Datasets:

marianna13
/

litarch

+{"file": "taid24_NBK299250/APPB.nxml", "text": "This appendix is not available from the publisher at this time.", "pairs": [], "interleaved": []}
+{"file": "taid24_NBK299250/AG1.nxml", "text": "\nAppendix A Acknowledgments\n\n\nAppendix B Data Abstraction Form\n\n\nAppendix C Evidence Tables\n", "pairs": [], "interleaved": []}
+{"file": "taid24_NBK299250/CH1.nxml", "text": "Background\nObstructive sleep apnea (OSA, sometimes characterized as obstructive sleep apnea syndrome) is now recognized to be a significant and serious public health problem. Researchers have estimated that approximately 2 percent to 4 percent of middle-aged women and men, respectively, have this condition; the majority (approximately 80 percent to 90 percent in one study1) remain undiagnosed. Undiagnosed and thus untreated, OSA is associated with significant morbidity and mortality, including excessive daytime somnolence, increased risk of automobile crashes, hypertension, cardiovascular disease, stroke, and metabolic abnormalities.\nEffective treatment modalities are available, primarily nocturnal continuous positive airway pressure (CPAP) and in some instances surgical procedures or dental appliances. These treatments, however, are expensive and have potential side effects, so they should not be applied without an accurately established diagnosis of OSA.\nMedical history and physical examination can provide an estimate of the likelihood of OSA, but they are not sufficient to establish the diagnosis or its severity. Therefore, patients suspected of having this condition must be evaluated with a diagnostic test that can provide a significant increase (\u201crule in\u201d) or decrease (\u201crule out\u201d) in the likelihood of the condition so that proper management can be implemented.\nUsing the reference standard polysomnography (PSG), which is a facility-based diagnostic intervention, is expensive. Because of limited facilities, waiting time for studies after the diagnosis is suspected on clinical grounds has been excessive in many areas of the country. Thus, various groups have attempted to develop portable technologies that can accurately classify patients as either having a very low likelihood of OSA and thus not need a PSG or having a very high likelihood of OSA and for whom management with a CPAP titration or other procedure should be initiated. The goal is to reduce the need for expensive laboratory testing while increasing the rapidity of diagnosis and initiation of appropriate management.\nThe first step in determining whether a portable monitoring device can achieve this goal is to determine its accuracy in characterizing the presence and severity of sleep-disordered breathing events relative to the reference standard PSG in a controlled study. This is the focus of most of the research papers published on portable monitors for OSA, and it was the main factor considered in the last systematic evidence review (described below). However, other considerations are also important in the overall assessment of whether the current technologies will be cost effective and provide adequate accuracy of diagnosis if applied to a large population of patients in unattended settings. These issues are examined further in the Discussion chapter of this report.\nEvaluating the Role of Home Testing of Obstructive Sleep Apnea\nAs noted, various portable devices have been developed over the past decade that are meant to be used as screening tools or replacements for the labor-intensive, complex, expensive, laboratory- or facility-based PSG for the evaluation of patients suspected of having OSA.\nThe American Sleep Disorders Association classified monitors used in diagnostic testing for sleep apnea into four types.2 Attended PSG, Type 1, is the gold standard, and the portable monitors fall into three types (2, 3, and 4) with fewer physiologic signals monitored in each subsequent type. The levels are briefly defined below to clarify the differences between them:\n\nType 1: Measures, at a minimum, eight channels \u2500 electroencephalogram (EEG), electro-oculogram (EOG), electrocardiogram (ECG), chin electromyogram (EMG), airflow, respiratory \u201ceffort,\u201d oxygen saturation (SaO2), and body position; it is attended in a laboratory setting.Type 2: Monitors a minimum of seven channels including EEG, EOG, chin EMG, ECG or heart rate, airflow, respiratory effort, and SaO2. This allows for sleep staging and measurement of total sleep time, and this information can be used to determine in the number of sleep-disordered breathing events per hour of sleep (e.g., the apnea/hypopnea index).Type 3: Includes a minimum of four channels and must monitor at least two channels of respiratory movement or respiratory movement and airflow to define an event; generally, no EEG signals are monitored.Level 4: Includes only one or two channels of physiologic signals and generally uses only one channel (either SaO2 or airflow) to define a sleep-disordered breathing event; no EEG signals are monitored.\n\nType 1: Measures, at a minimum, eight channels \u2500 electroencephalogram (EEG), electro-oculogram (EOG), electrocardiogram (ECG), chin electromyogram (EMG), airflow, respiratory \u201ceffort,\u201d oxygen saturation (SaO2), and body position; it is attended in a laboratory setting.\nType 2: Monitors a minimum of seven channels including EEG, EOG, chin EMG, ECG or heart rate, airflow, respiratory effort, and SaO2. This allows for sleep staging and measurement of total sleep time, and this information can be used to determine in the number of sleep-disordered breathing events per hour of sleep (e.g., the apnea/hypopnea index).\nType 3: Includes a minimum of four channels and must monitor at least two channels of respiratory movement or respiratory movement and airflow to define an event; generally, no EEG signals are monitored.\nLevel 4: Includes only one or two channels of physiologic signals and generally uses only one channel (either SaO2 or airflow) to define a sleep-disordered breathing event; no EEG signals are monitored.\nThe Original 2002 RTI-UNC Systematic Evidence Review\nIn 2002, RTI International (RTI) and the University of North Carolina at Chapel Hill (UNC) completed a systematic review of published articles on home diagnostic testing for sleep apnea in collaboration with three professional organizations: the American Academy of Sleep Medicine (ASSM), the American College of Chest Physicians (ACCP), and the American Thoracic Society (ATS). The RTI-UNC team conducted the literature search and prepared the evidence tables. Members of the three organizations analyzed the data and prepared three key publications.\nThe full evidence review was subsequently published in the journal Chest in October 2003.3 An article in the same issue of Chest measured agreement between diagnostic devices.4 Finally, new recommended clinical practice guidelines (practice parameters) from these three professional organizations for the use of portable monitoring devices in the investigation of suspected obstructive sleep apnea (SOSA) in adults was published in Sleep.5 An executive summary of the systematic review and practice parameters listed above was published in 2004 in the American Journal of Respiratory and Critical Care Medicine.6\nThe 2002 review compared all of the portable devices to the Type 1, attended, in-laboratory PSG. Although the accuracy of a single-night PSG in determining the presence or absence of clinically significant OSA does have certain limitations (see Discussion section of this report), to date no better standard has emerged. Therefore, this review also compares portable home monitoring devices to the PSG.\nThe most widely used measure to define the presence and severity of OSA by PSG is the apnea/hyponea index (AHI). Apneas in this calculation are events with complete cessation of airflow; hypopneas are events with decreases in airflow without complete cessation but with associated decreases in SaO2 or EEG arousals (or both) depending on the definition used by the researcher or clinician. The AHI is the number of disordered breathing events per hour of sleep calculated from the total number of apneas and hypopneas. Most studies use a lower cutoff level for the AHI to define the presence of OSA by PSG; increasing levels of AHI indicate increasing severity of OSA.\nIf portable monitoring does not allow for determination of sleep time, then the AHI cannot be calculated. Instead, researchers calculate the number of disordered breathing events per hour in bed or per hour of monitoring time and report this as the respiratory disturbance index (RDI). The RDI can be compared with the same measure calculated from a PSG.\nIn our original (2002) evidence report, we included 51 studies. Of these, four studies were of Type 2 devices, 12 of Type 3, and 35 of the Type 4. The joint ATS/ACCP/AASM summary of the evidence review and practice parameters stated that data were not adequate to recommend the clinical use of Type 2 portable monitors in either attended or unattended settings. Neither sensitivity nor specificity data were available; moreover, the number of studies of Type 2 devices was small. Overall, the level of evidence was low. Some Type 3 monitors appeared to be potentially acceptable in the attended laboratory setting, but six limitations were noted:\n\nCareful review of raw data is necessary (e.g., manual or a combination of automatic and manual scoring).The devices should be used only in populations that have been studied (e.g., those in a sleep clinic population) and should not be applied as generalized screening or in populations with significant comorbidity such as chronic obstructive pulmonary disease or congestive heart failure.AHI in these devices tends to underestimate the PSG-defined AHI because these devices do not measure sleep time.Symptomatic patients with a nondiagnostic or negative Type 3 study should undergo definitive evaluation to determine the cause of symptoms; if a sleep disorder requiring a sleep study is still a clinical consideration, then a full attended PSG should be used.Patients with a positive Type 3 study need a subsequent PSG if CPAP titration is needed.Type 3 portable monitors are not recommended for split-night studies.6\n\nCareful review of raw data is necessary (e.g., manual or a combination of automatic and manual scoring).\nThe devices should be used only in populations that have been studied (e.g., those in a sleep clinic population) and should not be applied as generalized screening or in populations with significant comorbidity such as chronic obstructive pulmonary disease or congestive heart failure.\nAHI in these devices tends to underestimate the PSG-defined AHI because these devices do not measure sleep time.\nSymptomatic patients with a nondiagnostic or negative Type 3 study should undergo definitive evaluation to determine the cause of symptoms; if a sleep disorder requiring a sleep study is still a clinical consideration, then a full attended PSG should be used.\nPatients with a positive Type 3 study need a subsequent PSG if CPAP titration is needed.\nType 3 portable monitors are not recommended for split-night studies.6\nIn essence, these Type 3 devices were not recommended for unattended use in the home. Type 4 devices were not recommended for diagnostic use or to assess the probability that a patient may or may not have OSA.\nOverall, portable devices were not recommended either for general screening or for patients with certain comorbid conditions. Manual scoring by \u201cphysicians with specific sleep training and familiarity with the devices and their limitations\u201d was recommended rather than use of the automated scoring available for some portable devices.\nPurpose of this Updated Evidence Report\nRTI is now assisting the Agency for Healthcare Quality and Research (AHRQ) to develop a summary of the available clinical and scientific evidence on home diagnosis of OSA since the last review in 2002. The Centers for Medicare and Medicaid Services (CMS) commissioned AHRQ to provide a technology assessment in preparation for a Medicare Coverage Advisory Committee (MCAC) meeting on September 28, 2004, at which the MCAC will review the evidence.\nCMS has provided one key question for this evidence review, with two subquestions based on the results of the main question:\n\nHow does the diagnostic test performance of unattended portable multi-channel home sleep testing compare to facility-based polysomnography in the diagnosis of obstructive sleep apnea?If unattended portable multi-channel home sleep testing is as effective as polysomnography in the diagnosis of obstructive sleep apnea, which parameters of sleep and cardiorespiratory function (i.e., sleep staging, body position, limb movements, respiratory effort, airflow, oxygen saturation, electrocardiogram) are required?If unattended portable multi-channel home sleep testing is as effective as polysomnography in the diagnosis of obstructive sleep apnea, what conditions (i.e., patient education, technician support) are required so that it is done correctly in the home?\n\nHow does the diagnostic test performance of unattended portable multi-channel home sleep testing compare to facility-based polysomnography in the diagnosis of obstructive sleep apnea?\nIf unattended portable multi-channel home sleep testing is as effective as polysomnography in the diagnosis of obstructive sleep apnea, which parameters of sleep and cardiorespiratory function (i.e., sleep staging, body position, limb movements, respiratory effort, airflow, oxygen saturation, electrocardiogram) are required?\nIf unattended portable multi-channel home sleep testing is as effective as polysomnography in the diagnosis of obstructive sleep apnea, what conditions (i.e., patient education, technician support) are required so that it is done correctly in the home?\nIn this updated review we report on 12 studies published since the last full review in 2002 that present data on the use of portable monitoring devices for evaluation of OSA.\nChapter 2 describes our methods for this update. Chapter 3 presents our findings from the updated evidence review and Chapter 4 discusses the implications of our results. Acknowledgments can be found in Appendix A; our data abstraction form is in Appendix B; evidence tables (one per study) appear in Appendix C.", "pairs": [], "interleaved": []}
+{"file": "taid24_NBK299250/CH4.nxml", "text": "Introduction\nIn interpreting these findings from our updated systematic review of the effectiveness of portable monitoring devices, as judged against that of in-facility polysomnography (PSG) for diagnosing obstructive sleep apnea (OSA), we call attention to three important questions. First, do the results of the more recently published studies using similar monitoring technologies differ significantly from those of the studies in the previous review done for the American Academy of Sleep Medicine (ASSM), the American College of Chest Physicians (ACCP), and the American Thoracic Society (ATS).3\u20135 Specifically, do recent studies of the use of portable monitors in the home indicate that accuracy in diagnosing OSA is better, worse, or unchanged from the levels of accuracy reported in previously published studies? Second, do new portable monitor technologies demonstrate significantly different effectiveness in accurately detecting OSA? Third, do these recently published studies have important limitations affecting either internal or external validity?\nIn updating our original evidence report, we presented in Chapter 3 results and findings from 12 studies that met our inclusion criteria. These 12 studies, which we fully reviewed, fell into four categories: Type 3 devices used in laboratory settings (four studies9,10,15,16); Type 3 devices tested in homes whether or not they were also tested in facilities (two studies15,16); Type 4 devices in laboratory settings (six studies12\u201314,17,18,20); and Type 4 devices tested in homes (whether or not in facilities, three studies11,14,19). We focused on the five studies with in-home testing, because the questions we were asked concerned the effectiveness of unattended monitoring in the home.\nWe discuss here the collective knowledge base from this newer work in the context of what was known after the previous review, drawing attention to the critical issues in evaluating portable monitors generally and the issues of evaluating performance against the reference standard (PSG). These are essentially issues of the internal validity of the studies we included. In addition, we consider factors that may affect the clinical usefulness of portable monitoring for OSA in the Medicare patient population; this is a question of the external validity or generalizability of these studies to the population of interest to CMS.\nCritical Issues in Evaluating Portable Monitor Studies\nThe effectiveness of home portable monitors is judged chiefly in terms of how well they correctly identify patients with and without clinically significant OSA. \u201cClinically significant\u201d turns on issues of severity, measured by (for instance) an apnea/hypopnea index (AHI) or a respiratory disturbance index (RDI). \u201cCorrect identification\u201d rests on whether the likelihood of a correct diagnosis or classification (as to the presence or absence of OSA) is better after the portable device test than it is before the test. Assessing this body of evidence requires appreciation of several limitations of published studies, as these problems place a ceiling on the level of internal validity (i.e., the extent to which these studies are free of systematic bias).\nPortable Testing in Laboratories or in Homes\nMost articles provided only comparisons of the results from portable monitoring done simultaneously with full PSG in the laboratory, i.e., \u201ca side-by-side\u201d study. Although this type of study does control for night-to-night variability in the important AHI (a measure of severity) observed by PSG, it does not provide information on the performance of the equipment unattended in patients\u2019 homes where usually no technical support is available (except, perhaps via a telephone help line).\nData Loss\nData loss in this context means that some or all portable monitoring measures for individual patients originally entered into the studies were not recorded in usable form, meaning that those patients had to be excluded from some or all analyses. Reported data loss in the home studies considered for this update ranged from 3 percent to 33 percent (in a subgroup). Moreover, at the upper end of this data loss range, many experts doing systematic reviews of clinical literature would probably regard the studies as being of only poor quality and perhaps not give them further consideration.\nOnly one home study directly compared the data loss rate between hook-up for the portable equipment by technicians and that by patients;11 the investigators reported a 7-percent loss for technician hook-up and 33-percent loss for patient hook-up. In the study using PAT technology,14 three of 28 (11 percent) of initial home studies set up by the patients were \u201crejected\u201d whereas only three of 102 (3 percent) of studies done in the laboratory with equipment hooked up by a technician were \u201crejected.\u201d\nThus, although only a limited amount of evidence in the reports reviewed addresses this issue, data loss appears to be greater when the patient performs the hook-up of the equipment.\nManual or Automated Scoring\nManual versus automated scoring remains a significant question. Some portable monitors report a score for respiratory disturbance derived from an automated scoring algorithm. Others provide data in a format which was later scored manually by a technician or physician.\nFour studies provided insights into this issue. Reported agreement between the PSG value of apneas plus hypopneas per hour of time in bed and that derived from the portable equipment was better for manual than automated scoring for Embletta data.15 The kappa statistic is a measure of agreement between two results beyond chance: the larger the value the better the agreement. In this study, kappa was 0.62 for manual scoring in the studies done simultaneously in the laboratory and 0.54 for those done in the home studies. The kappa statistics for the automated scoring were 0.28 and 0.10 indicating a poor agreement beyond chance. In another study, the investigators did not compare results for automated and manual scoring directly but did report area under the curve (AUC) for ROC curves derived from each method of scoring the portable data.11 The larger the AUC, the better the performance of the test. In this study, the AUC was 0.89 for manual scoring and 0.86 for automated scoring. A third study provided a Bland-Altman plot of agreement between AHI from the PSG and that from automated scoring of the data from the NovaSom portable equipment.16 Although the mean difference appeared to be small, the limits of agreement estimated by \u00b1 2 standard deviations (SD) were \u00b1 60 events per hour. Finally, the differences in AHI between the PSG and that for the Merlin portable equipment scored manually or with an automated method were reported in a simultaneous in laboratory study.10 The mean difference was \u22124 \u00b114 for the manual scoring and \u221224 \u00b1 30 for the automated scoring.\nIn short, results for automated methods of scoring respiratory events appear to provide less agreement with PSG results than do manual methods. That is, for portable monitors with data recordings that could be scored either manually or with automated algorithms (or both), manual scoring produced results with better concordance with PSG results. Several studies, however, apparently used only automated scoring of the portable device results, and one used a proprietary automatic scoring system.\nAs a related matter, all studies apparently used experienced technicians or physicians to interpret the PSG and the portable monitoring data when the latter could be scored manually, although not all studies reported on interpreter qualifications directly. This level of experience is critical to allow adequate detection of artifacts or situations in which patient data may be questionable.\nCore Aspects of Clinical Management\nAppropriate management of patients with sleep-related breathing disorders requires consideration of the clinical and physiologic consequences of sleep-disordered breathing, not just a classification of severity by AHI or RDI. The overall clinical diagnosis of OSA \u2500 and even more importantly the decision on the appropriate management of OSA when it is present \u2500 depends on additional factors. Therefore, evidence that an in-home portable test can measure an RDI as well as a PSG can in the same patient is not adequate in and of itself to evaluate the usefulness of that test clinically.\nThe medical history and examination supply crucial information in this regard, but also very important is the apparent impact of sleep-disordered breathing on sleep quality. This can include, for example, the amount of slow wave or \u201cdeep\u201d sleep and rapid eye movement (REM) sleep and the frequency of brief arousals and full awakenings. The portable monitors in this review did not include electroencephalogram (EEG) or electro-oculogram (EOG) signals, so investigators could not perform sleep staging or score EEG arousals or awakenings associated with respiratory or other events.\nSome portable monitors use other methods for estimating sleep (e.g., actigraphy). Although these measures may have significant overall correlation with total EEG sleep during the PSG, they cannot be used to stage sleep or to detect brief cortical arousals (\u201cmicro\u201d arousals) associated with respiratory or other events.\nOther Clinical Issues\nSpontaneous arousals may have an important impact on sleep quality. They can be correctly identified by manual scoring of a full PSG, but they may be missed or possibly inappropriately scored as primary respiratory-related events by automated portable scoring algorithms. The latter situation could arise if such arousals changed ventilation, for instance by increasing ventilation and oxygen saturation (SaO2) from the typically lower baseline (\u201casleep\u201d) values to the normally higher \u201cawake\u201d values followed by a reduction in ventilation and \u201cdesaturation\u201d to the sleeping baseline values again. This is especially problematic in patients with underlying heart or lung disease with relatively low baseline SaO2 levels.\nOther conditions producing arousals, such as Restless Legs Syndrome/Periodic Limb Movements of Sleep, are not detectable by portable monitors without electromyography (EMG) signals. These arousals may also be associated with changes in SaO2 and misinterpreted as primary respiratory-related events as described for spontaneous arousals.\nIf REM sleep is not appropriately detected, the clinicians\u2019 ability to assess the clinical impact of sleep-disordered breathing is reduced. For many patients with OSA, the severity of sleep-disordered breathing is much greater during REM sleep. The overall AHI may indicate a lower severity than clinicians might consider clinically pertinent, especially if the amount of REM sleep during the study is lower than the amount usually experienced during sleep at home unaffected by monitoring.\nLikewise, nonobstructive hypoventilation events with oxygen desaturations are common during REM sleep, especially in persons with underlying lung disease, obesity, or neuromuscular weakness. Without information on sleep stage and respiratory effort, these events may be misinterpreted as obstructive events consistent with OSA. Conversely, using time in bed or recording time rather than EEG-documented sleep time to calculate the respiratory disturbance index (RDI) may produce a spuriously low value if the patient has significant time awake during the study. Sleep efficiency is a measure of the amount of time the person was asleep during the testing, for example, from the time the patient was to go to sleep, \"lights out,\" to the awakening time \"lights on.\" Of the studies that took place in the laboratory only (side by side comparisons), there were two studies that reported sleep efficiencies ranging from 65.0 (\u00b1 20.9%, indicating a wide variability among the patients) to 76 (\u00b1 2%, standard error from the mean). One study reported that their home study sample had a mean sleep efficiency, as measured in the laboratory, of 82 (\u00b1 1% standard error from the mean). Thus, if the actual time were not known, the time used to calculate the time in bed could be 25% to 35% longer than actual sleep time.\nCo-existing Conditions\nComorbid conditions can have a significant impact on sleep and sleep-related respiratory abnormalities. Patients with underlying lung or heart disease are more likely to show significant oxygen desaturations with nonobstructive hypopneas. Also, periodic breathing with a central apnea component (Cheyne-Stokes breathing) is common in patients with significant heart failure or atrial fibrillation; if adequate measures of respiratory effort are not available, then these conditions may be mistaken for OSA.\nGenerally, the studies we reported here said little about coexisting conditions in these patient populations. One group noted that they had excluded patients using oxygen, those with certain current medications, and those who were \u201cmedically unstable.\u201d13\nNo in-home study gave information on whether comorbid conditions appeared to affect the rate of false-positive or false-negative cases. In sum, evidence is insufficient to draw conclusions about the effect of comorbidity on the effectiveness of portable devices as compared with that for PSG.\nSimilarity of Update and Prior Findings\nThe in-laboratory simultaneous studies in this review, which used technologies identical or similar to those in the studies reviewed by us for the AASM/ACCP/ATS, produced sensitivity and specificity results for diagnosing OSA similar to those previously reported. That is, the newer studies produced no meaningful changes in the level or quality of evidence for the effectiveness for home monitoring devices in diagnosing OSA. Chapter 1 summarized the earlier findings and conclusions.\nThree in-laboratory studies in this review used portable monitoring technologies identical or similar to those in the 2002 review;9,10,12 two produced sensitivity and specificity results for diagnosing OSA that were similar to those previously reported, and one reported slightly better results. We found no significant overall change in the level or quality of evidence for the effectiveness for home monitoring devices in diagnosing OSA when used in an attended laboratory setting.\nThe four in-home studies that employed technologies similar or identical to those in the AASM/ACCP/ATS review had sensitivity and specificity values similar to those previously reported.11,15,16,19 Thus, the newer studies yielded no major information that would change the previous basic conclusions about portable devices used in the home.\nThree studies used a device based on a technology not considered in the prior review \u2500 namely, peripheral arterial tonometry (PAT). Two were done only in the laboratory setting and did not demonstrate significantly better accuracy in diagnosing OSA than other devices.13,17 The only in-home study using PAT produced likelihood ratios indicating little effect of the test results on estimates of the probability of OSA.14\nIssues of Internal Validity of Reviewed Studies\nFactors that may falsely lower the apparent accuracy of home portable monitoring studies in detecting OSA were discussed in the original AASM/ACCP/ATS review. These include night-to-night variability, the lack of complete consensus on the definition of clinically significant respiratory events during a PSG, different sleep architecture and/or body position in the sleep laboratory different from that in the home, and the lack of clinical validity of a single AHI or RDI threshold to define the presence or absence of OSA when comparing PSG and home study results.\nThese issues are all still relevant to the studies reviewed in this update. In some instances these factors may have accounted for some of the difference in results observed between portable monitoring in the home and the in-laboratory PSG, thereby causing a spurious lowering of the reported diagnostic accuracy of the home study.\nOne way to interpret results of tests such as this is to determine whether, relative to the pretest probability, the testing changes the post-test probability (likelihood) that a condition, in this case OSA, is present or not. This determination is based, in part, on whether the test results are positive or negative. Likelihood ratios (LRs) for a test result from the device under investigation provide a convenient way to make this determination, Positive LRs (LR+) reflect the ratio of the percentage of patients with a disease correctly identified by the test result (true positives) to the percentage of patients without the disease who are misidentified (false positives); negative LRs (LR-) reflect the ratio of the percentage of patients with a disease who have a negative test result (false negatives) to the percentage of patients without the disease who have a negative test result (true negatives).\nLRs (positive or negative) of 1.0 could be said to reflect a useless test. LR+ values of 2 to 5 show modest effect of the test in ruling in a diagnosis (i.e., concluding the disease is present); those at the 5 to 10 level have strong effect. LR- values between 0.2 and 0.5 show a modest impact on ruling out a diagnosis (i.e., concluding that the disease is absent); those between 0.1 and 0.2, a moderate impact, and those less than 0.1, a strong effect. The original review and the accompanying paper on comparing diagnostic tests4 review these points in detail in the context of OSA.\nMixed evidence (one good study, two fair, one poor) showed that Type 3 devices, when used in an attended laboratory setting, can modestly increase or modestly decrease the likelihood of an accurate OSA diagnosis relative to pretest probabilities. Somewhat more questionable evidence (2 fair, 4 poor) suggests that Type 4 devices, when used in the attended laboratory setting, can also modestly increase and decrease the likelihood of correctly determining the presence or absence of OSA. Data loss (i.e., missing data for individual patients) of 10 percent to 20 percent should be expected in home studies.\nOne good study and one poor quality study indicate Type 3 monitoring devices, when used in unattended home settings, can both modestly increase and modestly decrease the probability of OSA relative to the probabilities before the testing.15,16 LRs in this context can be improved by using different thresholds for RDI on the portable test to increase (rule in) and decrease (rule out) the probability of OSA. However a relatively high proportion of patients (up to 40 percent) may then be \u201cunclassifiable\u201d and need further testing.\nThree studies (two fair, one poor quality) using Type 4 monitoring devices were done unattended in the home.11,14,19 These authors reported data indicating that these devices can modestly reduce the probability of OSA (e.g., reach a LR- of 0.2 or lower).11,19 Less evidence exists that these devices can increase the probability that OSA is present (e.g., reach or produce a LR+ of greater than 5). Overall, data loss of 3 percent to 20 percent was reported and up to 33 percent when the patient did the hook-up themselves. The only in-home study using nonstandard Type 4 monitoring technology (PAT14) received a fair quality rating and produced LRs indicating little effect of the test results on estimates of the probability of OSA.\nOnly one study addressed how clinical management decisions based on the portable test results would compare with those based on PSG results.11 In 10 of 44 cases (23 percent), the clinical decision on whether CPAP was indicated based on the interpretation of the portable results differed from that based on interpretation of the PSG. In six of the cases for which the management decisions differed between PSG and home study results, the home study was deemed a false-negative or a false-positive result.11 In the remaining 4 cases, the interpretation of the home studies agreed with that of the PSG about the presence of OSA, but the severity grading differed significantly and recommended therapy differed.\nThese studies were done in a highly selected patient populations with high prevalence rates of OSA by PSG (50 percent to 75 percent) and proportions of males (approximately 75 percent to 90 percent). The overall prevalence of comorbid conditions in the patients studied was typically not stated, and no characterization of comorbid conditions was given for those patients who were incorrectly classified (false positives and false negatives) or were \u201cunclassifiable\u201d by portable testing.\nExternal Validity or Generalizability\nOverall Generalizability\nThe published studies of portable monitors have several limitations in regard to generalizability of their results to less highly selected patients (i.e., populations with characteristics different from those of the samples studied). Studies in this update were done on patients identified as having a high pretest probability of having OSA by PSG; prevalence of OSA by PSG was generally 50 percent or greater, and in some cases the PSG prevalence rate approached 80 percent. Most of these studies had a majority of males and did not report the proportion of patients with significant comorbid conditions such as chronic obstructive pulmonary disease, asthma, congestive heart failure, or neuromuscular disorders.\nGeneralizability to the Medicare Beneficiary Population\nNo study specifically targeted an elderly population. Apart from that, applying findings in these studies to the Medicare population has several limitations over and above the issues raised with respect to overall generalizability and internal validity.\nFirst, the prevalence of reported excessive daytime sleepiness in the elderly is known to be high. For example, of 4,578 noninstitutionalized Medicare enrollees, 20 percent reported being \u201cusually sleepy in the day time.\u201d22 The prevalence of medical conditions associated with poor quality sleep and daytime sleepiness for reasons other than OSA is higher in the elderly than in younger populations. In one sample of 18,980 subjects, the prevalence of Restless Legs Syndrome increased with age (ages 40 to 49 years, 4.7 percent; ages 60 to 69 years, 8.3 percent; ages 70 to 79 years, 8.9 percent).23\nSecond, \u201cclassical\u201d signs and symptoms of SDB are less closely associated with OSA in the elderly than in other age groups. A study of 5,615 community-dwelling adults found that \u201cas age increased the magnitude of associations of SDB and body habitus, snoring and breathing pauses decreased\u201d (p. 893).24 The authors concluded that breathing pauses and obesity may be \u201cparticularly insensitive\u201d for identifying SDB in the elderly. Thus, the prevalence of true OSA in Medicare patients referred for sleep studies because of signs and symptoms such as obesity and excessive daytime sleepiness may be significantly different from the prevalence in the patient groups in the studies reviewed.\nFinally, we see no indications that adequate sleep study data are more difficult to obtain in unattended home studies in the elderly than in other groups. The Sleep Heart Health Study reported no significant effects of age or sex on the overall success rate in obtaining interpretable data,25 although these investigators did observe a significant decrease in the duration of adequate abdominal \u201ceffort\u201d signals in the elderly. Thus, the proportion of interpretable data obtainable in home studies on Medicare patients is likely to be similar to that from the patients in the study populations.\nHowever, the rates of false-positive and false-negative tests may differ because of the higher prevalence of comorbid conditions in the Medicare population.22,23 This may also occur because the prevalence of OSA in Medicare patients who are referred for study because of excessive daytime sleepiness may be lower than the OSA prevalence in the studied patients.1 This may also be true for patients referred by physicians without special training in sleep medicine, but no explicit information on this is available. Most reported studies derived patients from specialized sleep clinics; only one study in the review explicitly stated that patients were referred by \u201ccommunity physicians.\u201d16\nSummary\nThe key questions posed for this review asked how portable sleep testing devices compared to PSG in diagnosing OSA and, assuming equivalent effectiveness, what sleep and physiologic factors and what patient and technician conditions were important to measure or have in place.\nWe updated an earlier evidence report with a systematic literature search and in-depth review of 12 articles that met inclusion criteria for addressing these questions. Most articles covered technologies (of Types 3 and 4 only) that had been examined previously; three dealt with a single new technology. Most studies involved testing home devices against PSG in a sleep laboratory; five studies either wholly or partially examined home devices in the home.\nThis newer body of evidence does not materially change earlier findings regarding in-home devices for diagnosing OSA.3 Of the five in-home studies, two15,16 were done with Type 3 devices (one of good quality, and one of poor quality) and three11,14,19 with Type 4 devices (two fair and one poor quality). Information from the one in-home study of a new technology, of fair quality, gave little support for concluding that it was better than any other Type 4 device.\nChoices of cutoffs for determining OSA by AHI or RDI differed widely across these studies, making cross-study comparisons impossible. The better studies yielded sensitivity and specificity values (or LRs) that provided modest changes the probability of OSA over the pretest probability. In studies that directly compared automated vs. manual scoring of data from home monitoring devices, manual scoring correlated better with data from laboratory PSG.\nImproved sensitivities and specificities could be achieved by using two different thresholds to define results of a home test as \u201cpositive\u201d or \u201cnegative\u201d for OSA, but this left a large proportion of patients with \u201cindeterminate\u201d results. The clinical decision about the need for CPAP therapy based on the interpretation of the home study differed from that based on the PSG in 23 percent of cases in the one study which reported this type of comparison. No studies reported the effect of co-morbid conditions on the sensitivity or specificity of the home testing. The overall proportion of home studies with inadequate data averaged 13 percent but in one study data loss was as high as 33 percent when the patients performed the hookup compared to only 3 percent when hookup was done by technicians. Mean age for patients in the home studies ranged from 41.4 to 52.7 years. No information was presented on whether the sensitivity/specificity or the rate of data loss was associated with patient age. More evidence is needed to reach conclusions about the effect of comorbidities, age and patient versus technician performed hookup on the overall effectiveness of home studies in diagnosing OSA compared to an in-laboratory PSG.", "pairs": [], "interleaved": []}
+{"file": "taid24_NBK299250/APPA.nxml", "text": "We thank Elise Berliner, Ph.D., Task Order Officer for the Agency for Healthcare Research and Quality for her encouragement and support throughout this project. We extend our appreciation to Loraine Monroe, the RTI-UNC Evidence-based Practice Center document preparation specialist; to Tammeka Swinson, B.A., the EPC project manager, for considerable assistance with the literature search and abstracting tasks; and to Mark Howell, RTI librarian, for help with all searches.", "pairs": [], "interleaved": []}