Associations of Long-Term Physical Activity Trajectories With All-Cause Mortality in a General Population

Objectives: We investigated the associations of mean levels of leisure-time physical activity (LTPA) and latent LTPA trajectories with all-cause mortality risk. Methods: Trajectories of LTPA were established using group-based trajectory analysis with a latent class growth model in a population-based cohort between 1996 and 2014. A Cox-proportional hazard model was conducted to examine the associations of LTPA quintiles and LTPA trajectories with all-cause mortality. Results: A total of 21,211 participants (age 18-90 years) were analyzed (median follow-up 16.8 years). The study participants were divided into five groups according to percentiles of LTPA (<20th, 20th-<40th, 40th-<60th, 60th-<80th, ≥80th) and LTPA trajectories (low/stable, medium/stable, increasing, decreasing, and fluctuating), respectively. Participants with a decreasing trajectory did not have a significantly lower risk of all-cause mortality despite having the highest baseline level of LTPA. In contrast, participants with a medium/stable (HR 0.84, 95% CI 0.72-0.98, p = 0.031) or an increasing (HR 0.57, 95% CI 0.33-0.97, p = 0.037) trajectory had a significantly lower risk of all-cause mortality. Conclusion: Promotion of maintaining stable LTPA is beneficial for public health and survival.