Si-Ho Kim, Seung-Hwan Kim, Chang-Ho Chae
Background/Objectives: Low serum 25-hydroxyvitamin D [25(OH)D] has been proposed as a modifiable risk factor for depression, but the shape of the exposure-response relationship and its independence from sleep disturbance have not been characterised in a large occupational cohort with repeated measures. Methods: We analysed 246,751 health examinations of 62,989 Korean adults over 12 annual waves (2014-2025). Depressive symptoms were measured with the Center for Epidemiologic Studies Depression Scale (CES-D) and sleep quality with the Pittsburgh Sleep Quality Index (PSQI). Hierarchical linear and logistic regressions with cluster-robust standard errors (228,520 complete-case observations from 60,403 individuals) modelled serum 25-hydroxyvitamin D [25(OH)D], both continuously and categorically; restricted cubic splines tested non-linearity, and a random-intercept model estimated the intraclass correlation coefficient (ICC). Results: Mean 25(OH)D was 21.08 ± 8.71 ng/mL; 51.3% were deficient (<20 ng/mL) and 10.5% scored CES-D ≥ 16. The exposure-response relationship was non-linear (p = 0.001 for non-linearity). Relative to sufficiency (≥30 ng/mL), only severe deficiency (<10 ng/mL) was independently associated with depression (adjusted odds ratio [OR] 1.20, 95% confidence interval [CI] 1.09-1.32); deficiency (OR 1.05) and insufficiency (OR 1.02) were not. The per-1 ng/mL association was small (OR 0.996, 95% CI 0.993-0.999). PSQI score was the dominant predictor (OR 1.372 per point). The ICC was 0.743, and the within-person correlation (r = -0.008) was one-sixth of the between-person correlation (r = -0.046). Conclusions: The vitamin D-depression association is confined to severe deficiency rather than graded across the normal range, is small in absolute terms, and is substantially exceeded by that of sleep disturbance. Screening and repletion should target the severely deficient minority.