1166 Surgical Treatment for Obstructive Sleep Apnea and Cognition in Postmenopausal Women: A Multicohort Analysis

Abstract Introduction Obstructive sleep apnea (OSA) affects 1 billion people and, with aging, increases cognitive impairment risk. Few studies assess postmenopausal women after surgery. This study evaluates whether OSA surgery alters cognitive impairment risk versus non-surgical therapy in postmenopausal women. Methods Retrospective cohort using Atropos Health’s Evidence Network de-identified US datasets (Apollo and Fortuna). Eligible: women ≥40 years-old with incident OSA (≥2 ICD codes and a sleep test CPT code), BMI < 35 kg/m² in the prior year, and no prior investigated treatments or outcomes. The intervention group underwent surgery (partial glossectomy, palatopharyngoplasty, tonsillectomy, hypoglossal nerve stimulation) within 2 years of diagnosis; the non-surgical group received continuous positive airway pressure prescription. Follow-up began at therapy initiation. A Cox proportional hazards model estimated HRs (95% CIs) and restricted mean survival times (RMSTs) analysis were performed by treatment for cognitive impairment, memory loss, neurocognitive disorder, Alzheimer’s, vascular, Lewy body, frontotemporal, and other/unspecified dementia. Propensity score matching (PSM) balanced baseline covariates. Results Fortuna dataset: surgical n=2,896 versus non-surgical n=100,000; mean (SD) follow-up 1196.8 (758.9) versus 1487.6 (732.8) days; mean (SD) age 57.7 (10.7) versus 58.9 (10.3) years. No neurocognitive outcome differed between arms in unadjusted or PSM analyses. PSM HRs (95% CI): Alzheimer’s 0.44 (0.14–1.40), cognitive impairment 0.93 (0.57–1.52), neurocognitive disorder 0.73 (0.47–1.15), unspecified dementia 1.18 (0.65–2.14), vascular dementia 1.12 (0.23–5.54), composite 1.03 (0.82–1.29). Memory loss was slightly elevated in unmatched models but disappeared after adjustment (1.00, 0.76–1.32). Apollo dataset: non-surgical n=32,517 (mean[SD] age 63.7 [10.4]) versus surgical n=524 (60.6 [11]). BMI was available and balanced: 29.0 (3.6) vs 28.6 (3.5) kg/m². Follow-up 1592.9 (1010.9) versus 1093.0 (871.3) days. Results were similar to those previously described. Events were rare and HRs ~1.00, without significant differences in neurocognitive outcomes. RMSTs were nearly identical across groups for both cohorts. Conclusion Neurocognitive outcomes in postmenopausal women with OSA were equivalent among surgical and non-surgical strategies over a five-year horizon across two national cohorts. Findings should be interpreted with caution given the low event rates, and prospective studies with an older population and longer follow-up are needed to validate these initial findings. Support (if any)