Polysomnographic Measures of Sleep Architectural Disruption and Incident Atrial Fibrillation and Stroke in a Large Clinical Cohort

Introduction: Although epidemiologic studies have identified associations of architectural disruptions in sleep and atrial fibrillation (AF), inconsistencies in results limit understanding of longitudinal AF and stroke development in clinic-based cohorts. We hypothesize lower arousal index as a biomarker of prolonged respiratory events and objective measures of sleep disruption increase incident AF and stroke.

Methods: Cleveland Clinic patients (age>18) who underwent polysomnogram (PSG) or split studies 11/27/2004-12/30/2015 were examined. Arousal index, total sleep time (TST,hours), sleep efficiency (SE,%), and wake after sleep onset (WASO,hours) were assessed as predictors of incident AF. Cox proportional hazard models were fit with time from sleep study to AF diagnosis. Secondary analyses were conducted in the subset of patients with baseline AF, with time from sleep study to stroke. Covariates included age, sex, race, body mass index (BMI,kg/m2), cardiovascular risk factors (hypertension, diabetes mellitus, hyperlipidemia), heart failure, coronary artery disease, myocardial infarction, coronary artery bypass grafting, chronic obstructive pulmonary disease, tobacco use, and use of anti-arrhythmic drugs. Data were censored at date of last follow up or the 5-year mark.

Results: The sample comprised 43,634 patients: age 51.7±14.5, 51.9% male, 74.5% White, and 7.1%(n=3,090) with AF. Of those without baseline AF, 1,176(2.9%) developed 5-year incident AF. Incident AF increased by 9% (HR=1.09, 95%CI=1.04-1.13) for each hour of lost TST. For every 10 percentage point decreased SE, incident AF increased by 7% (HR=1.07, 95%CI=1.04-1.11). For every hour increased WASO, incident AF increased by 14% (HR=1.14, 95%CI=1.07-1.21). Arousal index conferred no statistically detectable change of incident AF. In secondary analyses, any association of arousal index and incident stroke was attenuated after accounting for confounding comorbidities (omnibus p=0.055).

Conclusion: Objective measures of disrupted sleep architecture predicted incident AF in this clinical cohort. However, arousal index was not associated with AF development, nor stroke development in secondary analyses. Further investigation is needed to elucidate the role of arousal in SDB, perhaps with hierarchal models to clarify the degree to which confounders attenuate or accentuate any relationship.

Support: Cleveland Clinic Neurological Institute Center for Outcomes Research & Education Pilot Grant, Neuroscience Transformative Research Resource Development Award