Abstract
BACKGROUND: Depression and anxiety frequently co-occur and exhibit overlapping symptoms and shared psychosocial risk and protective factors. Relationship-specific social support and strain are known to confer protective and risk effects, respectively, yet their role in shaping the symptom-level structure of depression and anxiety remains insufficiently understood. This study employed a network analytic approach to examine the structure of depression-anxiety comorbidity and its associations with perceived stress and relationship-specific social support and strain. We also examined whether these associations varied by level of perceived stress. METHODS: Data were drawn from a community sample of South Korean adults (N = 449; M(age) = 55.17 years; 49.67% male), all of whom were married and had at least one child and one sibling, based on targeted recruitment and inclusion criteria. Participants completed standardized measures of depression (CES-D), anxiety (GAD-7), perceived stress, and social support and strain across four relationship types (spouse, child, friend, sibling). Regularized partial correlation networks were estimated to examine symptom-to-symptom associations, links with psychosocial variables, and identification of central and bridge symptoms. Network Comparison Tests were conducted to evaluate structural differences between low and moderate-to-high perceived stress groups. RESULTS: Depressed affect (CES-D) and nervousness (GAD-7) emerged as the most central symptoms, while bridge symptoms included depressed affect, somatic complaints (CES-D), and trouble relaxing (GAD-7). Perceived stress was the most influential risk factor, and spousal support showed the strongest protective association, particularly with anhedonia (CES-D). In the integrated symptom-social network, stress was broadly linked to both depressive and anxiety symptoms, with the strongest edge observed between stress and anhedonia. Interpersonal problems (CES-D) also served as a key bridge node connecting social strain with internalizing symptoms. Compared to the low-stress group, the moderate-to-high stress group showed greater global network strength, denser clustering of depressive symptoms, and stronger cross-links with anxiety. Additionally, the protective associations between social support and anhedonia were more pronounced, and social strain was more strongly linked to interpersonal problems under elevated stress. CONCLUSIONS: This study highlights the complex symptom-level and social network pathways linking depression, anxiety, and relationship-specific support and strain, particularly under differing levels of perceived stress. Findings underscore the importance of considering stress as a moderator of social influences on internalizing symptoms and suggest that personalized interventions targeting central and bridge symptoms-and strengthening close social ties-may be especially beneficial for individuals under high stress.