Introduction: The frequency of obstructive sleep apnea (OSA) among stroke and TIA patients varies from 14% to 72% depending on apnea-hypopnea index (AHI) cut-off points and the study population. Various screening tools have been validated in unselected patients referred for diagnostic polysomnography (DP) but a screening tool that identifies stroke/TIA patients at highest risk of moderate-severe OSA is lacking.
Hypothesis: A simple screening tool that considers medical comorbidities and sleep-related symptoms can identify stroke/TIA patients who are most likely to have moderate-severe OSA.
Methods: Consecutive stroke and TIA patients referred by vascular neurology for DP from September 2012 to February 2015 were included in a retrospective analysis. Baseline clinical characteristics, DP results and outcomes were collected retrospectively. Moderate-severe OSA was defined as AHI≥15.
Results: Of 65 patients (mean age 61±14 years, 48% female, 51% African-American) included in this analysis, 19 (29.2%) were diagnosed with moderate-severe OSA. In multivariable analysis using backward elimination (entry and exit thresholds 0.15), predictors of moderate-severe OSA included history of Coronary artery disease, Hyperlipidemia, older Age, history of Atrial fibrillation, former or current Tobacco use and self- or family-reported Snoring or daytime Sleepiness (CHA2TS2). Variables excluded were sex, race, hypertension, diabetes, depression, chronic obstructive pulmonary disease, patent foramen ovale, body mass index, stroke subtype and wakeup onset . The CHA2TS2 model had excellent predictive power for moderate-severe OSA (AUC= 0.91). CHA2TS2 ≥5 had 100% sensitivity and 54% specificity for detecting moderate-severe OSA.
Conclusions: We developed a simple screening tool that can identify stroke and TIA patients who have high likelihood of having moderate or severe OSA identified on their DP. Prospective validation of CHA2TS2 score is currently underway.