Not all sleep matters equally: differential recovery functions of sleep dimensions linking work stress to depressive symptoms

Sleep is widely recognized as an important pathway linking work stress to depressive symptoms, yet the relative roles of individual sleep dimensions remain poorly understood. Using cross-sectional data from the 2025 Worker Health Survey in Gansu Province, China (n = 9,736), this study estimated the statistical indirect association between work stress and depressive symptoms through overall sleep problems using Bayesian mediation analysis. The estimated indirect association accounted for 27.8% of the total statistical association between work stress and depressive symptoms. The study further compared the relative contributions of sleep quality, sleep disturbance, sleep duration, and sleep latency using Bayesian parallel multiple mediation analysis under the assumed ordering of work stress, sleep, and depressive symptoms. Given the potential nonlinear association between sleep duration and depressive symptoms, both linear and nonlinear specifications were evaluated. The nonlinear model showed better predictive performance and indicated a U-shaped association between sleep duration and depressive symptoms. Accordingly, the relative contributions of the four sleep dimensions were interpreted based on the nonlinear parallel model. Sleep quality accounted for 50.7% of the estimated total indirect association, followed by sleep disturbance (37.2%), whereas sleep duration (9.2%) and sleep latency (2.9%) contributed substantially less. This overall pattern was broadly consistent across subgroups, although the relative contributions of individual sleep dimensions varied across demographic, occupational, socioeconomic, and geographic contexts. These findings suggest that different sleep dimensions may reflect distinct aspects of recovery rather than interchangeable indicators of the same construct. Considering sleep as a multidimensional process may improve understanding of the association between work stress and depressive symptoms and help inform the development and evaluation of more targeted workplace interventions.