Labor-condition indicators and recent depressed/down affect among Chinese workers, 2010–2023

Working time, employment-relation form, contract formalization, and task autonomy represent distinct aspects of work that broad labor-condition labels may obscure. We examined their cross-sectional associations with recent depressed/down affect across repeated population surveys in China. This repeated cross-sectional analysis harmonized nine waves of the Chinese General Social Survey (CGSS; 2010–2023). A single item measured feeling depressed or down at least sometimes during the previous four weeks. In exposure-specific samples of workers aged 18–59 years, weighted linear probability models estimated adjusted percentage-point differences for two weekly-hour thresholds, a combined employment-relation category (casual/no-fixed-employer or labor-dispatch work), written-contract status, and task autonomy. Models included survey-year and province indicators and province-clustered standard errors. The adjusted prevalence differences were 2.25% points (95% CI 0.95 to 3.55) for more than 44 h per week and 2.82% points (95% CI 1.12 to 4.51) for 60 h or more. The combined employment-relation estimate was 3.31% points (95% CI 0.71 to 5.92) for the primary outcome but attenuated to 0.88% points (95% CI -0.29 to 2.05; p = 0.135) under the stricter outcome threshold. Relative to fixed-employer employees, category-specific estimates were 4.87% points for casual/no-fixed-employer work and 0.73% points for labor dispatch; their direct contrast was imprecise. The written-contract estimate was imprecise, whereas the autonomy estimate was close to zero with a narrower interval. When 2011 and 2023 were excluded together, the estimates were 1.40% points (95% CI -0.06 to 2.85) for more than 44 h and 2.48% points (95% CI 0.45 to 4.51) for 60 h or more. The four labor-condition dimensions showed different patterns of estimates and sensitivity. Both working-hour contrasts were positive in pooled models, but the > 44-hour estimate was sensitive to excluding 2011 and 2023 together. The overlapping contrasts were not designed to test a dose-response relationship. The combined employment-relation estimate was sensitive to outcome definition, and the labor-dispatch estimate was inconclusive. These contemporaneous associations cannot establish temporal order or causal effects, but support measuring the four dimensions separately in longitudinal research.