Abstract Body (Do not enter title and authors here): Background: While prior research suggests a link between sleep disturbances and atrial fibrillation (AF), few studies have utilized objective sleep measurements, including sleep stages, in a large, diverse cohort. We examined the associations between sleep parameters derived from wearable devices and the incidence of AF in the All of Us Research Program.
Methods: This study included 52,952 participants with wearable-recorded sleep data and no baseline AF, followed for a median of 935 days. Sleep observations with mean total durations <3 or >16 hours were excluded. AF events (n = 437; 0.84%) were identified from linked electronic health records. Sleep parameters included total sleep duration, main and non-main session sleep, and durations of sleep stages. Cox proportional hazards models estimated hazard ratios (HRs) for incident AF. All reported HRs are from fully adjusted models for age, sex, race, smoking status, alcohol consumption, body mass index, physical activity, hypertension, diabetes mellitus, chronic obstructive pulmonary disease, and sleep disorders.
Results: Participants with incident AF were older (61.8 vs. 50.5 years), more likely to be male, smokers, and white, and had higher prevalences of hypertension, diabetes, COPD, and sleep disorders (all P < 0.001). Total sleep duration was associated with an increased risk of AF (HR _{per 1-hour increase}: 1.10, 95% CI: 1.01–1.19). Additionally, total sleep duration of more than 9 hours was significantly associated with an increased risk of AF (HR _{vs. 6-9 hours}: 1.47, 95% CI: 1.12–1.94; P-trend = 0.004), whereas short sleep duration (<6 hours) was not. Deep and REM sleep durations were inversely associated with AF incidence, while light sleep duration showed a modest positive association. Non-main session sleep of ≥2 hours was associated with an elevated AF risk (HR _{vs. 1–2 hours}: 1.64, 95% CI: 1.23–2.19; P-trend < 0.001).
Conclusion: Objective measures of sleep duration and architecture are independently associated with the incidence of AF. Specifically, prolonged total or non-main sleep, reduced deep and REM sleep, and increased light sleep may contribute to AF pathophysiology. These findings highlight the potential of wearable physiologic monitors in AF risk stratification and sleep-targeted prevention strategies.