Waking-hour movement patterns and all-cause mortality: A multi-dimensional accelerometer analysis in the UK Biobank.

Objective: To examine the independent and joint associations of time spent at each physical activity intensity level during waking hours, fragmentation, and bout accumulation patterns with all-cause mortality. Methods: We included 94,014 UK Biobank participants with valid wrist-worn accelerometer data. Waking-hour activity was characterized by light physical activity (LPA) time, moderate-to-vigorous physical activity (MVPA) time, activity fragmentation quantified by the active-to-sedentary transition probability (ASTP), and time accumulated in short (<5 min), medium (5-10 min), and long (≥10 min) bouts. Cox proportional hazards models, interpretable machine-learning analyses, and Gaussian mixture models were used. Results: Over a median follow-up of 6.9 years, 4901 deaths occurred. Both LPA and MVPA were inversely associated with mortality, with a stronger association for MVPA (hazard ratio per 10 min/day: 0.99 for LPA and 0.96 for MVPA). Higher ASTP was associated with higher mortality risk (hazard ratio per 0.01 increase: 1.02). After adjustment for total activity time, greater time in short bouts was associated with lower mortality risk, whereas greater time in long bouts was associated with higher risk for both LPA and MVPA. Five composite activity patterns were identified; the pattern with high activity time, low fragmentation, and a moderate proportion of long bouts showed the lowest mortality risk. Conclusion: Beyond total activity time, waking-hour fragmentation and bout accumulation patterns were independently associated with all-cause mortality. A balanced pattern with high activity time, low fragmentation, and a moderate proportion of long bouts was associated with the lowest mortality risk.