Cumulative ecological risk and depression–anxiety symptom networks in adolescents: a network comparison and intervention simulation study

Depression and anxiety commonly co-occur during adolescence, and cumulative ecological risk (CER) may play an important role in this comorbidity. However, it remains unclear whether CER is associated with differences in the symptom network structure of adolescent depression and anxiety. A cross-sectional survey was conducted among 1451 high school students from four provinces in China between September and December 2025. Depressive and anxiety symptoms were assessed using the Patient Health Questionnaire-9 and the Generalized Anxiety Disorder-7. A CER index was constructed from seven risk indicators across family, school, peer, and community domains. Participants were divided into high- and low-CER groups. Gaussian graphical models were estimated to characterize depression–anxiety symptom networks, and network comparison tests were used to examine between-group differences. Simulation-based analyses were further performed to estimate model-based projected changes in overall symptom burden under hypothetical reductions in individual symptoms. Adolescents exposed to high CER showed a substantially greater overall burden of depressive and anxiety symptoms than those with low CER. However, no significant between-group differences were detected in global network structure ( M = 0.137, p = 0.769) or global strength ( S = 0.035, p = 0.839). Using strength centrality as the primary index, the three highest-ranked nodes in the low-CER group were GAD5 (restlessness), PHQ4 (fatigue), and PHQ2 (loss of interest), whereas those in the high-CER group were GAD3 (excessive worry), PHQ4 (fatigue), and GAD2 (uncontrollable worry). Simulation-based analyses indicated that hypothetical reductions in GAD3, PHQ2, and PHQ1 were associated with the largest model-based projected reductions in overall symptom burden in both groups, with larger projected changes in the high-CER group. Cross-index comparisons showed partial convergence between the simulation-based ranking and conventional centrality indices, particularly for PHQ2 in the low-CER group and GAD3 in the high-CER group. CER was associated with more severe depression–anxiety symptoms in adolescents, but not with significant changes in the overall organization of the comorbid symptom network. These findings suggest that CER may primarily exacerbate symptom severity rather than reshape the global network architecture. The simulated intervention findings identify excessive worry, loss of interest, and depressed mood as hypothesis-generating candidate nodes for future longitudinal and intervention-oriented research. These model-based projections should not be interpreted as evidence of clinical efficacy or as direct treatment recommendations.