Abstract
Introduction
This study examined whether leisure-time physical activity (LTPA), occupational physical activity (OPA), and their combination are associated with risk of mental disorder–related long-term sickness absence (LSA).

Methods
This cohort study included 38,120 Japanese employees aged 20–59 years from the Japan Epidemiology Collaboration on Occupational Health Study. Baseline assessments were conducted primarily in 2011, with follow-up from 2012 through 2024. LTPA and OPA levels were self-reported at baseline. LSA (≥30 consecutive days) due to mental disorders was medically certified by physicians. Cox regression estimated hazard ratios (HRs) and 95% confidence intervals (CIs), adjusting for demographic, lifestyle, occupational, and health-related factors including depressive symptoms.

Results
During a mean follow-up of 9.2 years, 1,032 participants (2.7%) experienced LSA from mental disorders. LTPA showed an inverse but non-linear association with LSA risk (p for non-linearity=0.042) with lower risk apparent at low activity; activity below recommended levels was associated with reduced risk (HR 0.72, 95% CI 0.55–0.93). The inverse association of LTPA was evident when restricted to depressive disorders as the outcome (p for trend=0.029) and among sedentary workers (p for trend=0.027), although the interaction by OPA was non-significant (p=0.14). Higher OPA was linearly associated with lower risk (HR for physically active vs sedentary work = 0.50, 95% CI 0.37–0.69; p for linear trend<0.001).

Conclusions
LTPA showed inverse associations with mental disorder-related LSA, most evident for depressive disorders. Higher OPA was associated with reduced risk of mental disorder–related LSA, not clearly supporting the physical activity paradox.