumor heterogeneity[39]. The heatmaps of classification results overlaid on the tissue images in figures showed that mutated and wild-type regions are aggregated into separated regions. For example, the rightmost tissues in Figure 6C showed clear demarcation between TP53-mutated and wild-type regions. These results indicated that a tumor tissue can contain molecularly heterogeneous regions which can be easily visualized with the help of DL-based classifiers. The clear demarcation of molecularly heterogeneous regions in a tissue slide is an important advantage of a DL-based system and it can help the studies for the understanding of the prognostic and therapeutic values of the tumor heterogeneity without the application of the very expensive molecular tests such as multi-point single-cell sequencing.\nHowever, further studies are needed to build practical DL classifiers for mutation prediction. Our data showed that the performance was still unsatisfactory for verifying mutational status. The AUCs ranged from 0.661 to 0.862 for the TCGA dataset. The frequency of mutation in GC TCGA dataset was relatively low. The average mutation rate for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes was 8.28%. In our previous study for the mutation prediction in colorectal cancer TCGA dataset, the average mutation rate for APC, KRAS, PIK3CA, SMAD4, and TP53 genes was 39.18%[11]. Furthermore, the classifiers showed limited generalizability to the external dataset. Because DL critically depends on data for learning prominent features, it is generally recommended to build a large multinational dataset[1,2]. Therefore, to test whether the expanded dataset can improve the performance of the classifiers, new classifiers were trained using mixed data from the TCGA and SSMH datasets. As a result, the AUCs generally increased with the larger multinational datasets. These results suggest that we could improve the performance of DL-based mutation classifiers if a large multi-national and multi-institutional dataset can be built. One exception was the PIK3CA gene, which showed worse performance for the TCGA FFPE slides by a classifier trained with the mixed dataset. Although the reason for the decreased performance is unclear, we speculate that there are some different tissue features for the wild-type and mutated PIK3CA gene between the two datasets due to different ethnicities, which could negatively affect the feature learning process for the TCGA dataset. In addition, the numbers of patients with PIK3CA mutations were different; 64 and 11 for the TCGA and SSMH datasets, respectively. The different numbers of patients also hamper proper feature learning for the mixed dataset because data imbalance usually negatively affects the learning process. Furthermore, the studied tissues carry many additional mutations other than CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes. Because every tissue presented a different combination of mutations, the confounding effect of a mixture of different mutations on the tissue morphology would hamper the effective learning of features for the selected mutation. This factor also necessitates larger tissue datasets for proper learning of morphological features of specific mutations, irrespective of coexisting mutations. In our opinion, the datasets are still immature for building a prominent classifier for mutation prediction. Therefore, efforts to establish a larger tissue dataset with a mutation profile will help to understand the potential of DL-based mutation prediction systems. Recently, many countries have started to build nationwide datasets of pathologic tissue WSIs with genomic information. Therefore, we expect that the performance of DL-based mutation prediction can be greatly improved.\nAlthough we argued for the potential of DL-based mutation classifiers, there are important barriers to the adoption of DL-based assistant systems. First, the ‘black box’ nature of DL limits the interpretability of DL models and remains a significant barrier in their validation and adoption in clinics[18,40,41]. We could not trust a decision made by a DL model before we could clearly understand the basis of the decision. Therefore, a method for visualizing the features that determine the behavior of a DL model should be developed. Another barrier is the need for an individual DL system for an individual task. As described, separate systems should be built for tasks such as the classification of normal/tumor tissues for frozen and FFPE tissues. The classifier for each mutation should also be built separately. Furthermore, as shown in Figure 9, there was no compatibility between the different cancer types for the classification of genetic mutations. Therefore, many classifiers should be built to achieve optimal performance. It requires time to build many necessary classifiers to renovate current pathology workflows.", "Despite these limitations, DL has enormous potential for innovative medical practice. It can help capture important information by learning features automatically from the data that are waiting to be explored in the vast database of modern hospital information systems. This information will be used to determine the best medical practice and improve patient outcomes. The tissue slides of cancer patients contain important information on the prognosis of patients[42]; therefore, DL-based analysis of tissue slides has enormous potential for fine patient stratification in the era of precision medicine. Furthermore, its cost- and time-effective nature could help save the medical cost and decision time for patient care." ]

[ null, null, null, null, null, null, null ]

[ "INTRODUCTION", "MATERIALS AND METHODS", "Part I: Tests with The Cancer Genome Atlas whole-slide image datasets", "Part II: Tests on the external cohorts", "Statistical analysis", "RESULTS", "DISCUSSION", "CONCLUSION" ]

[ "Molecular tests to identify specific mutations in solid tumors have improved our ability to stratify cancer patients for more selective treatment regimens[1]. Therefore, molecular tests to detect various mutations are recommended for some tumors, including EGFR mutations in lung cancer, KRAS in colorectal cancer, and BRAF in melanoma. However, it is not routinely applied to cancer patients because molecular tests are not cost- and time-efficient[2]. Furthermore, the clinical significance of many mutations is still not well understood. For example, mutation profiling of gastric cancer (GC) is still proceeding, and the meaning of each mutation is not clearly understood[3]. GC is the fifth most common cancer and the third leading cause of cancer-related deaths worldwide[4]. It is important to evaluate the relationship between the mutational status and clinical characteristics of GC to improve the clinical outcomes of GC patients. Furthermore, many targeted drugs for treating various tumors are not effective in GC therapy because GC is not enriched with known driver mutations[5]. Therefore, research to characterize the roles of GC-related genes on the clinical behavior of tumors and the potential response to targeted therapies will have immense importance for the improvement of treatment response in GC[6]. A cost- and time-effective method to determine the mutational status of GC patients is necessary to promote these studies.\nRecently, deep learning (DL) has been increasingly implemented to predict the mutational status from hematoxylin and eosin (H and E)-stained tissue slides of various cancers[7-11]. The H and E-stained tissue slides were made for almost all cancer patients for basic diagnostic studies by pathologists[12]. Therefore, mutation prediction from the H and E-stained tissue slide based on a computational method can be a cost- and time-effective alternative tool for conventional molecular tests[13-15]. Although it has long been recognized that the morphological features of tissue architecture reflect the underlying molecular alterations[16,17], the features are not easily identifiable by human evaluators[18,19]. DL offers an alternative solution to overcome the limitations of a visual examination of tissue morphology by pathologists. By combining feature learning and model fitting in a unified step, DL can capture the most discriminative features for a given task directly from a large set of tissue images[20]. Digitization of tissue slides has been rapidly increasing after the approval of digitized whole-slide images (WSIs) for diagnostic purposes[21]. Digitized tissue data are rapidly accumulating with their associated mutational profiles. Therefore, the DL-based analysis of tissue slides for the mutational status of cancer tissues has immense potential as an alternative or complementary method for conventional molecular tests. \nBased on the potential of DL for the detection of mutations from digitized tissue slides, in a previous study, we successfully built DL-based classifiers for the prediction of mutational status of APC, KRAS, PIK3CA, SMAD4, and TP53 genes in colorectal cancer tissue slides[11]. This study investigated the feasibility of classifiers for mutations in the CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes in GC tissues. First, the classifiers were trained and tested for GC tissue slides from The Cancer Genome Atlas (TCGA). The generalizability of the classifiers was tested using an external dataset. Then, new classifiers were trained for combined datasets from TCGA and external datasets to investigate the effect of the extended datasets. The results suggest that it is feasible to predict mutational status directly from tissue slides with deep learning-based classifiers. Finally, as the classifiers for KRAS, PIK3CA, and TP53 mutations for both colorectal and GC were available, we also analyzed the generalizability of the DL-based mutation classifiers trained for different cancer types.", "Part I: Tests with The Cancer Genome Atlas whole-slide image datasets \nPatient cohort: The Cancer Genome Atlas (TCGA) provides extensive archives of digital pathology slides with multi-omics test results to test the possibility of tissue-based mutation detection[22]. After a carefully review of all the WSIs in the TCGA GC dataset (TCGA-STAD), we eliminated WSIs with poor scan quality and very small tumor contents. We selected slides from 25, 19, 34, 64, and 160 patients, which were confirmed to have mutations in CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. There were more than two slides for many patients in the TCGA dataset, with a maximum of four slides for some patients. However, in many cases, one or two slides contained only normal tissues. We excluded normal slides and selected a maximum of two tumor-containing slides per patient. The final number of frozen tissue slides was 34, 26, 50, 94, and 221 and that of formalin-fixed paraffin-embedded (FFPE) tissue slides was 27, 19, 34, 66, and 174 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. We selected 183 patients with wild-type CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes. Therefore, the same patients with wild-type genes for CDH1, ERBB2, KRAS, PIK3CA, and TP53 can be involved in the training of every classifier as a non-mutated group. This may help the comparison of the different classifiers more standardized because they all have the same group of patients as the wild-type group. The TCGA IDs of the patients in each group are listed in Supplementary Table 1. Our previous studies recognized that a DL model cannot perform optimally for both training and testing unless the dataset is forced to have similar amounts of data between classes[23]. Therefore, we limited the difference in patient numbers between the mutation and wild-type groups to less than 1.4 fold by random sampling. For example, only 35 of the 183 wild-type patients were randomly selected as the CDH1 wild-type group because there were only 25 CDH1 mutated patients. Ten-fold cross-validation was performed based on these randomly sampled wild-type patients. However, the classifiers yielded better results when the tumor patches from all wild-type patients other than the test sets were randomly sampled to match the 1.4 fold data ratio of wild-type/mutation groups for training, as this strategy could include a greater variety of tissue images. Therefore, we included all wild-type patients other than the test sets during training and randomly selected patients during testing.\n\nDeep learning model: In general, a WSI is too large to be analyzed simultaneously using a deep neural network. Therefore, the analysis results for small image patches are integrated for conclusion. We divided a WSI into non-overlapping patches of 360 × 360 pixel tissue images at 20× magnification to detect mutational status. To make the classification process fully automated, artifacts in the WSIs such as air bubbles, compression artifacts, out-of-focus blur, pen markings, tissue folding, and white background should be removed automatically. A simple convolutional neural network (CNN), termed as tissue/non-tissue classifier, was trained to discriminate these various artifacts all at once. The structure of the tissue/non-tissue classifier was described in our previous study[11]. The tissue/non-tissue classifier could filter out almost 99.9% of the improper tissue patches. Then, tissue patches classified as “improper” by the tissue/non-tissue classifier were removed, and the remaining “proper” tissue patches were collected. For the tumor or mutation classifiers described below, only proper tissue patches were analyzed (Figure 1).\n\nWorkflow for the fully automated prediction of mutation. Tissue image patches with tumor probability higher than 0.9 were selected by sequential application of the tissue/non-tissue and normal/tumor classifiers. Then the tumor patches were classified into the wild-type or mutated patches. The patch-level probabilities of mutation are averaged to yield the slide-level probability.\nMorphologic features reflecting mutations in specific genes might be expressed mainly in tumor tissues rather than normal tissues[24,25]. Therefore, tumor tissues should be separated from the WSI to predict the mutational status of the WSI. In a previous study, we successfully built normal/tumor classifiers for various tumors, including GC[26]. We concluded that frozen and FFPE slides should be separately analyzed using a deep neural network due to their different morphologic features. Thus, we adopted the normal/tumor classifiers for frozen and FFPE tissue slides from a previous study to delineate the normal/tumor gastric tissues for the frozen and FFPE slides of the TCGA-STAD dataset in the present study. Mutation classifiers were trained separately for the selected tumor patches for frozen and FFPE tissues. We selected tumor patches with a tumor probability higher than 0.9 to collect tissue patches with evident tumor features. We adopted a patient-level ten-fold cross-validation to completely characterize the TCGA-STAD dataset. Therefore, patients in each mutation/wild-type group for the five genes were separated into ten different folds, and one of the ten folds was used to test the classifiers trained with data from the other nine folds. Therefore, ten different classifiers were trained and tested for each group. The same label for all tumor tissue patches in a WSI as either ‘wild-type’ or ‘mutated’ were assigned based on the mutational status of the patient. Thereafter, the Inception-v3 model, a widely used CNN architecture, was trained to classify the tumor patches into ‘wild-type’ or ‘mutated’ tissues, as in our previous study on mutation prediction in colorectal cancer[11]. We fully trained the network from the beginning and did not adopt a transfer-learning scheme. The average probability of all tumor patches in a WSI was calculated to determine the slide-level mutation probability of a WSI. The Inception-v3 model was implemented using the TensorFlow DL library (http://tensorflow.org), and the network was trained with a mini-batch size of 128 and cross-entropy loss function as a loss function. For training, we used the RMSProp optimizer, with an initial learning rate of 0.1, weight decay of 0.9, momentum of 0.9, and epsilon of 1.0. Ten percent of the training slides were used as the validation dataset, and training was stopped when the loss for the validation data started to increase. Data augmentation techniques, including random horizontal/vertical flipping and random rotations by 90°, were applied to the tissue patches during training. Color normalization was applied to the tissue patches to avoid the effect of stain differences[27,28]. At least five classifiers were trained on each fold of mutation for the frozen and FFPE WSIs separately. The classifier with the best area under the curve (AUC) for the receiver operating characteristic (ROC) curves on the test dataset was included in the results. The ROC curves for fold with the lowest AUC, highest AUC, and the concatenated results for data from all ten folds are shown in the figures.\nIn summary, a WSI is analyzed as follows: 1. The whole slide is split into non-overlapping 360 × 360 pixel tissue patches, 2. Proper tissue patches are selected by tissue/non-tissue classifier, 3. Only tumor patches with tumor probability higher than 0.9 are selected, 4. High probability tumor patches are classified by each wild-type/mutation classifier, 5. The probabilities of tumor patches are averaged to obtain the slide-level mutation probability. The number of tissue patches used for the training of all mutation prediction models is summarized in Supplementary Table 2. The average number of training epochs for each classifier is summarized in Supplementary Table 3.\n\nPatient cohort: The Cancer Genome Atlas (TCGA) provides extensive archives of digital pathology slides with multi-omics test results to test the possibility of tissue-based mutation detection[22]. After a carefully review of all the WSIs in the TCGA GC dataset (TCGA-STAD), we eliminated WSIs with poor scan quality and very small tumor contents. We selected slides from 25, 19, 34, 64, and 160 patients, which were confirmed to have mutations in CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. There were more than two slides for many patients in the TCGA dataset, with a maximum of four slides for some patients. However, in many cases, one or two slides contained only normal tissues. We excluded normal slides and selected a maximum of two tumor-containing slides per patient. The final number of frozen tissue slides was 34, 26, 50, 94, and 221 and that of formalin-fixed paraffin-embedded (FFPE) tissue slides was 27, 19, 34, 66, and 174 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. We selected 183 patients with wild-type CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes. Therefore, the same patients with wild-type genes for CDH1, ERBB2, KRAS, PIK3CA, and TP53 can be involved in the training of every classifier as a non-mutated group. This may help the comparison of the different classifiers more standardized because they all have the same group of patients as the wild-type group. The TCGA IDs of the patients in each group are listed in Supplementary Table 1. Our previous studies recognized that a DL model cannot perform optimally for both training and testing unless the dataset is forced to have similar amounts of data between classes[23]. Therefore, we limited the difference in patient numbers between the mutation and wild-type groups to less than 1.4 fold by random sampling. For example, only 35 of the 183 wild-type patients were randomly selected as the CDH1 wild-type group because there were only 25 CDH1 mutated patients. Ten-fold cross-validation was performed based on these randomly sampled wild-type patients. However, the classifiers yielded better results when the tumor patches from all wild-type patients other than the test sets were randomly sampled to match the 1.4 fold data ratio of wild-type/mutation groups for training, as this strategy could include a greater variety of tissue images. Therefore, we included all wild-type patients other than the test sets during training and randomly selected patients during testing.\n\nDeep learning model: In general, a WSI is too large to be analyzed simultaneously using a deep neural network. Therefore, the analysis results for small image patches are integrated for conclusion. We divided a WSI into non-overlapping patches of 360 × 360 pixel tissue images at 20× magnification to detect mutational status. To make the classification process fully automated, artifacts in the WSIs such as air bubbles, compression artifacts, out-of-focus blur, pen markings, tissue folding, and white background should be removed automatically. A simple convolutional neural network (CNN), termed as tissue/non-tissue classifier, was trained to discriminate these various artifacts all at once. The structure of the tissue/non-tissue classifier was described in our previous study[11]. The tissue/non-tissue classifier could filter out almost 99.9% of the improper tissue patches. Then, tissue patches classified as “improper” by the tissue/non-tissue classifier were removed, and the remaining “proper” tissue patches were collected. For the tumor or mutation classifiers described below, only proper tissue patches were analyzed (Figure 1).\n\nWorkflow for the fully automated prediction of mutation. Tissue image patches with tumor probability higher than 0.9 were selected by sequential application of the tissue/non-tissue and normal/tumor classifiers. Then the tumor patches were classified into the wild-type or mutated patches. The patch-level probabilities of mutation are averaged to yield the slide-level probability.\nMorphologic features reflecting mutations in specific genes might be expressed mainly in tumor tissues rather than normal tissues[24,25]. Therefore, tumor tissues should be separated from the WSI to predict the mutational status of the WSI. In a previous study, we successfully built normal/tumor classifiers for various tumors, including GC[26]. We concluded that frozen and FFPE slides should be separately analyzed using a deep neural network due to their different morphologic features. Thus, we adopted the normal/tumor classifiers for frozen and FFPE tissue slides from a previous study to delineate the normal/tumor gastric tissues for the frozen and FFPE slides of the TCGA-STAD dataset in the present study. Mutation classifiers were trained separately for the selected tumor patches for frozen and FFPE tissues. We selected tumor patches with a tumor probability higher than 0.9 to collect tissue patches with evident tumor features. We adopted a patient-level ten-fold cross-validation to completely characterize the TCGA-STAD dataset. Therefore, patients in each mutation/wild-type group for the five genes were separated into ten different folds, and one of the ten folds was used to test the classifiers trained with data from the other nine folds. Therefore, ten different classifiers were trained and tested for each group. The same label for all tumor tissue patches in a WSI as either ‘wild-type’ or ‘mutated’ were assigned based on the mutational status of the patient. Thereafter, the Inception-v3 model, a widely used CNN architecture, was trained to classify the tumor patches into ‘wild-type’ or ‘mutated’ tissues, as in our previous study on mutation prediction in colorectal cancer[11]. We fully trained the network from the beginning and did not adopt a transfer-learning scheme. The average probability of all tumor patches in a WSI was calculated to determine the slide-level mutation probability of a WSI. The Inception-v3 model was implemented using the TensorFlow DL library (http://tensorflow.org), and the network was trained with a mini-batch size of 128 and cross-entropy loss function as a loss function. For training, we used the RMSProp optimizer, with an initial learning rate of 0.1, weight decay of 0.9, momentum of 0.9, and epsilon of 1.0. Ten percent of the training slides were used as the validation dataset, and training was stopped when the loss for the validation data started to increase. Data augmentation techniques, including random horizontal/vertical flipping and random rotations by 90°, were applied to the tissue patches during training. Color normalization was applied to the tissue patches to avoid the effect of stain differences[27,28]. At least five classifiers were trained on each fold of mutation for the frozen and FFPE WSIs separately. The classifier with the best area under the curve (AUC) for the receiver operating characteristic (ROC) curves on the test dataset was included in the results. The ROC curves for fold with the lowest AUC, highest AUC, and the concatenated results for data from all ten folds are shown in the figures.\nIn summary, a WSI is analyzed as follows: 1. The whole slide is split into non-overlapping 360 × 360 pixel tissue patches, 2. Proper tissue patches are selected by tissue/non-tissue classifier, 3. Only tumor patches with tumor probability higher than 0.9 are selected, 4. High probability tumor patches are classified by each wild-type/mutation classifier, 5. The probabilities of tumor patches are averaged to obtain the slide-level mutation probability. The number of tissue patches used for the training of all mutation prediction models is summarized in Supplementary Table 2. The average number of training epochs for each classifier is summarized in Supplementary Table 3.\nPart II: Tests on the external cohorts \nPatient cohort: GC tissue slides were collected from 96 patients who had previously undergone surgical resection at Seoul St. Mary’s Hospital between 2017 and 2020 (SSMH dataset). An Aperio slide scanner (Leica Biosystems) was used to scan the FFPE slides. The Institutional Review Board of the College of Medicine at the Catholic University of Korea approved this study (KC19SESI0787). \n\nMutation prediction on SSMH dataset: For CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, 6, 6, 12, 11, and 39 patients were confirmed to have the mutations, respectively. Thirty-eight patients had wild-type genes for all five genes. For CDH1, ERBB2, KRAS, and PIK3CA genes, we selected the number of wild-type patients to be 1.4 times that of mutated patients. For TP53, all 38 patients with wild-type genes were enrolled. The normal/tumor classifier for TCGA FFPE tissues was also used to discriminate the tumor tissue patches of SSMH WSIs. Our previous study showed that the normal/tumor classifier for TCGA-STAD was valid for SSMH FFPE slides[29]. First, the mutational status of the SSMH slides was analyzed by classifiers trained on TCGA-STAD FFPE WSIs. Subsequently, new classifiers were trained using both TCGA and SSMH FFPE tissues. Patient-level three-fold cross validation was applied to the SSMH datasets because the number of mutated patients was not sufficient for ten-fold cross-validation.\n\nPatient cohort: GC tissue slides were collected from 96 patients who had previously undergone surgical resection at Seoul St. Mary’s Hospital between 2017 and 2020 (SSMH dataset). An Aperio slide scanner (Leica Biosystems) was used to scan the FFPE slides. The Institutional Review Board of the College of Medicine at the Catholic University of Korea approved this study (KC19SESI0787). \n\nMutation prediction on SSMH dataset: For CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, 6, 6, 12, 11, and 39 patients were confirmed to have the mutations, respectively. Thirty-eight patients had wild-type genes for all five genes. For CDH1, ERBB2, KRAS, and PIK3CA genes, we selected the number of wild-type patients to be 1.4 times that of mutated patients. For TP53, all 38 patients with wild-type genes were enrolled. The normal/tumor classifier for TCGA FFPE tissues was also used to discriminate the tumor tissue patches of SSMH WSIs. Our previous study showed that the normal/tumor classifier for TCGA-STAD was valid for SSMH FFPE slides[29]. First, the mutational status of the SSMH slides was analyzed by classifiers trained on TCGA-STAD FFPE WSIs. Subsequently, new classifiers were trained using both TCGA and SSMH FFPE tissues. Patient-level three-fold cross validation was applied to the SSMH datasets because the number of mutated patients was not sufficient for ten-fold cross-validation.\nStatistical analysis To demonstrate the performance of each classifier, the ROC curves and their AUCs are presented in the figures. For the concatenated results from all ten folds, 95% confidence intervals (CIs) were also presented using the percentile bootstrap method. In addition, the accuracy, sensitivity, specificity, and F1 score of the classification results of mutation prediction models with cutoff values for maximal Youden index (sensitivity + specificity - 1) were presented. We used a permutation test with 1000 iterations to compare the differences between the two paired or unpaired ROC curves when a comparison was necessary[30]. Statistical significance was set at P < 0.05.\nTo demonstrate the performance of each classifier, the ROC curves and their AUCs are presented in the figures. For the concatenated results from all ten folds, 95% confidence intervals (CIs) were also presented using the percentile bootstrap method. In addition, the accuracy, sensitivity, specificity, and F1 score of the classification results of mutation prediction models with cutoff values for maximal Youden index (sensitivity + specificity - 1) were presented. We used a permutation test with 1000 iterations to compare the differences between the two paired or unpaired ROC curves when a comparison was necessary[30]. Statistical significance was set at P < 0.05.", "\nPatient cohort: The Cancer Genome Atlas (TCGA) provides extensive archives of digital pathology slides with multi-omics test results to test the possibility of tissue-based mutation detection[22]. After a carefully review of all the WSIs in the TCGA GC dataset (TCGA-STAD), we eliminated WSIs with poor scan quality and very small tumor contents. We selected slides from 25, 19, 34, 64, and 160 patients, which were confirmed to have mutations in CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. There were more than two slides for many patients in the TCGA dataset, with a maximum of four slides for some patients. However, in many cases, one or two slides contained only normal tissues. We excluded normal slides and selected a maximum of two tumor-containing slides per patient. The final number of frozen tissue slides was 34, 26, 50, 94, and 221 and that of formalin-fixed paraffin-embedded (FFPE) tissue slides was 27, 19, 34, 66, and 174 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. We selected 183 patients with wild-type CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes. Therefore, the same patients with wild-type genes for CDH1, ERBB2, KRAS, PIK3CA, and TP53 can be involved in the training of every classifier as a non-mutated group. This may help the comparison of the different classifiers more standardized because they all have the same group of patients as the wild-type group. The TCGA IDs of the patients in each group are listed in Supplementary Table 1. Our previous studies recognized that a DL model cannot perform optimally for both training and testing unless the dataset is forced to have similar amounts of data between classes[23]. Therefore, we limited the difference in patient numbers between the mutation and wild-type groups to less than 1.4 fold by random sampling. For example, only 35 of the 183 wild-type patients were randomly selected as the CDH1 wild-type group because there were only 25 CDH1 mutated patients. Ten-fold cross-validation was performed based on these randomly sampled wild-type patients. However, the classifiers yielded better results when the tumor patches from all wild-type patients other than the test sets were randomly sampled to match the 1.4 fold data ratio of wild-type/mutation groups for training, as this strategy could include a greater variety of tissue images. Therefore, we included all wild-type patients other than the test sets during training and randomly selected patients during testing.\n\nDeep learning model: In general, a WSI is too large to be analyzed simultaneously using a deep neural network. Therefore, the analysis results for small image patches are integrated for conclusion. We divided a WSI into non-overlapping patches of 360 × 360 pixel tissue images at 20× magnification to detect mutational status. To make the classification process fully automated, artifacts in the WSIs such as air bubbles, compression artifacts, out-of-focus blur, pen markings, tissue folding, and white background should be removed automatically. A simple convolutional neural network (CNN), termed as tissue/non-tissue classifier, was trained to discriminate these various artifacts all at once. The structure of the tissue/non-tissue classifier was described in our previous study[11]. The tissue/non-tissue classifier could filter out almost 99.9% of the improper tissue patches. Then, tissue patches classified as “improper” by the tissue/non-tissue classifier were removed, and the remaining “proper” tissue patches were collected. For the tumor or mutation classifiers described below, only proper tissue patches were analyzed (Figure 1).\n\nWorkflow for the fully automated prediction of mutation. Tissue image patches with tumor probability higher than 0.9 were selected by sequential application of the tissue/non-tissue and normal/tumor classifiers. Then the tumor patches were classified into the wild-type or mutated patches. The patch-level probabilities of mutation are averaged to yield the slide-level probability.\nMorphologic features reflecting mutations in specific genes might be expressed mainly in tumor tissues rather than normal tissues[24,25]. Therefore, tumor tissues should be separated from the WSI to predict the mutational status of the WSI. In a previous study, we successfully built normal/tumor classifiers for various tumors, including GC[26]. We concluded that frozen and FFPE slides should be separately analyzed using a deep neural network due to their different morphologic features. Thus, we adopted the normal/tumor classifiers for frozen and FFPE tissue slides from a previous study to delineate the normal/tumor gastric tissues for the frozen and FFPE slides of the TCGA-STAD dataset in the present study. Mutation classifiers were trained separately for the selected tumor patches for frozen and FFPE tissues. We selected tumor patches with a tumor probability higher than 0.9 to collect tissue patches with evident tumor features. We adopted a patient-level ten-fold cross-validation to completely characterize the TCGA-STAD dataset. Therefore, patients in each mutation/wild-type group for the five genes were separated into ten different folds, and one of the ten folds was used to test the classifiers trained with data from the other nine folds. Therefore, ten different classifiers were trained and tested for each group. The same label for all tumor tissue patches in a WSI as either ‘wild-type’ or ‘mutated’ were assigned based on the mutational status of the patient. Thereafter, the Inception-v3 model, a widely used CNN architecture, was trained to classify the tumor patches into ‘wild-type’ or ‘mutated’ tissues, as in our previous study on mutation prediction in colorectal cancer[11]. We fully trained the network from the beginning and did not adopt a transfer-learning scheme. The average probability of all tumor patches in a WSI was calculated to determine the slide-level mutation probability of a WSI. The Inception-v3 model was implemented using the TensorFlow DL library (http://tensorflow.org), and the network was trained with a mini-batch size of 128 and cross-entropy loss function as a loss function. For training, we used the RMSProp optimizer, with an initial learning rate of 0.1, weight decay of 0.9, momentum of 0.9, and epsilon of 1.0. Ten percent of the training slides were used as the validation dataset, and training was stopped when the loss for the validation data started to increase. Data augmentation techniques, including random horizontal/vertical flipping and random rotations by 90°, were applied to the tissue patches during training. Color normalization was applied to the tissue patches to avoid the effect of stain differences[27,28]. At least five classifiers were trained on each fold of mutation for the frozen and FFPE WSIs separately. The classifier with the best area under the curve (AUC) for the receiver operating characteristic (ROC) curves on the test dataset was included in the results. The ROC curves for fold with the lowest AUC, highest AUC, and the concatenated results for data from all ten folds are shown in the figures.\nIn summary, a WSI is analyzed as follows: 1. The whole slide is split into non-overlapping 360 × 360 pixel tissue patches, 2. Proper tissue patches are selected by tissue/non-tissue classifier, 3. Only tumor patches with tumor probability higher than 0.9 are selected, 4. High probability tumor patches are classified by each wild-type/mutation classifier, 5. The probabilities of tumor patches are averaged to obtain the slide-level mutation probability. The number of tissue patches used for the training of all mutation prediction models is summarized in Supplementary Table 2. The average number of training epochs for each classifier is summarized in Supplementary Table 3.", "\nPatient cohort: GC tissue slides were collected from 96 patients who had previously undergone surgical resection at Seoul St. Mary’s Hospital between 2017 and 2020 (SSMH dataset). An Aperio slide scanner (Leica Biosystems) was used to scan the FFPE slides. The Institutional Review Board of the College of Medicine at the Catholic University of Korea approved this study (KC19SESI0787). \n\nMutation prediction on SSMH dataset: For CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, 6, 6, 12, 11, and 39 patients were confirmed to have the mutations, respectively. Thirty-eight patients had wild-type genes for all five genes. For CDH1, ERBB2, KRAS, and PIK3CA genes, we selected the number of wild-type patients to be 1.4 times that of mutated patients. For TP53, all 38 patients with wild-type genes were enrolled. The normal/tumor classifier for TCGA FFPE tissues was also used to discriminate the tumor tissue patches of SSMH WSIs. Our previous study showed that the normal/tumor classifier for TCGA-STAD was valid for SSMH FFPE slides[29]. First, the mutational status of the SSMH slides was analyzed by classifiers trained on TCGA-STAD FFPE WSIs. Subsequently, new classifiers were trained using both TCGA and SSMH FFPE tissues. Patient-level three-fold cross validation was applied to the SSMH datasets because the number of mutated patients was not sufficient for ten-fold cross-validation.", "To demonstrate the performance of each classifier, the ROC curves and their AUCs are presented in the figures. For the concatenated results from all ten folds, 95% confidence intervals (CIs) were also presented using the percentile bootstrap method. In addition, the accuracy, sensitivity, specificity, and F1 score of the classification results of mutation prediction models with cutoff values for maximal Youden index (sensitivity + specificity - 1) were presented. We used a permutation test with 1000 iterations to compare the differences between the two paired or unpaired ROC curves when a comparison was necessary[30]. Statistical significance was set at P < 0.05.", "Tissue patches with high tumor probability were automatically collected from a WSI by sequentially applying the tissue/non-tissue and normal/tumor classifiers to 360 × 360 pixels tissue image patches (Figure 1). Then, classifiers to distinguish the mutational status of CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes in the tumor tissue patches from the TCGA-STAD frozen and FFPE WSI datasets were separately trained with a patient-level ten-fold cross-validation scheme. \nThe classification results of the TCGA-STAD WSIs are presented in Figures 2 to 6 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes. Results for the frozen and FFPE tissues are presented in the upper and lower part of each figure, respectively. Panels A and C demonstrated the representative binary heatmaps of tissue patches classified as wild-type or mutated tissues. The WSIs with gene mutation correctly classified as mutation, with wild-type gene correctly classified as wild-type, with gene mutation falsely classified as wild-type, and with wild-type gene falsely classified as mutation are presented from left to right for panels A and C. The binary heatmaps were drawn with the wild-type/mutation discrimination threshold set to 0.5. We simply set the threshold to 0.5, because every classifier for different folds had different optimal thresholds. Slide-level ROC curves for folds with the lowest and highest AUCs are presented to demonstrate the differences in the performance between folds (left and middle ROC curves in each figure). Finally, the slide-level ROC curves for the concatenated results from all ten folds were used to infer the overall performance (right ROC curves). The results for the CDH1 gene are shown in Figure 2. The AUCs per fold ranged from 0.833 to 1.000 for frozen WSIs and from 0.833 to 1.000 for FFPE WSIs. The AUCs for the concatenated results were 0.842 (95%CI: 0.749-0.936) and 0.781 (95%CI: 0.645-0.917) for frozen and FFPE WSIs, respectively. For ERBB2 (Figure 3), the lowest and highest AUCs per fold were 0.667 and 1.000, respectively, for both frozen and FFPE WSIs. The concatenated AUCs were 0.751 (95%CI: 0.631-0.871) and 0.661 (95%CI: 0.501-0.821), respectively. For the KRAS gene (Figure 4), the AUCs per fold were between 0.775 and 1.000 for frozen WSIs and between 0.750 and 1.000 for FFPE WSIs. The concatenated AUCs were 0.793 (95%CI: 0.706-0.879) and 0.858 (95%CI: 0.738-0.979) for frozen and FFPE WSIs, respectively. For the PIK3CA gene (Figure 5), the concatenated AUC for the frozen WSIs was 0.862 (95%CI: 0.809-0.916), with a range of 0.705 to 0.990. For FFPE WSIs, the lowest and highest AUCs per fold were 0.675 and 1.000, respectively, yielding a concatenated AUC of 0.828 (95%CI: 0.750-0.907). Lastly, the results for the TP53 gene are presented in Figure 6. The AUCs per fold were between 0.666 to 0.810 for frozen WSIs and between 0.702 to 0.847 for FFPE WSIs. The concatenated AUCs were 0.727 (95%CI: 0.683-0.771) and 0.727 (95%CI: 0.671-0.784) for frozen and FFPE WSIs, respectively. For the colorectal cancer dataset from TCGA, mutation classification results for frozen tissues were better than those for FFPE tissues in some genes[11]. However, there were no significant differences between the frozen and FFPE tissues in the TCGA-STAD dataset (P = 0.491, 0.431, 0.187, 0.321, and 0.613 between the concatenated AUCs for the frozen and FFPE tissues by Venkatraman’s permutation test for unpaired ROC curves for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively. For a clearer assessment of the performance of each model, the accuracy, sensitivity, specificity, and F1 score of the classification results are presented in Table 1.\n\nClassification results of CDH1 gene in the The Cancer Genome Atlas gastric cancer dataset. A: Representative whole slide images (WSIs) of the frozen slides with CDH1 gene mutation correctly classified as mutation, with wild-type gene correctly classified as wild-type, with gene mutation falsely classified as wild-type, and with wild-type gene falsely classified as mutation, from left to right; B: Receiver operating characteristic (ROC) curves for the fold with lowest area under the curve (AUC), for the fold with highest AUC, and for the concatenated results of all ten folds, from left to right, obtained with the classifiers trained with the frozen tissues; C and D: Same as A and B but the results were for the formalin-fixed paraffin-embedded WSIs. CDH1-M: CDH1 mutated, CDH1-W: CDH1 wild-type.\n\nClassification results of ERBB2 gene in the The Cancer Genome Atlas gastric cancer dataset. The configuration of the figure is the same as in Figure 2. A and B: Upper panels are results for the frozen tissue and lower panels; C and D: Results for the formalin-fixed paraffin-embedded tissues. ERBB2-M: ERBB2 mutated, ERBB2-W: ERBB2 wild-type.\n\nClassification results of KRAS gene in the The Cancer Genome Atlas gastric cancer dataset. The configuration of the figure is the same as in Figure 2. A and B: Upper panels are results for the frozen tissue and lower panels; C and D: Results for the formalin-fixed paraffin-embedded tissues. KRAS-M: KRAS mutated, KRAS-W: KRAS wild-type.\n\nClassification results of PIK3CA gene in the The Cancer Genome Atlas gastric cancer dataset. The configuration of the figure is the same as in Figure 2. A and B: Upper panels are results for the frozen tissue and lower panels; C and D: Results for the formalin-fixed paraffin-embedded tissues. PIK3CA-M: PIK3CA mutated, PIK3CA-W: PIK3CA wild-type.\n\nClassification results of TP53 gene in the The Cancer Genome Atlas gastric cancer dataset. The configuration of the figure is the same as in Figure 2. A and B: Upper panels are results for the frozen tissue and lower panels; C and D: Results for the formalin-fixed paraffin-embedded tissues. TP53-M: TP53 mutated, TP53-W: TP53 wild-type.\nAccuracy, sensitivity, specificity, and F1 score of the classification results of mutation prediction models for the The Cancer Genome Atlas datasets\nTCGA: The Cancer Genome Atlas; FFPE: Formalin-fixed paraffin-embedded.\nThe performance of a DL model on an external dataset should be tested to validate the generalizability of the trained model. Therefore, we collected GC FFPE WSIs with matching mutation data from Seoul St. Mary’s Hospital (SSMH dataset). The normal/tumor classifier for TCGA-STAD FFPE tissues was also applied to select tissue patches with high tumor probabilities. Thereafter, the mutation classifier for each gene trained on the TCGA-STAD FFPE tissues was tested on the SSMH dataset. The slide-level ROC curves for the CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes are presented in Supplementary Figure 1. The AUCs for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes were 0.667, 0.630, 0.657, 0.688, and 0.572, respectively. For the KRAS, PIK3CA, and TP53 genes, the performance of the TCGA-trained mutation classifiers on the SSMH dataset were worse than that of the TCGA dataset (P = 0.389, P = 0.849, P < 0.05, P < 0.05, and P < 0.05 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively, by Venkatraman’s permutation test for unpaired ROC curves). These results demonstrate that the mutation classifiers trained with TCGA-STAD WSI datasets had limited generalizability. It is of interest if the performance can be enhanced by training the classifiers with expanded datasets, including both TCGA and SSMH datasets. Cancer tissues from different ethnic groups can show different features[16,19]; therefore, the performance of the classifiers can be improved by mixing the datasets. When the classifiers trained with the mixed datasets were used, the performance on the SSMH dataset was generally improved because the SSMH data were included in the training data in this setting (Figures 7 and 8). The AUCs became 0.778, 0.833, 0.838, 0.761, and 0.775 for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively (P = 0.234, P < 0.05, P < 0.05, P = 0.217, and P < 0.05 between the ROCs of classification results by classifiers trained on the TCGA-STAD dataset and mixed dataset for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively, by Venkatraman’s permutation test for paired ROC curves). Furthermore, the performance on the TCGA-STAD FFPE dataset was also generally improved by the new classifiers trained on both datasets, except for the PIK3CA gene, which showed worse results (Supplementary Figure 2). The AUCs were 0.918, 0.872, 0.885, 0.766, and 0.820 for the CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, respectively (P < 0.05, P < 0.05, P = 0.216, P < 0.05, and P < 0.05 compared with the TCGA-trained classifiers by Venkatraman’s permutation test for paired ROC curves). The accuracy, sensitivity, specificity, and F1 score of the classification results of mutation prediction models trained with both SSMH and TCGA datasets are presented in Supplementary Table 4.\n\nThe classifiers trained with both The Cancer Genome Atlas and SSMH data were used to predict the mutation of CDH1 (A and B), ERBB2 (C and D), and KRAS (E and F) genes. Representative binary heatmaps of the whole slide images (WSIs) correctly classified as mutation, correctly classified as wild-type, falsely classified as wild-type, and falsely classified as mutation were presented. Receiver operating characteristic curves for the folds with the lowest and highest area under the curve and the concatenated ten folds were also presented for each gene. CDH1-M: CDH1 mutated, CDH1-W: CDH1 wild-type, ERBB2-M: ERBB2 mutated, ERBB2-W: ERBB2 wild-type, KRAS-M: KRAS mutated, KRAS-W: KRAS wild-type.\n\nMutation prediction of PIK3CA (A and B) and TP53 (C and D) genes for the SSMH gastric cancer tissue slides by the classifiers trained with both The Cancer Genome Atlas and SSMH data. The configuration of the figure is the same as in figure 7. PIK3CA-M: PIK3CA mutated, PIK3CA-W: PIK3CA wild-type, TP53-M: TP53 mutated, TP53-W: TP53 wild-type.\nAnother interesting question is whether the DL-based classifiers for mutational status can be compatible with other types of cancers. We already built the mutation classifiers for KRAS, PIK3CA, and TP53 genes in the colorectal cancer dataset of TCGA in a previous study[11]. Therefore, we tested whether the classifiers trained on colorectal cancer can distinguish the mutational status in GC. As shown in Figure 9, the classifiers trained to discriminate the mutational status of KRAS, PIK3CA, and TP53 genes in the FFPE tissues of colorectal cancer approximately failed to distinguish the mutational status in the FFPE tissues of the TCGA-STAD dataset, with AUCs of 0.458, 0.550, and 0.538 for the KRAS, PIK3CA, and TP53 genes, respectively. The results indicate that the tissue morphologic features reflecting the wild-type and mutated genes are relatively different between cancers originating from different organs. \n\nMutation prediction of KRAS, PIK3CA, and TP53 genes for the The Cancer Genome Atlas gastric cancer tissue slides by the classifiers trained with The Cancer Genome Atlas colorectal cancer tissues. Receiver operating characteristic curves of the classification results for each gene were presented.", "Recently, many drugs targeting specific biological molecules have been introduced to improve the survival of patients with advanced GC[31]. However, patient stratification strategies to maximize the treatment response of these new drugs are not yet well established. Targeted therapies can yield different responses depending on the mutational status of genes in patients with cancer[32]. To overcome this complexity, clinical trials for new drugs have begun to adopt the umbrella platform strategy, which assigns treatment arms based on the mutational status of cancer patients[6,33]. Therefore, data regarding the mutational status of cancer patients is essential for patient stratification in modern-day medicine. However, molecular tests to detect gene mutations are still not affordable for all cancer patients. If cost- and time-effective alternative methods for mutation detection can be introduced, it will promote prospective clinical trials and retrospective studies to correlate the treatment response with the mutational profiles of cancer patients, which can be retrospectively obtained from clinical data and stored tissue samples. Therefore, the new cost- and time-effective methods will help to establish molecular stratification of cancer patients that can be used to determine effective treatment and improve clinical outcomes[34].\nCancer tissue slides are made and stored for most cancer patients. As a result, DL-based mutation prediction from the tissue slides can be a good candidate for alternative methods. It has been well recognized that the molecular alterations are manifested as morphologic changes in tissue architecture[35]. For example, some morphological features in GC tissues have been associated with specific mutations, including CDH1 and KRAS genes[36,37]. Although it is impractical to quantitatively assess these features for the detection of mutations by visual inspection, DL can learn and distinguish subtle discriminative features for mutation detection in various cancer tissues[7-11]. This study demonstrated the feasibility of DL-based prediction of mutations in CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes, which are prevalent in both the TCGA and SSMH GC datasets, from tissue slide images of GC. Other studies have also shown that mutations in these genes are frequently observed in GC[3,5]. Furthermore, many studies have attempted to evaluate the prognostic value of these mutations[3,5,38]. However, the clinical relevance of these mutations for prognosis and treatment response has not been completely determined because the studies often presented discordant results. Various factors, including the relatively low incidence of mutation, small study size, and ethnicity of the studied groups, may have contributed to the inconsistent study results. Although it is still unclear how specific mutations are involved in the prognosis and treatment response in GC patients, further studies for the fine molecular stratification of patients based on mutational status are ongoing[6]. DL-based mutation prediction from the tissue slides could provide valuable tools to support these efforts because the mutational status can be promptly obtained with minimal cost from the existing H and E-stained tissue slides. \nFurthermore, DL-based classifiers can provide important information for the study of tumor heterogeneity[39]. The heatmaps of classification results overlaid on the tissue images in figures showed that mutated and wild-type regions are aggregated into separated regions. For example, the rightmost tissues in Figure 6C showed clear demarcation between TP53-mutated and wild-type regions. These results indicated that a tumor tissue can contain molecularly heterogeneous regions which can be easily visualized with the help of DL-based classifiers. The clear demarcation of molecularly heterogeneous regions in a tissue slide is an important advantage of a DL-based system and it can help the studies for the understanding of the prognostic and therapeutic values of the tumor heterogeneity without the application of the very expensive molecular tests such as multi-point single-cell sequencing.\nHowever, further studies are needed to build practical DL classifiers for mutation prediction. Our data showed that the performance was still unsatisfactory for verifying mutational status. The AUCs ranged from 0.661 to 0.862 for the TCGA dataset. The frequency of mutation in GC TCGA dataset was relatively low. The average mutation rate for CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes was 8.28%. In our previous study for the mutation prediction in colorectal cancer TCGA dataset, the average mutation rate for APC, KRAS, PIK3CA, SMAD4, and TP53 genes was 39.18%[11]. Furthermore, the classifiers showed limited generalizability to the external dataset. Because DL critically depends on data for learning prominent features, it is generally recommended to build a large multinational dataset[1,2]. Therefore, to test whether the expanded dataset can improve the performance of the classifiers, new classifiers were trained using mixed data from the TCGA and SSMH datasets. As a result, the AUCs generally increased with the larger multinational datasets. These results suggest that we could improve the performance of DL-based mutation classifiers if a large multi-national and multi-institutional dataset can be built. One exception was the PIK3CA gene, which showed worse performance for the TCGA FFPE slides by a classifier trained with the mixed dataset. Although the reason for the decreased performance is unclear, we speculate that there are some different tissue features for the wild-type and mutated PIK3CA gene between the two datasets due to different ethnicities, which could negatively affect the feature learning process for the TCGA dataset. In addition, the numbers of patients with PIK3CA mutations were different; 64 and 11 for the TCGA and SSMH datasets, respectively. The different numbers of patients also hamper proper feature learning for the mixed dataset because data imbalance usually negatively affects the learning process. Furthermore, the studied tissues carry many additional mutations other than CDH1, ERBB2, KRAS, PIK3CA, and TP53 genes. Because every tissue presented a different combination of mutations, the confounding effect of a mixture of different mutations on the tissue morphology would hamper the effective learning of features for the selected mutation. This factor also necessitates larger tissue datasets for proper learning of morphological features of specific mutations, irrespective of coexisting mutations. In our opinion, the datasets are still immature for building a prominent classifier for mutation prediction. Therefore, efforts to establish a larger tissue dataset with a mutation profile will help to understand the potential of DL-based mutation prediction systems. Recently, many countries have started to build nationwide datasets of pathologic tissue WSIs with genomic information. Therefore, we expect that the performance of DL-based mutation prediction can be greatly improved.\nAlthough we argued for the potential of DL-based mutation classifiers, there are important barriers to the adoption of DL-based assistant systems. First, the ‘black box’ nature of DL limits the interpretability of DL models and remains a significant barrier in their validation and adoption in clinics[18,40,41]. We could not trust a decision made by a DL model before we could clearly understand the basis of the decision. Therefore, a method for visualizing the features that determine the behavior of a DL model should be developed. Another barrier is the need for an individual DL system for an individual task. As described, separate systems should be built for tasks such as the classification of normal/tumor tissues for frozen and FFPE tissues. The classifier for each mutation should also be built separately. Furthermore, as shown in Figure 9, there was no compatibility between the different cancer types for the classification of genetic mutations. Therefore, many classifiers should be built to achieve optimal performance. It requires time to build many necessary classifiers to renovate current pathology workflows.", "Despite these limitations, DL has enormous potential for innovative medical practice. It can help capture important information by learning features automatically from the data that are waiting to be explored in the vast database of modern hospital information systems. This information will be used to determine the best medical practice and improve patient outcomes. The tissue slides of cancer patients contain important information on the prognosis of patients[42]; therefore, DL-based analysis of tissue slides has enormous potential for fine patient stratification in the era of precision medicine. Furthermore, its cost- and time-effective nature could help save the medical cost and decision time for patient care." ]

[ null, "methods", null, null, null, null, null, null ]

[ "Gastric cancer", "Mutation", "Deep learning", "Digital pathology", "Formalin-fixed paraffin-embedded" ]

[ 691, 1503, 278, 119, 2275, 1394, 119 ]

[ "tissue", "mutation", "tumor", "patients", "wild", "wild type", "type", "patches", "classifiers", "slides" ]

[ "tumor mutation classifiers", "mutations colorectal gc", "cancer patients molecular", "gastric cancer gc", "profiling gastric cancer" ]

[ "Adult", "Anemia, Iron-Deficiency", "Ferric Compounds", "Humans", "Hypophosphatemia", "Inflammatory Bowel Diseases", "Iron", "Norway", "Quality of Life" ]

[ "INTRODUCTION", "Study design and patient population", "Study assessments and data collection", "Study outcomes", "Statistical analysis", "Ethical considerations", "RESULTS", "Serum phosphate and urinary excretion of phosphate", "FGF23", "Vitamin D", "Calcium and PTH", "Respiratory muscle function tests", "SF-36", "VAS scores", "DISCUSSION", "CONCLUSION" ]

[ "Iron replacement therapy is often needed in patients with inflammatory bowel disease (IBD) because iron deficiency (ID) and ID anaemia (IDA) occur frequently in this patient group[1-3]. A large proportion of IBD patients experience intolerance to oral iron[4]. Additionally, it is asserted that oral iron can lead to an exacerbation of inflammation in the bowel mucosa due to a local effect on the enterocytes[5-7]. Therefore, administration of high-dose iron as an intravenous infusion is an effective, suitable and convenient treatment option in IBD. Ferric carboxymaltose (FCM; Ferinject®; Vifor Pharma) and ferric derisomaltose (FDI), previously known as iron isomaltoside (Monofer®; Pharmacosmos A/S), are the most widely used preparations in Europe when high-dose intravenous iron is indicated.\nIn a recent publication, we described a high incidence of hypophosphatemia in IBD patients who had received treatment with FCM[8]. The mechanism behind the development of hypophosphatemia has been described by Wolf et al[9], but probably is not yet fully understood, and has not been investigated in patients with IBD. Fibroblast growth factor 23 (FGF23) is a small peptide hormone, synthesized by osteocytes, which regulates phosphate and vitamin D homeostasis[9]. FGF23 consists of a biologically-active component (full-length, intact FGF23) and inactive C-terminal fragments (C-terminal FGF23). FCM causes an increase in intact FGF23, which triggers the pathophysiological cascade of renal phosphate wasting, suppressed levels of 1,25-dihydroxyvitamin D, and secondary hyperparathyroidism[9]. In contrast, FDI does not appear to induce increased intact FGF23 levels, and is associated with a low incidence of hypophosphatemia[9].\nModerate to severe hypophosphatemia over time, as well as acute severe hypophosphatemia, can lead to serious complications, e.g., respiratory failure, haemolysis, left ventricular failure, and rhabdomyolysis[10-13]. Development of osteomalacia with pseudo-fractures has been found in patients with sustained hypophosphatemia [14-17]. However, there are uncertainties with regard to both the frequency of symptoms and the clinical impact of hypophosphatemia.\nReduced quality of life (QoL) is common and well-documented in IBD patients due to chronic inflammation in the gut and the occurrence of extra-intestinal manifestations[18,19]. Therefore, addressing additional symptoms and implications of hypophosphatemia in this patient group is a challenge, and no specific questionnaire related to hypophosphatemia is available.\nIn this short-term study, we aimed to investigate the mechanisms causing the development of hypophosphatemia in IBD patients, with ID or IDA, who received one high-dose (1000 mg) infusion of iron. Moreover, we aimed to document symptoms and clinical manifestations related to hypophosphatemia.", "This prospective observational study was conducted between February 1, 2017 and July 1, 2018. The study design and patient recruitment have previously been described in detail (Detlie et al[8]). In brief, adult IBD patients (> 18 years) diagnosed with ID or IDA (according to European Crohn’s and Colitis Organisation guidelines)[2] were recruited at two separate study sites in the southeast region of Norway and treated with either FCM or FDI.\nEligible patients were prescribed 1000 mg of high-dose intravenous iron, FCM (50 mg/mL) or iron derisomaltose (100 mg/mL), administered as a single dose. Patients who had received high-dose intravenous iron treatment or a packed red blood cell transfusion within 3 mo of study entry, or for whom high-dose intravenous iron treatment was contraindicated, were not included in the study.\nEnrolment continued until at least 50 consecutive patients with complete adherence to the study protocol were recruited at each site (a total of more than 100 patients) (Supplementary Figure 1). The enrolment period was followed by a prospective observation period, which lasted ≤ 7 wk for each patient and included three study visits.\nStudy inclusion was performed at baseline, at which time intravenous iron treatment was administered. Patients attended the clinic at week 2 (10-15 d) and at week 6 (5-7 wk) following intravenous iron treatment. Each patient could receive only one infusion within an approximate 2-mo period after consenting to study participation.", "Blood analysis at each study visit included ionised calcium, creatinine, phosphate, parathyroid hormone (PTH) and vitamin D (25-hydroxyvitamin D).\nBlood samples were also frozen and sent to Medizinische Universität Innsbruck, Universitätsklinik für Innere Medizin I, for analysis of 1,25-dihydroxyvitamin D, intact and C-terminal FGF23. The Kainos FGF-23 ELISA Kit was used for the FGF23 analysis. The assay for intact FGF23 measures only full-length peptide, whereas the assay for C-terminal FGF23 measures full-length peptide and the C-terminal fragments thereby representing total FGF23.\nSpot urine samples were collected at each study visit and analysed for urine phosphate and urine creatinine. A calculation of the fractional excretion of phosphate rate (FEPO4) was then performed using the formula, FEPO4 = (urine phosphate × plasma creatinine × 100)/(plasma phosphate × urine creatinine). Oslo University Hospital Ullevål used the Roche analysis method (Roche/Hitachi Cobas® C systems PHOS2 and CREP2) while Akershus University Hospital used the Vitros analysis (VITROS® MicroSlide Assay 5.1 FS Diluent Pack 3). The slight sensitivity difference between the two analytical methods was minimized by recalculating FEPO4 using the above-mentioned formula.\nSymptoms that might be related to hypophosphatemia were assessed at each of the three study visits using the MicroRPMTM (CareFusion) test to determine respiratory muscle function by measuring maximum inspiratory and maximum expiratory pressure, a health-related QoL questionnaire (36-item short form health survey, SF-36), and a visual analogue scale (VAS).\nFor the MicroRPMTM respiratory function test, patients were asked to inhale and exhale as hard as possible. The test was repeated three times at every visit, and the best result of the three attempts was registered.\nThe SF-36 is a generic, self-administered questionnaire containing 36 items[20]. The items are divided into eight multi-item scales that reflect general health, physical functioning, role limitations due to physical problems, bodily pain, vitality, mental health, social functioning, and role limitations due to emotional problems. Each scale is transformed into a 1-100 scale, where a lower score represents more disability. The processing of raw SF-36 data into results was executed according to the SF-36 scoring algorithms[21].\nThe VAS is a 10 cm line on which the patient is asked to place a vertical mark to indicate the level of intensity of a symptom that best fits his or her experience. Scores range from 0-100 (mm) where a higher score represents greater symptom intensity. The VAS was used to assess general weakness, fatigue, joint pain, joint stiffness, muscle pain, bone and skeletal pain, and difficulties performing daily activities.\nAll demographic information was collected from patients’ medical records and was entered into an electronic case report form.\nThe study was completed when all enrolled patients had received intravenous iron administration, had attended all three study visits, and had fulfilled the requirements of the study protocol.", "Serum phosphate, PTH, 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D, ionised calcium, creatinine, intact and C-terminal FGF23, and FEPO4 (urine phosphate and urine creatinine) were measured in order to assess possible mechanisms behind the development of hypophosphatemia after intravenous iron treatment. Results from the FCM treatment group were compared with results from the FDI treatment group.\nHypophosphatemia was defined as a serum phosphate level < 0.8 mmol/L (< 2.5 mg/dL). The clinical impact of hypophosphatemia was evaluated at week 2 and week 6 using the respiratory muscle function test, SF-36, and the VAS score. In relation to the assessment of clinical impact, the hypophosphatemia group was defined as patients experiencing hypophosphatemia at both week 2 and week 6. Results for patients with hypophosphatemia were compared with results for patients without hypophosphatemia, independent of treatment group.", "This study was designed to achieve 80% power to detect a difference in the primary outcome, which was the incidence of hypophosphatemia (previously described by Detlie et al[8]). Hence, the MicroRPMTM respiratory test, SF-36, and VAS scores were not used to justify sample size.\nData are presented descriptively, as mean with SD or 95% confidence intervals for continuous variables, and as the number of exposed patients (with proportions) for categorical variables. Hypothesis tests for differences in change between treatment groups, change from baseline, and groups with or without hypophosphatemia, were conducted using paired t-tests. All analyses were performed in R. A P value of < 0.05 was considered significant.", "The study protocol was approved by the relevant local regulatory and ethical committees and adhered to the applicable laws on data protection. A study registration application was sent to the EudraCT system with the application No. 2016-003476-41, but the application was deemed unnecessary since there were no indications of a medical intervention study.\nAll patients gave informed consent before inclusion into the study, and the study was performed in accordance with the principles for post-authorisation safety studies, according to Good Clinical Practice guidelines.\nAll biological material obtained from patients was destroyed after analysis, as were the frozen blood samples sent to the Medical University of Innsbruck.\nStudy nurses were blinded to the results of laboratory findings but, for safety reasons, the primary investigator at each study centre was not blinded.", "Of the 130 patients screened for this study, 106 patients (52 patients at Oslo University Hospital Ullevål and 54 patients at Akershus University Hospital) were included in the analyses. Demographic and clinical characteristics of the patients have previously been described[8].\nData for serum phosphate, FEPO4, intact and C-terminal FGF23, 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D, ionised calcium, and PTH at baseline and at each study visit are shown in Table 1. A sub-analysis of the same data, stratified according to hypophosphatemia status (with/without) at week 2 and at week 6, is shown in Table 2.\nDescriptive data for laboratory parameters at baseline, at week 2 and at week 6\nNormal phosphate levels: > 0.8 mmol/L = > 2.48 mg/dL; Mild hypophosphatemia 0.79-0.6 mmol/L = 2.44-1.86 mg/dL; Moderate hypophosphatemia 0.59-0.32 mmol/L = 1.83-0.99 mg/dL; Severe hypophosphatemia < 0.32 mmol/L = < 0.99 mg/dL. CI: Confidence interval; FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone; N/A: Not applicable.\nLaboratory parameters for patients stratified by hypophosphatemia status (with/without) at week 2 and at week 6\nNormal phosphate levels: > 0.8 mmol/L = > 2.48 mg/dL; Mild hypophosphatemia 0.79-0.6 mmol/L = 2.44-1.86 mg/dL; Moderate hypophosphatemia 0.59-0.32 mmol/L = 1.83-0.99 mg/dL; Severe hypophosphatemia < 0.32 mmol/L = < 0.99 mg/dL. FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone; N/A: Not applicable.\nSerum phosphate and urinary excretion of phosphate As previously described, following treatment with FCM, hypophosphatemia was present in 72.5% (37/51) of patients at week 2, and in 21.6% (11/51) of patients at week 6. In comparison, in the FDI treatment group, 11.3% (6/53) of patients had hypophosphatemia at week 2, and 3.7% (2/54) at week 6. There were no new incidences of hypophosphatemia at week 6. The difference in incidence was highly significant at both week 2 and 6 (P < 0.001 and P < 0.013, respectively)[8]. These findings are consistent with the mean urine FEPO4 that was significantly (P = 0.004) higher at week 2 in the FCM treatment group compared with the FDI treatment group, and still elevated (though declining) at week 6 in the FCM group (Table 1 and Figure 1A). In the sub-analysis, the FDI-treated patients with hypophosphatemia (n = 6) had numerically increased FEPO4 (Table 2).\n\nmean ± SD change from baseline in laboratory parameters in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Fractional excretion of phosphate; B: Intact fibroblast growth factor 23; C: C-terminal fibroblast growth factor 23; D: 25-Hydroxyvitamin D; E: 1,25-Hydroxyvitamin D; F: Ionised calcium; G: Parathyroid hormone. FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone.\nPatients in both treatment groups without hypophosphatemia at week 2 also experienced an increase in FEPO4 at week 2 compared to baseline values, but urinary phosphate excretion declined again at week 6 in these patients.\nAs previously described, following treatment with FCM, hypophosphatemia was present in 72.5% (37/51) of patients at week 2, and in 21.6% (11/51) of patients at week 6. In comparison, in the FDI treatment group, 11.3% (6/53) of patients had hypophosphatemia at week 2, and 3.7% (2/54) at week 6. There were no new incidences of hypophosphatemia at week 6. The difference in incidence was highly significant at both week 2 and 6 (P < 0.001 and P < 0.013, respectively)[8]. These findings are consistent with the mean urine FEPO4 that was significantly (P = 0.004) higher at week 2 in the FCM treatment group compared with the FDI treatment group, and still elevated (though declining) at week 6 in the FCM group (Table 1 and Figure 1A). In the sub-analysis, the FDI-treated patients with hypophosphatemia (n = 6) had numerically increased FEPO4 (Table 2).\n\nmean ± SD change from baseline in laboratory parameters in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Fractional excretion of phosphate; B: Intact fibroblast growth factor 23; C: C-terminal fibroblast growth factor 23; D: 25-Hydroxyvitamin D; E: 1,25-Hydroxyvitamin D; F: Ionised calcium; G: Parathyroid hormone. FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone.\nPatients in both treatment groups without hypophosphatemia at week 2 also experienced an increase in FEPO4 at week 2 compared to baseline values, but urinary phosphate excretion declined again at week 6 in these patients.\nFGF23 There was a significant (P < 0.001) increase in intact FGF23 from baseline to week 2 after infusion of FCM, compared with the FDI treatment group (Table 1 and Figure 1B). At week 6, intact FGF23 values in the FCM treatment group had returned close to baseline. In comparison, after FDI treatment no such increases were found (Figure 1B). At baseline, the serum concentration of C-terminal FGF23 was higher in the FDI treatment group than in the FCM treatment group (Table 1), and declined after FDI infusion (Figure 1C). This high value at baseline was not seen in the FCM treatment group (Table 1), which is probably compatible with the less severe ID/IDA seen in the FCM group[8].\nIn the sub-analysis, for the FCM-treated patients with hypophosphatemia, intact FGF23 was significantly increased compared with FCM-treated patients without hypophosphatemia at week 2 and at week 6 (Table 2). In the FCM-treated patients who had normal phosphate, intact FGF23 was not increased at week 2 or week 6.\nFor FDI-treated patients, the sub-analysis showed that there was no significant difference in mean intact FGF23 Levels between patients with/without hypophosphatemia at week 2 or at week 6 (Table 2). At week 2, only one FDI-treated patient with hypophosphatemia had significantly increased intact FGF23; the other five patients with hypophosphatemia had minimal change in their intact FGF23 values.\nThere was a significant (P < 0.001) increase in intact FGF23 from baseline to week 2 after infusion of FCM, compared with the FDI treatment group (Table 1 and Figure 1B). At week 6, intact FGF23 values in the FCM treatment group had returned close to baseline. In comparison, after FDI treatment no such increases were found (Figure 1B). At baseline, the serum concentration of C-terminal FGF23 was higher in the FDI treatment group than in the FCM treatment group (Table 1), and declined after FDI infusion (Figure 1C). This high value at baseline was not seen in the FCM treatment group (Table 1), which is probably compatible with the less severe ID/IDA seen in the FCM group[8].\nIn the sub-analysis, for the FCM-treated patients with hypophosphatemia, intact FGF23 was significantly increased compared with FCM-treated patients without hypophosphatemia at week 2 and at week 6 (Table 2). In the FCM-treated patients who had normal phosphate, intact FGF23 was not increased at week 2 or week 6.\nFor FDI-treated patients, the sub-analysis showed that there was no significant difference in mean intact FGF23 Levels between patients with/without hypophosphatemia at week 2 or at week 6 (Table 2). At week 2, only one FDI-treated patient with hypophosphatemia had significantly increased intact FGF23; the other five patients with hypophosphatemia had minimal change in their intact FGF23 values.\nVitamin D There were no significant differences between the treatment groups in the concentration of 25-hydroxyvitamin D throughout the study period (Figure 1D). However, the sub-analysis showed that 25-hydroxyvitamin D concentrations were lower at week 6 in the two FDI-treated patients with hypophosphatemia when compared with baseline concentrations within the same group (Tables 1 and 2). At week 2, 1,25-dihydroxyvitamin D concentrations were significantly lower in patients who received FCM compared with patients who received FDI (Table 1). In the FCM-treatment group, the mean concentration of 1,25-dihydroxyvitamin D returned to baseline at week 6 (Table 1 and Figure 1E). However, the sub-analysis revealed that, for the FCM-treated patients with hypophosphatemia, low 1,25-dihydroxyvitamin D levels persisted at week 6 (Table 2). In the subgroups of patients without hypophosphatemia, 1,25-dihydroxyvitamin D levels were relatively unchanged.\nIn our cohort, we identified 36 patients (34.0%) with vitamin D deficiency (25-hydroxyvitamin D < 50 nmol/L) at baseline; 10 of these patients had severe vitamin D deficiency (25-hydroxyvitamin D < 30 nmol/L). The distribution of these patients was equal in the two treatment groups, as well as equally distributed across disease states – ulcerative colitis and Crohn’s disease. Moreover, we found no association between low levels of vitamin D and development of hypophosphatemia.\nThere were no significant differences between the treatment groups in the concentration of 25-hydroxyvitamin D throughout the study period (Figure 1D). However, the sub-analysis showed that 25-hydroxyvitamin D concentrations were lower at week 6 in the two FDI-treated patients with hypophosphatemia when compared with baseline concentrations within the same group (Tables 1 and 2). At week 2, 1,25-dihydroxyvitamin D concentrations were significantly lower in patients who received FCM compared with patients who received FDI (Table 1). In the FCM-treatment group, the mean concentration of 1,25-dihydroxyvitamin D returned to baseline at week 6 (Table 1 and Figure 1E). However, the sub-analysis revealed that, for the FCM-treated patients with hypophosphatemia, low 1,25-dihydroxyvitamin D levels persisted at week 6 (Table 2). In the subgroups of patients without hypophosphatemia, 1,25-dihydroxyvitamin D levels were relatively unchanged.\nIn our cohort, we identified 36 patients (34.0%) with vitamin D deficiency (25-hydroxyvitamin D < 50 nmol/L) at baseline; 10 of these patients had severe vitamin D deficiency (25-hydroxyvitamin D < 30 nmol/L). The distribution of these patients was equal in the two treatment groups, as well as equally distributed across disease states – ulcerative colitis and Crohn’s disease. Moreover, we found no association between low levels of vitamin D and development of hypophosphatemia.\nCalcium and PTH Ionised calcium values dropped significantly from baseline to week 2 in the FCM treatment group compared with the FDI treatment group (P < 0.012) but stayed within normal range. The mean values in the FCM group had increased by week 6, but the between-group difference was still significant (P < 0.044). Calcium values remained stable throughout the study in the FDI treatment group (Figure 1F), and in the subgroup of FCM-treated patients who did not develop hypophosphatemia. The sub-analysis showed that there was a numerically lower level of ionised calcium in the FDI-treated patients with hypophosphatemia than in the FDI-treated patients without hypophosphatemia (Table 2).\nPTH values were elevated (> 7 pmol/L) in 28 patients (26.4%) at baseline; the distribution was similar between treatment groups. PTH concentrations were similar between treatment groups at baseline, and no significant between-group differences were observed in mean PTH concentrations at week 2, and at week 6 (Table 1). PTH values increased in both treatment groups at week 2 and decreased again at week 6 (Figure 1G). The sub-analysis indicated that the increase in PTH in both treatment groups was mainly driven by the patients who developed hypophosphatemia, with significant differences at week 2 and week 6 for the FCM-treated patients with hypophosphatemia compared to FCM-treated patients without hypophosphatemia (Table 2).\nIonised calcium values dropped significantly from baseline to week 2 in the FCM treatment group compared with the FDI treatment group (P < 0.012) but stayed within normal range. The mean values in the FCM group had increased by week 6, but the between-group difference was still significant (P < 0.044). Calcium values remained stable throughout the study in the FDI treatment group (Figure 1F), and in the subgroup of FCM-treated patients who did not develop hypophosphatemia. The sub-analysis showed that there was a numerically lower level of ionised calcium in the FDI-treated patients with hypophosphatemia than in the FDI-treated patients without hypophosphatemia (Table 2).\nPTH values were elevated (> 7 pmol/L) in 28 patients (26.4%) at baseline; the distribution was similar between treatment groups. PTH concentrations were similar between treatment groups at baseline, and no significant between-group differences were observed in mean PTH concentrations at week 2, and at week 6 (Table 1). PTH values increased in both treatment groups at week 2 and decreased again at week 6 (Figure 1G). The sub-analysis indicated that the increase in PTH in both treatment groups was mainly driven by the patients who developed hypophosphatemia, with significant differences at week 2 and week 6 for the FCM-treated patients with hypophosphatemia compared to FCM-treated patients without hypophosphatemia (Table 2).\nRespiratory muscle function tests In the comparison of patients who developed hypophosphatemia vs those who did not develop hypophosphatemia, independent of treatment group, no significant differences were observed in the respiratory muscle function test results. The differences between patients with hypophosphatemia and those with normal phosphate values were minimal and the standard deviation was wide in both groups (Figure 2).\n\nmean ± SD changes from baseline in respiratory pressure in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Inspiratory pressure; B: Expiratory pressure.\nIn the comparison of patients who developed hypophosphatemia vs those who did not develop hypophosphatemia, independent of treatment group, no significant differences were observed in the respiratory muscle function test results. The differences between patients with hypophosphatemia and those with normal phosphate values were minimal and the standard deviation was wide in both groups (Figure 2).\n\nmean ± SD changes from baseline in respiratory pressure in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Inspiratory pressure; B: Expiratory pressure.\nSF-36 The results of the SF-36 QoL assessment are presented in Table 3. Overall, there were no significant differences between patient groups with or without hypophosphatemia at baseline and at any time point during the study. The mean scores at baseline in both treatment groups were generally low.\nDescriptive 36-item short form health survey scores for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; SF-36: 36-Item short form health survey.\nThe results of the SF-36 QoL assessment are presented in Table 3. Overall, there were no significant differences between patient groups with or without hypophosphatemia at baseline and at any time point during the study. The mean scores at baseline in both treatment groups were generally low.\nDescriptive 36-item short form health survey scores for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; SF-36: 36-Item short form health survey.\nVAS scores There were no significant differences in VAS scores between the groups of patients with/without hypophosphatemia at week 2 and at week 6 (Table 4). Overall, VAS scores were elevated at baseline. However, the group of patients who developed hypophosphatemia had lower VAS scores at baseline for the items joint pain, muscle pain, and bone and skeletal pain, compared to the group of patients who did not develop hypophosphatemia; between-group differences were not significant for these items. There was, however, a significant between-group difference (P < 0.001) at baseline for the VAS joint stiffness item score, with lower values in the group of patients who developed hypophosphatemia.\nDescriptive visual analogue scale score for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; VAS: Visual analogue scale.\nThere were no significant differences in VAS scores between the groups of patients with/without hypophosphatemia at week 2 and at week 6 (Table 4). Overall, VAS scores were elevated at baseline. However, the group of patients who developed hypophosphatemia had lower VAS scores at baseline for the items joint pain, muscle pain, and bone and skeletal pain, compared to the group of patients who did not develop hypophosphatemia; between-group differences were not significant for these items. There was, however, a significant between-group difference (P < 0.001) at baseline for the VAS joint stiffness item score, with lower values in the group of patients who developed hypophosphatemia.\nDescriptive visual analogue scale score for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; VAS: Visual analogue scale.", "As previously described, following treatment with FCM, hypophosphatemia was present in 72.5% (37/51) of patients at week 2, and in 21.6% (11/51) of patients at week 6. In comparison, in the FDI treatment group, 11.3% (6/53) of patients had hypophosphatemia at week 2, and 3.7% (2/54) at week 6. There were no new incidences of hypophosphatemia at week 6. The difference in incidence was highly significant at both week 2 and 6 (P < 0.001 and P < 0.013, respectively)[8]. These findings are consistent with the mean urine FEPO4 that was significantly (P = 0.004) higher at week 2 in the FCM treatment group compared with the FDI treatment group, and still elevated (though declining) at week 6 in the FCM group (Table 1 and Figure 1A). In the sub-analysis, the FDI-treated patients with hypophosphatemia (n = 6) had numerically increased FEPO4 (Table 2).\n\nmean ± SD change from baseline in laboratory parameters in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Fractional excretion of phosphate; B: Intact fibroblast growth factor 23; C: C-terminal fibroblast growth factor 23; D: 25-Hydroxyvitamin D; E: 1,25-Hydroxyvitamin D; F: Ionised calcium; G: Parathyroid hormone. FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone.\nPatients in both treatment groups without hypophosphatemia at week 2 also experienced an increase in FEPO4 at week 2 compared to baseline values, but urinary phosphate excretion declined again at week 6 in these patients.", "There was a significant (P < 0.001) increase in intact FGF23 from baseline to week 2 after infusion of FCM, compared with the FDI treatment group (Table 1 and Figure 1B). At week 6, intact FGF23 values in the FCM treatment group had returned close to baseline. In comparison, after FDI treatment no such increases were found (Figure 1B). At baseline, the serum concentration of C-terminal FGF23 was higher in the FDI treatment group than in the FCM treatment group (Table 1), and declined after FDI infusion (Figure 1C). This high value at baseline was not seen in the FCM treatment group (Table 1), which is probably compatible with the less severe ID/IDA seen in the FCM group[8].\nIn the sub-analysis, for the FCM-treated patients with hypophosphatemia, intact FGF23 was significantly increased compared with FCM-treated patients without hypophosphatemia at week 2 and at week 6 (Table 2). In the FCM-treated patients who had normal phosphate, intact FGF23 was not increased at week 2 or week 6.\nFor FDI-treated patients, the sub-analysis showed that there was no significant difference in mean intact FGF23 Levels between patients with/without hypophosphatemia at week 2 or at week 6 (Table 2). At week 2, only one FDI-treated patient with hypophosphatemia had significantly increased intact FGF23; the other five patients with hypophosphatemia had minimal change in their intact FGF23 values.", "There were no significant differences between the treatment groups in the concentration of 25-hydroxyvitamin D throughout the study period (Figure 1D). However, the sub-analysis showed that 25-hydroxyvitamin D concentrations were lower at week 6 in the two FDI-treated patients with hypophosphatemia when compared with baseline concentrations within the same group (Tables 1 and 2). At week 2, 1,25-dihydroxyvitamin D concentrations were significantly lower in patients who received FCM compared with patients who received FDI (Table 1). In the FCM-treatment group, the mean concentration of 1,25-dihydroxyvitamin D returned to baseline at week 6 (Table 1 and Figure 1E). However, the sub-analysis revealed that, for the FCM-treated patients with hypophosphatemia, low 1,25-dihydroxyvitamin D levels persisted at week 6 (Table 2). In the subgroups of patients without hypophosphatemia, 1,25-dihydroxyvitamin D levels were relatively unchanged.\nIn our cohort, we identified 36 patients (34.0%) with vitamin D deficiency (25-hydroxyvitamin D < 50 nmol/L) at baseline; 10 of these patients had severe vitamin D deficiency (25-hydroxyvitamin D < 30 nmol/L). The distribution of these patients was equal in the two treatment groups, as well as equally distributed across disease states – ulcerative colitis and Crohn’s disease. Moreover, we found no association between low levels of vitamin D and development of hypophosphatemia.", "Ionised calcium values dropped significantly from baseline to week 2 in the FCM treatment group compared with the FDI treatment group (P < 0.012) but stayed within normal range. The mean values in the FCM group had increased by week 6, but the between-group difference was still significant (P < 0.044). Calcium values remained stable throughout the study in the FDI treatment group (Figure 1F), and in the subgroup of FCM-treated patients who did not develop hypophosphatemia. The sub-analysis showed that there was a numerically lower level of ionised calcium in the FDI-treated patients with hypophosphatemia than in the FDI-treated patients without hypophosphatemia (Table 2).\nPTH values were elevated (> 7 pmol/L) in 28 patients (26.4%) at baseline; the distribution was similar between treatment groups. PTH concentrations were similar between treatment groups at baseline, and no significant between-group differences were observed in mean PTH concentrations at week 2, and at week 6 (Table 1). PTH values increased in both treatment groups at week 2 and decreased again at week 6 (Figure 1G). The sub-analysis indicated that the increase in PTH in both treatment groups was mainly driven by the patients who developed hypophosphatemia, with significant differences at week 2 and week 6 for the FCM-treated patients with hypophosphatemia compared to FCM-treated patients without hypophosphatemia (Table 2).", "In the comparison of patients who developed hypophosphatemia vs those who did not develop hypophosphatemia, independent of treatment group, no significant differences were observed in the respiratory muscle function test results. The differences between patients with hypophosphatemia and those with normal phosphate values were minimal and the standard deviation was wide in both groups (Figure 2).\n\nmean ± SD changes from baseline in respiratory pressure in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Inspiratory pressure; B: Expiratory pressure.", "The results of the SF-36 QoL assessment are presented in Table 3. Overall, there were no significant differences between patient groups with or without hypophosphatemia at baseline and at any time point during the study. The mean scores at baseline in both treatment groups were generally low.\nDescriptive 36-item short form health survey scores for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; SF-36: 36-Item short form health survey.", "There were no significant differences in VAS scores between the groups of patients with/without hypophosphatemia at week 2 and at week 6 (Table 4). Overall, VAS scores were elevated at baseline. However, the group of patients who developed hypophosphatemia had lower VAS scores at baseline for the items joint pain, muscle pain, and bone and skeletal pain, compared to the group of patients who did not develop hypophosphatemia; between-group differences were not significant for these items. There was, however, a significant between-group difference (P < 0.001) at baseline for the VAS joint stiffness item score, with lower values in the group of patients who developed hypophosphatemia.\nDescriptive visual analogue scale score for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; VAS: Visual analogue scale.", "Our study indicates that FGF23 plays an important role in the development of hypophosphatemia in IBD patients treated with FCM. In these patients, a high level of intact FGF23, an increased excretion of phosphate in the urine, a decrease of 1,25-dihydroxyvitamin D and of serum calcium levels, and a slight elevation of PTH, was demonstrated.\nPrevious clinical trials of FCM have shown similar results[9,22]. However, for the most part, these studies have been conducted in healthy and, predominantly, female populations. The role of FGF23 has also been described in earlier publications[23-26]. Regulation of phosphate concentrations in the body seems to be strongly influenced by intact FGF23, which reduces phosphate reabsorption in the proximal tubules in the kidneys and inhibits production of 1,25-dihydroxyvitamin D, probably by inhibiting the activity of the enzyme 25-hydroxyvitamin D-1a-hydroxylase and increased expression of 24-hydroxylase[24,26]. Our findings suggest that FCM could have a direct impact on cleavage of FGF23, resulting in a high level of intact FGF23 and consequent phosphate wasting. This might also explain why baseline phosphate level does not predict the development of mild or severe hypophosphatemia, due to the inappropriate excretion of available phosphate in the urine, following FCM treatment[8]. We also observed a decrease in 1,25-dihydroxyvitamin D (the active vitamin D metabolite), a decrease in ionised calcium, and development of secondary hyperparathyroidism. This might explain why some patients treated with FCM still had hypophosphatemia six weeks after treatment, when the intact FGF23 values had normalized (Table 2) since elevated PTH promotes excretion of phosphate in the urine[9,27,28].\nThe majority of patients in the FCM treatment group developed hypophosphatemia at week 2. The remaining patients did not develop hypophosphatemia and had unchanged levels of intact FGF23. So, there is a clear association between the development of high levels of intact FGF23 and hypophosphatemia. Therefore, it can only be speculated that there might be some individual factors related to the handling of FCM that cause the majority of patients treated with FCM to develop hypophosphatemia, whereas others do not. Neither is it known if any individual patient would develop hypophosphatemia on subsequent administrations of FCM, or if the effect of FCM treatment on phosphate wasting is indiscriminate. Perhaps some patients are protected against the influence of FCM on the enzyme responsible for FGF23 protein cleavage. From our results, we postulate that the mechanism of FCM-induced hypophosphatemia is not related to IBD; instead, it appears to be independently connected to the drug itself.\nA few patients who received treatment with FDI also developed hypophosphatemia but, unlike those receiving FCM, these patients did not on average have significantly elevated intact FGF23 Levels when assessed at week 2, which would suggest a different underlying mechanism. A transient increase in intact FGF23 during the first 2 wk in patients experiencing hypophosphatemia cannot be ruled out, as data were not collected during this time period. A numerical increase in PTH was observed at week 2 along with decreased ionised calcium, and decreased 25-hydroxyvitamin D at week 6. It is not clear whether these observations are the result of a transient increase of intact FGF23 during the first 2 wk, or solely a physiological response to a rapid correction of ID, or simply an artefact due to the low numbers of FDI patients who developed hypophosphatemia. The general physiological response of mineral metabolism markers to rapid ID correction is not fully elucidated and is an area of further research.\nAn important observation is that 34% of the study population was vitamin D deficient at baseline with 25-hydroxyvitamin D values < 50 nmol/L and, perhaps more interestingly, 24% of the patients had PTH values compatible with secondary hyperparathyroidism. These findings were equally distributed between the two treatment groups. This disturbance in vitamin D metabolism is unlikely to be a consequence of previous iron infusions since no patients received high-dose intravenous iron treatment during the 6 mo prior to inclusion in this study. The high prevalence of vitamin D deficiency at baseline is in agreement with previous studies of patients with IBD[29]. However, it is important to note that, in our study, many of the samples were taken during the winter months when sun exposure is reduced in Norway, and individuals could therefore be expected to be somewhat vitamin D deficient during this time. Nevertheless, this finding is important since both hypophosphatemia and vitamin D deficiency can contribute to the development of metabolic bone disease, including osteomalacia.\nGuidelines regarding hypophosphatemia diagnosis, treatment, and follow-up are available, but the possible risk or incidence rate of developing hypophosphatemia with symptoms or complications are rarely mentioned[30]. A risk of developing respiratory failure, rhabdomyolysis, and left ventricular failure due to severe hypophosphatemia has been reported in case series[10]. More recent data also predict an increased risk of developing osteomalacia, especially in long-standing hypophosphatemia[14,15]. What is less well known is the number of patients developing more subtle, but identifiable, symptoms related to hypophosphatemia that are experienced as troublesome and might influence QoL.\nWith respect to the clinical impact of hypophosphatemia, measuring forced inspiratory and expiratory respiratory pressure can be used as a proxy to assess the physical effect of hypophosphatemia on skeletal and proximal muscles. There are no specific questionnaires available to evaluate the clinical impact of hypophosphatemia. The SF-36 is, however, one of the most commonly applied QoL questionnaires used world-wide in health surveys. Additionally, the VAS score can be used as a general assessment of impact of symptoms, such as fatigue, general weakness, bone and skeletal pain, and joint and muscle conditions. In our study, these three methods were applied to assess clinical impact in patients who developed hypophosphatemia compared to those who did not develop hypophosphatemia. All three methods failed to demonstrate significant differences in clinical impact following one administration of high-dose intravenous iron in this short-term study.\nWe hypothesize several reasons that might explain these results. In addition to the fact that a type II error cannot be excluded, it can be speculated that the positive effect of the correction of ID or IDA plays a more important role than any short-term negative clinical impact of hypophosphatemia and, hence, the effects of hypophosphatemia would be difficult to discern in our study. Another challenge is that IBD, ID, IDA and hypophosphatemia are associated with similar symptoms and, possibly, similar impacts on daily life. Indeed, assessing the specific impact of hypophosphatemia with questionnaires would, therefore, prove difficult. Since the SF-36 and the VAS questionnaires are not disease- or population-specific, there may be uncertainty surrounding the reliability of the results. Additionally, the patient cohort had more than one dynamic medical condition, with overlapping symptoms, and patients were observed in a longitudinal manner. Certainly, it would be almost impossible to determine which disease state or co-morbidity is reflecting improvement or worsening of clinical status.\nThe already affected baseline recordings in SF-36 and the VAS score should not go unnoticed. These findings mirror previous studies of IBD populations[31], and reflect the reduced QoL and the intensity of symptoms that these patients experience in general. Finally, the fact that we did not detect clinical consequences in patients who developed hypophosphatemia suggests that, in order to detect overt symptoms and complications, the population size needs to be larger than our sample, as one might expect such complications to be relatively rare. Hence, this needs to be taken into account when considering the expectation of finding significant changes in the clinical outcomes in this study.", "In summary, our study has implicated the small peptide hormone FGF23 in the development of hypophosphatemia in IBD patients treated with FCM. An increase in intact FGF23 occurs, which probably results in phosphate wasting in the urine. Assessment of symptoms did not exclude, nor did they demonstrate, any short-term clinical impact of hypophosphatemia in IBD patients treated for ID or IDA with high-dose intravenous iron." ]

[ null, null, null, null, null, null, null, null, null, null, null, null, null, null, null, null ]

[ "INTRODUCTION", "MATERIALS AND METHODS", "Study design and patient population", "Study assessments and data collection", "Study outcomes", "Statistical analysis", "Ethical considerations", "RESULTS", "Serum phosphate and urinary excretion of phosphate", "FGF23", "Vitamin D", "Calcium and PTH", "Respiratory muscle function tests", "SF-36", "VAS scores", "DISCUSSION", "CONCLUSION" ]

[ "Iron replacement therapy is often needed in patients with inflammatory bowel disease (IBD) because iron deficiency (ID) and ID anaemia (IDA) occur frequently in this patient group[1-3]. A large proportion of IBD patients experience intolerance to oral iron[4]. Additionally, it is asserted that oral iron can lead to an exacerbation of inflammation in the bowel mucosa due to a local effect on the enterocytes[5-7]. Therefore, administration of high-dose iron as an intravenous infusion is an effective, suitable and convenient treatment option in IBD. Ferric carboxymaltose (FCM; Ferinject®; Vifor Pharma) and ferric derisomaltose (FDI), previously known as iron isomaltoside (Monofer®; Pharmacosmos A/S), are the most widely used preparations in Europe when high-dose intravenous iron is indicated.\nIn a recent publication, we described a high incidence of hypophosphatemia in IBD patients who had received treatment with FCM[8]. The mechanism behind the development of hypophosphatemia has been described by Wolf et al[9], but probably is not yet fully understood, and has not been investigated in patients with IBD. Fibroblast growth factor 23 (FGF23) is a small peptide hormone, synthesized by osteocytes, which regulates phosphate and vitamin D homeostasis[9]. FGF23 consists of a biologically-active component (full-length, intact FGF23) and inactive C-terminal fragments (C-terminal FGF23). FCM causes an increase in intact FGF23, which triggers the pathophysiological cascade of renal phosphate wasting, suppressed levels of 1,25-dihydroxyvitamin D, and secondary hyperparathyroidism[9]. In contrast, FDI does not appear to induce increased intact FGF23 levels, and is associated with a low incidence of hypophosphatemia[9].\nModerate to severe hypophosphatemia over time, as well as acute severe hypophosphatemia, can lead to serious complications, e.g., respiratory failure, haemolysis, left ventricular failure, and rhabdomyolysis[10-13]. Development of osteomalacia with pseudo-fractures has been found in patients with sustained hypophosphatemia [14-17]. However, there are uncertainties with regard to both the frequency of symptoms and the clinical impact of hypophosphatemia.\nReduced quality of life (QoL) is common and well-documented in IBD patients due to chronic inflammation in the gut and the occurrence of extra-intestinal manifestations[18,19]. Therefore, addressing additional symptoms and implications of hypophosphatemia in this patient group is a challenge, and no specific questionnaire related to hypophosphatemia is available.\nIn this short-term study, we aimed to investigate the mechanisms causing the development of hypophosphatemia in IBD patients, with ID or IDA, who received one high-dose (1000 mg) infusion of iron. Moreover, we aimed to document symptoms and clinical manifestations related to hypophosphatemia.", "Study design and patient population This prospective observational study was conducted between February 1, 2017 and July 1, 2018. The study design and patient recruitment have previously been described in detail (Detlie et al[8]). In brief, adult IBD patients (> 18 years) diagnosed with ID or IDA (according to European Crohn’s and Colitis Organisation guidelines)[2] were recruited at two separate study sites in the southeast region of Norway and treated with either FCM or FDI.\nEligible patients were prescribed 1000 mg of high-dose intravenous iron, FCM (50 mg/mL) or iron derisomaltose (100 mg/mL), administered as a single dose. Patients who had received high-dose intravenous iron treatment or a packed red blood cell transfusion within 3 mo of study entry, or for whom high-dose intravenous iron treatment was contraindicated, were not included in the study.\nEnrolment continued until at least 50 consecutive patients with complete adherence to the study protocol were recruited at each site (a total of more than 100 patients) (Supplementary Figure 1). The enrolment period was followed by a prospective observation period, which lasted ≤ 7 wk for each patient and included three study visits.\nStudy inclusion was performed at baseline, at which time intravenous iron treatment was administered. Patients attended the clinic at week 2 (10-15 d) and at week 6 (5-7 wk) following intravenous iron treatment. Each patient could receive only one infusion within an approximate 2-mo period after consenting to study participation.\nThis prospective observational study was conducted between February 1, 2017 and July 1, 2018. The study design and patient recruitment have previously been described in detail (Detlie et al[8]). In brief, adult IBD patients (> 18 years) diagnosed with ID or IDA (according to European Crohn’s and Colitis Organisation guidelines)[2] were recruited at two separate study sites in the southeast region of Norway and treated with either FCM or FDI.\nEligible patients were prescribed 1000 mg of high-dose intravenous iron, FCM (50 mg/mL) or iron derisomaltose (100 mg/mL), administered as a single dose. Patients who had received high-dose intravenous iron treatment or a packed red blood cell transfusion within 3 mo of study entry, or for whom high-dose intravenous iron treatment was contraindicated, were not included in the study.\nEnrolment continued until at least 50 consecutive patients with complete adherence to the study protocol were recruited at each site (a total of more than 100 patients) (Supplementary Figure 1). The enrolment period was followed by a prospective observation period, which lasted ≤ 7 wk for each patient and included three study visits.\nStudy inclusion was performed at baseline, at which time intravenous iron treatment was administered. Patients attended the clinic at week 2 (10-15 d) and at week 6 (5-7 wk) following intravenous iron treatment. Each patient could receive only one infusion within an approximate 2-mo period after consenting to study participation.\nStudy assessments and data collection Blood analysis at each study visit included ionised calcium, creatinine, phosphate, parathyroid hormone (PTH) and vitamin D (25-hydroxyvitamin D).\nBlood samples were also frozen and sent to Medizinische Universität Innsbruck, Universitätsklinik für Innere Medizin I, for analysis of 1,25-dihydroxyvitamin D, intact and C-terminal FGF23. The Kainos FGF-23 ELISA Kit was used for the FGF23 analysis. The assay for intact FGF23 measures only full-length peptide, whereas the assay for C-terminal FGF23 measures full-length peptide and the C-terminal fragments thereby representing total FGF23.\nSpot urine samples were collected at each study visit and analysed for urine phosphate and urine creatinine. A calculation of the fractional excretion of phosphate rate (FEPO4) was then performed using the formula, FEPO4 = (urine phosphate × plasma creatinine × 100)/(plasma phosphate × urine creatinine). Oslo University Hospital Ullevål used the Roche analysis method (Roche/Hitachi Cobas® C systems PHOS2 and CREP2) while Akershus University Hospital used the Vitros analysis (VITROS® MicroSlide Assay 5.1 FS Diluent Pack 3). The slight sensitivity difference between the two analytical methods was minimized by recalculating FEPO4 using the above-mentioned formula.\nSymptoms that might be related to hypophosphatemia were assessed at each of the three study visits using the MicroRPMTM (CareFusion) test to determine respiratory muscle function by measuring maximum inspiratory and maximum expiratory pressure, a health-related QoL questionnaire (36-item short form health survey, SF-36), and a visual analogue scale (VAS).\nFor the MicroRPMTM respiratory function test, patients were asked to inhale and exhale as hard as possible. The test was repeated three times at every visit, and the best result of the three attempts was registered.\nThe SF-36 is a generic, self-administered questionnaire containing 36 items[20]. The items are divided into eight multi-item scales that reflect general health, physical functioning, role limitations due to physical problems, bodily pain, vitality, mental health, social functioning, and role limitations due to emotional problems. Each scale is transformed into a 1-100 scale, where a lower score represents more disability. The processing of raw SF-36 data into results was executed according to the SF-36 scoring algorithms[21].\nThe VAS is a 10 cm line on which the patient is asked to place a vertical mark to indicate the level of intensity of a symptom that best fits his or her experience. Scores range from 0-100 (mm) where a higher score represents greater symptom intensity. The VAS was used to assess general weakness, fatigue, joint pain, joint stiffness, muscle pain, bone and skeletal pain, and difficulties performing daily activities.\nAll demographic information was collected from patients’ medical records and was entered into an electronic case report form.\nThe study was completed when all enrolled patients had received intravenous iron administration, had attended all three study visits, and had fulfilled the requirements of the study protocol.\nBlood analysis at each study visit included ionised calcium, creatinine, phosphate, parathyroid hormone (PTH) and vitamin D (25-hydroxyvitamin D).\nBlood samples were also frozen and sent to Medizinische Universität Innsbruck, Universitätsklinik für Innere Medizin I, for analysis of 1,25-dihydroxyvitamin D, intact and C-terminal FGF23. The Kainos FGF-23 ELISA Kit was used for the FGF23 analysis. The assay for intact FGF23 measures only full-length peptide, whereas the assay for C-terminal FGF23 measures full-length peptide and the C-terminal fragments thereby representing total FGF23.\nSpot urine samples were collected at each study visit and analysed for urine phosphate and urine creatinine. A calculation of the fractional excretion of phosphate rate (FEPO4) was then performed using the formula, FEPO4 = (urine phosphate × plasma creatinine × 100)/(plasma phosphate × urine creatinine). Oslo University Hospital Ullevål used the Roche analysis method (Roche/Hitachi Cobas® C systems PHOS2 and CREP2) while Akershus University Hospital used the Vitros analysis (VITROS® MicroSlide Assay 5.1 FS Diluent Pack 3). The slight sensitivity difference between the two analytical methods was minimized by recalculating FEPO4 using the above-mentioned formula.\nSymptoms that might be related to hypophosphatemia were assessed at each of the three study visits using the MicroRPMTM (CareFusion) test to determine respiratory muscle function by measuring maximum inspiratory and maximum expiratory pressure, a health-related QoL questionnaire (36-item short form health survey, SF-36), and a visual analogue scale (VAS).\nFor the MicroRPMTM respiratory function test, patients were asked to inhale and exhale as hard as possible. The test was repeated three times at every visit, and the best result of the three attempts was registered.\nThe SF-36 is a generic, self-administered questionnaire containing 36 items[20]. The items are divided into eight multi-item scales that reflect general health, physical functioning, role limitations due to physical problems, bodily pain, vitality, mental health, social functioning, and role limitations due to emotional problems. Each scale is transformed into a 1-100 scale, where a lower score represents more disability. The processing of raw SF-36 data into results was executed according to the SF-36 scoring algorithms[21].\nThe VAS is a 10 cm line on which the patient is asked to place a vertical mark to indicate the level of intensity of a symptom that best fits his or her experience. Scores range from 0-100 (mm) where a higher score represents greater symptom intensity. The VAS was used to assess general weakness, fatigue, joint pain, joint stiffness, muscle pain, bone and skeletal pain, and difficulties performing daily activities.\nAll demographic information was collected from patients’ medical records and was entered into an electronic case report form.\nThe study was completed when all enrolled patients had received intravenous iron administration, had attended all three study visits, and had fulfilled the requirements of the study protocol.\nStudy outcomes Serum phosphate, PTH, 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D, ionised calcium, creatinine, intact and C-terminal FGF23, and FEPO4 (urine phosphate and urine creatinine) were measured in order to assess possible mechanisms behind the development of hypophosphatemia after intravenous iron treatment. Results from the FCM treatment group were compared with results from the FDI treatment group.\nHypophosphatemia was defined as a serum phosphate level < 0.8 mmol/L (< 2.5 mg/dL). The clinical impact of hypophosphatemia was evaluated at week 2 and week 6 using the respiratory muscle function test, SF-36, and the VAS score. In relation to the assessment of clinical impact, the hypophosphatemia group was defined as patients experiencing hypophosphatemia at both week 2 and week 6. Results for patients with hypophosphatemia were compared with results for patients without hypophosphatemia, independent of treatment group.\nSerum phosphate, PTH, 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D, ionised calcium, creatinine, intact and C-terminal FGF23, and FEPO4 (urine phosphate and urine creatinine) were measured in order to assess possible mechanisms behind the development of hypophosphatemia after intravenous iron treatment. Results from the FCM treatment group were compared with results from the FDI treatment group.\nHypophosphatemia was defined as a serum phosphate level < 0.8 mmol/L (< 2.5 mg/dL). The clinical impact of hypophosphatemia was evaluated at week 2 and week 6 using the respiratory muscle function test, SF-36, and the VAS score. In relation to the assessment of clinical impact, the hypophosphatemia group was defined as patients experiencing hypophosphatemia at both week 2 and week 6. Results for patients with hypophosphatemia were compared with results for patients without hypophosphatemia, independent of treatment group.\nStatistical analysis This study was designed to achieve 80% power to detect a difference in the primary outcome, which was the incidence of hypophosphatemia (previously described by Detlie et al[8]). Hence, the MicroRPMTM respiratory test, SF-36, and VAS scores were not used to justify sample size.\nData are presented descriptively, as mean with SD or 95% confidence intervals for continuous variables, and as the number of exposed patients (with proportions) for categorical variables. Hypothesis tests for differences in change between treatment groups, change from baseline, and groups with or without hypophosphatemia, were conducted using paired t-tests. All analyses were performed in R. A P value of < 0.05 was considered significant.\nThis study was designed to achieve 80% power to detect a difference in the primary outcome, which was the incidence of hypophosphatemia (previously described by Detlie et al[8]). Hence, the MicroRPMTM respiratory test, SF-36, and VAS scores were not used to justify sample size.\nData are presented descriptively, as mean with SD or 95% confidence intervals for continuous variables, and as the number of exposed patients (with proportions) for categorical variables. Hypothesis tests for differences in change between treatment groups, change from baseline, and groups with or without hypophosphatemia, were conducted using paired t-tests. All analyses were performed in R. A P value of < 0.05 was considered significant.\nEthical considerations The study protocol was approved by the relevant local regulatory and ethical committees and adhered to the applicable laws on data protection. A study registration application was sent to the EudraCT system with the application No. 2016-003476-41, but the application was deemed unnecessary since there were no indications of a medical intervention study.\nAll patients gave informed consent before inclusion into the study, and the study was performed in accordance with the principles for post-authorisation safety studies, according to Good Clinical Practice guidelines.\nAll biological material obtained from patients was destroyed after analysis, as were the frozen blood samples sent to the Medical University of Innsbruck.\nStudy nurses were blinded to the results of laboratory findings but, for safety reasons, the primary investigator at each study centre was not blinded.\nThe study protocol was approved by the relevant local regulatory and ethical committees and adhered to the applicable laws on data protection. A study registration application was sent to the EudraCT system with the application No. 2016-003476-41, but the application was deemed unnecessary since there were no indications of a medical intervention study.\nAll patients gave informed consent before inclusion into the study, and the study was performed in accordance with the principles for post-authorisation safety studies, according to Good Clinical Practice guidelines.\nAll biological material obtained from patients was destroyed after analysis, as were the frozen blood samples sent to the Medical University of Innsbruck.\nStudy nurses were blinded to the results of laboratory findings but, for safety reasons, the primary investigator at each study centre was not blinded.", "This prospective observational study was conducted between February 1, 2017 and July 1, 2018. The study design and patient recruitment have previously been described in detail (Detlie et al[8]). In brief, adult IBD patients (> 18 years) diagnosed with ID or IDA (according to European Crohn’s and Colitis Organisation guidelines)[2] were recruited at two separate study sites in the southeast region of Norway and treated with either FCM or FDI.\nEligible patients were prescribed 1000 mg of high-dose intravenous iron, FCM (50 mg/mL) or iron derisomaltose (100 mg/mL), administered as a single dose. Patients who had received high-dose intravenous iron treatment or a packed red blood cell transfusion within 3 mo of study entry, or for whom high-dose intravenous iron treatment was contraindicated, were not included in the study.\nEnrolment continued until at least 50 consecutive patients with complete adherence to the study protocol were recruited at each site (a total of more than 100 patients) (Supplementary Figure 1). The enrolment period was followed by a prospective observation period, which lasted ≤ 7 wk for each patient and included three study visits.\nStudy inclusion was performed at baseline, at which time intravenous iron treatment was administered. Patients attended the clinic at week 2 (10-15 d) and at week 6 (5-7 wk) following intravenous iron treatment. Each patient could receive only one infusion within an approximate 2-mo period after consenting to study participation.", "Blood analysis at each study visit included ionised calcium, creatinine, phosphate, parathyroid hormone (PTH) and vitamin D (25-hydroxyvitamin D).\nBlood samples were also frozen and sent to Medizinische Universität Innsbruck, Universitätsklinik für Innere Medizin I, for analysis of 1,25-dihydroxyvitamin D, intact and C-terminal FGF23. The Kainos FGF-23 ELISA Kit was used for the FGF23 analysis. The assay for intact FGF23 measures only full-length peptide, whereas the assay for C-terminal FGF23 measures full-length peptide and the C-terminal fragments thereby representing total FGF23.\nSpot urine samples were collected at each study visit and analysed for urine phosphate and urine creatinine. A calculation of the fractional excretion of phosphate rate (FEPO4) was then performed using the formula, FEPO4 = (urine phosphate × plasma creatinine × 100)/(plasma phosphate × urine creatinine). Oslo University Hospital Ullevål used the Roche analysis method (Roche/Hitachi Cobas® C systems PHOS2 and CREP2) while Akershus University Hospital used the Vitros analysis (VITROS® MicroSlide Assay 5.1 FS Diluent Pack 3). The slight sensitivity difference between the two analytical methods was minimized by recalculating FEPO4 using the above-mentioned formula.\nSymptoms that might be related to hypophosphatemia were assessed at each of the three study visits using the MicroRPMTM (CareFusion) test to determine respiratory muscle function by measuring maximum inspiratory and maximum expiratory pressure, a health-related QoL questionnaire (36-item short form health survey, SF-36), and a visual analogue scale (VAS).\nFor the MicroRPMTM respiratory function test, patients were asked to inhale and exhale as hard as possible. The test was repeated three times at every visit, and the best result of the three attempts was registered.\nThe SF-36 is a generic, self-administered questionnaire containing 36 items[20]. The items are divided into eight multi-item scales that reflect general health, physical functioning, role limitations due to physical problems, bodily pain, vitality, mental health, social functioning, and role limitations due to emotional problems. Each scale is transformed into a 1-100 scale, where a lower score represents more disability. The processing of raw SF-36 data into results was executed according to the SF-36 scoring algorithms[21].\nThe VAS is a 10 cm line on which the patient is asked to place a vertical mark to indicate the level of intensity of a symptom that best fits his or her experience. Scores range from 0-100 (mm) where a higher score represents greater symptom intensity. The VAS was used to assess general weakness, fatigue, joint pain, joint stiffness, muscle pain, bone and skeletal pain, and difficulties performing daily activities.\nAll demographic information was collected from patients’ medical records and was entered into an electronic case report form.\nThe study was completed when all enrolled patients had received intravenous iron administration, had attended all three study visits, and had fulfilled the requirements of the study protocol.", "Serum phosphate, PTH, 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D, ionised calcium, creatinine, intact and C-terminal FGF23, and FEPO4 (urine phosphate and urine creatinine) were measured in order to assess possible mechanisms behind the development of hypophosphatemia after intravenous iron treatment. Results from the FCM treatment group were compared with results from the FDI treatment group.\nHypophosphatemia was defined as a serum phosphate level < 0.8 mmol/L (< 2.5 mg/dL). The clinical impact of hypophosphatemia was evaluated at week 2 and week 6 using the respiratory muscle function test, SF-36, and the VAS score. In relation to the assessment of clinical impact, the hypophosphatemia group was defined as patients experiencing hypophosphatemia at both week 2 and week 6. Results for patients with hypophosphatemia were compared with results for patients without hypophosphatemia, independent of treatment group.", "This study was designed to achieve 80% power to detect a difference in the primary outcome, which was the incidence of hypophosphatemia (previously described by Detlie et al[8]). Hence, the MicroRPMTM respiratory test, SF-36, and VAS scores were not used to justify sample size.\nData are presented descriptively, as mean with SD or 95% confidence intervals for continuous variables, and as the number of exposed patients (with proportions) for categorical variables. Hypothesis tests for differences in change between treatment groups, change from baseline, and groups with or without hypophosphatemia, were conducted using paired t-tests. All analyses were performed in R. A P value of < 0.05 was considered significant.", "The study protocol was approved by the relevant local regulatory and ethical committees and adhered to the applicable laws on data protection. A study registration application was sent to the EudraCT system with the application No. 2016-003476-41, but the application was deemed unnecessary since there were no indications of a medical intervention study.\nAll patients gave informed consent before inclusion into the study, and the study was performed in accordance with the principles for post-authorisation safety studies, according to Good Clinical Practice guidelines.\nAll biological material obtained from patients was destroyed after analysis, as were the frozen blood samples sent to the Medical University of Innsbruck.\nStudy nurses were blinded to the results of laboratory findings but, for safety reasons, the primary investigator at each study centre was not blinded.", "Of the 130 patients screened for this study, 106 patients (52 patients at Oslo University Hospital Ullevål and 54 patients at Akershus University Hospital) were included in the analyses. Demographic and clinical characteristics of the patients have previously been described[8].\nData for serum phosphate, FEPO4, intact and C-terminal FGF23, 25-hydroxyvitamin D, 1,25-dihydroxyvitamin D, ionised calcium, and PTH at baseline and at each study visit are shown in Table 1. A sub-analysis of the same data, stratified according to hypophosphatemia status (with/without) at week 2 and at week 6, is shown in Table 2.\nDescriptive data for laboratory parameters at baseline, at week 2 and at week 6\nNormal phosphate levels: > 0.8 mmol/L = > 2.48 mg/dL; Mild hypophosphatemia 0.79-0.6 mmol/L = 2.44-1.86 mg/dL; Moderate hypophosphatemia 0.59-0.32 mmol/L = 1.83-0.99 mg/dL; Severe hypophosphatemia < 0.32 mmol/L = < 0.99 mg/dL. CI: Confidence interval; FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone; N/A: Not applicable.\nLaboratory parameters for patients stratified by hypophosphatemia status (with/without) at week 2 and at week 6\nNormal phosphate levels: > 0.8 mmol/L = > 2.48 mg/dL; Mild hypophosphatemia 0.79-0.6 mmol/L = 2.44-1.86 mg/dL; Moderate hypophosphatemia 0.59-0.32 mmol/L = 1.83-0.99 mg/dL; Severe hypophosphatemia < 0.32 mmol/L = < 0.99 mg/dL. FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone; N/A: Not applicable.\nSerum phosphate and urinary excretion of phosphate As previously described, following treatment with FCM, hypophosphatemia was present in 72.5% (37/51) of patients at week 2, and in 21.6% (11/51) of patients at week 6. In comparison, in the FDI treatment group, 11.3% (6/53) of patients had hypophosphatemia at week 2, and 3.7% (2/54) at week 6. There were no new incidences of hypophosphatemia at week 6. The difference in incidence was highly significant at both week 2 and 6 (P < 0.001 and P < 0.013, respectively)[8]. These findings are consistent with the mean urine FEPO4 that was significantly (P = 0.004) higher at week 2 in the FCM treatment group compared with the FDI treatment group, and still elevated (though declining) at week 6 in the FCM group (Table 1 and Figure 1A). In the sub-analysis, the FDI-treated patients with hypophosphatemia (n = 6) had numerically increased FEPO4 (Table 2).\n\nmean ± SD change from baseline in laboratory parameters in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Fractional excretion of phosphate; B: Intact fibroblast growth factor 23; C: C-terminal fibroblast growth factor 23; D: 25-Hydroxyvitamin D; E: 1,25-Hydroxyvitamin D; F: Ionised calcium; G: Parathyroid hormone. FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone.\nPatients in both treatment groups without hypophosphatemia at week 2 also experienced an increase in FEPO4 at week 2 compared to baseline values, but urinary phosphate excretion declined again at week 6 in these patients.\nAs previously described, following treatment with FCM, hypophosphatemia was present in 72.5% (37/51) of patients at week 2, and in 21.6% (11/51) of patients at week 6. In comparison, in the FDI treatment group, 11.3% (6/53) of patients had hypophosphatemia at week 2, and 3.7% (2/54) at week 6. There were no new incidences of hypophosphatemia at week 6. The difference in incidence was highly significant at both week 2 and 6 (P < 0.001 and P < 0.013, respectively)[8]. These findings are consistent with the mean urine FEPO4 that was significantly (P = 0.004) higher at week 2 in the FCM treatment group compared with the FDI treatment group, and still elevated (though declining) at week 6 in the FCM group (Table 1 and Figure 1A). In the sub-analysis, the FDI-treated patients with hypophosphatemia (n = 6) had numerically increased FEPO4 (Table 2).\n\nmean ± SD change from baseline in laboratory parameters in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Fractional excretion of phosphate; B: Intact fibroblast growth factor 23; C: C-terminal fibroblast growth factor 23; D: 25-Hydroxyvitamin D; E: 1,25-Hydroxyvitamin D; F: Ionised calcium; G: Parathyroid hormone. FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone.\nPatients in both treatment groups without hypophosphatemia at week 2 also experienced an increase in FEPO4 at week 2 compared to baseline values, but urinary phosphate excretion declined again at week 6 in these patients.\nFGF23 There was a significant (P < 0.001) increase in intact FGF23 from baseline to week 2 after infusion of FCM, compared with the FDI treatment group (Table 1 and Figure 1B). At week 6, intact FGF23 values in the FCM treatment group had returned close to baseline. In comparison, after FDI treatment no such increases were found (Figure 1B). At baseline, the serum concentration of C-terminal FGF23 was higher in the FDI treatment group than in the FCM treatment group (Table 1), and declined after FDI infusion (Figure 1C). This high value at baseline was not seen in the FCM treatment group (Table 1), which is probably compatible with the less severe ID/IDA seen in the FCM group[8].\nIn the sub-analysis, for the FCM-treated patients with hypophosphatemia, intact FGF23 was significantly increased compared with FCM-treated patients without hypophosphatemia at week 2 and at week 6 (Table 2). In the FCM-treated patients who had normal phosphate, intact FGF23 was not increased at week 2 or week 6.\nFor FDI-treated patients, the sub-analysis showed that there was no significant difference in mean intact FGF23 Levels between patients with/without hypophosphatemia at week 2 or at week 6 (Table 2). At week 2, only one FDI-treated patient with hypophosphatemia had significantly increased intact FGF23; the other five patients with hypophosphatemia had minimal change in their intact FGF23 values.\nThere was a significant (P < 0.001) increase in intact FGF23 from baseline to week 2 after infusion of FCM, compared with the FDI treatment group (Table 1 and Figure 1B). At week 6, intact FGF23 values in the FCM treatment group had returned close to baseline. In comparison, after FDI treatment no such increases were found (Figure 1B). At baseline, the serum concentration of C-terminal FGF23 was higher in the FDI treatment group than in the FCM treatment group (Table 1), and declined after FDI infusion (Figure 1C). This high value at baseline was not seen in the FCM treatment group (Table 1), which is probably compatible with the less severe ID/IDA seen in the FCM group[8].\nIn the sub-analysis, for the FCM-treated patients with hypophosphatemia, intact FGF23 was significantly increased compared with FCM-treated patients without hypophosphatemia at week 2 and at week 6 (Table 2). In the FCM-treated patients who had normal phosphate, intact FGF23 was not increased at week 2 or week 6.\nFor FDI-treated patients, the sub-analysis showed that there was no significant difference in mean intact FGF23 Levels between patients with/without hypophosphatemia at week 2 or at week 6 (Table 2). At week 2, only one FDI-treated patient with hypophosphatemia had significantly increased intact FGF23; the other five patients with hypophosphatemia had minimal change in their intact FGF23 values.\nVitamin D There were no significant differences between the treatment groups in the concentration of 25-hydroxyvitamin D throughout the study period (Figure 1D). However, the sub-analysis showed that 25-hydroxyvitamin D concentrations were lower at week 6 in the two FDI-treated patients with hypophosphatemia when compared with baseline concentrations within the same group (Tables 1 and 2). At week 2, 1,25-dihydroxyvitamin D concentrations were significantly lower in patients who received FCM compared with patients who received FDI (Table 1). In the FCM-treatment group, the mean concentration of 1,25-dihydroxyvitamin D returned to baseline at week 6 (Table 1 and Figure 1E). However, the sub-analysis revealed that, for the FCM-treated patients with hypophosphatemia, low 1,25-dihydroxyvitamin D levels persisted at week 6 (Table 2). In the subgroups of patients without hypophosphatemia, 1,25-dihydroxyvitamin D levels were relatively unchanged.\nIn our cohort, we identified 36 patients (34.0%) with vitamin D deficiency (25-hydroxyvitamin D < 50 nmol/L) at baseline; 10 of these patients had severe vitamin D deficiency (25-hydroxyvitamin D < 30 nmol/L). The distribution of these patients was equal in the two treatment groups, as well as equally distributed across disease states – ulcerative colitis and Crohn’s disease. Moreover, we found no association between low levels of vitamin D and development of hypophosphatemia.\nThere were no significant differences between the treatment groups in the concentration of 25-hydroxyvitamin D throughout the study period (Figure 1D). However, the sub-analysis showed that 25-hydroxyvitamin D concentrations were lower at week 6 in the two FDI-treated patients with hypophosphatemia when compared with baseline concentrations within the same group (Tables 1 and 2). At week 2, 1,25-dihydroxyvitamin D concentrations were significantly lower in patients who received FCM compared with patients who received FDI (Table 1). In the FCM-treatment group, the mean concentration of 1,25-dihydroxyvitamin D returned to baseline at week 6 (Table 1 and Figure 1E). However, the sub-analysis revealed that, for the FCM-treated patients with hypophosphatemia, low 1,25-dihydroxyvitamin D levels persisted at week 6 (Table 2). In the subgroups of patients without hypophosphatemia, 1,25-dihydroxyvitamin D levels were relatively unchanged.\nIn our cohort, we identified 36 patients (34.0%) with vitamin D deficiency (25-hydroxyvitamin D < 50 nmol/L) at baseline; 10 of these patients had severe vitamin D deficiency (25-hydroxyvitamin D < 30 nmol/L). The distribution of these patients was equal in the two treatment groups, as well as equally distributed across disease states – ulcerative colitis and Crohn’s disease. Moreover, we found no association between low levels of vitamin D and development of hypophosphatemia.\nCalcium and PTH Ionised calcium values dropped significantly from baseline to week 2 in the FCM treatment group compared with the FDI treatment group (P < 0.012) but stayed within normal range. The mean values in the FCM group had increased by week 6, but the between-group difference was still significant (P < 0.044). Calcium values remained stable throughout the study in the FDI treatment group (Figure 1F), and in the subgroup of FCM-treated patients who did not develop hypophosphatemia. The sub-analysis showed that there was a numerically lower level of ionised calcium in the FDI-treated patients with hypophosphatemia than in the FDI-treated patients without hypophosphatemia (Table 2).\nPTH values were elevated (> 7 pmol/L) in 28 patients (26.4%) at baseline; the distribution was similar between treatment groups. PTH concentrations were similar between treatment groups at baseline, and no significant between-group differences were observed in mean PTH concentrations at week 2, and at week 6 (Table 1). PTH values increased in both treatment groups at week 2 and decreased again at week 6 (Figure 1G). The sub-analysis indicated that the increase in PTH in both treatment groups was mainly driven by the patients who developed hypophosphatemia, with significant differences at week 2 and week 6 for the FCM-treated patients with hypophosphatemia compared to FCM-treated patients without hypophosphatemia (Table 2).\nIonised calcium values dropped significantly from baseline to week 2 in the FCM treatment group compared with the FDI treatment group (P < 0.012) but stayed within normal range. The mean values in the FCM group had increased by week 6, but the between-group difference was still significant (P < 0.044). Calcium values remained stable throughout the study in the FDI treatment group (Figure 1F), and in the subgroup of FCM-treated patients who did not develop hypophosphatemia. The sub-analysis showed that there was a numerically lower level of ionised calcium in the FDI-treated patients with hypophosphatemia than in the FDI-treated patients without hypophosphatemia (Table 2).\nPTH values were elevated (> 7 pmol/L) in 28 patients (26.4%) at baseline; the distribution was similar between treatment groups. PTH concentrations were similar between treatment groups at baseline, and no significant between-group differences were observed in mean PTH concentrations at week 2, and at week 6 (Table 1). PTH values increased in both treatment groups at week 2 and decreased again at week 6 (Figure 1G). The sub-analysis indicated that the increase in PTH in both treatment groups was mainly driven by the patients who developed hypophosphatemia, with significant differences at week 2 and week 6 for the FCM-treated patients with hypophosphatemia compared to FCM-treated patients without hypophosphatemia (Table 2).\nRespiratory muscle function tests In the comparison of patients who developed hypophosphatemia vs those who did not develop hypophosphatemia, independent of treatment group, no significant differences were observed in the respiratory muscle function test results. The differences between patients with hypophosphatemia and those with normal phosphate values were minimal and the standard deviation was wide in both groups (Figure 2).\n\nmean ± SD changes from baseline in respiratory pressure in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Inspiratory pressure; B: Expiratory pressure.\nIn the comparison of patients who developed hypophosphatemia vs those who did not develop hypophosphatemia, independent of treatment group, no significant differences were observed in the respiratory muscle function test results. The differences between patients with hypophosphatemia and those with normal phosphate values were minimal and the standard deviation was wide in both groups (Figure 2).\n\nmean ± SD changes from baseline in respiratory pressure in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Inspiratory pressure; B: Expiratory pressure.\nSF-36 The results of the SF-36 QoL assessment are presented in Table 3. Overall, there were no significant differences between patient groups with or without hypophosphatemia at baseline and at any time point during the study. The mean scores at baseline in both treatment groups were generally low.\nDescriptive 36-item short form health survey scores for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; SF-36: 36-Item short form health survey.\nThe results of the SF-36 QoL assessment are presented in Table 3. Overall, there were no significant differences between patient groups with or without hypophosphatemia at baseline and at any time point during the study. The mean scores at baseline in both treatment groups were generally low.\nDescriptive 36-item short form health survey scores for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; SF-36: 36-Item short form health survey.\nVAS scores There were no significant differences in VAS scores between the groups of patients with/without hypophosphatemia at week 2 and at week 6 (Table 4). Overall, VAS scores were elevated at baseline. However, the group of patients who developed hypophosphatemia had lower VAS scores at baseline for the items joint pain, muscle pain, and bone and skeletal pain, compared to the group of patients who did not develop hypophosphatemia; between-group differences were not significant for these items. There was, however, a significant between-group difference (P < 0.001) at baseline for the VAS joint stiffness item score, with lower values in the group of patients who developed hypophosphatemia.\nDescriptive visual analogue scale score for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; VAS: Visual analogue scale.\nThere were no significant differences in VAS scores between the groups of patients with/without hypophosphatemia at week 2 and at week 6 (Table 4). Overall, VAS scores were elevated at baseline. However, the group of patients who developed hypophosphatemia had lower VAS scores at baseline for the items joint pain, muscle pain, and bone and skeletal pain, compared to the group of patients who did not develop hypophosphatemia; between-group differences were not significant for these items. There was, however, a significant between-group difference (P < 0.001) at baseline for the VAS joint stiffness item score, with lower values in the group of patients who developed hypophosphatemia.\nDescriptive visual analogue scale score for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; VAS: Visual analogue scale.", "As previously described, following treatment with FCM, hypophosphatemia was present in 72.5% (37/51) of patients at week 2, and in 21.6% (11/51) of patients at week 6. In comparison, in the FDI treatment group, 11.3% (6/53) of patients had hypophosphatemia at week 2, and 3.7% (2/54) at week 6. There were no new incidences of hypophosphatemia at week 6. The difference in incidence was highly significant at both week 2 and 6 (P < 0.001 and P < 0.013, respectively)[8]. These findings are consistent with the mean urine FEPO4 that was significantly (P = 0.004) higher at week 2 in the FCM treatment group compared with the FDI treatment group, and still elevated (though declining) at week 6 in the FCM group (Table 1 and Figure 1A). In the sub-analysis, the FDI-treated patients with hypophosphatemia (n = 6) had numerically increased FEPO4 (Table 2).\n\nmean ± SD change from baseline in laboratory parameters in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Fractional excretion of phosphate; B: Intact fibroblast growth factor 23; C: C-terminal fibroblast growth factor 23; D: 25-Hydroxyvitamin D; E: 1,25-Hydroxyvitamin D; F: Ionised calcium; G: Parathyroid hormone. FCM: Ferric carboxymaltose; FDI: Ferric derisomaltose; FEPO4: Fractional excretion of phosphate; FGF23: Fibroblast growth factor 23; PTH: Parathyroid hormone.\nPatients in both treatment groups without hypophosphatemia at week 2 also experienced an increase in FEPO4 at week 2 compared to baseline values, but urinary phosphate excretion declined again at week 6 in these patients.", "There was a significant (P < 0.001) increase in intact FGF23 from baseline to week 2 after infusion of FCM, compared with the FDI treatment group (Table 1 and Figure 1B). At week 6, intact FGF23 values in the FCM treatment group had returned close to baseline. In comparison, after FDI treatment no such increases were found (Figure 1B). At baseline, the serum concentration of C-terminal FGF23 was higher in the FDI treatment group than in the FCM treatment group (Table 1), and declined after FDI infusion (Figure 1C). This high value at baseline was not seen in the FCM treatment group (Table 1), which is probably compatible with the less severe ID/IDA seen in the FCM group[8].\nIn the sub-analysis, for the FCM-treated patients with hypophosphatemia, intact FGF23 was significantly increased compared with FCM-treated patients without hypophosphatemia at week 2 and at week 6 (Table 2). In the FCM-treated patients who had normal phosphate, intact FGF23 was not increased at week 2 or week 6.\nFor FDI-treated patients, the sub-analysis showed that there was no significant difference in mean intact FGF23 Levels between patients with/without hypophosphatemia at week 2 or at week 6 (Table 2). At week 2, only one FDI-treated patient with hypophosphatemia had significantly increased intact FGF23; the other five patients with hypophosphatemia had minimal change in their intact FGF23 values.", "There were no significant differences between the treatment groups in the concentration of 25-hydroxyvitamin D throughout the study period (Figure 1D). However, the sub-analysis showed that 25-hydroxyvitamin D concentrations were lower at week 6 in the two FDI-treated patients with hypophosphatemia when compared with baseline concentrations within the same group (Tables 1 and 2). At week 2, 1,25-dihydroxyvitamin D concentrations were significantly lower in patients who received FCM compared with patients who received FDI (Table 1). In the FCM-treatment group, the mean concentration of 1,25-dihydroxyvitamin D returned to baseline at week 6 (Table 1 and Figure 1E). However, the sub-analysis revealed that, for the FCM-treated patients with hypophosphatemia, low 1,25-dihydroxyvitamin D levels persisted at week 6 (Table 2). In the subgroups of patients without hypophosphatemia, 1,25-dihydroxyvitamin D levels were relatively unchanged.\nIn our cohort, we identified 36 patients (34.0%) with vitamin D deficiency (25-hydroxyvitamin D < 50 nmol/L) at baseline; 10 of these patients had severe vitamin D deficiency (25-hydroxyvitamin D < 30 nmol/L). The distribution of these patients was equal in the two treatment groups, as well as equally distributed across disease states – ulcerative colitis and Crohn’s disease. Moreover, we found no association between low levels of vitamin D and development of hypophosphatemia.", "Ionised calcium values dropped significantly from baseline to week 2 in the FCM treatment group compared with the FDI treatment group (P < 0.012) but stayed within normal range. The mean values in the FCM group had increased by week 6, but the between-group difference was still significant (P < 0.044). Calcium values remained stable throughout the study in the FDI treatment group (Figure 1F), and in the subgroup of FCM-treated patients who did not develop hypophosphatemia. The sub-analysis showed that there was a numerically lower level of ionised calcium in the FDI-treated patients with hypophosphatemia than in the FDI-treated patients without hypophosphatemia (Table 2).\nPTH values were elevated (> 7 pmol/L) in 28 patients (26.4%) at baseline; the distribution was similar between treatment groups. PTH concentrations were similar between treatment groups at baseline, and no significant between-group differences were observed in mean PTH concentrations at week 2, and at week 6 (Table 1). PTH values increased in both treatment groups at week 2 and decreased again at week 6 (Figure 1G). The sub-analysis indicated that the increase in PTH in both treatment groups was mainly driven by the patients who developed hypophosphatemia, with significant differences at week 2 and week 6 for the FCM-treated patients with hypophosphatemia compared to FCM-treated patients without hypophosphatemia (Table 2).", "In the comparison of patients who developed hypophosphatemia vs those who did not develop hypophosphatemia, independent of treatment group, no significant differences were observed in the respiratory muscle function test results. The differences between patients with hypophosphatemia and those with normal phosphate values were minimal and the standard deviation was wide in both groups (Figure 2).\n\nmean ± SD changes from baseline in respiratory pressure in inflammatory bowel disease patients with iron deficiency/iron deficiency anaemia treated with a single 1000 mg intravenous dose of ferric carboxymaltose or ferric derisomaltose. A: Inspiratory pressure; B: Expiratory pressure.", "The results of the SF-36 QoL assessment are presented in Table 3. Overall, there were no significant differences between patient groups with or without hypophosphatemia at baseline and at any time point during the study. The mean scores at baseline in both treatment groups were generally low.\nDescriptive 36-item short form health survey scores for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; SF-36: 36-Item short form health survey.", "There were no significant differences in VAS scores between the groups of patients with/without hypophosphatemia at week 2 and at week 6 (Table 4). Overall, VAS scores were elevated at baseline. However, the group of patients who developed hypophosphatemia had lower VAS scores at baseline for the items joint pain, muscle pain, and bone and skeletal pain, compared to the group of patients who did not develop hypophosphatemia; between-group differences were not significant for these items. There was, however, a significant between-group difference (P < 0.001) at baseline for the VAS joint stiffness item score, with lower values in the group of patients who developed hypophosphatemia.\nDescriptive visual analogue scale score for patient groups with/without hypophosphatemia independent of treatment group\nHypophosphatemia defined as serum phosphate < 0.8 mmol/L (< 2.5 mg/dL). \nDifferences are normal phosphate group minus hypophosphatemia group. CI: Confidence interval; VAS: Visual analogue scale.", "Our study indicates that FGF23 plays an important role in the development of hypophosphatemia in IBD patients treated with FCM. In these patients, a high level of intact FGF23, an increased excretion of phosphate in the urine, a decrease of 1,25-dihydroxyvitamin D and of serum calcium levels, and a slight elevation of PTH, was demonstrated.\nPrevious clinical trials of FCM have shown similar results[9,22]. However, for the most part, these studies have been conducted in healthy and, predominantly, female populations. The role of FGF23 has also been described in earlier publications[23-26]. Regulation of phosphate concentrations in the body seems to be strongly influenced by intact FGF23, which reduces phosphate reabsorption in the proximal tubules in the kidneys and inhibits production of 1,25-dihydroxyvitamin D, probably by inhibiting the activity of the enzyme 25-hydroxyvitamin D-1a-hydroxylase and increased expression of 24-hydroxylase[24,26]. Our findings suggest that FCM could have a direct impact on cleavage of FGF23, resulting in a high level of intact FGF23 and consequent phosphate wasting. This might also explain why baseline phosphate level does not predict the development of mild or severe hypophosphatemia, due to the inappropriate excretion of available phosphate in the urine, following FCM treatment[8]. We also observed a decrease in 1,25-dihydroxyvitamin D (the active vitamin D metabolite), a decrease in ionised calcium, and development of secondary hyperparathyroidism. This might explain why some patients treated with FCM still had hypophosphatemia six weeks after treatment, when the intact FGF23 values had normalized (Table 2) since elevated PTH promotes excretion of phosphate in the urine[9,27,28].\nThe majority of patients in the FCM treatment group developed hypophosphatemia at week 2. The remaining patients did not develop hypophosphatemia and had unchanged levels of intact FGF23. So, there is a clear association between the development of high levels of intact FGF23 and hypophosphatemia. Therefore, it can only be speculated that there might be some individual factors related to the handling of FCM that cause the majority of patients treated with FCM to develop hypophosphatemia, whereas others do not. Neither is it known if any individual patient would develop hypophosphatemia on subsequent administrations of FCM, or if the effect of FCM treatment on phosphate wasting is indiscriminate. Perhaps some patients are protected against the influence of FCM on the enzyme responsible for FGF23 protein cleavage. From our results, we postulate that the mechanism of FCM-induced hypophosphatemia is not related to IBD; instead, it appears to be independently connected to the drug itself.\nA few patients who received treatment with FDI also developed hypophosphatemia but, unlike those receiving FCM, these patients did not on average have significantly elevated intact FGF23 Levels when assessed at week 2, which would suggest a different underlying mechanism. A transient increase in intact FGF23 during the first 2 wk in patients experiencing hypophosphatemia cannot be ruled out, as data were not collected during this time period. A numerical increase in PTH was observed at week 2 along with decreased ionised calcium, and decreased 25-hydroxyvitamin D at week 6. It is not clear whether these observations are the result of a transient increase of intact FGF23 during the first 2 wk, or solely a physiological response to a rapid correction of ID, or simply an artefact due to the low numbers of FDI patients who developed hypophosphatemia. The general physiological response of mineral metabolism markers to rapid ID correction is not fully elucidated and is an area of further research.\nAn important observation is that 34% of the study population was vitamin D deficient at baseline with 25-hydroxyvitamin D values < 50 nmol/L and, perhaps more interestingly, 24% of the patients had PTH values compatible with secondary hyperparathyroidism. These findings were equally distributed between the two treatment groups. This disturbance in vitamin D metabolism is unlikely to be a consequence of previous iron infusions since no patients received high-dose intravenous iron treatment during the 6 mo prior to inclusion in this study. The high prevalence of vitamin D deficiency at baseline is in agreement with previous studies of patients with IBD[29]. However, it is important to note that, in our study, many of the samples were taken during the winter months when sun exposure is reduced in Norway, and individuals could therefore be expected to be somewhat vitamin D deficient during this time. Nevertheless, this finding is important since both hypophosphatemia and vitamin D deficiency can contribute to the development of metabolic bone disease, including osteomalacia.\nGuidelines regarding hypophosphatemia diagnosis, treatment, and follow-up are available, but the possible risk or incidence rate of developing hypophosphatemia with symptoms or complications are rarely mentioned[30]. A risk of developing respiratory failure, rhabdomyolysis, and left ventricular failure due to severe hypophosphatemia has been reported in case series[10]. More recent data also predict an increased risk of developing osteomalacia, especially in long-standing hypophosphatemia[14,15]. What is less well known is the number of patients developing more subtle, but identifiable, symptoms related to hypophosphatemia that are experienced as troublesome and might influence QoL.\nWith respect to the clinical impact of hypophosphatemia, measuring forced inspiratory and expiratory respiratory pressure can be used as a proxy to assess the physical effect of hypophosphatemia on skeletal and proximal muscles. There are no specific questionnaires available to evaluate the clinical impact of hypophosphatemia. The SF-36 is, however, one of the most commonly applied QoL questionnaires used world-wide in health surveys. Additionally, the VAS score can be used as a general assessment of impact of symptoms, such as fatigue, general weakness, bone and skeletal pain, and joint and muscle conditions. In our study, these three methods were applied to assess clinical impact in patients who developed hypophosphatemia compared to those who did not develop hypophosphatemia. All three methods failed to demonstrate significant differences in clinical impact following one administration of high-dose intravenous iron in this short-term study.\nWe hypothesize several reasons that might explain these results. In addition to the fact that a type II error cannot be excluded, it can be speculated that the positive effect of the correction of ID or IDA plays a more important role than any short-term negative clinical impact of hypophosphatemia and, hence, the effects of hypophosphatemia would be difficult to discern in our study. Another challenge is that IBD, ID, IDA and hypophosphatemia are associated with similar symptoms and, possibly, similar impacts on daily life. Indeed, assessing the specific impact of hypophosphatemia with questionnaires would, therefore, prove difficult. Since the SF-36 and the VAS questionnaires are not disease- or population-specific, there may be uncertainty surrounding the reliability of the results. Additionally, the patient cohort had more than one dynamic medical condition, with overlapping symptoms, and patients were observed in a longitudinal manner. Certainly, it would be almost impossible to determine which disease state or co-morbidity is reflecting improvement or worsening of clinical status.\nThe already affected baseline recordings in SF-36 and the VAS score should not go unnoticed. These findings mirror previous studies of IBD populations[31], and reflect the reduced QoL and the intensity of symptoms that these patients experience in general. Finally, the fact that we did not detect clinical consequences in patients who developed hypophosphatemia suggests that, in order to detect overt symptoms and complications, the population size needs to be larger than our sample, as one might expect such complications to be relatively rare. Hence, this needs to be taken into account when considering the expectation of finding significant changes in the clinical outcomes in this study.", "In summary, our study has implicated the small peptide hormone FGF23 in the development of hypophosphatemia in IBD patients treated with FCM. An increase in intact FGF23 occurs, which probably results in phosphate wasting in the urine. Assessment of symptoms did not exclude, nor did they demonstrate, any short-term clinical impact of hypophosphatemia in IBD patients treated for ID or IDA with high-dose intravenous iron." ]

[ null, "methods", null, null, null, null, null, null, null, null, null, null, null, null, null, null, null ]

[ "Iron deficiency", "Hypophosphatemia", "Inflammatory bowel disease", "Ferric carboxymaltose", "Ferric derisomaltose" ]

[ 520, 290, 568, 164, 133, 151, 3554, 340, 284, 274, 271, 109, 116, 186, 1424, 76 ]

[ "patients", "hypophosphatemia", "week", "treatment", "group", "study", "fcm", "phosphate", "fgf23", "baseline" ]

[ "iron intravenous", "iron treatment administered", "iron intravenous infusion", "disease ibd iron", "ibd iron deficiency" ]

[ "Humans", "Respiration, Artificial", "Glasgow Coma Scale", "Pneumonia, Ventilator-Associated", "Respiratory Insufficiency", "Pulmonary Disease, Chronic Obstructive" ]

[ "2. Materials and methods", "2.1. Subjects", "2.1.1. A flow chart of the study design is shown in Figure 1\n.", "2.2. Methods", "2.2.1. Sequential invasive-noninvasive ventilation.", "2.2.2. Modified GCS score.[13]", "2.2.3. Observational indicators.", "3.1. Comparison of baseline clinical data", "3.2. Comparison of oxygenation indicators", "3.3. Comparison of related medical indicators", "Author contributions" ]

[ "2.1. Subjects The present study was designed as a prospective randomized controlled trial. COPD patients with respiratory failure who underwent endotracheal intubation and IMV between June 2017 and June 2020 were recruited from 4 hospitals in China: Wenling Hospital Affiliated with Wenzhou Medical University, the First Affiliated Hospital of Wenzhou Medical University, Changsha Central Hospital, and the First People’s Hospital of Jingmen. According to previous sample size calculations, the minimum number of patients that needed to be recruited for each group was 76.\nThe inclusion criteria were as follows: Patients receiving IMV for Respiratory failure, COPD diagnosed according to the diagnostic criteria published by the Respiratory Diseases Branch of the Chinese Medical Association in 2007,[11] with partial oxygen pressure < 60 mm Hg (1 mm Hg = 0.133 kPa) on blood gas analysis and no absolute contraindications for NIV[12]; Age ≥ 18 years old.\nThe exclusion criteria were as follows: Acute respiratory failure caused by stroke (acute stage), acute pulmonary embolism, or acute cardiogenic pulmonary edema; Death within 3 days; Active upper gastrointestinal bleeding; Family members terminated treatment; Admission to the ICU in the past 3 months.\n2.1.1. A flow chart of the study design is shown in Figure 1\n. The study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\nThe study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\nThe present study was designed as a prospective randomized controlled trial. COPD patients with respiratory failure who underwent endotracheal intubation and IMV between June 2017 and June 2020 were recruited from 4 hospitals in China: Wenling Hospital Affiliated with Wenzhou Medical University, the First Affiliated Hospital of Wenzhou Medical University, Changsha Central Hospital, and the First People’s Hospital of Jingmen. According to previous sample size calculations, the minimum number of patients that needed to be recruited for each group was 76.\nThe inclusion criteria were as follows: Patients receiving IMV for Respiratory failure, COPD diagnosed according to the diagnostic criteria published by the Respiratory Diseases Branch of the Chinese Medical Association in 2007,[11] with partial oxygen pressure < 60 mm Hg (1 mm Hg = 0.133 kPa) on blood gas analysis and no absolute contraindications for NIV[12]; Age ≥ 18 years old.\nThe exclusion criteria were as follows: Acute respiratory failure caused by stroke (acute stage), acute pulmonary embolism, or acute cardiogenic pulmonary edema; Death within 3 days; Active upper gastrointestinal bleeding; Family members terminated treatment; Admission to the ICU in the past 3 months.\n2.1.1. A flow chart of the study design is shown in Figure 1\n. The study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\nThe study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\n2.2. Methods 2.2.1. Sequential invasive-noninvasive ventilation. The patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\nThe patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\n2.2.2. Modified GCS score.[13] Eye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\nEye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\n2.2.3. Observational indicators. The baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\nThe baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\n2.2.1. Sequential invasive-noninvasive ventilation. The patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\nThe patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\n2.2.2. Modified GCS score.[13] Eye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\nEye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\n2.2.3. Observational indicators. The baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\nThe baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\n2.3. Statistical methods All data were analyzed with SPSS 25.0 (for Windows; IBM Corp). Continuous variables are expressed as the mean ± (SD),categorical variablesare expressed as the ratios (n%). The independent sample t test was used for comparisons between the 2 groups. The paired sample t test was used for comparisons within a group. The chi squared test was used for analysis of the categorical variables. Differences were considered statistically significant when P < .05.\nAll data were analyzed with SPSS 25.0 (for Windows; IBM Corp). Continuous variables are expressed as the mean ± (SD),categorical variablesare expressed as the ratios (n%). The independent sample t test was used for comparisons between the 2 groups. The paired sample t test was used for comparisons within a group. The chi squared test was used for analysis of the categorical variables. Differences were considered statistically significant when P < .05.", "The present study was designed as a prospective randomized controlled trial. COPD patients with respiratory failure who underwent endotracheal intubation and IMV between June 2017 and June 2020 were recruited from 4 hospitals in China: Wenling Hospital Affiliated with Wenzhou Medical University, the First Affiliated Hospital of Wenzhou Medical University, Changsha Central Hospital, and the First People’s Hospital of Jingmen. According to previous sample size calculations, the minimum number of patients that needed to be recruited for each group was 76.\nThe inclusion criteria were as follows: Patients receiving IMV for Respiratory failure, COPD diagnosed according to the diagnostic criteria published by the Respiratory Diseases Branch of the Chinese Medical Association in 2007,[11] with partial oxygen pressure < 60 mm Hg (1 mm Hg = 0.133 kPa) on blood gas analysis and no absolute contraindications for NIV[12]; Age ≥ 18 years old.\nThe exclusion criteria were as follows: Acute respiratory failure caused by stroke (acute stage), acute pulmonary embolism, or acute cardiogenic pulmonary edema; Death within 3 days; Active upper gastrointestinal bleeding; Family members terminated treatment; Admission to the ICU in the past 3 months.\n2.1.1. A flow chart of the study design is shown in Figure 1\n. The study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\nThe study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.", "The study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.", "2.2.1. Sequential invasive-noninvasive ventilation. The patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\nThe patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\n2.2.2. Modified GCS score.[13] Eye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\nEye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\n2.2.3. Observational indicators. The baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\nThe baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.", "The patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.", "Eye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.", "The baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.", "The total number of patients was 296, including 175 men and 121 women. The age range was from 20 to 87 years old, and the average age was 52.3 ± 8.7 years. There was no significant difference in gender, age, baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, body mass index, heart rate, respiratory rate, MBP, OI, PaO2, PaCO2, or blood pH value between groups A and B. Likewise, there were no significant differences with respect to concomitant diseases such as cardiovascular diseases, cerebrovascular diseases, diabetes, and chronic kidney disease between the 2 groups (Table 2).\nComparison of the baseline clinical data between the 2 groups\nAPACHE = acute physiology and chronic health enquiry, BMI = body mass index, GCS = Glasgow coma scale, MBP = mean blood pressure, OI = oxygenation index, PaCO2 = arterial partial pressure of, PaO2 = arterial partial pressure of O2.", "Groups A and B (switching point modified GCS ≥ 15) had similar MBP, OI, PaO2, and PaCO2 before extubation. Three hours after NIV, the MBP remained stable in both groups A and B. The OI and PaO2 decreased slightly whereas the PaCO2 increased slightly in both groups, but the change did not reach statistical significance. There was no significant difference between switching when modified GCS ≥ 13 (group A) and when modified GCS ≥ 15 (group B). The MBP, OI, PaO2, and PaCO2 values at 3 hours after NIV were similar in groups A and B (P = .864, .730, .425, .784, see Table 3).\nComparison of the MBP, OI, PaO2, and PaCO2 between the 2 groups.\nMBP = mean blood pressure, NIV = noninvasive ventilation, OI = oxygenation index, PaCO2 = arterial partial pressure of CO2, PaO2 = arterial partial pressure of O2.", "Only a few patients required re-intubation in both groups (6.1% in group A and 10.1% in group B). Although the re-intubation rate was slightly higher in group B (modified GCS ≥ 15 used as a switching point), the difference did not reach statistical significance (P = .201). However, the length of hospital stay and duration of IMV in group A (modified GCS ≥ 13 used as a switching point) were significantly shorter than those in group B (P = .041, <.001). The duration of IMV was 2.7 days shorter in group A and the length of hospital stay was 6.9 days shorter. The incidence of VAP in group A was only 34% of that seen in group B (P = .001, see Table 4).\nComparison of the related medical indicators between the 2 groups.\nIMV = invasive mechanical ventilation, VAP = ventilator-associated pneumonia.", "Conceptualization: Jin-Bo Zhang, Li-Hong Li, Jin-Qiang Zhu, Shi-Fang Zhou, Ji-Hong Ma, Zhi-Qiang Li, Xiao-Qin Lin.\nData curation: Li-Hong Li, Jin-Qiang Zhu, Shi-Fang Zhou, Zhi-Qiang Li, Xiao-Qin Lin.\nFormal analysis: Jin-Bo Zhang, Ji-Hong Ma, Xiao-Hong Jin.\nFunding acquisition: Xiao-Hong Jin.\nInvestigation: Jin-Bo Zhang, Xiao-Hong Jin.\nProject administration: Jin-Bo Zhang, Zhi-Qiang Li.\nResources: Zhi-Qiang Li.\nSoftware: Li-Hong Li.\nSupervision: Shi-Fang Zhou.\nVisualization: Jin-Qiang Zhu, Zhi-Qiang Li, Xiao-Qin Lin.\nWriting – original draft: Jin-Bo Zhang, Jin-Qiang Zhu, Shi-Fang Zhou, Ji-Hong Ma, Zhi-Qiang Li.\nWriting – review & editing: Jin-Bo Zhang, Li-Hong Li, Shi-Fang Zhou, Xiao-Hong Jin." ]

[ "methods", null, null, "methods", null, null, null, null, null, null, null ]

[ "1. Introduction", "2. Materials and methods", "2.1. Subjects", "2.1.1. A flow chart of the study design is shown in Figure 1\n.", "2.2. Methods", "2.2.1. Sequential invasive-noninvasive ventilation.", "2.2.2. Modified GCS score.[13]", "2.2.3. Observational indicators.", "2.3. Statistical methods", "3. Results", "3.1. Comparison of baseline clinical data", "3.2. Comparison of oxygenation indicators", "3.3. Comparison of related medical indicators", "4. Discussion", "5. Conclusions", "Author contributions" ]

[ "Bronchopulmonary infections are responsible for 80% to 90% of the acute exacerbations of chronic obstructive pulmonary disease (COPD).[1] Severe acute exacerbations can lead to respiratory failure, which requires treatment with invasive mechanical ventilation (IMV). However, long term use of IMV is associated with hazards such as ventilator-associated pneumonia (VAP) and ventilator-associated lung injury, which significantly affect the prognosis of patients.[2,3] noninvasive mechanical ventilation (NIV) is associated with fewer negative effects than IMV. Sequential invasive-noninvasive ventilation is widely used as an effective treatment for COPD complicated with respiratory failure.[4] The key to successful sequential therapy is finding the best switching point from IMV to NIV. In China, the control window for pulmonary infection (PIC) is often used as the switching point for sequential invasive-noninvasive ventilation with good results. However, PIC diagnosis relies on chest X-ray findings, which often lag behind clinical manifestations and are too focused on infectious factors while ignoring other causes. Internationally, it is common to perform a weaning test after 48 hours of tracheal intubation and IMV and if no signs of spontaneous breathing are observed, NIV is started immediately after removal of the tracheal tube. The problem with using 48 hours after IMV as the switching point is that individual differences and the characteristics of the noninvasive ventilator are ignored. Some clinical trials have found that in COPD patients with respiratory failure, NIV is beneficial and harmless if the patient has good consciousness and cooperation.[5] The modified Glasgow coma scale (GCS) can objectively and quantitatively reflect the overall clinical status of COPD patients with respiratory failure. Some studies have reported positive effects from using the modified GCS to guide sequential invasive-noninvasive ventilation and a score ≥ 15 as the switching point.[6] The modified GCS is widely used clinically in China and has been shown to be an objective and quantitative measure of the clinical condition.[7,8] Our previous studies showed that using a modified GCS score ≥ 13 as the switching point resulted in more benefits.[9,10] The aim of the present study was to compare the clinical results from using a modified GCS score ≥ 13 or 15 points as the switching point for sequential invasive-noninvasive ventilation in COPD patients with respiratory failure.", "2.1. Subjects The present study was designed as a prospective randomized controlled trial. COPD patients with respiratory failure who underwent endotracheal intubation and IMV between June 2017 and June 2020 were recruited from 4 hospitals in China: Wenling Hospital Affiliated with Wenzhou Medical University, the First Affiliated Hospital of Wenzhou Medical University, Changsha Central Hospital, and the First People’s Hospital of Jingmen. According to previous sample size calculations, the minimum number of patients that needed to be recruited for each group was 76.\nThe inclusion criteria were as follows: Patients receiving IMV for Respiratory failure, COPD diagnosed according to the diagnostic criteria published by the Respiratory Diseases Branch of the Chinese Medical Association in 2007,[11] with partial oxygen pressure < 60 mm Hg (1 mm Hg = 0.133 kPa) on blood gas analysis and no absolute contraindications for NIV[12]; Age ≥ 18 years old.\nThe exclusion criteria were as follows: Acute respiratory failure caused by stroke (acute stage), acute pulmonary embolism, or acute cardiogenic pulmonary edema; Death within 3 days; Active upper gastrointestinal bleeding; Family members terminated treatment; Admission to the ICU in the past 3 months.\n2.1.1. A flow chart of the study design is shown in Figure 1\n. The study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\nThe study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\nThe present study was designed as a prospective randomized controlled trial. COPD patients with respiratory failure who underwent endotracheal intubation and IMV between June 2017 and June 2020 were recruited from 4 hospitals in China: Wenling Hospital Affiliated with Wenzhou Medical University, the First Affiliated Hospital of Wenzhou Medical University, Changsha Central Hospital, and the First People’s Hospital of Jingmen. According to previous sample size calculations, the minimum number of patients that needed to be recruited for each group was 76.\nThe inclusion criteria were as follows: Patients receiving IMV for Respiratory failure, COPD diagnosed according to the diagnostic criteria published by the Respiratory Diseases Branch of the Chinese Medical Association in 2007,[11] with partial oxygen pressure < 60 mm Hg (1 mm Hg = 0.133 kPa) on blood gas analysis and no absolute contraindications for NIV[12]; Age ≥ 18 years old.\nThe exclusion criteria were as follows: Acute respiratory failure caused by stroke (acute stage), acute pulmonary embolism, or acute cardiogenic pulmonary edema; Death within 3 days; Active upper gastrointestinal bleeding; Family members terminated treatment; Admission to the ICU in the past 3 months.\n2.1.1. A flow chart of the study design is shown in Figure 1\n. The study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\nThe study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\n2.2. Methods 2.2.1. Sequential invasive-noninvasive ventilation. The patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\nThe patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\n2.2.2. Modified GCS score.[13] Eye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\nEye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\n2.2.3. Observational indicators. The baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\nThe baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\n2.2.1. Sequential invasive-noninvasive ventilation. The patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\nThe patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\n2.2.2. Modified GCS score.[13] Eye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\nEye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\n2.2.3. Observational indicators. The baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\nThe baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\n2.3. Statistical methods All data were analyzed with SPSS 25.0 (for Windows; IBM Corp). Continuous variables are expressed as the mean ± (SD),categorical variablesare expressed as the ratios (n%). The independent sample t test was used for comparisons between the 2 groups. The paired sample t test was used for comparisons within a group. The chi squared test was used for analysis of the categorical variables. Differences were considered statistically significant when P < .05.\nAll data were analyzed with SPSS 25.0 (for Windows; IBM Corp). Continuous variables are expressed as the mean ± (SD),categorical variablesare expressed as the ratios (n%). The independent sample t test was used for comparisons between the 2 groups. The paired sample t test was used for comparisons within a group. The chi squared test was used for analysis of the categorical variables. Differences were considered statistically significant when P < .05.", "The present study was designed as a prospective randomized controlled trial. COPD patients with respiratory failure who underwent endotracheal intubation and IMV between June 2017 and June 2020 were recruited from 4 hospitals in China: Wenling Hospital Affiliated with Wenzhou Medical University, the First Affiliated Hospital of Wenzhou Medical University, Changsha Central Hospital, and the First People’s Hospital of Jingmen. According to previous sample size calculations, the minimum number of patients that needed to be recruited for each group was 76.\nThe inclusion criteria were as follows: Patients receiving IMV for Respiratory failure, COPD diagnosed according to the diagnostic criteria published by the Respiratory Diseases Branch of the Chinese Medical Association in 2007,[11] with partial oxygen pressure < 60 mm Hg (1 mm Hg = 0.133 kPa) on blood gas analysis and no absolute contraindications for NIV[12]; Age ≥ 18 years old.\nThe exclusion criteria were as follows: Acute respiratory failure caused by stroke (acute stage), acute pulmonary embolism, or acute cardiogenic pulmonary edema; Death within 3 days; Active upper gastrointestinal bleeding; Family members terminated treatment; Admission to the ICU in the past 3 months.\n2.1.1. A flow chart of the study design is shown in Figure 1\n. The study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.\nThe study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.", "The study protocol was approved by the Ethics Committee of the Wenling Hospital Affiliated with the Wenzhou Medical University. All enrolled patients signed the informed consent form.\nFlowchart of the study design.", "2.2.1. Sequential invasive-noninvasive ventilation. The patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\nThe patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.\n2.2.2. Modified GCS score.[13] Eye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\nEye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.\n2.2.3. Observational indicators. The baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.\nThe baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.", "The patients were randomly divided into 2 groups (A and B) with equal sizes using the random number table method. All patients were treated with broad-spectrum antibiotics, spasmolytic, anti-asthma medication, glucocorticoids, expectorant, nutritional support, analgesia and sedation, homeostasis maintenance therapy, etc. Ventilator parameters were adjusted according to the results of arterial blood gas analysis and disease progression. The ventilation mode was synchronous intermittent mandatory ventilation + pressure support ventilation (PSV) or auxiliary/control ventilation (Amax C). The respiratory rate was 13:18 beats/minute, the tidal volume (VT) was 8:10 mL/kg, and the arterial partial pressure of carbon dioxide (PaCO2) was maintained at 35 to 50 mm Hg. The inhaled oxygen concentration (FiO2) and positive end-expiratory pressure were adjusted to keep blood oxygen saturation (SpO2) ≥ 90%.\nOnce the patients reached the switching point, they were removed from the invasive ventilator, extubated, and switched to a noninvasive ventilator (Philips Medical Systems, Inc). The initial parameters were positive inspiratory airway pressure at 12 to 14 cm H2O (1 cm H2O = 0.098 kPa) and positive expiratory airway pressure (EPAP) at 5 cm H2O. Appropriate levels were gradually reached within 5 to 20 minutes. The switching point for group A was defined as when modified GCS stabilized at ≥ 13 for 3 hours; the switching point for group B was defined as when modified GCS stabilized at ≥ 15 for 3 hours.", "Eye opening: 4 points for spontaneous eye opening, 3 points for eye opening in response to voice, 2 points for eye opening in response to pain, and 1 point for inability to open the eyes; The best motor response: 6 points for obeying commands, 5 points for locating pain, 4 points for withdrawal response to pain, 3 points for flexor response to pain, 2 points for extensor response to pain, and 1 point for immobility; Language response: 5 points for correct speech comprehension, 4 points for slow speech comprehension, 3 points for misunderstanding speech, 2 points for response to loud calls, and 1 point for no response to speech. The sum of the 3 components comprised the modified GCS score. The modified GCS was evaluated every morning starting from the 3rd day of endotracheal intubation and IMV. The detailed definition of the modified GCS is shown in Table 1.\nModified GCS score.\nThe modified GCS evaluates 3 parameters: Eye, language, and motor responses. The separate scores for these 3 parameters as well as the sum of the scores are considered. The lowest possible overall modified GCS score is 3 (deep coma or death), whereas the highest is 15 (fully conscious person).\nGCS = Glasgow coma scale, N/A = not applicable.", "The baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, mean blood pressure, and oxygenation index (OI) were recorded the next morning following endotracheal intubation under non-sedated conditions. MBP, OI, arterial partial pressure of oxygen (PaO2), and PaCO2 were recorded both before extubation and 3 hours after NIV. The time of IMV, incidence of re-intubation, incidence of VAP, and total length of stay were recorded in both groups.\nVAP was defined in patients with IMV for more than 48 hours who showed the following signs within 48 hours after extubation: new or progressive infiltrative lesions on chest X-ray and 2 of the following 3 clinical indicators (leukocytes > 12 × 109/L, body temperature greater than or equal to 38.3°C, and purulent bronchial secretion).\nRe-intubation criteria[14]: Blood pH ≤ 7.20 and PaCO2 showing a progressive increasing trend; Difficult-to-correct hypoxic state; Severe symptoms of disturbance of consciousness such as coma, lethargy or delirium; Respiratory or cardiac arrest; Respiratory depression or severe dyspnea.", "All data were analyzed with SPSS 25.0 (for Windows; IBM Corp). Continuous variables are expressed as the mean ± (SD),categorical variablesare expressed as the ratios (n%). The independent sample t test was used for comparisons between the 2 groups. The paired sample t test was used for comparisons within a group. The chi squared test was used for analysis of the categorical variables. Differences were considered statistically significant when P < .05.", "3.1. Comparison of baseline clinical data The total number of patients was 296, including 175 men and 121 women. The age range was from 20 to 87 years old, and the average age was 52.3 ± 8.7 years. There was no significant difference in gender, age, baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, body mass index, heart rate, respiratory rate, MBP, OI, PaO2, PaCO2, or blood pH value between groups A and B. Likewise, there were no significant differences with respect to concomitant diseases such as cardiovascular diseases, cerebrovascular diseases, diabetes, and chronic kidney disease between the 2 groups (Table 2).\nComparison of the baseline clinical data between the 2 groups\nAPACHE = acute physiology and chronic health enquiry, BMI = body mass index, GCS = Glasgow coma scale, MBP = mean blood pressure, OI = oxygenation index, PaCO2 = arterial partial pressure of, PaO2 = arterial partial pressure of O2.\nThe total number of patients was 296, including 175 men and 121 women. The age range was from 20 to 87 years old, and the average age was 52.3 ± 8.7 years. There was no significant difference in gender, age, baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, body mass index, heart rate, respiratory rate, MBP, OI, PaO2, PaCO2, or blood pH value between groups A and B. Likewise, there were no significant differences with respect to concomitant diseases such as cardiovascular diseases, cerebrovascular diseases, diabetes, and chronic kidney disease between the 2 groups (Table 2).\nComparison of the baseline clinical data between the 2 groups\nAPACHE = acute physiology and chronic health enquiry, BMI = body mass index, GCS = Glasgow coma scale, MBP = mean blood pressure, OI = oxygenation index, PaCO2 = arterial partial pressure of, PaO2 = arterial partial pressure of O2.\n3.2. Comparison of oxygenation indicators Groups A and B (switching point modified GCS ≥ 15) had similar MBP, OI, PaO2, and PaCO2 before extubation. Three hours after NIV, the MBP remained stable in both groups A and B. The OI and PaO2 decreased slightly whereas the PaCO2 increased slightly in both groups, but the change did not reach statistical significance. There was no significant difference between switching when modified GCS ≥ 13 (group A) and when modified GCS ≥ 15 (group B). The MBP, OI, PaO2, and PaCO2 values at 3 hours after NIV were similar in groups A and B (P = .864, .730, .425, .784, see Table 3).\nComparison of the MBP, OI, PaO2, and PaCO2 between the 2 groups.\nMBP = mean blood pressure, NIV = noninvasive ventilation, OI = oxygenation index, PaCO2 = arterial partial pressure of CO2, PaO2 = arterial partial pressure of O2.\nGroups A and B (switching point modified GCS ≥ 15) had similar MBP, OI, PaO2, and PaCO2 before extubation. Three hours after NIV, the MBP remained stable in both groups A and B. The OI and PaO2 decreased slightly whereas the PaCO2 increased slightly in both groups, but the change did not reach statistical significance. There was no significant difference between switching when modified GCS ≥ 13 (group A) and when modified GCS ≥ 15 (group B). The MBP, OI, PaO2, and PaCO2 values at 3 hours after NIV were similar in groups A and B (P = .864, .730, .425, .784, see Table 3).\nComparison of the MBP, OI, PaO2, and PaCO2 between the 2 groups.\nMBP = mean blood pressure, NIV = noninvasive ventilation, OI = oxygenation index, PaCO2 = arterial partial pressure of CO2, PaO2 = arterial partial pressure of O2.\n3.3. Comparison of related medical indicators Only a few patients required re-intubation in both groups (6.1% in group A and 10.1% in group B). Although the re-intubation rate was slightly higher in group B (modified GCS ≥ 15 used as a switching point), the difference did not reach statistical significance (P = .201). However, the length of hospital stay and duration of IMV in group A (modified GCS ≥ 13 used as a switching point) were significantly shorter than those in group B (P = .041, <.001). The duration of IMV was 2.7 days shorter in group A and the length of hospital stay was 6.9 days shorter. The incidence of VAP in group A was only 34% of that seen in group B (P = .001, see Table 4).\nComparison of the related medical indicators between the 2 groups.\nIMV = invasive mechanical ventilation, VAP = ventilator-associated pneumonia.\nOnly a few patients required re-intubation in both groups (6.1% in group A and 10.1% in group B). Although the re-intubation rate was slightly higher in group B (modified GCS ≥ 15 used as a switching point), the difference did not reach statistical significance (P = .201). However, the length of hospital stay and duration of IMV in group A (modified GCS ≥ 13 used as a switching point) were significantly shorter than those in group B (P = .041, <.001). The duration of IMV was 2.7 days shorter in group A and the length of hospital stay was 6.9 days shorter. The incidence of VAP in group A was only 34% of that seen in group B (P = .001, see Table 4).\nComparison of the related medical indicators between the 2 groups.\nIMV = invasive mechanical ventilation, VAP = ventilator-associated pneumonia.", "The total number of patients was 296, including 175 men and 121 women. The age range was from 20 to 87 years old, and the average age was 52.3 ± 8.7 years. There was no significant difference in gender, age, baseline Acute Physiology and Chronic Health Enquiry II score, modified GCS score, body mass index, heart rate, respiratory rate, MBP, OI, PaO2, PaCO2, or blood pH value between groups A and B. Likewise, there were no significant differences with respect to concomitant diseases such as cardiovascular diseases, cerebrovascular diseases, diabetes, and chronic kidney disease between the 2 groups (Table 2).\nComparison of the baseline clinical data between the 2 groups\nAPACHE = acute physiology and chronic health enquiry, BMI = body mass index, GCS = Glasgow coma scale, MBP = mean blood pressure, OI = oxygenation index, PaCO2 = arterial partial pressure of, PaO2 = arterial partial pressure of O2.", "Groups A and B (switching point modified GCS ≥ 15) had similar MBP, OI, PaO2, and PaCO2 before extubation. Three hours after NIV, the MBP remained stable in both groups A and B. The OI and PaO2 decreased slightly whereas the PaCO2 increased slightly in both groups, but the change did not reach statistical significance. There was no significant difference between switching when modified GCS ≥ 13 (group A) and when modified GCS ≥ 15 (group B). The MBP, OI, PaO2, and PaCO2 values at 3 hours after NIV were similar in groups A and B (P = .864, .730, .425, .784, see Table 3).\nComparison of the MBP, OI, PaO2, and PaCO2 between the 2 groups.\nMBP = mean blood pressure, NIV = noninvasive ventilation, OI = oxygenation index, PaCO2 = arterial partial pressure of CO2, PaO2 = arterial partial pressure of O2.", "Only a few patients required re-intubation in both groups (6.1% in group A and 10.1% in group B). Although the re-intubation rate was slightly higher in group B (modified GCS ≥ 15 used as a switching point), the difference did not reach statistical significance (P = .201). However, the length of hospital stay and duration of IMV in group A (modified GCS ≥ 13 used as a switching point) were significantly shorter than those in group B (P = .041, <.001). The duration of IMV was 2.7 days shorter in group A and the length of hospital stay was 6.9 days shorter. The incidence of VAP in group A was only 34% of that seen in group B (P = .001, see Table 4).\nComparison of the related medical indicators between the 2 groups.\nIMV = invasive mechanical ventilation, VAP = ventilator-associated pneumonia.", "Acute respiratory distress syndrome with respiratory failure is associated with a mortality rate of over 40% despite nearly 20 years of research and the adoption of protective lung ventilation strategies and optimal management practices.[15–17] Patients with long-term endotracheal intubation easily develop lower respiratory tract infection and VAP. Bacterial contamination along the tracheobronchial tree, sputum suction, and downflow of the air bag can lead to the aggravation of lower respiratory tract infection, prolongation of the duration of IMV, and difficulties in weaning from the ventilator.[18,19] Long-term IMV is also associated with other complications such as airway injury and tracheoesophageal fistula, which have significant impacts on the overall survival.[20] The prognosis is significantly worse if the mode of mechanical ventilation cannot be adjusted in a timely and reasonably manner.[21] In the present study, a sequential invasive-noninvasive ventilation strategy was employed to enable early extubation and minimize the duration of IMV as much as possible, thereby avoiding serious complications and improving the outcomes.\nAccurate identification of the optimal time point for switching to NIV from IMV is the key to successful sequential ventilation.[22] In 2007, the European Respiratory Association, American Thoracic Association, European Association of intensive Care Medicine, American Association of critical Care Medicine (SCCM), and French Institute of Terminology Revision (SRLF) all advocated the use of the autonomous breathing test (SBT) as an important diagnostic test to judge the success of weaning and indicated that the duration of SBT should be 30 to 120 minutes.[23] However, the optimal SBT time for mechanically ventilated patients with different underlying diseases has not been determined. Later studies found that SBT often led to the delayed withdrawal of IMV and increased the risk of VAP, which subsequently increased mortality.[24] In China, the switching point is identified based on the clinical picture of PIC. The problem with this approach is that it does not take into account the lag in imaging presentation behind clinical manifestations and noninfectious factors (the cause of acute exacerbation of COPD is difficult to determine)[25]; therefore, PIC is not suitable for all patients with acute exacerbation of COPD. Some countries use 48 hours after IMV as the switching point, but this ignores the patients’ state of consciousness, compliance, individual differences, ethnic differences, and other factors.\nTo be able to use NIV, patients are required to have a clear consciousness, a certain degree of cooperation and understanding, and good compliance. Assessment of the consciousness state of patients is therefore important. The GCS score is widely used for the assessment of consciousness. However, intubated patients cannot speak, which makes it impossible to use the verbal response portion of the original GCS score. In the modified GCS, the verbal response portion is replaced with the understanding of language and speech to better suit the conditions of an intubated patient. Luo Xianhai et al[26] proposed the use of the modified GCS score to guide sequential invasive-noninvasive ventilation and used a score ≥ 10 as the switching point. Zheng Dawei et al[6] proposed that a modified GCS score of 15 can be used as the switching point for invasive-noninvasive sequential therapy, which can significantly improve the therapeutic effect for COPD patients with respiratory failure. Other researchers have also attempted to use a modified GCS score of 15 as the switching point.\nIn this study, the authors found that a modified GCS score that is stable for 3 hours at ≥ 13 is a better switching point for sequential invasive-noninvasive ventilation than a score ≥ 15. Although there was no difference in MBP, OI, PaO2, and PaCO2 between the 2 groups at 3 hours after extubation and NIV, the duration of IMV, length of hospital stay, and incidence of VAP was all lower in group A, the group in which a modified GCS score ≥ 13 was used as the switching point. After early extubation and switching to NIV, the patients experienced no discomfort, restlessness, or pain. Voice communication and self-consciousness improved and there was also increased trust and coordination among the medical staff. The success rate of withdrawal was high, which shortened the time of hospitalization and reduced medical expenses.\nThere are several limitations to this study. The modified GCS is widely used clinically in China, but is less well-known outside China. Only objective data were collected in the study, and the subjective comfort of the patients was not studied. Patient discomfort and pain should be evaluated in future studies.", "In conclusion, in patients with COPD who develop respiratory failure and require IMV, it is safe and feasible to switch to NIV if the modified GCS score remains stable at ≥ 13 for 3 hours. Using a modified GCS score ≥ 13 as the switching point for sequential invasive-noninvasive ventilation therapy can reduce the duration of IMV, duration of hospital stay, and the incidence of VAP compared to using a modified GCS score ≥ 15.", "Conceptualization: Jin-Bo Zhang, Li-Hong Li, Jin-Qiang Zhu, Shi-Fang Zhou, Ji-Hong Ma, Zhi-Qiang Li, Xiao-Qin Lin.\nData curation: Li-Hong Li, Jin-Qiang Zhu, Shi-Fang Zhou, Zhi-Qiang Li, Xiao-Qin Lin.\nFormal analysis: Jin-Bo Zhang, Ji-Hong Ma, Xiao-Hong Jin.\nFunding acquisition: Xiao-Hong Jin.\nInvestigation: Jin-Bo Zhang, Xiao-Hong Jin.\nProject administration: Jin-Bo Zhang, Zhi-Qiang Li.\nResources: Zhi-Qiang Li.\nSoftware: Li-Hong Li.\nSupervision: Shi-Fang Zhou.\nVisualization: Jin-Qiang Zhu, Zhi-Qiang Li, Xiao-Qin Lin.\nWriting – original draft: Jin-Bo Zhang, Jin-Qiang Zhu, Shi-Fang Zhou, Ji-Hong Ma, Zhi-Qiang Li.\nWriting – review & editing: Jin-Bo Zhang, Li-Hong Li, Shi-Fang Zhou, Xiao-Hong Jin." ]

[ "intro", "methods", null, null, "methods", null, null, null, "methods", "results", null, null, null, "discussion", "conclusions", null ]

[ "chronic obstructive pulmonary disease", "Glasgow coma scale", "mechanical", "respiratory failure", "ventilation" ]

[ 3823, 304, 36, 1516, 285, 250, 211, 181, 179, 179, 221 ]

[ "gcs", "modified", "modified gcs", "points", "pressure", "point", "patients", "response", "score", "group" ]

[ "successful sequential ventilation", "protective lung ventilation", "respiratory failure copd", "disease progression ventilation", "imv respiratory failure" ]

[ "Automobile Driving", "Caffeine", "Central Nervous System Stimulants", "Coffee", "Cross-Over Studies", "Double-Blind Method", "Female", "Humans", "Male", "Time Factors", "Wakefulness", "Young Adult" ]

[ "Subjects", "Study design", "Treatments", "STISIM highway driving test", "Subjective assessments", "Statistical analysis", "Driving test", "Subjective driving assessments", "Subjective sleepiness" ]

[ "Twenty-four adult healthy volunteers (12 males and 12 females) were recruited by means of public advertisements at and around Utrecht University campus. Subjects were included if they were healthy volunteers, moderate caffeine drinkers (two to four cups a day), non-smokers, had a body mass index between 21 and 30, possessed a valid driver’s license for at least 3 years, and drove more than 5,000 km per year.\nSleep disturbances were assessed with the SLEEP-50 questionnaire (Spoormaker et al. 2005), and excessive daytime sleepiness was examined using the Epworth Sleepiness Scale (ESS; Johns 1991), filled out by participants on the screening day.\nBefore the start of each test day, urine samples were collected to test for drugs of abuse including amphetamines (including MDMA), barbiturates, cannabinoids, benzodiazepines, cocaine, and opiates (Instant-View, Alfa Scientific Designs Inc.) and pregnancy in female subjects (β-HCG test). To test for the presence of alcohol, the Dräger Alcotest 7410 Breath Analyzer was used. From 24 h before the start of the test day until completion of the test day, alcohol consumption was not permitted. Caffeinated beverages were not allowed from awakening on test days until the end of the tests.", "Participants were screened and familiarized with the test procedures during a training day. When meeting all inclusion and passing all exclusion criteria, subjects performed a practice session in the STISIM driving simulator and completed the Simulator Sickness Questionnaire (Kennedy et al. 1993) to identify possible simulator sickness. Included subjects were randomly assigned to a treatment order comprising decaffeinated coffee and caffeinated coffee (80 mg) administered during a break.\nUpon arrival, possible use of drugs or alcohol, pregnancy, illness, and medication were checked. In addition, quality and duration of sleep was assessed using the 14-item Groningen Sleep Quality Scale (Mulder-Hajonides van der Meulen et al. 1980). When all criteria were met, a 120-min drive in the STISIM driving simulator was conducted. Thereafter, a 15-min break was scheduled in which subjects received the double-blind treatment. After the break, another 120-min driving session was performed. Every hour, subjective assessments of driving quality, driving style, mental effort to perform the test, and sleepiness were conducted. Test sessions were scheduled at the same time for each subject, either in the morning (0800–1300 hours) or in the afternoon (1300–1700 hours).", "This study aimed to mimic the effect of a cup of coffee drivers consume when having a break along the highway. Treatments were 2.68 g of Nescafé Gold® instant coffee containing 80 mg caffeine or 2.68 g of Nescafé Gold® decaffeinated coffee dissolved in 180 ml boiled water. To confirm that each cup of coffee contained 80 mg of caffeine, the amount of caffeine in the instant coffee was determined with high-performance liquid chromatography (HPLC; Shimadzu LC-10AT VP equipped with UV–Vis detector). The column was a reversed-phase Select B column Lichrocart HPLC C18, 5 μm, length, 0.125 m, Ø = 4.6 mm. All of the procedures were carried out isocratic. The separation was done at room temperature. Caffeine and the spiked matrices were separated with a mobile phase of 20% MeOH and 10 mM HClO4, at a flow rate of 0.5 mL/min. The injection volume was 5 μL, and the detection was carried out at 273 nm. The mean (SD) amount of caffeine per gram Nescafé Goud instant coffee samples (n = 10) was 29.79 (0.656) mg/g. The mean amount of caffeine in decaffeinated coffee was 0.79 mg/g. The accuracy of determinations was 98.1% (SD, 0.56). Because both the precision and the accuracy met up to the requirement demands, all of the results of this HPLC determination can be concluded with certainty.\nTreatments were administered double-blind, and a nose clip was worn to enhance treatment blinding. Drinks were consumed within 5 min, starting from 5 min after onset of the break.", "Driving tests were performed in a fixed-base driving simulator employing STISIM Drive™ (version M300, Systems Technology Inc., Hawthorne, CA, USA). This is an interactive system in which the roadway scenery is projected on a screen (2.10 × 1.58 m), 1.90 m in front of the center of the steering wheel of the car unit (Mets et al. 2011a). The 100-km highway driving test scenarios were developed (EyeOctopus BV) in accordance with Dutch traffic situations, including a two-lane highway in each direction and a monotonous environment with trees, occasional hills and bridges, and other traffic. The duration of each 100-km scenario is approximately 60 min. Two scenarios (200 km) were conducted before a 15-min break, and two other scenarios (200 km) thereafter (Mets et al. 2011a).\nSubjects were instructed to drive with a steady lateral position within the right, slower, traffic lane with a constant speed of 95 km/h. Overtaking slower-moving vehicles was allowed. During blinded editing, these maneuvers were removed from the data, before statistical analysis of the “clean” data. The primary outcome variable was the standard deviation of lateral position (SDLP, centimeters), expressing the weaving of the car (Verster and Roth 2011). The standard deviation of speed (SDS, kilometers per hour) was the secondary outcome measure. Mean speed (MS, kilometers per hour) and mean lateral position (MLP, cm) were control variables.", "After each hour of driving, questionnaires were administered on subjective sleepiness and driving performance. Subjective sleepiness was measured by means of the Karolinska Sleepiness Scale (KSS), ranging from 1 (very alert) to 9 (very sleepy, fighting sleep) (Åkerstedt and Gillberg 1990).\nDriving task-related questionnaires comprised mental effort to perform the driving test (Meijman et al. 1986; Zijlstra and Van Doorn 1985), subjective driving quality, and driving style (McCormick et al. 1987). Completing the questionnaires took approximately 2 min, after which, the driving task was immediately resumed.", "Statistical analyses were performed with SPSS, version 19. For each variable, mean (SD) was computed for each subsequent hour. Data of the first 2 h were compared, to confirm that no significant differences between the treatment days were present before the break and treatment administration. To determine whether caffeinated coffee has an effect on driving performance, data from the third and fourth hour were compared using a general linear model for repeated measures (two-tailed, p ≤ 0.05).", "Figure 1 shows the effect of caffeinated coffee consumption on driving performance. No significant differences in SDLP were observed before the break. However, both in the first (F\n(1,23) = 5.8; p = 0.024) and in the second hour (F\n(1,23) = 6.4; p = 0.019) after the break, caffeinated coffee significantly reduced SDLP.Fig. 1Standard deviation of lateral position (SDLP). Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nStandard deviation of lateral position (SDLP). Asterisks indicate significant difference compared to placebo (p < 0.05)\nIn line, caffeinated coffee significantly reduced SD speed in the third (F\n(1,23) = 5.8; p = 0.024) and fourth hour (F\n(1,23) = 13.0; p = 0.001) of driving (see Fig. 2).Fig. 2Standard deviation of speed (SDS). Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nStandard deviation of speed (SDS). Asterisks indicate significant difference compared to placebo (p < 0.05)\nNo effects were found on mean speed or mean lateral position, confirming that subjects performed the test according to the instructions.", "Compared to decaffeinated coffee, caffeinated coffee improved subjective driving quality in the third hour of driving (F\n(1,23) = 10.5; p = 0.004), but not in the fourth (F\n(1,23) = 2.6; p = n.s.). Subjects indicated that the mental effort needed to perform the test after caffeinated coffee was significantly reduced in the third (F\n(1,23) = 11.4; p = 0.003) and fourth hour of driving (F\n(1,23) = 5.9; p = 0.023). In addition, drivers rated their driving quality as significantly more considerate, responsible, and safer in the caffeinated coffee condition (see Table 1).", "After the break with caffeinated coffee, drivers reported significantly lower sleepiness scores as compared to the break with decaffeinated coffee. This effect was significant both in the third (F\n(1,23) = 18.5; p < 0.001) and the fourth hour of driving (F\n(1,23) = 11.9; p = 0.002) (see Fig. 3).Fig. 3Karolinska sleepiness scale. Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nKarolinska sleepiness scale. Asterisks indicate significant difference compared to placebo (p < 0.05)" ]

[ null, null, null, null, null, null, null, null, null ]

[ "Introduction", "Materials and methods", "Subjects", "Study design", "Treatments", "STISIM highway driving test", "Subjective assessments", "Statistical analysis", "Results", "Driving test", "Subjective driving assessments", "Subjective sleepiness", "Discussion" ]

[ "Drowsy driving is an important cause of traffic accidents (Connor et al. 2002; Horne and Reyner 1995; Maycock 1996), and therefore, the development of effective countermeasures is essential. Consuming a cup of coffee is one of the most commonly used ways to combat driver sleepiness. An estimated 80% of the population consumes caffeine-containing beverages, often on a daily basis (Fredholm et al. 1999; Heckman et al. 2010). Caffeine (1,3,7-trimethylxanthine) is rapidly and completely absorbed in the body within approximately 45 min (Blanchard and Sawers 1983). It reaches its peak plasma concentration within 15 to 120 min after intake (Arnaud 1987), averaging around 30 min (O’Connell and Zurzola 1984; Blanchard and Sawers 1983). Its elimination half life is 1.5 to 9.5 h (Arnaud 1987; Bonati et al. 1982). Although additional mechanisms of action are involved, it is now believed that caffeine’s stimulant effects are exerted by antagonizing adenosine, primarily by blocking the adenosine A1 and A2A receptors. Adenosine is considered to be a mediator of sleep (Dunwiddie and Mansino 2001; Fredholm et al. 1999).\nA great number of studies have demonstrated effects of caffeine on mood and performance (Childs and De Wit 2006; Christopher et al. 2005; Haskell et al. 2005; Lieberman et al. 1987; Olson et al. 2010). However, the effects are complex and depend on the specific tasks examined, dosages, subjects, and test conditions (Lorist and Tops 2003). Overall, caffeine was found to be specifically effective in restoring performance to baseline levels when individuals are in a state of low arousal, such as seen during the dip in the circadian rhythm, after sleep restriction, and in fatigued subjects (Nehlig 2010; Smith 2002). Indeed, many people consume coffee with the purpose to refresh or stay awake, for example, when driving a car (Anund et al. 2008; Vanlaar et al. 2008).\nSeveral driving studies showed that caffeine improves performance and decreases subjective sleepiness both in driving simulators (Biggs et al. 2007; Brice and Smith 2001; De Valck and Cluydts 2001; Horne and Reyner 1996; Regina et al. 1974; Reyner and Horne 1997, 2000) and on the road (Philip et al. 2006; Sagaspe et al. 2007). Most of these studies tested sleep-restricted subjects. In addition, relatively high dosages of caffeine (100–300 mg) were examined. These studies showed that relatively high dosages of caffeine had a positive effect on driving performance and reduced driver sleepiness. In real life however, it is more likely that a driver consumes only one cup of coffee (80 mg of caffeine) during a break, before continuing driving. Up to now, the effects on driving performance of lower dosages of caffeine, e.g., a regular cup of coffee, have not been examined.\nTherefore, the objective of this study was to examine the effects of one cup of coffee (80 mg caffeine) on prolonged simulated highway driving in non-sleep deprived individuals. Various traffic safety organizations advise drivers to take a 15-min break after 2 h of driving. The protocol used in the current study (2 h driving, a 15-min break with or without consuming caffeinated coffee, followed by 2 h of driving) was based on this advice.", "This study was a double-blind, randomized, placebo-controlled, cross-over study. The study was conducted according to the ICH Guidelines for “Good Clinical Practice,” and the Declaration of Helsinki and its latest amendments. Written informed consent was obtained from the participants before taking part in the study. The study was approved by the Institutional Review Board; no medical ethical approval was required to conduct the study.\n Subjects Twenty-four adult healthy volunteers (12 males and 12 females) were recruited by means of public advertisements at and around Utrecht University campus. Subjects were included if they were healthy volunteers, moderate caffeine drinkers (two to four cups a day), non-smokers, had a body mass index between 21 and 30, possessed a valid driver’s license for at least 3 years, and drove more than 5,000 km per year.\nSleep disturbances were assessed with the SLEEP-50 questionnaire (Spoormaker et al. 2005), and excessive daytime sleepiness was examined using the Epworth Sleepiness Scale (ESS; Johns 1991), filled out by participants on the screening day.\nBefore the start of each test day, urine samples were collected to test for drugs of abuse including amphetamines (including MDMA), barbiturates, cannabinoids, benzodiazepines, cocaine, and opiates (Instant-View, Alfa Scientific Designs Inc.) and pregnancy in female subjects (β-HCG test). To test for the presence of alcohol, the Dräger Alcotest 7410 Breath Analyzer was used. From 24 h before the start of the test day until completion of the test day, alcohol consumption was not permitted. Caffeinated beverages were not allowed from awakening on test days until the end of the tests.\nTwenty-four adult healthy volunteers (12 males and 12 females) were recruited by means of public advertisements at and around Utrecht University campus. Subjects were included if they were healthy volunteers, moderate caffeine drinkers (two to four cups a day), non-smokers, had a body mass index between 21 and 30, possessed a valid driver’s license for at least 3 years, and drove more than 5,000 km per year.\nSleep disturbances were assessed with the SLEEP-50 questionnaire (Spoormaker et al. 2005), and excessive daytime sleepiness was examined using the Epworth Sleepiness Scale (ESS; Johns 1991), filled out by participants on the screening day.\nBefore the start of each test day, urine samples were collected to test for drugs of abuse including amphetamines (including MDMA), barbiturates, cannabinoids, benzodiazepines, cocaine, and opiates (Instant-View, Alfa Scientific Designs Inc.) and pregnancy in female subjects (β-HCG test). To test for the presence of alcohol, the Dräger Alcotest 7410 Breath Analyzer was used. From 24 h before the start of the test day until completion of the test day, alcohol consumption was not permitted. Caffeinated beverages were not allowed from awakening on test days until the end of the tests.\n Study design Participants were screened and familiarized with the test procedures during a training day. When meeting all inclusion and passing all exclusion criteria, subjects performed a practice session in the STISIM driving simulator and completed the Simulator Sickness Questionnaire (Kennedy et al. 1993) to identify possible simulator sickness. Included subjects were randomly assigned to a treatment order comprising decaffeinated coffee and caffeinated coffee (80 mg) administered during a break.\nUpon arrival, possible use of drugs or alcohol, pregnancy, illness, and medication were checked. In addition, quality and duration of sleep was assessed using the 14-item Groningen Sleep Quality Scale (Mulder-Hajonides van der Meulen et al. 1980). When all criteria were met, a 120-min drive in the STISIM driving simulator was conducted. Thereafter, a 15-min break was scheduled in which subjects received the double-blind treatment. After the break, another 120-min driving session was performed. Every hour, subjective assessments of driving quality, driving style, mental effort to perform the test, and sleepiness were conducted. Test sessions were scheduled at the same time for each subject, either in the morning (0800–1300 hours) or in the afternoon (1300–1700 hours).\nParticipants were screened and familiarized with the test procedures during a training day. When meeting all inclusion and passing all exclusion criteria, subjects performed a practice session in the STISIM driving simulator and completed the Simulator Sickness Questionnaire (Kennedy et al. 1993) to identify possible simulator sickness. Included subjects were randomly assigned to a treatment order comprising decaffeinated coffee and caffeinated coffee (80 mg) administered during a break.\nUpon arrival, possible use of drugs or alcohol, pregnancy, illness, and medication were checked. In addition, quality and duration of sleep was assessed using the 14-item Groningen Sleep Quality Scale (Mulder-Hajonides van der Meulen et al. 1980). When all criteria were met, a 120-min drive in the STISIM driving simulator was conducted. Thereafter, a 15-min break was scheduled in which subjects received the double-blind treatment. After the break, another 120-min driving session was performed. Every hour, subjective assessments of driving quality, driving style, mental effort to perform the test, and sleepiness were conducted. Test sessions were scheduled at the same time for each subject, either in the morning (0800–1300 hours) or in the afternoon (1300–1700 hours).\n Treatments This study aimed to mimic the effect of a cup of coffee drivers consume when having a break along the highway. Treatments were 2.68 g of Nescafé Gold® instant coffee containing 80 mg caffeine or 2.68 g of Nescafé Gold® decaffeinated coffee dissolved in 180 ml boiled water. To confirm that each cup of coffee contained 80 mg of caffeine, the amount of caffeine in the instant coffee was determined with high-performance liquid chromatography (HPLC; Shimadzu LC-10AT VP equipped with UV–Vis detector). The column was a reversed-phase Select B column Lichrocart HPLC C18, 5 μm, length, 0.125 m, Ø = 4.6 mm. All of the procedures were carried out isocratic. The separation was done at room temperature. Caffeine and the spiked matrices were separated with a mobile phase of 20% MeOH and 10 mM HClO4, at a flow rate of 0.5 mL/min. The injection volume was 5 μL, and the detection was carried out at 273 nm. The mean (SD) amount of caffeine per gram Nescafé Goud instant coffee samples (n = 10) was 29.79 (0.656) mg/g. The mean amount of caffeine in decaffeinated coffee was 0.79 mg/g. The accuracy of determinations was 98.1% (SD, 0.56). Because both the precision and the accuracy met up to the requirement demands, all of the results of this HPLC determination can be concluded with certainty.\nTreatments were administered double-blind, and a nose clip was worn to enhance treatment blinding. Drinks were consumed within 5 min, starting from 5 min after onset of the break.\nThis study aimed to mimic the effect of a cup of coffee drivers consume when having a break along the highway. Treatments were 2.68 g of Nescafé Gold® instant coffee containing 80 mg caffeine or 2.68 g of Nescafé Gold® decaffeinated coffee dissolved in 180 ml boiled water. To confirm that each cup of coffee contained 80 mg of caffeine, the amount of caffeine in the instant coffee was determined with high-performance liquid chromatography (HPLC; Shimadzu LC-10AT VP equipped with UV–Vis detector). The column was a reversed-phase Select B column Lichrocart HPLC C18, 5 μm, length, 0.125 m, Ø = 4.6 mm. All of the procedures were carried out isocratic. The separation was done at room temperature. Caffeine and the spiked matrices were separated with a mobile phase of 20% MeOH and 10 mM HClO4, at a flow rate of 0.5 mL/min. The injection volume was 5 μL, and the detection was carried out at 273 nm. The mean (SD) amount of caffeine per gram Nescafé Goud instant coffee samples (n = 10) was 29.79 (0.656) mg/g. The mean amount of caffeine in decaffeinated coffee was 0.79 mg/g. The accuracy of determinations was 98.1% (SD, 0.56). Because both the precision and the accuracy met up to the requirement demands, all of the results of this HPLC determination can be concluded with certainty.\nTreatments were administered double-blind, and a nose clip was worn to enhance treatment blinding. Drinks were consumed within 5 min, starting from 5 min after onset of the break.\n STISIM highway driving test Driving tests were performed in a fixed-base driving simulator employing STISIM Drive™ (version M300, Systems Technology Inc., Hawthorne, CA, USA). This is an interactive system in which the roadway scenery is projected on a screen (2.10 × 1.58 m), 1.90 m in front of the center of the steering wheel of the car unit (Mets et al. 2011a). The 100-km highway driving test scenarios were developed (EyeOctopus BV) in accordance with Dutch traffic situations, including a two-lane highway in each direction and a monotonous environment with trees, occasional hills and bridges, and other traffic. The duration of each 100-km scenario is approximately 60 min. Two scenarios (200 km) were conducted before a 15-min break, and two other scenarios (200 km) thereafter (Mets et al. 2011a).\nSubjects were instructed to drive with a steady lateral position within the right, slower, traffic lane with a constant speed of 95 km/h. Overtaking slower-moving vehicles was allowed. During blinded editing, these maneuvers were removed from the data, before statistical analysis of the “clean” data. The primary outcome variable was the standard deviation of lateral position (SDLP, centimeters), expressing the weaving of the car (Verster and Roth 2011). The standard deviation of speed (SDS, kilometers per hour) was the secondary outcome measure. Mean speed (MS, kilometers per hour) and mean lateral position (MLP, cm) were control variables.\nDriving tests were performed in a fixed-base driving simulator employing STISIM Drive™ (version M300, Systems Technology Inc., Hawthorne, CA, USA). This is an interactive system in which the roadway scenery is projected on a screen (2.10 × 1.58 m), 1.90 m in front of the center of the steering wheel of the car unit (Mets et al. 2011a). The 100-km highway driving test scenarios were developed (EyeOctopus BV) in accordance with Dutch traffic situations, including a two-lane highway in each direction and a monotonous environment with trees, occasional hills and bridges, and other traffic. The duration of each 100-km scenario is approximately 60 min. Two scenarios (200 km) were conducted before a 15-min break, and two other scenarios (200 km) thereafter (Mets et al. 2011a).\nSubjects were instructed to drive with a steady lateral position within the right, slower, traffic lane with a constant speed of 95 km/h. Overtaking slower-moving vehicles was allowed. During blinded editing, these maneuvers were removed from the data, before statistical analysis of the “clean” data. The primary outcome variable was the standard deviation of lateral position (SDLP, centimeters), expressing the weaving of the car (Verster and Roth 2011). The standard deviation of speed (SDS, kilometers per hour) was the secondary outcome measure. Mean speed (MS, kilometers per hour) and mean lateral position (MLP, cm) were control variables.\n Subjective assessments After each hour of driving, questionnaires were administered on subjective sleepiness and driving performance. Subjective sleepiness was measured by means of the Karolinska Sleepiness Scale (KSS), ranging from 1 (very alert) to 9 (very sleepy, fighting sleep) (Åkerstedt and Gillberg 1990).\nDriving task-related questionnaires comprised mental effort to perform the driving test (Meijman et al. 1986; Zijlstra and Van Doorn 1985), subjective driving quality, and driving style (McCormick et al. 1987). Completing the questionnaires took approximately 2 min, after which, the driving task was immediately resumed.\nAfter each hour of driving, questionnaires were administered on subjective sleepiness and driving performance. Subjective sleepiness was measured by means of the Karolinska Sleepiness Scale (KSS), ranging from 1 (very alert) to 9 (very sleepy, fighting sleep) (Åkerstedt and Gillberg 1990).\nDriving task-related questionnaires comprised mental effort to perform the driving test (Meijman et al. 1986; Zijlstra and Van Doorn 1985), subjective driving quality, and driving style (McCormick et al. 1987). Completing the questionnaires took approximately 2 min, after which, the driving task was immediately resumed.\n Statistical analysis Statistical analyses were performed with SPSS, version 19. For each variable, mean (SD) was computed for each subsequent hour. Data of the first 2 h were compared, to confirm that no significant differences between the treatment days were present before the break and treatment administration. To determine whether caffeinated coffee has an effect on driving performance, data from the third and fourth hour were compared using a general linear model for repeated measures (two-tailed, p ≤ 0.05).\nStatistical analyses were performed with SPSS, version 19. For each variable, mean (SD) was computed for each subsequent hour. Data of the first 2 h were compared, to confirm that no significant differences between the treatment days were present before the break and treatment administration. To determine whether caffeinated coffee has an effect on driving performance, data from the third and fourth hour were compared using a general linear model for repeated measures (two-tailed, p ≤ 0.05).", "Twenty-four adult healthy volunteers (12 males and 12 females) were recruited by means of public advertisements at and around Utrecht University campus. Subjects were included if they were healthy volunteers, moderate caffeine drinkers (two to four cups a day), non-smokers, had a body mass index between 21 and 30, possessed a valid driver’s license for at least 3 years, and drove more than 5,000 km per year.\nSleep disturbances were assessed with the SLEEP-50 questionnaire (Spoormaker et al. 2005), and excessive daytime sleepiness was examined using the Epworth Sleepiness Scale (ESS; Johns 1991), filled out by participants on the screening day.\nBefore the start of each test day, urine samples were collected to test for drugs of abuse including amphetamines (including MDMA), barbiturates, cannabinoids, benzodiazepines, cocaine, and opiates (Instant-View, Alfa Scientific Designs Inc.) and pregnancy in female subjects (β-HCG test). To test for the presence of alcohol, the Dräger Alcotest 7410 Breath Analyzer was used. From 24 h before the start of the test day until completion of the test day, alcohol consumption was not permitted. Caffeinated beverages were not allowed from awakening on test days until the end of the tests.", "Participants were screened and familiarized with the test procedures during a training day. When meeting all inclusion and passing all exclusion criteria, subjects performed a practice session in the STISIM driving simulator and completed the Simulator Sickness Questionnaire (Kennedy et al. 1993) to identify possible simulator sickness. Included subjects were randomly assigned to a treatment order comprising decaffeinated coffee and caffeinated coffee (80 mg) administered during a break.\nUpon arrival, possible use of drugs or alcohol, pregnancy, illness, and medication were checked. In addition, quality and duration of sleep was assessed using the 14-item Groningen Sleep Quality Scale (Mulder-Hajonides van der Meulen et al. 1980). When all criteria were met, a 120-min drive in the STISIM driving simulator was conducted. Thereafter, a 15-min break was scheduled in which subjects received the double-blind treatment. After the break, another 120-min driving session was performed. Every hour, subjective assessments of driving quality, driving style, mental effort to perform the test, and sleepiness were conducted. Test sessions were scheduled at the same time for each subject, either in the morning (0800–1300 hours) or in the afternoon (1300–1700 hours).", "This study aimed to mimic the effect of a cup of coffee drivers consume when having a break along the highway. Treatments were 2.68 g of Nescafé Gold® instant coffee containing 80 mg caffeine or 2.68 g of Nescafé Gold® decaffeinated coffee dissolved in 180 ml boiled water. To confirm that each cup of coffee contained 80 mg of caffeine, the amount of caffeine in the instant coffee was determined with high-performance liquid chromatography (HPLC; Shimadzu LC-10AT VP equipped with UV–Vis detector). The column was a reversed-phase Select B column Lichrocart HPLC C18, 5 μm, length, 0.125 m, Ø = 4.6 mm. All of the procedures were carried out isocratic. The separation was done at room temperature. Caffeine and the spiked matrices were separated with a mobile phase of 20% MeOH and 10 mM HClO4, at a flow rate of 0.5 mL/min. The injection volume was 5 μL, and the detection was carried out at 273 nm. The mean (SD) amount of caffeine per gram Nescafé Goud instant coffee samples (n = 10) was 29.79 (0.656) mg/g. The mean amount of caffeine in decaffeinated coffee was 0.79 mg/g. The accuracy of determinations was 98.1% (SD, 0.56). Because both the precision and the accuracy met up to the requirement demands, all of the results of this HPLC determination can be concluded with certainty.\nTreatments were administered double-blind, and a nose clip was worn to enhance treatment blinding. Drinks were consumed within 5 min, starting from 5 min after onset of the break.", "Driving tests were performed in a fixed-base driving simulator employing STISIM Drive™ (version M300, Systems Technology Inc., Hawthorne, CA, USA). This is an interactive system in which the roadway scenery is projected on a screen (2.10 × 1.58 m), 1.90 m in front of the center of the steering wheel of the car unit (Mets et al. 2011a). The 100-km highway driving test scenarios were developed (EyeOctopus BV) in accordance with Dutch traffic situations, including a two-lane highway in each direction and a monotonous environment with trees, occasional hills and bridges, and other traffic. The duration of each 100-km scenario is approximately 60 min. Two scenarios (200 km) were conducted before a 15-min break, and two other scenarios (200 km) thereafter (Mets et al. 2011a).\nSubjects were instructed to drive with a steady lateral position within the right, slower, traffic lane with a constant speed of 95 km/h. Overtaking slower-moving vehicles was allowed. During blinded editing, these maneuvers were removed from the data, before statistical analysis of the “clean” data. The primary outcome variable was the standard deviation of lateral position (SDLP, centimeters), expressing the weaving of the car (Verster and Roth 2011). The standard deviation of speed (SDS, kilometers per hour) was the secondary outcome measure. Mean speed (MS, kilometers per hour) and mean lateral position (MLP, cm) were control variables.", "After each hour of driving, questionnaires were administered on subjective sleepiness and driving performance. Subjective sleepiness was measured by means of the Karolinska Sleepiness Scale (KSS), ranging from 1 (very alert) to 9 (very sleepy, fighting sleep) (Åkerstedt and Gillberg 1990).\nDriving task-related questionnaires comprised mental effort to perform the driving test (Meijman et al. 1986; Zijlstra and Van Doorn 1985), subjective driving quality, and driving style (McCormick et al. 1987). Completing the questionnaires took approximately 2 min, after which, the driving task was immediately resumed.", "Statistical analyses were performed with SPSS, version 19. For each variable, mean (SD) was computed for each subsequent hour. Data of the first 2 h were compared, to confirm that no significant differences between the treatment days were present before the break and treatment administration. To determine whether caffeinated coffee has an effect on driving performance, data from the third and fourth hour were compared using a general linear model for repeated measures (two-tailed, p ≤ 0.05).", "A total of 24 subjects (12 males and 12 females) completed the study. Their mean (SD) age was 23.2 (1.6) years old; on average, they consumed 2.5 (0.7) caffeinated drinks per day, had a mean (SD) body mass index of 23.9 (2.7), possessed a valid driver’s license for 58.8 (17.9) months, and on average drove 12,979 (SD, 10,785) km per year. All subjects reported normal sleep quality and duration on the nights before the test days with no differences observed between the two test conditions. Results from the study are summarized in Table 1. There were no significant order effects (caffeinated–decaffeinated coffee versus decaffeinated–caffeinated coffee) or time-of-testing effects (a.m. versus p.m.).Table 1Effects of caffeinated coffee in comparison to decaf on simulated driving performance and subjective sleepinessTimeDecaffeinated coffeeCaffeinated coffeeDriving test resultsStandard deviation of lateral position (cm)121.43 (4.37)22.11 (3.67)223.65 (5.90)24.13 (4.76)322.92 (4.61)21.08 (3.74)*423.69 (4.72)22.41 (4.37)*Standard deviation of speed (km/h)10.85 (0.44)0.88 (0.35)20.98 (0.51)1.1 (0.61)31.03 (0.72)0.78 (0.34)*41.15 (0.77)0.87 (0.56)*Mean lateral position (cm)1−18.04 (12.71)−18.03 (10.47)2−19.24 (12.60)−18.98 (9.98)3−18.63 (12.31)−20.16 (11.05)4−18.17 (11.54)−18.93 (10.80)Mean speed (km/h)195.40 (0.19)95.42 (0.21)295.46 (0.16)95.40 (0.26)395.44 (0.31)95.54 (0.18)495.53 (0.15)95.54 (0.25)Subjective driving assessmentsDriving quality19.75 (3.66)9.08 (4.10)29.01 (2.81)8.48 (3.46)39.70 (3.89)11.84 (2.82)*49.23 (3.02)10.60 (3.41)Mental effort15.33 (2.30)5.70 (2.47)25.84 (2.76)6.39 (2.50)35.89 (2.82)4.50 (2.36)*45.72 (2.38)4.90 (2.93)*Subjective sleepiness scoresKarolinska sleepiness scaleBaseline3.25 (0.94)3.33 (0.87)16.08 (1.67)5.83 (2.16)26.17 (1.95)6.29 (1.97)36.13 (2.11)4.21 (1.47)*45.79 (1.59)4.54 (1.86)*Mean (SD) is shown for each parameter. Driving quality ranges from 0 (“I drove exceptionally poorly”) to 20 (“I drove exceptionally well”). For mental effort, higher scores indicate higher effort; higher KSS scores indicate increased subjective sleepiness*p < 0.05 compared to decaf\n\nEffects of caffeinated coffee in comparison to decaf on simulated driving performance and subjective sleepiness\nMean (SD) is shown for each parameter. Driving quality ranges from 0 (“I drove exceptionally poorly”) to 20 (“I drove exceptionally well”). For mental effort, higher scores indicate higher effort; higher KSS scores indicate increased subjective sleepiness\n*p < 0.05 compared to decaf\n Driving test Figure 1 shows the effect of caffeinated coffee consumption on driving performance. No significant differences in SDLP were observed before the break. However, both in the first (F\n(1,23) = 5.8; p = 0.024) and in the second hour (F\n(1,23) = 6.4; p = 0.019) after the break, caffeinated coffee significantly reduced SDLP.Fig. 1Standard deviation of lateral position (SDLP). Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nStandard deviation of lateral position (SDLP). Asterisks indicate significant difference compared to placebo (p < 0.05)\nIn line, caffeinated coffee significantly reduced SD speed in the third (F\n(1,23) = 5.8; p = 0.024) and fourth hour (F\n(1,23) = 13.0; p = 0.001) of driving (see Fig. 2).Fig. 2Standard deviation of speed (SDS). Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nStandard deviation of speed (SDS). Asterisks indicate significant difference compared to placebo (p < 0.05)\nNo effects were found on mean speed or mean lateral position, confirming that subjects performed the test according to the instructions.\nFigure 1 shows the effect of caffeinated coffee consumption on driving performance. No significant differences in SDLP were observed before the break. However, both in the first (F\n(1,23) = 5.8; p = 0.024) and in the second hour (F\n(1,23) = 6.4; p = 0.019) after the break, caffeinated coffee significantly reduced SDLP.Fig. 1Standard deviation of lateral position (SDLP). Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nStandard deviation of lateral position (SDLP). Asterisks indicate significant difference compared to placebo (p < 0.05)\nIn line, caffeinated coffee significantly reduced SD speed in the third (F\n(1,23) = 5.8; p = 0.024) and fourth hour (F\n(1,23) = 13.0; p = 0.001) of driving (see Fig. 2).Fig. 2Standard deviation of speed (SDS). Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nStandard deviation of speed (SDS). Asterisks indicate significant difference compared to placebo (p < 0.05)\nNo effects were found on mean speed or mean lateral position, confirming that subjects performed the test according to the instructions.\n Subjective driving assessments Compared to decaffeinated coffee, caffeinated coffee improved subjective driving quality in the third hour of driving (F\n(1,23) = 10.5; p = 0.004), but not in the fourth (F\n(1,23) = 2.6; p = n.s.). Subjects indicated that the mental effort needed to perform the test after caffeinated coffee was significantly reduced in the third (F\n(1,23) = 11.4; p = 0.003) and fourth hour of driving (F\n(1,23) = 5.9; p = 0.023). In addition, drivers rated their driving quality as significantly more considerate, responsible, and safer in the caffeinated coffee condition (see Table 1).\nCompared to decaffeinated coffee, caffeinated coffee improved subjective driving quality in the third hour of driving (F\n(1,23) = 10.5; p = 0.004), but not in the fourth (F\n(1,23) = 2.6; p = n.s.). Subjects indicated that the mental effort needed to perform the test after caffeinated coffee was significantly reduced in the third (F\n(1,23) = 11.4; p = 0.003) and fourth hour of driving (F\n(1,23) = 5.9; p = 0.023). In addition, drivers rated their driving quality as significantly more considerate, responsible, and safer in the caffeinated coffee condition (see Table 1).\n Subjective sleepiness After the break with caffeinated coffee, drivers reported significantly lower sleepiness scores as compared to the break with decaffeinated coffee. This effect was significant both in the third (F\n(1,23) = 18.5; p < 0.001) and the fourth hour of driving (F\n(1,23) = 11.9; p = 0.002) (see Fig. 3).Fig. 3Karolinska sleepiness scale. Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nKarolinska sleepiness scale. Asterisks indicate significant difference compared to placebo (p < 0.05)\nAfter the break with caffeinated coffee, drivers reported significantly lower sleepiness scores as compared to the break with decaffeinated coffee. This effect was significant both in the third (F\n(1,23) = 18.5; p < 0.001) and the fourth hour of driving (F\n(1,23) = 11.9; p = 0.002) (see Fig. 3).Fig. 3Karolinska sleepiness scale. Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nKarolinska sleepiness scale. Asterisks indicate significant difference compared to placebo (p < 0.05)", "Figure 1 shows the effect of caffeinated coffee consumption on driving performance. No significant differences in SDLP were observed before the break. However, both in the first (F\n(1,23) = 5.8; p = 0.024) and in the second hour (F\n(1,23) = 6.4; p = 0.019) after the break, caffeinated coffee significantly reduced SDLP.Fig. 1Standard deviation of lateral position (SDLP). Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nStandard deviation of lateral position (SDLP). Asterisks indicate significant difference compared to placebo (p < 0.05)\nIn line, caffeinated coffee significantly reduced SD speed in the third (F\n(1,23) = 5.8; p = 0.024) and fourth hour (F\n(1,23) = 13.0; p = 0.001) of driving (see Fig. 2).Fig. 2Standard deviation of speed (SDS). Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nStandard deviation of speed (SDS). Asterisks indicate significant difference compared to placebo (p < 0.05)\nNo effects were found on mean speed or mean lateral position, confirming that subjects performed the test according to the instructions.", "Compared to decaffeinated coffee, caffeinated coffee improved subjective driving quality in the third hour of driving (F\n(1,23) = 10.5; p = 0.004), but not in the fourth (F\n(1,23) = 2.6; p = n.s.). Subjects indicated that the mental effort needed to perform the test after caffeinated coffee was significantly reduced in the third (F\n(1,23) = 11.4; p = 0.003) and fourth hour of driving (F\n(1,23) = 5.9; p = 0.023). In addition, drivers rated their driving quality as significantly more considerate, responsible, and safer in the caffeinated coffee condition (see Table 1).", "After the break with caffeinated coffee, drivers reported significantly lower sleepiness scores as compared to the break with decaffeinated coffee. This effect was significant both in the third (F\n(1,23) = 18.5; p < 0.001) and the fourth hour of driving (F\n(1,23) = 11.9; p = 0.002) (see Fig. 3).Fig. 3Karolinska sleepiness scale. Asterisks indicate significant difference compared to placebo (p < 0.05)\n\nKarolinska sleepiness scale. Asterisks indicate significant difference compared to placebo (p < 0.05)", "This study demonstrates that one cup of caffeinated coffee (80 mg caffeine) significantly improves driving performance and reduces driver sleepiness.\nBoth lane keeping (SDLP) and speed maintenance were improved up to 2 h after caffeine consumption. The effect on SDLP of caffeinated coffee, compared to placebo, is comparable to changes observed with a blood alcohol concentration of 0.05% (Mets et al. 2011b), i.e., the legal limit for driving in many countries, but in the opposite direction. Hence, the improvement by caffeinated coffee can be regarded as clinically relevant.\nThe magnitude of driving improvement observed after coffee consumption was comparable to the improvement seen in a driving study with Red Bull® Energy Drink using the same design and driving test (Mets et al. 2011a), and on-road driving studies showing improvement after administration of methylphenidate to patients with attention-deficit hyperactivity disorder (Verster et al. 2008).\nThe improvement in objective performance was accompanied by improvement in subjective assessments of sleepiness and driving performance. An average decrease of almost 2 points (out of 7) on the KSS scale was observed after the intake of caffeinated coffee as compared to decaffeinated coffee. The average KSS score was 6 (“some signs of sleepiness”) in the decaffeinated coffee condition compared to 4 (“rather alert”) in the caffeinated coffee condition. These findings are in agreement with the pharmacokinetic profile of caffeine (Tmax ≈ 30 min; T1/2 > 2 h), as well as with the known actions of caffeine as a sleepiness countermeasure with the ability to restore performance to baseline.\nUp to now, higher dosages of caffeine (150–250 mg, comparable to two to three cups of regular coffee) have been shown to be effective in counteracting sleep restriction (<5 h spent in bed) when driving in the early morning (Reyner and Horne 2000) and in the early afternoon (Horne and Reyner 1996; Reyner and Horne 1997). A moderate caffeine dosage (100 mg) decreased drifting out of lane and reduced subjective sleepiness in drivers who had slept for no more than 4 h (Biggs et al. 2007). Furthermore, caffeine (3 mg/kg, approximately 225 mg in a 75-kg adult) improved steering accuracy in non-fatigued volunteers (Brice and Smith 2001). Slow release caffeine capsules (300 mg) had similar effects (De Valck and Cluydts 2001). Interestingly, caffeine decreased lane drifting both in individuals who had spent 4.5 h in bed and in those who had spent 7.5 h in bed, while effects on speed maintenance, fatigue, and sleepiness were only observed after 4.5 h spent in bed (De Valck and Cluydts 2001). Two on-the-road driving studies on a public highway in France confirmed these findings and showed that relatively high dosages of caffeine (200 mg) improved nighttime driving both in young and in middle-aged drivers (Philip et al. 2006; Sagaspe et al. 2007).\nThe current results are in agreement with these studies, but further show that lower caffeine content found in one regular cup of coffee also significantly improves driving performance and reduces driver sleepiness. In addition, where studies find effects on lane keeping, the current study shows that speed maintenance is also affected, indicating more pronounced effects on vehicle control. The importance of this finding is evident, since it can be assumed that in order to refresh, most drivers consume only one cup of coffee during a break, instead of three or four. Driving simulator research has several limitations, which are discussed elsewhere (e.g., Mets et al 2011b; Verster and Roth 2011, 2012). Therefore, it is important to replicate and confirm our findings in an on-the-road driving study. Furthermore, subjects who were tested in the afternoon may have had an additional effect of the afternoon dip in the circadian rhythm. For this reason, each subject had his test days at the same time of day. However, no significant differences were found between subjects tested in the morning and those tested in the afternoon. Further studies could examine if a low dose of caffeine has similar effects on (professional) drivers who are sleep-restricted or shifted their day–night rhythm, since current studies have only been performed with higher dosages of caffeine.\nIn conclusion, the present study demonstrates that one cup of caffeinated coffee (80 mg caffeine) has a positive effect on continuous highway driving in non-sleep restricted, healthy volunteers." ]

[ "introduction", "materials|methods", null, null, null, null, null, null, "results", null, null, null, "discussion" ]

[ "Caffeine", "Automobile driving", "Fatigue", "Sleepiness" ]

[ 245, 236, 322, 302, 116, 95, 255, 148, 114 ]

[ "driving", "coffee", "test", "caffeine", "sleepiness", "caffeinated", "break", "caffeinated coffee", "compared", "subjects" ]

[ "caffeine sleepiness countermeasure", "break consuming caffeinated", "caffeine conclusion", "effects caffeine", "consuming caffeinated coffee" ]