Associations of sleep duration and sleep-timing patterns with depressive mood and suicidal outcomes in Korean adolescents

Abstract Background Sleep duration, sleep timing, and weekday–weekend sleep-timing differences are interrelated characteristics associated with adolescent mental health. Categorizing continuous measures and interpreting mutually adjusted estimates as independent effects may obscure nonlinear patterns and their shared derivation. We examined associations of corrected mid-sleep on free days (MSFsc), weekly average sleep duration, and the absolute weekday–weekend sleep midpoint difference with depressive mood and suicidal outcomes. Methods This cross-sectional study used data from the nationally representative 2025 Korea Youth Risk Behavior Survey. Descriptive analyses included 48,426 adolescents with valid sleep-derived variables, and adjusted models included 48,416 complete cases. Survey-weighted logistic regression with restricted cubic splines evaluated nonlinear associations of the three continuous sleep descriptors with depressive mood, suicidal ideation, suicide plan, and suicide attempt. Secondary analyses examined categorical sleep-timing groups, sequential adjustment models, and exploratory combined sleep-timing group–sleep-duration profiles. Results Weekly average sleep duration and the absolute weekday–weekend sleep midpoint difference were associated with all four outcomes. Sleep duration was nonlinear for all outcomes, with higher odds at shorter durations. The midpoint difference was nonlinear only for depressive mood and was otherwise generally monotonic. MSFsc showed overall and nonlinear associations with depressive mood, suicidal ideation, and suicide plan, but not suicide attempt in the primary analysis (overall p = 0.054; nonlinearity p = 0.993); this result was sensitive to exclusion of extreme MSFsc values. Categorical sleep-timing group estimates changed substantially after adjustment for sleep duration and midpoint difference. In exploratory combined-profile analyses, < 5 h of sleep was associated with higher odds across all sleep-timing groups, but only 4 of 72 direct contrasts remained significant after false-discovery-rate adjustment. Conclusions Very short sleep duration and larger weekday–weekend sleep midpoint differences showed the most consistent associations with depressive mood and suicidal outcomes. MSFsc associations varied across outcomes and were less robust for suicide attempt. Because the sleep descriptors share timing information, mutually adjusted estimates should be interpreted as conditional associations rather than independent causal effects. Longitudinal studies using more detailed circadian and sleep measures are needed.