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

Disclosures

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

In This Article

Methods

The University of Michigan institutional review board approved all study procedures.

Data Sources and Study Population

National Health and Aging Trends Study. Data were obtained from Round 3 of the National Health and Aging Trends Study (NHATS), a nationally representative, longitudinal survey of Medicare beneficiaries designed to assess the effect of aging (http://www.nhats.org/). Funded by the National Institute on Aging (U01AG032947), NHATS has performed annual face–to–face interviews in beneficiaries' residences since 2011. The NHATS protocol includes assessment of physical and cognitive capacity, specific health conditions, disability, pain, mood, symptom severity and frequency, well–being, mobility, accommodations, self–care, social support activities, demographic characteristics, and socioeconomic characteristics. Proxy respondents are used if a participant is unable to answer NHATS questions. Completed with an 88% response rate,[25] Round 3 included interviews from 5,097 participants, who through survey weights, represented 32,639,407 older Americans. Survey weights accounted for differential selection probabilities and potential nonresponse bias.

NHATS Sleep Module

In 2013, NHATS asked also questions about sleep disturbances and symptoms of sleep–disordered breathing. Six of these "sleep module" items, which are very similar to items of the (STOP–BANG) questionnaire,[26] were adapted for use in this study. The STOP–BANG is a validated, 8–item screening instrument that assesses characteristics known to confer risk for OSA which form the acronym "STOP–BANG" (Snoring, Tiredness, Observed apneas, high blood Pressure, BMI, Age, Neck circumference, Gender). Item scores (1/0) are based on yes/no answers.[27,28] The sensitivity of a STOP–BANG score of 3 or greater was 83.9% to predict OSA (apnea hypopnea index (AHI) >5), 92.9% to predict moderate to severe OSA (AHI >15), and 100% to predict severe OSA (AHI >30).[28] In general, a score of 0 to 2 is considered low risk, 3 to 4 moderate risk, and 5 or greater (or a score of 3 that include specific combinations of STOP–BANG items) high risk.[29] The utility of the STOP–BANG has been widely demonstrated in a variety of large samples, many of which included high proportions of elderly adults.[27,30–34] The NHATS sleep module was administered to a random subset of 1,052 Round 3 participants, which, through sampling weights, corresponds to 7,082,963 beneficiaries (Figure 1). Responses to sleep–specific items from this module were used to estimate the proportion of Medicare enrollees at risk for OSA. For primary analyses, participants were considered to be at risk for OSA if they scored 3 points or more on the surrogate NHATS STOP–BANG items (Supplementary Table S1). As all respondents scored at least 1 on the STOP–BANG based on the age item (≥50, which could differentially affect estimation of OSA risk within our pool of respondents), an additional dataset in which the age item was dropped from the score was also created for exploratory analyses. Additional NHATS sleep module items included questions about initial and sleep maintenance insomnia, sleep duration, hypnotic use, napping frequency, and napping duration (Supplementary Table S2).

Figure 1.

Proportion of obstructive sleep apnea (OSA) recognition and treatment in at–risk National Health and Aging Trends Study participants; 1,052 respondents reflect the unweighted sample frequencies that represent 7,082,963 older Americans in the general population. PSG=polysomnography; HSAT=home sleep apnea testing.

Linkage to Claims Data

NHATS data were linked to Medicare fee–for–service claims files to identify sleep module participants (n=1,052) who received formal OSA evaluations, diagnosis, and treatment. The linked dataset allowed us to estimate the proportion of beneficiaries whose claims included Current Procedural Terminology codes for in–laboratory polysomnography (PSG, the criterion standard method for OSA diagnosis) or home sleep apnea testing (HSAT), an International Classification of Diseases, Ninth Revision (ICD–9)–coded OSA diagnosis, and Healthcare Common Procedure Coding System codes for positive airway pressure (PAP) equipment (Supplementary Table S3). To minimize the likelihood of including prevalent OSA cases diagnosed before 2011, beneficiaries with existing PAP claims or OSA diagnosis codes who did not have PSG or HSAT claims during the 2011 to 2013 observation period were excluded from analyses.

Statistical Methods

Descriptive statistics procedures for complex survey data (chi–square) were used to examine demographic and health characteristics for all Round 3 participants with linked fee–for–service claims. In sleep module participants with available claims, for primary analyses, we estimated the proportion of participants at risk of OSA, the proportion of at–risk participants who were evaluated for OSA with HSAT or in–laboratory PSG, the proportion of at–risk participants diagnosed with OSA after HSAT or PSG, and the proportion of at–risk participants with OSA who were prescribed PAP treatment. These steps were repeated for low–risk participants (surrogate STOP–BANG score ≤ 2).

Bivariate logistic regression models were constructed to examine associations of clinically relevant characteristics postulated to be associated with ICD–9–coded OSA diagnoses amoing the full Round 3 fee–for–service Medicare linked sample (n=3,195).

Clinically relevant independent variables associated with OSA diagnosis (p<.15 in bivariate analyses) were included in a multivariable logistic regression model with OSA diagnosis as the dependent variable. Independent variables included age (categorical); sex; marital status; presence of bothersome pain; body mass index (BMI); use of a mobility device; diabetes mellitus; cardiovascular disease; and a composite variable defined as positive if participants endorsed one or more of hypertension, congestive heart failure, myocardial infarction, or stroke.

Bivariate logistic regression models were also used to explore other characteristics not captured in the STOP–BANG items (comorbidities, independent mobility) that could influence the likelihood of OSA evaluation with sleep studies in at–risk respondents.

All analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, NC).

processing....