We wished to determine if obstructive sleep apnea (OSA) is associated with increased left ventricular mass (LVM) and impaired left ventricular diastolic function (LVDF) independently of coexisting obesity, hypertension (HTN), and diabetes mellitus (DM). Patients without primary cardiac disease, referred for evaluation of OSA (n = 533), had overnight polysomnography and Doppler echocardiography while awake. Patients were divided, according to the apnea–hypopnea index (AHI), into an OSA group (AHI ⩾ 5/h, n = 353) and a non-OSA group (AHI < 5/h, n = 180). In men, LVM was greater in the OSA group (98.9 ± 25.6 versus 92.3 ± 22.5 g/m, p = 0.023) despite exclusion of those with HTN and DM. A similar trend was noted in women. Regression analysis revealed that LVM was correlated with body mass index (BMI) ( β = 0.480, p < 0.0005), age ( β = 0.16, p = 0.001), and the presence of HTN ( β = 0.137, p = 0.003) in men and with BMI ( β = 0.501, p < 0.0005) in women, but not with AHI or oxygen saturation during sleep. The ratio of peak early filling velocity to peak late filling velocity (E/A), an index of LVDF, was similar in both groups (1.28 ± 0.32 versus 1.34 ± 0.31, p = 0.058); it was correlated with age ( β = − 0.474, p < 0.0005), but not with AHI or oxygen saturation during sleep. We conclude that OSA is not associated with increased LVM or impaired LVDF independently of obesity, HTN, or advancing age.
The development of cardiovascular disease is one of the most serious long-term complications of untreated obstructive sleep apnea (OSA). Potential cardiovascular complications of OSA include systemic hypertension (1), pulmonary hypertension and right heart failure (2), cardiac arrhythmias (3), and coronary artery disease (4). There are conflicting data in the literature on the association between OSA and left ventricular hypertrophy. Although some investigators have reported that patients with OSA develop left ventricular hypertrophy independently of hypertension (HTN) (5), others have found no difference in left ventricular mass (LVM) between patients with OSA and appropriate control subjects (6, 7). All of these study populations were relatively small, which prevented rigorous statistical analysis. Left ventricular hypertrophy, regardless of the underlying cause, is associated with increased cardiovascular morbidity and mortality (8). Left ventricular diastolic dysfunction has not been assessed in patients with OSA, but it is regarded as an early sign of myocardial disease and an important determinant of symptoms and clinical outcome in patients with cardiovascular disease (9, 10). However, patients with OSA often have coexisting disorders which have been associated with increased LVM and diastolic dysfunction such as obesity (11), HTN (12), and diabetes mellitus (DM) (13, 14). Consequently, the objective of our study was to determine whether OSA is associated with increased LVM or impaired diastolic function independently of these coexisting conditions.
We studied consecutive patients who were referred for evaluation of snoring and possible sleep apnea between April 1990 and December 1998. All patients came to the sleep laboratory for a diagnostic overnight sleep study. Before the sleep study, each patient completed a questionnaire regarding medical and sleep history and current medications. This information was subsequently reviewed by a physician during a follow-up visit to the sleep clinic. Additional data that was collected on the night of the sleep study included patients' demographics and anthropomorphic measurements. On the morning after the sleep study, each patient had Doppler echocardiography performed. Both the sonographer and the reporting cardiologist were blinded to the patient's sleep study findings.
Overnight polysomnography was performed on all patients. This included two-channel electroencephalogram (EEG) (C3-A2, C4-A1), electrooculogram (EOG), submental electromyogram (EMG), electrocardiogram (ECG) and heart rate, airflow, respiratory effort, and arterial oxygen saturation. (See online data supplement.) Sleep stage was scored manually according to the criteria of Rechtschaffen and Kales (15) and apneas and hypopneas were scored according to conventional criteria. (See online data supplement.)
M-mode, two-dimensional and Doppler echocardiography were performed from the standard parasternal and apical views in the resting state, in the supine or left lateral position. This provided measurements of LVM which was corrected for the patient's height (LVM/ht), left ventricular systolic function, and left ventricular diastolic function (LVDF), reflected by the ratio of peak early filling velocity to peak late filling velocity (E/A). (See online data supplement.)
All statistical testing was performed at a 95% significance level (alpha = 0.05). Transformations were performed on some variables (identified in the text) in order to make their distribution more symmetric and appropriate for our analysis. Student's t test was used to compare differences in means of continuous variables. Chi-square test was used to compare differences in proportions of categorical variables. Bivariate correlation analysis was used to rule out significant colinearity between variables, which was defined as a Pearson correlation coefficient greater than 80%. Multivariate regression analysis was used to identify determinants of LVM and E/A ratio and to evaluate the interaction between indices of OSA (apnea–hypopnea index) [AHI] and SaO2 during sleep) and known determinants of LVM. The strength of these relationships was expressed as the beta coefficient (β) and p value.
All patients in whom technical difficulty precluded accurate interpretation of the echocardiographic results were excluded from the study. This accounted for 9.4% of our echocardiograms (92 studies) which were evenly divided between patients with and without OSA. The remaining 883 patients were eligible for enrollment in our study protocol. Patients who were considered to have a primary cardiac abnormality were excluded from subsequent analysis. This consisted of those who had a history of coronary artery disease (angina, myocardial infarction, or coronary artery bypass surgery) (n = 63), echocardiographic evidence of valvular heart disease (n = 306) or impaired left ventricular systolic function (estimated left ventricular ejection fraction < 60%) (n = 45). These diagnoses were not mutually exclusive. Of the 45 patients who were excluded because of impaired left ventricular systolic function, 22 had OSA (AHI ⩾ 5) and 23 did not have OSA (AHI < 5). The remaining 533 patients were divided into two groups (Table 1) based on the AHI during the overnight sleep study: those with an AHI of 5 or more per hour of sleep (OSA group) and those with an AHI less than 5 per hour (non-OSA group). Individuals were considered hypertensive or diabetic if they reported those conditions in the medical history or if they were receiving medications for the treatment of hypertension or diabetes (insulin or oral hypoglycemic agents), respectively. The majority (78%) of patients were men and 66% had OSA. The OSA group were older, heavier, had a higher prevalence of hypertension, and experienced more severe hypoxemia during sleep.
First, we compared LVM, corrected for height, between those with OSA and those without OSA (Table 2). We excluded subjects with coexisting HTN (n = 97) and DM (n = 23) because these disorders have the potential to increase LVM independently of OSA. These diagnoses were not mutually exclusive and nine patients had both HTN and DM. Analysis of the remaining 422 patients revealed that LVM was significantly greater in men with OSA than in those without OSA. Although a similar trend was seen in women, the difference was not statistically significant (p = 0.155). Both men and women with OSA were older, heavier, and experienced more severe hypoxemia during sleep than those without OSA. A small number of patients in the non-OSA group experienced hypoxemia during sleep because of coexisting respiratory disease. It is theoretically possible, although unlikely, that this could have reduced the difference in LVM between the two groups. This was addressed by repeating the analysis after exclusion of patients from the non-OSA group with nocturnal hypoxemia (defined as SaO2 < 90% for more than 10% of the total sleep time). As shown in Table E1 in the online data supplement, the results were unchanged.
|OSA (n = 224)||Non-OSA (n = 108)||OSA (n = 43)||Non-OSA (n = 47)|
|Age||46.1 ± 11.1†||43.5 ± 10.7||48.9 ± 10.9‡||41.4 ± 12|
|BMI kg/m2||32.3 ± 6.6§||28.5 ± 5||34.9 ± 8.2‖||31.3 ± 8.8|
|AHI/h||34 ± 25.5§||0.76 ± 1.4||25 ± 22§||0.71 ± 1.4|
|SaO2 < 90% (%TST)||19.6 ± 25.3§||6.57 ± 18.8||13.7 ± 22.5¶||5.54 ± 16.2|
|LVM/ht, g/m||98.9 ± 25.6**||92.3 ± 22.5||90.3 ± 21.12-164||83.9 ± 21.2|
In the second stage of our analysis, we examined the determinants of LVM in the entire population (OSA and non-OSA groups, n = 533). The potential determinants included age, body mass index (BMI), AHI, the percentage of time during sleep that oxygen saturation was less than 90% (SaO2 < 90%), and the presence of HTN and DM. Multivariate regression analysis was performed using transformed measurements, with LVM/ht as the dependent variable. As described in Methods, we also ensured that a strong degree of multicollinearity did not exist between the independent variables. In men, LVM/ht was positively correlated with BMI (β = 0.480, p < 0.0005), age (β = 0.16, p = 0.001), and the presence of HTN (β = 0.137, p = 0.003). No significant correlation was found between LVM/ht and SaO2 < 90% (β = 0.006, p = 0.915), AHI (β = −0.063, p = 0.229), and the presence of DM (β = 0.014, p = 0.759). In women, LVM/ht was positively correlated with BMI (β = 0.501, p < 0.0005), but not with age (β = 0.011, p = 0.920), AHI (β = 0.087, p = 0.365), SaO2 < 90% (β = 0.088, p = 0.457), or the presence of HTN (β = −0.021, p = 0.819) or DM (β = 0.054, p = 0.535).
In the third stage of our analysis, we looked for synergism between OSA and the potential determinants of LVM that were outlined in the previous paragraph. Once again, this analysis was performed on the entire population (OSA and non-OSA groups, n = 553). The AHI and SaO2 < 90% were used to reflect the severity of sleep apnea. A second linear regression was performed which evaluated the synergistic impact of AHI and SaO2 < 90% on the potential determinants of LVM. In men, the combination of SaO2 < 90% and DM were positively correlated with LVM (β = 0.315, p = 0.027) although, in isolation, neither was significantly correlated with LVM. To confirm this finding, a reduced model was created, which removed all interactions and main effects not involved in significant interactions from the equation; this model also demonstrated a significant interaction between SaO2 < 90% and DM. In women, we found no evidence of synergism between AHI or SaO2 < 90% and potential determinants of LVM.
The impact of OSA on LVDF, reflected by the E/A ratio, was also evaluated. Patients with other potential risk factors for impaired LVDF were excluded from this analysis, which included those with coronary artery disease (n = 63), impaired LV systolic function (echocardiographic evidence of LV ejection fraction < 60%) (n = 45), valvular heart disease (n = 306), DM (n = 16), HTN (n = 73), and tachyarrhythmia (heart rate > 80 at the time of echocardiography) (n = 356). These diagnoses were not mutually exclusive and many patients had more than one exclusion criterion. Of the remaining 399 patients, the majority (79%) were men and 52% had OSA (Table 3). Mean age was similar in both groups, and patients with OSA were heavier and had more severe hypoxemia during sleep. There was a marginal difference between the E/A ratio in the two groups, which did not reach statistical significance (p = 0.058). Potential determinants of the E/A ratio were evaluated by performing multivariate regression analysis between transformed values of the E/A ratio (dependent variable), and transformed values of BMI, age, AHI, and SaO2 < 90% (independent variables). Tests for multicollinearity were also performed which found no evidence that relationships between the independent variables may lead to inaccurate results. The E/A ratio was negatively correlated with age (β = −0.474, p < 0.0005), but was not significantly correlated with BMI (β = −0.142, p = 0.066), AHI (β = −0.040, p = 0.643), or SaO2 < 90% (β = −0.172, p = 0.066).
|n (M/F)||210 (178/32)||189 (137/52)|
|Age||47.4 ± 11.4||45.6 ± 11.8|
|BMI, kg/m2||32 ± 6.6†||30.3 ± 6.14|
|AHI/h||30.4 ± 24‡||0.52 ± 1.11|
|SaO2 < 90% (%TST)||16.9 ± 23.6‡||8.4 ± 18|
|E/A ratio||1.28 ± 0.32§||1.34 ± 0.31|
Because LVM was most strongly correlated with BMI, we performed an additional comparison between men with and without OSA who were not obese, were not taking medications, and did not have coexisting DM or HTN (see Table E2 in the online data supplement). There was no difference either in LVM or E/A ratio between nonobese men with and without OSA, which is consistent with our previous findings that OSA does not increase LVM independently of obesity. All data are available on the ATS website.
The primary objective of our study was to assess the effect of OSA on LVM and diastolic function. We tried to distinguish the effect of OSA from other potential risk factors for increased LVM and diastolic dysfunction which commonly coexist with OSA, including coronary artery disease, impaired LV systolic function, valvular heart disease, obesity, aging, HTN, and diabetes. We used both AHI and the degree of hypoxemia during sleep (SaO2 < 90%) as indices of the severity of OSA.
Our initial analysis showed that LVM, corrected for the patient's height, was higher in men with OSA than in those without OSA. A similar trend was seen in women but did not reach statistical significance, possibly because of the smaller sample size. Patients with OSA were heavier, older, and had a higher prevalence of HTN than patients without OSA. We postulated that these conditions, rather than sleep apnea itself, might have been responsible for increased LVM in the OSA group. This hypothesis was supported by the findings of our regression analysis, which showed that, in men, LVM was correlated with BMI, age, and the presence of HTN, but was not correlated with AHI or SaO2 < 90%. In women, LVM was correlated only with BMI. Furthermore, when we compared OSA and non-OSA subjects who were matched for BMI, we found no difference in LVM (Table E2). We also found minimal evidence of synergism between OSA and other potential determinants of increased LVM. In men, there was an interaction between SaO2 < 90% and DM, that is, whereas neither of these two factors alone was associated with LVM, their combined presence was correlated with increased LVM. We found similar results in our assessment of LVDF. The E/A ratio was marginally lower (indicating more impaired LVDF) in patients with OSA compared with those without OSA but the difference did not reach statistical significance. Furthermore, the E/A ratio was significantly correlated with age alone, which is a recognized phenomenon (16), but not with AHI or SaO2 < 90%.
The most likely explanation for the relative absence of effects of DM on LVM was the relatively low prevalence of DM in our patient population, and the fact that this prevalence was similar in patients with OSA (4.5%) and non-OSA (3.9%). In addition, we have previously observed that the impact of DM on LVM is predominantly through an interaction with obesity and age rather than a direct and independent effect (17). The small impact of HTN on LVM can be explained by a number of factors. First, the prevalence of HTN was relatively low in both the OSA group (20.7%) and the non-OSA group (13.3%). Second, the impact of treatment of HTN in our patients, before enrollment in the study, may have reduced our ability to find an association between the presence of HTN and echocardiographic abnormalities.
It has been suggested that OSA contributes to the development of LV hypertrophy (5). The proposed causes include associated changes in LV afterload, intermittent hypoxemia, and recurrent arousals during sleep. Left ventricular afterload increases during sleep in patients with OSA because of the combined effects of increased negative intrathoracic pressure, associated with attempted breathing against an occluded upper airway (18), and increased systemic blood pressure (19) associated with elevated sympathetic nervous system activity, hypoxemia, and arousal from sleep (20, 21). Forced inspiration against increased airway resistance during wakefulness (Mueller maneuver) raises aortic transmural pressure, thereby increasing aortic stiffness (22) and left ventricular systolic load (23). Left ventricular early diastolic filling has also been reported to decrease in normal subjects during acute intrathoracic pressure changes, including a Mueller maneuver (24). Virolainen and associates (25) also examined the effects of Mueller maneuver on LV hemodynamics and suggested that a reduced rate of LV relaxation contributes to impairment of LV early diastolic filling. Isovolumic relaxation time of the LV has also been shown to increase in the presence of either HTN-related (26) or age-dependent (27) increase in aortic stiffness.
Hedner and colleagues (5) were one of the first group of investigators to report that OSA causes LV hypertrophy in a study that compared 61 men with OSA and 61 male control subjects. The OSA group were heavier and 50% had systemic HTN. They reported that LVM, indexed to body surface area, was approximately 15% higher among normotensive OSA patients than in normotensive control subjects, despite comparison of subjects with matching BMI. However, only 10 subjects (all in the OSA group) had overnight polysomnography, which limits the reliability of the diagnosis of OSA, and none of the control subjects had an objective assessment of their breathing during sleep. More recently, Noda and coworkers (28) reported echocardiographic evidence of left ventricular hypertrophy in 50% of patients with an AHI > 20/h compared with 21.4% in those with an AHI < 20/h. However, the sample size was small (51 patients) and all patients with LV hypertrophy had systemic HTN. In contrast, Davies and coworkers (6) did not find a significant difference in LVM, determined by echocardiography, between 19 patients with OSA, 19 nonapneic snorers, and 38 control subjects matched for age, sex, BMI, smoking habits, and alcohol intake.
Our study has a number of unique features. First, the sample size is much greater than any comparable, previous study, which increases the power of our statistical analysis and enables us to compare the impact of multiple factors on LVM and diastolic function. Second, unlike most previous studies, a significant number of female subjects were enrolled, which enables us to make gender comparisons on our outcome measurements. Third, all of our subjects had comprehensive overnight polysomnography. This is critical to rule out occult OSA, which may be undetected by history and physical examination in up to 30% of obese subjects (29). Consequently, we are confident that our study compares snorers with and without sleep apnea. Finally, we enrolled a large number of nonhypertensive subjects, which enabled us to evaluate the impact of factors, other than HTN, on LVM and diastolic function.
Our study also has a number of limitations. First, the cross-sectional design does not permit us to evaluate the longitudinal impact of OSA on LVM and diastolic function. However, given the natural history of OSA and the severity of sleep apnea in many of our patients, it is likely that OSA was present for a significant period of time before diagnosis. Second, our study was performed on a clinic rather than a community-based population, which raises the possibility of referral bias. The male/female ratio of our study population was 4.6/1, which is similar to that in some community-based population studies (30). In addition, the clinical features of our primary outcome measurements, namely LVM and diastolic function, are subtle, if not imperceptible, and consequently were unlikely to influence referral of the patient for an overnight sleep study. Finally, studies on patients with OSA recruited from both the sleep clinic and community have reported similar age, BMI, and symptoms (31-33). Third, the impact of cardiac medications could not be evaluated because neither these medications nor the duration of therapy were controlled in the study. However, only a minority of our patients were taking cardiac medications, which was reflected by the low prevalence of HTN (Table 1), and most of them were not included in our comparison of LVM (Table 2) and diastolic function (Table 3) because we excluded those with HTN and primary cardiac disease. Consequently, we do not believe that cardiac medications had a significant impact on our results.
In summary, in a large clinic population, we have found that OSA does not independently increase LVM or impair LVDF. Although LVM is higher in patients with OSA, this is predominantly related to coexisting obesity, in addition to the effects of aging and the presence of HTN.
|1.||Peppard PE, Young T, Palta M, Skatrud JProspective study of the association between sleep-disordered breathing and hypertension. N Engl J Med342200013781384|
|2.||Bradley TD, Rutherford R, Grossman RF, Lue F, Zamel N, Moldofsky H, Phillipson EARole of daytime hypoxemia in the pathogenesis of right heart failure in obstructive sleep apnea syndrome. Am Rev Respir Dis1311985835839|
|3.||Tilkian AG, Guilleminault C, Schroeder JS, Lehrman KL, Simons FB, Dement WCSleep induced apnea syndrome: prevalence of cardiac arrhythmia and their reversal after tracheostomy. Am J Med631977348358|
|4.||Schafer H, Koehler U, Ewig S, Hasper E, Tasci S, Luderitz BObstructive sleep apnea as a risk marker in coronary artery disease. Cardiology9219997984|
|5.||Hedner J, Ejnell H, Caidahl KLeft ventricular hypertrophy independent of hypertension in patients with obstructive sleep apnea. J Hypertens81990941946|
|6.||Davies RJ, Crosby J, Prothero A, Stradling JRAmbulatory blood pressure and left ventricular hypertrophy in subjects with untreated obstructive sleep apnea and snoring, compared with matched control subjects, and their response to treatment. Clin Sci861994417424|
|7.||Hanly P, Sasson Z, Zuberi N, Alderson MVentricular function in snorers and patients with obstructive sleep apnea. Chest1021992100105|
|8.||Quinones MA, Greenberg BH, Kopelen HA, Koilpillai C, Limacher M, Shindler DM, Shelton BJ, Weiner DHEchocardiographic predictors of clinical outcome in patients with left ventricular dysfunction enrolled in the SOLVD registry and trials: significance of left ventricular hypertrophy. Studies of Left Ventricular Dysfunction. J Am Coll Cardiol35200012371244|
|9.||Shen WF, Tribouilloy C, Rey JL, Baudhuin JJ, Boey S, Dufosse H, Lesbre JPPrognostic significance of Doppler-derived left ventricular diastolic filling variables in dilated cardiomyopathy. Am Heart J124199215241533|
|10.||Xie GY, Berk MR, Smith MD, Gurley JS, De Maria ANPrognostic value of Doppler transmitral flow patterns in patients with congestive heart failure. J Am Coll Cardiol241992132139|
|11.||Sasson Z, Rasooly Y, Gupta R, Rasooly ILeft atrial enlargement in healthy obese: prevalence and relation to left ventricular mass and diastolic function. Can J Cardiol121996257263|
|12.||Schulman SP, Weiss JL, Becker LC, Gottleib SO, Woodruff KM, Weisfeldt ML, Gerstenblith GThe effects of antihypertensive therapy on left ventricular mass in elderly patients. N Engl J Med322199013501356|
|13.||Galderisi M, Anderson KM, Wilson PWF, Levy DEchocardiographic evidence of the existence of a distinct diabetic cardiomyopathy. Am J Cardiol6819918589|
|14.||Ruddy TD, Shumak SL, Liu PP, Barnie A, Seawright SJ, McLaughlin PR, Zinman BThe relationship of cardiac diastolic dysfunction to concurrent hormonal and metabolic status in type 1 diabetes mellitus. J Clin Endocrinol Metab661988113118|
|15.||Kales A, Rechtschaffen A. A Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects. National Institute of Neurological Disease and Blindness, Bethesda, MD. NIH Publication No. 204, 1968.|
|16.||Kuo LC, Quinones MA, Rokey R, Sartori M, Abinader EG, Zoghbi WAQuantification of atrial contribution to left ventricular filling by pulsed Doppler echocardiography and the effect of age in normal and diseased hearts. Am J Cardiol59198711741178|
|17.||Kuperstein R, Hanly P, Niroumand M, O'Connor C, Perl J, Thornley K, Sasson ZImportance of obesity and hypertension to the relation between age and left ventricular mass. J Am Coll Cardiol352000265A|
|18.||Buda AJ, Pinsky MR, Ingles NB, Daughters GT, Stinson EB, Alderman ELEffect of intrathoracic pressure on left ventricular performance. N Engl J Med3011979453459|
|19.||Hedner J, Enjell H, Sellgren J, Hedner T, Wallin G. Is high and fluctuating muscle nerve sympathetic activity in the sleep apnea syndrome of pathogenic importance for the development of hypertension? J Hypertens 1988;6(Suppl 4):S529–S531.|
|20.||Coccagna G, Mantovani M, Brignanai F, Parchi C, Lugaresi EContinuous recordings of pulmonary and arterial pressure during sleep in syndromes of hypersomnolence with periodic breathing. Bull Eur Physiopath Respir8197211591172|
|21.||Somers VK, Mark AL, Abbound FMInteraction of baroreceptor and chemoreceptor reflex control of sympathetic nerve activity in normal humans. J Clin Invest87199119531957|
|22.||Virolainen J, Kupari MAge-dependent increase in aortic stiffness during negative intrathoracic pressure in healthy subjects. Am J Cardiol711993878882|
|23.||Condos WR, Latham RD, Hoadley SD, Pasipoularides AHemodynamics of Mueller maneuver in man: right and left heart micromanometry and Doppler echocardiography. Circulation76198710231028|
|24.||Iliceto S, Dambrosio M, Sorino M, D'Ambrosio A, Amico A, Fiore T, Rizzon PEffects of acute intrathoracic pressure changes on left ventricular geometry and filling. Am Heart J1161988455465|
|25.||Virolainen J, Ventila M, Turto H, Kupari MEffect of negative intrathoracic pressure on left ventricular pressure dynamics and relaxation. J Appl Physiol791995455460|
|26.||Chang K-C, Tseng YZ, Kuo TS, Chen HIImpaired left ventricular relaxation and arterial stiffness in patients with essential hypertension. Clin Sci Lond871994641647|
|27.||Ochi H, Shimada T, Ikuma I, Morioka S, Moriyama KEffect of a decrease in aortic compliance on isovolumic relaxation period of the left ventricle in man. Am J Noninvasive Cardiol51991149154|
|28.||Noda A, Okada T, Yasuma F, Nakashima N, Yokota MCardiac hypertrophy in obstructive sleep apnea syndrome. Chest107199515381544|
|29.||Narkiewicz K, Philippe JH, van de Borne PJH, Cooley RL, Dyken ME, Somers VKSympathetic activity in obese subjects with and without obstructive sleep apnea. Circulation981998772776|
|30.||Kripke DF, Ancoli-Israel S, Klauber MR, Wingard DL, Mason WT, Mullaney DJPrevalence of sleep-disordered breathing in ages 40–60 years: a population-based survey. Sleep2019976576|
|31.||Olson LG, King MT, Hensley MJ, Saunders NAA community study of snoring and sleep-disordered breathing: health outcomes. Am J Respir Crit Care Med1521995717720|
|32.||Bearpark HM, Elliott L, Grunstein RR, Cullen S, Schneider H, Althaus W, Sullivan CSnoring and sleep apnea: a population based study in Australian men. Am J Respir Crit Care Med151199514591465|
|33.||Kump K, Whale C, Tishler PAssessment of the validity and utility of a sleep symptom questionnaire. Am J Respir Crit Care Med1501994736741|
This article has an online data supplement, which is accessible from this issue's table of contents online at www.atsjournals.org.