Sleep Disordered Breathing and Subjective Sleepiness in the Elderly: A Deadly Combination?

In this issue of SLEEP, Gooneratne and colleagues¹ describe a prospective community-based study demonstrating increased mortality risk among older adults with moderate-severe sleep-disordered breathing (SDB) with excessive daytime sleepiness (EDS). This provocative study adds to a field in which previous research has presented conflicting evidence on whether sleep apnea in the geriatric population increases mortality risk. Although several clinic-based and community-based studies have demonstrated an association between severe SDB and decreased survival (Table 1),^2–13 for the most part, this finding has been limited to middle-aged adults and men. As the authors suggest, detecting a link between SDB and mortality among older persons is potentially confounded by various other comorbidities that contribute to mortality with increasing age. For example, in the Sleep Heart Health Study (SHSS), a community-based cohort where elderly subjects were included, the effect of SDB on mortality lost significance with age.⁹ Moreover, some evidence suggests that mortality risk secondary to SDB may even decrease with age.⁵ Aside from Ancoli-Israel and colleagues who demonstrated that SDB is a risk factor for mortality among female nursing home patients,¹⁴ many other studies have excluded the elderly entirely. Given the conflicting data available, clinicians are left feeling uneasy regarding the sequelae of SDB among older patients.

Table 1.

Clinic-based and community-based studies reporting an association between moderate to severe SDB and decreased survival

Author, year, Country	Design	Number of participants (% men)	Age upon enrollment*	Mean years of follow-up	Risk of all-cause mortality in moderate to severe OSA compared to controls¶
Partinen,8 1988 U.S.	Clinic-based	198 (96)	51.3 ± 11.3	5	11 per 100 patients per 5 years vs. 0 per 100 (conservative vs. surgical treatment)
He,3 1988 U.S.	Clinic-based, retrospective	385 (100)	51.6 ±12	8	AI > 20: 63% survival AI < 20: 96% survival
Young,11 1998 U.S.	Clinic-based	354 (74)	Men: 50.4 Women: 51.1	5	AHI > 20: Men 8%, Women 33% AHI < 5: Men 11% , Women 10%
Lavie,13 2005 Israel	Clinic-based	14,589 (100)	48.4 ± 12.3	4.7	AHI ≥ 40: 10.22 per 1,000 patient-years AHI ≤ 10: 2.19 per 1,000 patient-years
Yaggi,10 2005 U.S.	Clinic-based, longitudinal	1,022 (71)	OSA: 60.9 Controls: 58.7	3.4	HR: 1.97 (95% CI, 1.12-3.48)** for stroke or death
Marin,6 2005 Spain	Clinic-based, longitudinal	1651 (100)	OSA: 49.9 ± 7.2 Control: 49.6 ± 8.1	10	OR: 2.87 (95% CI, 1.17-7.51)**
Campos-Rodriguez,2 2005 Spain	Clinic-based, retrospective	871 (81)	55.4 ± 10.6	4	CPAP < 1 h/d: 85.5% CPAP > 6 h/d : 96.4%
Marshall,7 2008 Australia	Busselton Health Study community-based, longitudinal	380 (73)	OSA: 55.1 ± 8 Controls: 52.6 ± 7.5	13.4	HR: 6.24 (95% CI, 2.01-19.39)**
Young,12 2008 U.S.	Wisconsin Sleep Cohort Study, community-based, longitudinal	1,522 (55)	48 ± 8	18	HR: 3.0 (95% CI, 1.4-6.3)**
Punjabi,9 2009 U.S.	Sleep Heart Health Study community-based, longitudinal	6,294 (47)	62.9 ± 11.0	8.2	HR: 1.46 (95% CI, 1.14-1.86)**
Gooneratne,1 2011 U.S.	Community-based, longitudinal	289 (26)	78.0 ± 6.3	13.8	AHI ≥ 20 and EDS: HR 2.28 (95% CI, 1.46-3.57)** AHI ≥ 20 and no EDS: HR 0.74 (95% CI, 0.39-1.38)**

HR, Hazard ratio (95% confidence interval); OR, Odds ratio; AI, apnea index; AHI, apnea-hypopnea index; EDS, excessive daytime sleepiness;

^*expressed as mean ± SD unless otherwise stated;

^**fully adjusted model;

^¶unless otherwise noted.

The study by Gooneratne et al.¹ helps to explain some of this heterogeneity in the sleep literature by including significantly older subjects (mean age on enrollment, 78 years) with longer follow-up (mean 13.8 years) and exploring the potential mediating effect of subjective sleepiness (a factor that has been independently associated with mortality in the elderly).¹⁵ The authors dichotomized subjects with SDB into 2 groups: apneahypopnea index (AHI) ≥ 20 or < 20, similar to other reports that identified that the cutoff of AHI ≥ 20 was associated with a higher mortality.^3,11 Gooneratne et al. observed that participants with AHI ≥ 20, with concomitant subjective EDS, had an increased risk of mortality; however, the same severity of SDB without EDS did not impact survival.¹ Subjective EDS was assessed with a questionnaire that quantified how many times per week subjects felt sleepy (i.e., sleepy at least 3-4 times per week). EDS was also evaluated objectively with a multiple sleep latency test (MSLT) that followed 2 nights of polysomnography. Interestingly, the association between SDB and mortality did not persist when the objective definition for EDS (i.e., mean sleep onset latency < 8 minutes) was added to subjective reports. It is important to note that subjective complaints of EDS are not as specific as MSLT data for physiologic sleepiness and may also be measuring other determinants (e.g., depression; though patients with depression were excluded for the present study). Importantly, these findings emphasize the need for clinicians to pay particular attention to symptoms of EDS in the elderly, as this, in combination with the presence or absence of SDB, may impact mortality, and thus clinical decision-making regarding treatment. In addition to increased mortality risk, SDB with EDS has been identified in younger populations as being an important contributor to poor outcomes, such as increased metabolic risk¹⁶ and hypertension.¹⁷ Thus, SDB with EDS may represent a phenotype in the elderly that requires more clinical investigation.

Another strength of the study by Gooneratne et al.¹ is the inclusion of a cohort with a high percentage of women (74%) in comparison to preceding studies. Increased mortality risk among women with SDB has not been consistently demonstrated previously, but this may partly be due to underrepresentation or even exclusion of women in several studies (Table 1). Identifying this potential gender-bias, Young and colleagues examined a clinic-based sample and observed an increased mortality risk in women with OSA (mean age 50.7 years, mean AHI 50.7).¹¹ These results differed from other studies in that the women enrolled had severe OSA, thus potentially increasing the likelihood that they might die from the disease. In contrast, the SHHS cohort did not demonstrate increased mortality risk in women with OSA.⁹ However, in the SHHS cohort, only a minority of women (approximately 6%) had moderate to severe SDB. In contrast, Gooneratne et al. had a majority of women in their cohort (i.e. 74%, n = 214) and an average follow-up period of 13.8 years.¹ Of the entire cohort of 289 subjects included in the study, 66 (22.8%) had SDB. Although it is not stated how many of these with SDB were women or how many had moderate-severe OSA, there was an increase in mortality in this elderly cohort even after adjusting for male gender. Perhaps part of the discrepancy in the Gooneratne study¹ compared to the SHHS cohort⁹ may be explained by the former having a longer follow-up period, which may be required to detect a signal of increased mortality in women. Additionally, the older age of the women, and the stratification of the presence or absence of EDS in the Gooneratne study may also have played a role.

Gooneratne and colleagues¹ assessed mortality through the social security death index but did not further investigate underlying cause of death, such as death secondary to cardiovascular disease. Cardiovascular mortality is significantly increased among middle-aged patients with SDB,¹² but the mechanisms of mortality among older persons (e.g., cardiovascular, metabolic, cancer-related, driving-related) remain underexamined. Similarly, the study was not powered to explore the relative importance of central sleep apnea (known to increase with age¹⁸) compared to obstructive sleep apnea in terms of their respective associations with mortality.

Before finally concluding whether or not OSA leads to increased mortality risk in elderly populations, further prospective randomized-controlled studies with continuous positive airway pressure (CPAP) are needed to determine if mortality risk can be ameliorated with therapy. Also of importance will be the investigation of the role that subjective EDS will play in risk stratification and the mechanisms by which it may mediate mortality. Finally, what still remains unknown is what to do with patients who have milder sleep apnea, since although this subtype does not appear to be associated with increased mortality, other outcome measures may indicate treatment of milder disease in the elderly.

DISCLOSURE STATEMENT

The authors have indicated no financial conflicts of interest.