O010 Sleep duration and timing predict next-day exercise likelihood and duration: an analysis of 6-million person-nights

Abstract Introduction Sleep and exercise are pillars of health. However, the temporal dynamics between these two behaviours remain poorly understood. This study examined the independent between- and within-person associations of sleep duration and onset on next-day exercise likelihood and duration in a large, longitudinal sample of adults under free-living conditions. Methods Participants (N=19,977; 6,593,237 person-nights) wore a validated wrist-worn biometric device (WHOOP Strap) for one year. Objective sleep metrics were derived from the wrist-worn device and exercises were logged on a smartphone application. We fit generalized additive mixed models to assess the between- and within-person associations between sleep and exercise variables, adjusted for age, gender, weekday, seasonal effects, biometric feedback, and autocorrelated errors. Results At the between-person level, longer average sleep duration and later average sleep onset were each independently associated with reduced exercise likelihood and duration (ps<.001). At the within-person level, sleeping 1 to 1.75 fewer hours and going to sleep 2 to 2.5 hours earlier than usual were each independently associated with increased next-day exercise likelihood and duration (ps<.001). However, further sleep loss and earlier sleep onset advance predicted worse next-day exercise outcomes. Sleeping longer and going to sleep later than usual were negatively associated with next-day exercise (ps<.001). Discussion Together, our findings indicate that contrary to symbiotic models of sleep and exercise, individuals may sacrifice time in one health behaviour for time in the other. Interventions aimed at promoting exercise adherence should carefully consider the temporal dynamics between the duration and timing of sleep and next-day exercise outcomes.