Joint associations of accelerometer-measured sleep duration and physical activity with cardiovascular disease and all-cause mortality: a longitudinal cohort study.
Level 3 - non-randomized controlled study
Prospective observational cohort study
PubMed 42386465 · doi:10.1016/j.numecd.2026.104849
What was done
This longitudinal cohort study analyzed 87,879 UK Biobank participants with accelerometer-measured sleep duration and physical activity. Sleep duration was categorized into insufficient (<7 h/day), optimal (≥7 to <10 h/day), and prolonged (≥10 h/day). Physical activity was evaluated as moderate-to-vigorous physical activity (MVPA) and light physical activity (LPA). Cox proportional hazards models and restricted cubic splines evaluated independent and joint associations, including additive interactions, with incident cardiovascular disease (CVD) and all-cause mortality.
What was found
Significant additive interactions between sleep duration and physical activity were observed for all outcomes. Among participants with low MVPA, suboptimal sleep compared with optimal sleep was associated with increased CVD risk for prolonged sleep (HR 1.17, 95% CI: 1.01 to 1.36) and increased all-cause mortality for insufficient sleep (HR 1.51, 95% CI: 1.21 to 1.89) and prolonged sleep (HR 1.20, 95% CI: 1.05 to 1.36). In the high MVPA group, participants with suboptimal sleep had lower risks compared to the reference group (optimal sleep with low MVPA): prolonged sleep had lower CVD risk (HR 0.76, 95% CI: 0.60 to 0.97), and mortality was lower for insufficient (HR 0.59, 95% CI: 0.40 to 0.85) and prolonged sleep (HR 0.70, 95% CI: 0.56 to 0.88). Similar patterns were observed for LPA.
Why it matters
These findings suggest that higher physical activity levels, whether moderate-to-vigorous or light, may mitigate the elevated cardiovascular and mortality risks associated with inadequate or excessive sleep duration.
Limits
The observational design precludes causal inference and remains susceptible to unmeasured confounding and reverse causality. Accelerometer monitoring reflected a single snapshot in time. The abstract does not report duration of follow-up, absolute risk numbers, or specific covariate adjustments, and the UK Biobank cohort is subject to healthy volunteer selection bias.