Virtual Poster Presentation Assignment

CLINICAL RESEARCH STUDY

Agreement Between Results of Home Sleep Testing for Obstructive Sleep Apnea with and Without a Sleep Specialist

R. Nisha Aurora, MD, MHS,a Nirupama Putcha, MD, MHS,a Rachel Swartz, BA,a Naresh M. Punjabi, MD, PhDa,b

Departments of aMedicine and bEpidemiology, Johns Hopkins University, Baltimore, Md.

Funding: Sup stitutes of Health.

Conflict of Int Respironics for re

Authorship: A the manuscript.

Requests for r PhD, Division of Hopkins Universi Baltimore, MD 21

E-mail address

0002-9343/$ -see http://dx.doi.org/1

ABSTRACT

BACKGROUND: Obstructive sleep apnea is a prevalent yet underdiagnosed condition associated with car- diovascular morbidity and mortality. Home sleep testing offers an efficient means for diagnosing obstructive sleep apnea but has been deployed primarily in clinical samples with a high pretest probability. The present study sought to assess whether obstructive sleep apnea can be diagnosed with home sleep testing in a nonreferred sample without involvement of a sleep medicine specialist. METHODS: A study of community-based adults with untreated obstructive sleep apnea was undertaken. Misclassification of disease severity according to home sleep testing with and without involvement of a sleep medicine specialist was assessed, and agreement was characterized using scatter plots, Pearson’s correlation coefficient, Bland-Altman analysis, and the k statistic. Analyses were also conducted to assess whether any observed differences varied as a function of pretest probability of obstructive sleep apnea or subjective sleepiness. RESULTS: The sample consisted of 191 subjects, with more than half (56.5%) having obstructive sleep apnea. Without involvement of a sleep medicine specialist, obstructive sleep apnea was not identified in only 5.8% of the sample. Analyses comparing the categorical assessment of disease severity with and without a sleep medicine specialist showed that in total, 32 subjects (16.8%) were misclassified. Agreement in the disease severity with and without a sleep medicine specialist was not influenced by the pretest probability or daytime sleep tendency. CONCLUSION: Obstructive sleep apnea can be reliably identified with home sleep testing in a nonreferred sample, irrespective of the pretest probability of the disease. � 2016 Elsevier Inc. All rights reserved. � The American Journal of Medicine (2016) 129, 725-730 KEYWORDS: Diagnosis; Home sleep testing; Obstructive sleep apnea

It is well established that identifying modifiable risk factors for chronic diseases and intervening early can help slow disease progression and possibly even reduce associated morbidity and mortality.1 Obstructive sleep apnea, which

ported by a grant (HL075078) from the National In-

erest: NMP has received grant support from Resmed and search unrelated to this article. ll authors had access to and played a role in the writing

eprints should be addressed to Naresh M. Punjabi, MD, Pulmonary, Critical Care, and Sleep Medicine, Johns ty School of Medicine, 5501 Hopkins Bayview Circle, 224. : [email protected]

front matter � 2016 Elsevier Inc. All rights reserved. 0.1016/j.amjmed.2016.02.015

affects approximately 24% of adult men and 9% of adult women in the general population,2 is an independent risk factor for incident hypertension and cardiovascular dis- ease.3,4 Yet, more than 80% of those with obstructive sleep apnea remain undiagnosed.5,6 Given the available evidence that treatment of obstructive sleep apnea can decrease blood pressure and possibly mitigate adverse cardiovascular end- points,7,8 early case identification has clinical and public health merit. The advent of home sleep testing has been a useful addition to the diagnostic armamentarium for obstructive sleep apnea because it offers a simple, objective, and economical means for assessing those that may be affected. However, current guidelines recommend that home sleep testing be restricted to only those patients with a high pretest probability for obstructive sleep apnea and that a

726 The American Journal of Medicine, Vol 129, No 7, July 2016

sleep trained provider be involved in the review and inter- pretation of the recording. Although self- or interview- administered questionnaires, such as the Berlin9 and the STOP-BANG10 questionnaires, are useful in identifying those with a high pretest probability for obstructive sleep apnea, these instruments have varying levels of sensitivity,

CLINICAL SIGNIFICANCE

� Identification of obstructive sleep apnea using home sleep testing in a non- referred sample without involvement of a sleep specialist has not been routinely performed.

� Obstructive sleep apnea was missed in less than 6% when home sleep testing was conducted without a sleep specialist.

� Pretest probability of obstructive sleep apnea or subjective sleepiness does not alter diagnostic accuracy of home sleep testing when a sleep specialist is not involved.

specificity, and predictive value depending on the specific popula- tion in which they are used, the underlying prevalence of disease in the study sample, and disease severity.11,12 More importantly, self-reported data are inherently limited because many of those affected with obstructive sleep apnea underreport symptoms of snoring and cessation of breathing during sleep, or they do not have a bedpartner to aid in identifying signs or symptoms. Clinical pre- diction models and morphometric measures, which represent the next layer of techniques to assess risk, are simple to implement but have limited sensitivity and spec- ificity.11,12 Thus, many patients with obstructive sleep apnea

remain undiagnosed and are at risk for the associated cardio- metabolic sequelae. The present study sought to evaluate whether assessment of obstructive sleep apnea can be per- formed with home sleep testing independent of a sleep medicine specialist in a nonreferred sample that may have an admixture of low and high pretest probability for obstructive sleep apnea. It was hypothesized that there would be a high degree of agreement between the results derived from home sleep testing for obstructive sleep apnea with and without involvement of a sleep medicine specialist.

METHODS

Study Sample and Screening Assessments Case identification of undiagnosed obstructive sleep apnea was undertaken for the present study in a general commu- nity sample from the local BaltimoreeWashington area. Recruitment was based on newspaper advertisements and mailed invitations using a commercially available electronic mailing list. A total of 827 respondents expressed interest in participating in the study and were screened by a telephone interview that was conducted by trained research assistants using a structured questionnaire. Eligibility criteria included age between of 21 and 80 years and no prior treatment for obstructive sleep apnea. Those with a history of prior upper airway surgery for obstructive sleep apnea (n ¼ 45) or treatment with an oral appliance (n ¼ 22) or positive pres- sure therapy (n ¼ 557) were not eligible. Shift work, a preference for an early or delayed sleep schedule suggestive

of a circadian rhythm disorder, or restless legs syndrome were also considered exclusionary (n ¼ 12). Thus, 636 people responding to the study advertisements were considered as screen failures. A total of 191 eligible subjects completed a telephone questionnaire that included de- mographic information, such as age, sex, and race, as well

as the self-reported medical his- tory on prevalent hypertension, high cholesterol, asthma, chronic obstructive lung disease, coronary artery disease, and hypothyroid- ism. The Epworth Sleepiness Scale, which assesses average sleep propensity with 8 questions on the likelihood of dozing in various situations on a 4-point scale (0-3), was administered to all those that enrolled in the study. An Epworth Sleepiness Scale score of �11 out of a maximum 24 is considered subjectively sleepy.13 Finally, all those con- sented to the study completed the Berlin Questionnaire, a validated 3-part questionnaire used to assess obstructive sleep apnea risk (low vs high).9 The research protocol

was approved by the Johns Hopkins University institutional review board on human research (IRB approval number NA_00036672).

Assessment of Obstructive Sleep Apnea To examine differences in detection of obstructive sleep apnea and severity classification with and without involve- ment of a sleep medicine specialist, the ApneaLink Plus (ResMed, San Diego, Calif), a type III portable monitoring device, was used for home sleep testing in volunteers who qualified for the study. Apnealink Plus (type III), which has been previously validated,14 was given to study subjects by research assistants who were not registered poly- somnologists. Nasal airflow was recorded with a nasal cannula connected to a pressure transducer. Pulse oximetry was used to assess oxyhemoglobin saturation, and respira- tory effort was measured with a pneumatic sensor attached to an effort belt. At least 4 hours of interpretable recording time was required for inclusion in the study. The scoring software for ApneaLink Plus was configured to identify disordered breathing events according to standard criteria as follows. Apneas were identified if there was a 90% or greater reduction in airflow for at least 10 seconds, and hypopneas were identified if there was a �30% reduction in airflow for at least 10 seconds that was associated with an oxyhemoglobin desaturation of at least 4%. The apneaehypopnea index, the disease-defining metric for obstructive sleep apnea, was the number of apneas and hypopneas per hour of recording time. Disease severity was

Aurora et al Home Sleep Testing for Obstructive Sleep Apnea 727

categorized according to commonly used clinical cut points, as follows: <5.0 events per hour (normal), 5.0-14.9 events per hour (mild), 15.0-29.9 events per hour (moderate), and �30 events per hour (severe). To compare diagnostic clas- sification with and without sleep specialist involvement, the apneaehypopnea index was derived as follows. First, the apneaehypopnea index was determined without a sleep medicine specialist by subjecting each sleep recording to the Apnealink Plus software to determine the presence and severity of obstructive sleep apnea. The sleep recording was also independently scored by a board-certified sleep physi- cian (RNA) who was blinded to any of the other results but used the aforementioned criteria for scoring apneas and hypopneas. Specifically, the sleep physician did not have access to results of the automated ApneaLink Plus assess- ments. Thus, two independent measures of disease severity were derived.

Statistical Analysis Agreement in disease severity with and without involve- ment by a sleep medicine specialist was characterized as follows. Using the apneaehypopnea index as a continuous measure, initial assessments of agreement used bivariate scatter plots and Pearson’s correlation coefficient. In addition, Bland-Altman analysis15 and the average differ- ence in the apneaehypopnea index with and without involvement of a sleep specialist were used to characterize the bias between the 2 assessments. The k statistic was also used to examine whether classification of disease severity (normal, mild, moderate, and severe) differed when comparing the apneaehypopnea index obtained with and without a sleep specialist. Additional analyses were con- ducted to assess whether the agreement varied as a function of pretest probability of obstructive sleep apnea risk or daytime sleepiness. All analyses were performed by using

Figure 1 Scatterplot (left) and Bland-Altman plot ( and without sleep specialist involvement.

the SAS 9.1 statistical software system (SAS Institute, Cary, NC).

RESULTS The study sample consisted of 191 eligible subjects with a median age of 53.2 years (25th-75th percentile, 47.3-59.9). The mean body mass index was 28.4 kg/m2 (standard de- viation, 4.9 kg/m2), and men constituted 63.4% of the sample. Self-reported prevalent medical conditions in the sample were as follows: hypertension (24.6%), high cholesterol (9.2%), asthma (4.7%), and coronary artery disease (0.5%). The overall failure rate with home sleep testing was 2.2%, which represents the sample proportion that required retesting due to less than 4 hours of inter- pretable recording of oxygen saturation and airflow signal. Loss of the oximetry signal was the most common reason for failure. A majority of the subjects (73.8%) did not report subjective sleepiness (Epworth Sleepiness Scale score <11). According to the responses on the Berlin Question- naire, 93 subjects (48.7%) and 98 subjects (51.3%) were categorized as having a low and high pretest probability of obstructive sleep apnea, respectively. Using the sleep medicine specialist as the reference, obstructive sleep apnea was present is 56.5% of the sample, and the median apneaehypopnea index was 6.2 events per hour (25th-75th percentile, 2.3-15.6). Of those with obstructive sleep apnea, 51.9% had mild, 34.2% had moderate, and 13.9% had se- vere disease. Figure 1 shows the apneaehypopnea index values derived with and without involvement of a sleep medicine specialist and the associated Bland-Altman plot. Overall, the correlation between the 2 derived apneaehypopnea indices was high (r ¼ 0.96; 95% confi- dence interval [CI], 0.95-0.97), and the average bias was low (D apneaehypopnea index, 1.8 events per hour; 95% CI, 1.1-2.4).

right) of the apneaehypopnea index (AHI) with

Figure 3 Percent concordance with and without sleep specialist involvement stratified by subjective sleepiness and obstructive sleep apnea (OSA) risk according to the Berlin Questionnaire.

728 The American Journal of Medicine, Vol 129, No 7, July 2016

On the basis of the apneaehypopnea index derived without a sleep medicine specialist, obstructive sleep apnea was not identified in only 5.8% of the sample. Analyses comparing the categorical assessment of disease severity with and without involvement of a sleep medicine specialist showed that, in total, 32 subjects (16.8%) were misclassified (Figure 2). Of these, 10 subjects (5.2%) were classified as having less severe disease, and 22 subjects (11.5%) were classified as having greater disease severity using the sleep specialist as the reference, yielding a k statistic of 0.75 (95% CI, 0.69-0.79), indicating substantial agreement between the 2 assessments.

To examine whether the level of agreement between the 2 diagnostic approaches varied as a function of subjective sleepiness or the pretest probability of obstructive sleep apnea, subgroup analyses were conducted. Stratified ana- lyses that examined misclassification of disease severity as a function of subjective sleep tendency showed that of the 50 subjects characterized as subjectively sleepy with an Epworth Sleepiness Scale score �11, 7 (14.0%) were misclassified (Figure 3). Disease severity class was underestimated in 10.0% of sleepy subjects when a sleep medicine specialist was not involved. Similarly, in those without subjective sleepiness (Epworth Sleepiness Scale score <11), 12.1% of the 141 subjects were classified to

Figure 2 Classification of obstructive sleep apnea (OSA) severity (top) and concordance (bottom) with and without sleep specialist involvement.

a less severe obstructive sleep apnea category when there was no sleep medicine specialist involved. The difference in disease severity misclassification by subjective sleepiness was not statistically significant (10.0% vs 12.1%, P ¼ .43). Analyses were also conducted to examine whether there was heterogeneity in misclassification as a function of pretest probability as assessed by the Berlin Questionnaire. As with self- reported sleepiness, misclassification of disease severity did not differ between high-risk and low-risk Berlin categories (P ¼ .83). In the high-risk group (n ¼ 98), 16 subjects overall (16.3%) were misclassified, whereas in the low-risk group (n ¼ 93), 16 subjects (17.2%) were mis- classified (Figure 3).

DISCUSSION The present study indicates that case identification of obstructive sleep apnea using home sleep testing in a non- referred sample can be effectively accomplished irrespective of the pretest probability. Although overall misclassification in identification and disease classification with and without a sleep specialist was 16.8%, the diagnosis was missed in less than 6% of our community-based sample when a sleep specialist was not involved in any capacity. Moreover, home sleep testing without a sleep specialist appropriately iden- tified obstructive sleep apnea irrespective of whether sub- jects were categorized as low-risk or high-risk for obstructive sleep apnea, as determined by the Berlin Ques- tionnaire, or whether there was subjectively sleepiness. Thus, risk stratification before portable sleep testing does not alter the diagnostic accuracy of home sleep testing that does not include a sleep specialist.

The results reported herein extend the available evidence on home sleep testing for obstructive sleep apnea. Currently there is a dearth of information on the use of home sleep

Aurora et al Home Sleep Testing for Obstructive Sleep Apnea 729

testing in a nonreferred sample that includes subjects with high and low pretest probability for obstructive sleep apnea. On the basis of current recommendations, the preponder- ance of the available literature has assessed the use of home sleep testing primarily in clinic-based populations with a high pretest probability of the disease. In the only available study exploring the use of home sleep testing in subjects with a low pretest probability,16 it was demonstrated that 3 nights of home sleep testing can effectively and economi- cally diagnose obstructive sleep apnea. However, given that a sleep clinic sample was assessed and the enrolled patients were not subjected to a standardized pretest assessment of risk, the results of that study are not generalizable to other patient samples. In contrast, the results of our study are more applicable to a larger pool of patients that may benefit from diagnostic sleep testing because of the nonreferred nature of our sample, of which approximately half was at low risk. Although the role of healthcare providers not trained in sleep medicine has also been previously examined, the primary focus of the published data has been on stream- lining management in patients that are referred for clinical care.17 Thus, the results of the present study are distinct and expand the limited empirical evidence base by demon- strating that obstructive sleep apnea can be reliably diag- nosed without a sleep-trained provider even in those with a low pretest probability, the group most likely to have false negatives and consequently remain undiagnosed.

The use of automated diagnostic tools to improve screening and diagnostic efficiency in clinical practice is not novel. An example that has leveraged such tools is screening for breast cancer with mammography. Previous large-scale prospective studies and clinical trials have demonstrated that computer-aided analysis can reduce the need for specialized expertise for screening mammograms and help improve breast cancer detection.18-20 A similar conceptual framework might also be considered in the early identifi- cation of other chronic diseases, such as obstructive sleep apnea. Given that cardiovascular disease remains the leading cause of morbidity and mortality,21 it is reasonable to identify key potential risk factors and address them pre- emptively. The prevalence of obstructive sleep apnea in those affected by hypertension, heart failure, coronary artery disease, stroke, and atrial fibrillation is alarmingly high, with estimates ranging from 12% to 83% depending on the un- derlying disorder.22,23 Furthermore, the prevalence of obstructive sleep apnea in type 2 diabetes, another estab- lished risk factor for cardiovascular disease, is estimated to be 50%-80%.24 Thus, it is reasonable to consider testing for those with a low clinical suspicion but with associated comorbidities such as type 2 diabetes and heart disease, given that treatment of obstructive sleep apnea with positive airway pressure therapy has been shown to improve car- diovascular outcomes.7,25 Moreover, the noted association between motor vehicle accidents and obstructive sleep ap- nea26 is a major public safety concern further highlighting the need to screen those at risk. Thus, the ability to economically and expeditiously identify obstructive sleep

apnea in various patient population subsets by home sleep testing has important implications on healthcare costs, public safety, and medical outcomes. These implications are potentially even more substantial if simple tools for case identification can be deployed by healthcare providers of any specialty.

The present study has several strengths and limitations that merit discussion. First, the fairly large sample size from a community-based setting spanning the full spectrum of disease severity increases the generalizability of the reported findings. Second, the inclusion of subjects with both high and low pretest probability of obstructive sleep apnea, as well as sleepy and nonsleepy subjects, substantiates the utility of home sleep testing as a diagnostic tool with broad clinical utility. Third, there is a dearth of information regarding the use of home sleep testing in asymptomatic or low-risk individuals. The present investigation helps fill this knowledge gap in that particular subgroup. Limitations of our study include the use of a single expert who reviewed the raw recordings. Variability among sleep experts in their visual assessment of the recordings is well documented27-29

and can affect the observed misclassification. Moreover, there are a large number and types of home sleep testing devices, each having a distinct approach for identifying disordered breathing events. Thus, results from a single device, as used in the present study, cannot be extrapolated to other devices, and the heterogeneity of performance across devices cannot be determined. Moreover, the present study did not examine treatment prescription which may require participation of a physician experienced in sleep medicine. Nonetheless, it has been previously demonstrated that treatment-related outcomes for obstructive sleep apnea are similar when comparing a model that has a sleep specialist with one that does not.17 Finally, it is important to recognize that home sleep testing with portable monitoring can yield false-negative results compared with a full- montage, in-laboratory sleep study because the severity assessment is based on total recording time and not total sleep time. These limitations notwithstanding, the findings of the present study demonstrate that testing for obstructive sleep apnea with portable sleep monitoring may be effec- tively conducted irrespective of the pretest probability. There is little doubt that early treatment for symptomatic obstructive sleep apnea is warranted,30-32 and home moni- toring by those without specialty training in sleep medicine is feasible and can increase access to diagnosis and early case identification. The use of an objective test that is simple to attain, even in those at low risk, represents a paradigm shift in testing for obstructive sleep apnea.

References 1. Grundy SM. Cardiovascular and metabolic risk factors: How can we

improve outcomes in the high-risk patient? Am J Med. 2007;120(9 suppl 1):S3-S8.

2. Young T, Palta M, Dempsey J, et al. The occurrence of sleep- disordered breathing among middle-aged adults. N Engl J Med. 1993;328(17):1230-1235.

730 The American Journal of Medicine, Vol 129, No 7, July 2016

3. Gottlieb DJ, Yenokyan G, Newman AB, et al. Prospective study of obstructive sleep apnea and incident coronary heart disease and heart failure: The Sleep Heart Health Study. Circulation. 2010;122(4): 352-360.

4. O’Connor GT, Caffo B, Newman AB, et al. Prospective study of sleep- disordered breathing and hypertension: The Sleep Heart Health Study. Am J Respir Crit Care Med. 2009;179(12):1159-1164.

5. Kapur V, Strohl KP, Redline S, et al. Underdiagnosis of sleep apnea syndrome in U.S. communities. Sleep Breath. 2002;6(2):49-54.

6. Young T, Evans L, Finn L, et al. Estimation of the clinically diagnosed proportion of sleep apnea syndrome in middle-aged men and women. Sleep. 1997;20(9):705-706.

7. Marin JM, Carrizo SJ, Vicente E, et al. Long-term cardiovascular outcomes in men with obstructive sleep apnoea-hypopnoea with or without treatment with continuous positive airway pressure: An observational study. Lancet. 2005;365(9464):1046-1053.

8. Marin JM, Agusti A, Villar I, et al. Association between treated and untreated obstructive sleep apnea and risk of hypertension. JAMA. 2012;307(20):2169-2176.

9. Netzer NC, Stoohs RA, Netzer CM, et al. Using the Berlin Question- naire to identify patients at risk for the sleep apnea syndrome. Ann Intern Med. 1999;131(7):485-491.

10. Chung F, Yegneswaran B, Liao P, et al. STOP questionnaire: A tool to screen patients for obstructive sleep apnea. Anesthesiology. 2008;108(5):812-821.

11. Pang KP, Terris DJ. Screening for obstructive sleep apnea: An evidence-based analysis. Am J Otolaryngol. 2006;27(2):112-118.

12. Ramachandran SK, Josephs LA. A meta-analysis of clinical screening tests for obstructive sleep apnea. Anesthesiology. 2009;110(4):928-939.

13. Johns MW. A new method for measuring daytime sleepiness: The Epworth sleepiness scale. Sleep. 1991;14(6):540-545.

14. Dawson A, Loving RT, Gordon RM, et al. Type III home sleep testing versus pulse oximetry: Is the respiratory disturbance index better than the oxygen desaturation index to predict the apnoea-hypopnoea index measured during laboratory polysomnography? BMJ Open. 2015;5(6): e007956.

15. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1(8476): 307-310.

16. Guerrero A, Embid C, Isetta V, et al. Management of sleep apnea without high pretest probability or with comorbidities by three nights of portable sleep monitoring. Sleep. 2014;37(8):1363-1373.

17. Chai-Coetzer CL, Antic NA, Rowland LS, et al. Primary care vs specialist sleep center management of obstructive sleep apnea and

daytime sleepiness and quality of life: A randomized trial. JAMA. 2013;309(10):997-1004.

18. Gilbert FJ, Astley SM, Gillan MG, et al. Single reading with computer- aided detection for screening mammography. N Engl J Med. 2008;359(16):1675-1684.

19. Gromet M. Comparison of computer-aided detection to double reading of screening mammograms: Review of 231,221 mammograms. AJR Am J Roentgenol. 2008;190(4):854-859.

20. Morton MJ, Whaley DH, Brandt KR, et al. Screening mammograms: Interpretation with computer-aided detectioneprospective evaluation. Radiology. 2006;239(2):375-383.

21. Murphy SL, Xu J, Kochanek KD. Deaths: Final data for 2010. Natl Vital Stat Rep. 2013;61(4):1-117.

22. Bradley TD, Floras JS. Obstructive sleep apnoea and its cardiovascular consequences. Lancet. 2009;373(9657):82-93.

23. Gonzaga C, Bertolami A, Bertolami M, et al. Obstructive sleep apnea, hypertension and cardiovascular diseases. J Hum Hypertens. 2015;29(12):705-712.

24. Foster GD, Sanders MH, Millman R, et al. Obstructive sleep apnea among obese patients with type 2 diabetes. Diabetes Care. 2009;32(6): 1017-1019.

25. Gottlieb DJ, Punjabi NM, Mehra R, et al. CPAP versus oxygen in obstructive sleep apnea. N Engl J Med. 2014;370(24):2276-2285.

26. Tregear S, Reston J, Schoelles K, et al. Obstructive sleep apnea and risk of motor vehicle crash: Systematic review and meta-analysis. J Clin Sleep Med. 2009;5(6):573-581.

27. Rosenberg RS, Van HS. The American Academy of Sleep Medicine inter-scorer reliability program: Sleep stage scoring. J Clin Sleep Med. 2013;9(1):81-87.

28. Rosenberg RS, Van HS. The American Academy of Sleep Medicine Inter-scorer Reliability program: Respiratory events. J Clin Sleep Med. 2014;10(4):447-454.

29. Suzuki M, Saigusa H, Chiba S, et al. Discrepancy in polysomnography scoring for a patient with obstructive sleep apnea hypopnea syndrome. Tohoku J Exp Med. 2005;206(4):353-360.

30. Marshall NS, Wong KK, Liu PY, et al. Sleep apnea as an independent risk factor for all-cause mortality: The Busselton Health Study. Sleep. 2008;31(8):1079-1085.

31. Punjabi NM, Caffo BS, Goodwin JL, et al. Sleep-disordered breathing and mortality: A prospective cohort study. PLoS Med. 2009;6(8): e1000132.

32. Young T, Finn L, Peppard PE, et al. Sleep disordered breathing and mortality: Eighteen-year follow-up of the Wisconsin sleep cohort. Sleep. 2008;31(8):1071-1078.