(A-317) Investigating the association between glucagon-like peptide 1 receptor agonists and obstructive sleep apnea incidence using United States Medicare data 2007-2019

Background: Obstructive sleep apnea (OSA) is highly prevalent among patients with type 2 diabetes (T2DM) and is linked with adverse neurocognitive and cardiovascular sequelae. Despite the widespread use of glucagon-like peptide 1 receptor agonists (GLP-1RA) in T2DM and obesity, their potential role in mitigating OSA risk remains largely unexplored.

Objectives: To estimate the comparative effect of GLP-1RA and versus sodium-glucose cotransporter-2 inhibitors (SGLT-2i) on OSA incidence among US older adults with type 2 diabetes.

Methods: We implemented an active comparator new user design using a random 20% sample of Medicare Fee-for-Service beneficiaries aged 66+ years with two consecutive prescription claims for GLP-1RA or SGLT-2i between Jan 1, 2008, and Dec 31, 2018. New users were defined as individuals with a first prescription fill of GLP-1RA or SGLT-2i (a guideline recommended treatment alternative) after a preceding washout period of at least 12 months without a prescription for either drug class. We excluded patients without a second prescription for the same drug class within days’ supply plus 30 days grace period. Follow-up began three months after the second prescription date, and continued until the first occurrence of the outcome, treatment discontinuation, switching to or augmentation with comparator, death from any cause, termination of Medicare enrollment, or Dec 31, 2019. OSA diagnosis was defined by ≥1 primary ICD diagnosis codes (ICD-9-CM: 327.23 or ICD-10-CM: G47.33) within 60 days of a polysomnography (sleep test). We estimated adjusted 3-year cumulative risk differences of OSA with 95% confidence intervals (CIs) using standardized morbidity ratio (SMR) weighted Kaplan-Meier survival functions.

Results: We included 5,295 new users of GLP-1RA (mean age 71.6 years, male: 49.9%) and 9,417 new users of SGLT-2i (mean age 72.2 years, male: 55.6%). GLP-1RA new users were generally sicker (except ischemic heart disease), more likely to be female (50% vs. 44%), on long-acting insulin (5.3% vs. 1.4%), and have codes for obesity (35% vs. 26%). The 3-year crude risks of OSA were 5.12 per 100 vs 3.32 per 100 (RD = 1.80 per 100, 95% CI: [0.08, 3.52]). After applying SMR weights, the adjusted risk difference was -0.04 per 100 (95% CI: -2.29, 2.22).

Conclusions: Our study of U.S. Medicare beneficiaries with diabetes suggests that the initiation of GLP-1 receptor agonists is associated with a decreased 3-year risk of obstructive sleep apnea. Accounting for unmeasured confounding by BMI, a strong risk factor for OSA, will likely impact our estimate and will further be evaluated in bias analyses using linked Medicare-EHR data.