Abstract TP179: Selecting an Outcome for Sleep Apnea Treatment Trials in Stroke: A Comparative Analysis

Abstract

Introduction: Post-stroke sleep apnea (SA) affects the majority of stroke patients and is associated with poorer stroke outcomes. To select appropriate endpoints for a future SA treatment trial in stroke patients, we assessed the predictive ability between a variety of important outcome measures and SA severity, as measured by the apnea-hypopnea index (AHI), within a population-based stroke study. Methods: Ischemic stroke patients enrolled in the Brain Attack Surveillance in Corpus Christi (BASIC) project were offered SA screening shortly after stroke with a portable sleep apnea test (ApneaLink Plus). Apnea-hypopnea index (AHI) was calculated as the sum of apneas and hypopneas during the nocturnal recording. Subjects had baseline information collected from chart abstraction and interview, and had 3-month outcomes assessed in-person including an activities of daily living (ADL)/instrumental ADL(IADL) scale, NIHSS, Modified Mini-Mental State exam (3MS), and 12-item Stroke-specific Quality of Life scale (SSQOL). Predictive R 2 (a measure of how well models predict responses for new observations) from linear regression models was used to compare predictive ability of models with and without AHI, adjusted for significant predictors of outcome. Results: Within 455 subjects, the predictive R 2 was 31.7% for ADL/IADL, 23.2% for SSQOL, 25.0% for the 3MS, and 11.7% for NIHSS in the model without AHI. When AHI was included in the models, the predictive R 2 increased to 32.8% for ADL/IADL and increased to 11.9% for NIHSS; however, the predictive R 2 decreased by 0.1% for SSQOL and decreased by 0.2% for 3MS. Conclusions: In future trials that test SA treatment’s effect on stroke outcomes at 3 months, improvement is most likely to be observed from a functional outcome, such as that measured by the ADL/IADL scale.