Recognition and Diagnosis of Obstructive Sleep Apnea in Older Americans

Tiffany J. Braley, MD, MS; Galit Levi Dunietz, PhD, MPH; Ronald D. Chervin, MD, MS; Lynda D. Lisabeth, PhD, MPH; Lesli E. Skolarus, MD, MS; James F. Burke, MD, MS


J Am Geriatr Soc. 2018;66(7):1296-1302. 

In This Article


This study of a nationally representative sample of Medicare beneficiaries suggests that a large proportion of older Americans who are at risk of OSA are not evaluated for this condition. Using NHATS items similar to those in the STOP–BANG questionnaire, 56% were found to be at risk of OSA. Excluding age as a risk factor, 20% of older beneficiaries still met criteria for moderate OSA risk. Of at–risk individuals who received evaluations with PSG or HSAT, 94% received a diagnosis of OSA, and 82% of these received treatment with PAP. Our study provides new evidence that, in older Americans, increased risk for OSA is common; is seldom investigated; and when investigated, is usually confirmed and treated. These data also invite speculation that older adults may be vulnerable to disparities in clinical evaluation for OSA. Follow–on work that validates the full STOP–BANG questionnaire and demonstrates health consequences that efficient OSA detection in older individuals may offset will be necessary to shape future OSA screening guidelines for older adults, but these data provide an important first step in highlighting the national scope of OSA risk, recognition, and treatment in older Americans.

The magnitude and effect of OSA underevaluation in the United States, particularly in older adults, is unknown.[35–37] Perhaps the most comprehensive study of sleep–disordered breathing and predictive clinical correlates that focused solely on older Americans was completed in the 1980s. In that study, interviews and HSAT were conducted in older San Diego residents.[38] Depending on HSAT measures used, at least 24% had evidence of sleep–disordered breathing. As with the STOP–BANG, features most reliably associated with sleep–disordered breathing included BMI, sex, and sleepiness frequency. Older adults accounted for approximately 60% of the sample in the Sleep Heart Health Study—a cohort study designed to investigate OSA as a risk factor for cardiovascular diseases.[39] In the same study, male sex, snoring, and breathing pauses were also identified as predictors of OSA,[20] although the scope of national gaps of OSA evaluation in older adults was not a focus of these studies, and factors that may prompt or dissuade providers from referring older adults for OSA evaluations have not been sufficiently explored.

Although traditional correlates of OSA (snoring, sleepiness, hypertension) are recognized triggers for PSG and HSAT in the general population,[40] many of these characteristics are also attributed to normal aging. Consequently, older adults who exhibit these characteristics may be more likely than their middle–aged counterparts to escape sleep evaluations. Older adults may also be less likely to seek medical attention for symptoms that signal OSA in younger individuals[20] or more likely to experience sequelae not classically associated with OSA in younger individuals.[4,41–44] Accordingly, our analyses examined factors associated with OSA diagnosis (and likelihood of evaluation) outside of the STOP–BANG construct in the entire sample of 3,195 participants. As with previous studies,[38,45] male sex and BMI remained strong predictors of OSA diagnosis. Additional noteworthy correlates of OSA diagnosis included bothersome pain, use of a mobility device, diabetes mellitus, and being married or cohabiting (Table 3), as well as poor overall health and independent use of a vehicle (Results). Potential factors associated with likelihood of OSA evaluation included independent use of a vehicle and need for mobility assistance in and around the house. The above associations allow speculation that characteristics that the STOP–BANG does not capture, including symptoms associated with sleep disturbances (pain), input from a bed partner or caregiver, or transportation barriers could confer OSA risk or influence likelihood of evaluation in older adults. In this regard, other sleep–specific characteristics that the NHATS sleep module examines, such as insomnia, short sleep duration, and frequency of hypnotic use (Supplementary Table S2) could explain the high prevalence of OSA diagnosis in the 2% of low–risk respondents who received OSA evaluations with PSG or HSATs. Definitive conclusions cannot be drawn regarding snoring or hypersomnolence as predictors of OSA from our logistic regression models, because NHATS items pertaining to these symptoms were asked only of participants in the sleep module, which precluded their inclusion in logistic regression models. Furthermore, although only 8% of sleep module respondents were classified as snorers, complex item phrasing (Supplemental Table S1) may have discouraged a positive response. Additional studies are necessary to assess the prevalence of snoring and its association with OSA in older Americans.

These findings raise questions about the implications of OSA screening using the STOP–BANG in older adults. Although the STOP–BANG has not been formally validated in adults aged 65 and older, previous validation studies that have included many older individuals[27,30–32] and the large proportion of respondents with surrogate STOP–BANG scores of 3 or more who received an OSA diagnosis (94%) after PSG or HSAT in this sample combine to suggest that this instrument could offer adequate positive predictive value for OSA screening in older adults.

Our findings also raise questions about the predictive value of other factors and symptoms outside of the STOP–BANG. For example, it is hypothesized that age–related weakness of pharyngeal dilator muscles, thought to arise from vibratory trauma to the oropharynx, contributes to OSA risk in older adults. Although snoring, which the STOP–BANG captures, could contribute to this phenomenon, the STOP–BANG does not directly assess signs of pharyngeal muscle weakness.[46–48] Furthermore, the STOP–BANG is designed to capture those who are male and obese, even though OSA is not strongly associated with a sex predilection in older adults and obesity is not as frequently encountered in elderly adults. Future research that focuses on assessment of additional predictors of OSA not included in the STOP–BANG as screening tools could reduce underestimation of this condition in older women and nonobese adults.

Although additional studies are necessary to determine whether complications of OSA in older adults parallel those of middle–aged adults, recent studies that have included older adults would suggest that a substantial proportion of older adults, and men in particular, may be subject to the same cardiovascular risks and all–cause mortality.[49–51] Furthermore, older adults may be more vulnerable to other OSA–related consequences, including falls, cognitive impairment, and dementia.[52–54] Prior work has shown a link between OSA and cognitive dysfunction in individuals with dementia,[55] and some studies suggest that continuous PAP (CPAP) may improve performance in specific cognitive domains or delay cognitive decline.[55,56] Confirmation of effects of CPAP on cognitive function and cardiovascular risk would provide support for population–based OSA screening for older adults.

To our knowledge, this is the first study to characterize the national scope of OSA risk and likelihood of evaluation in older Americans. Strengths include a large representative sample of Medicare beneficiaries and high response rate (88%). Linkage of to Medicare fee–for–service claims files allowed for coupling of beneficiary characteristics (collected through in–person interviews) with objective claims data regarding PSG or HSAT, OSA diagnosis, and Medicare prescriptions of PAP equipment. The use of sampling weights allowed generalization of results to the population.

Potential limitations should be acknowledged. Although 3 of the 6 NHATS surrogate STOP–BANG items were objectively measured (age, sex, BMI), the remaining items required reformatting to allow uniform scoring. Although these items closely resemble STOP–BANG items, these adapted items have not been validated formally. It is also possible that absence of a neck circumference item could have excluded some at–risk participants. To examine the influence on overall OSA risk of allowance of 2 points for the NHATS snoring item (which queried the presence of snoring OR gasping or choking within the same item), we performed sensitivity analyses in subjects with positive responses to this item. Our analyses suggest that the majority of participants who were given 2 points for a positive snoring item (92%) would have still received a surrogate STOP–BANG score of 3 or more, even if they had been assigned only 1 point for the snoring item. In addition, when more specific but less sensitive STOP–BANG scoring methods were applied to our sleep module sample,[57] a similarly low proportion of at–risk participants (2%) were evaluated for OSA. Potential barriers to OSA evaluations of at–risk older adults require further exploration; that said, we acknowledge that data not available in NHATS could in part influence clinical decisions to refer for PSG or HSAT. Finally, it is possible that excluding Medicare Advantage beneficiaries (36% of the sample) could reduce generalizability.

Our data highlight a national gap in evaluation of one of the most rapidly growing demographic groups in the United States. If older individuals with OSA are subject to the same benefits of treatment as middle–aged adults, research that addresses potential causes and solutions for these gaps and demonstrates the value of OSA screening in older adults could offer a vital opportunity to improve one of the nation's top health problems.