Nonlinear associations of BMI and body dissatisfaction with depression and an exploratory decomposition of the indirect association: a cross-sectional study of Chinese university students.

OBJECTIVE: This study aimed to examine the non-linear associations of body mass index (BMI) and body dissatisfaction with depressive symptoms among Chinese university students, and to explore how the indirect association between BMI and depressive symptoms through body dissatisfaction varies across the BMI range. METHODS: A cross-sectional design was used, and 7,928 full-time students were recruited from eight universities in China. Depressive symptoms were assessed with the Zung Self-Rating Depression Scale (SDS); body dissatisfaction was operationalised as the absolute difference between measured BMI and self-reported ideal BMI. Analyses comprised restricted cubic spline (RCS) models, an exploratory non-linear decomposition of the indirect association, generalised additive models (GAM) as a sensitivity analysis, a supplementary receiver operating characteristic (ROC) analysis, and k-means clustering. Because measured BMI enters the calculation of body dissatisfaction, the exposure and the proposed mediator are not mathematically independent; all decomposition results are therefore reported as exploratory. RESULTS: Both BMI and body dissatisfaction were non-linearly associated with depressive symptoms (both P < 0.01). In the exploratory decomposition, the estimated indirect association between BMI and depressive symptoms through body dissatisfaction attenuated progressively as BMI increased. In the supplementary ROC analysis, the BMI value maximising the Youden index was 23.11 kg/m² (AUC = 0.768); this is a sample-specific, internally derived discrimination point rather than a screening threshold. A five-stratum descriptive matrix combining BMI groups with body-dissatisfaction clusters was constructed, and the observed prevalence of depressive symptoms ranged from 14.4% to 100.0% across strata. CONCLUSION: BMI and body dissatisfaction were non-linearly associated with depressive symptoms in this sample, and the magnitude of the indirect association through body dissatisfaction varied across the BMI range. Because the body-dissatisfaction indicator is mathematically derived from measured BMI and the design was cross-sectional, these results describe statistical associations and should not be read as evidence of a causal mediating mechanism. The risk-stratification matrix was developed and evaluated in the same dataset without internal or external validation; it is therefore exploratory and requires independent validation before any screening application.