Abstract
Background: Depressive symptoms in older adults are associated with adverse psychosocial outcomes. Understanding how depressive symptoms interrelate can enhance intervention strategies. While network analysis has advanced our comprehension of depressive symptom structure, few studies have explored dynamic interactions in older populations. This study examined both cross-sectional and longitudinal networks of depressive symptoms in older adults to identify core symptoms and symptom interactions over time. Methods: Participants aged 60 and older with complete two-wave data (baseline: 2018; follow-up: 2020) from the China Health and Retirement Longitudinal Study (CHARLS) were included (N = 6621). Depressive symptoms were assessed using the 10-item Center for Epidemiologic Studies Depression Scale (CESD-10), administered face-to-face by trained interviewers. Cross-sectional networks were estimated using the Ising model for each time point, and a cross-lagged panel network (CLPN) model was applied to examine longitudinal symptom interactions over time. Network accuracy and stability were assessed through bootstrap procedures. Results: Participants had a mean age of 67.34 years, 52% male, and 93.7% Han ethnicity. "Felt depressed" (r (s) = 1.244 at Wave 1, r (s) = 1.251 at Wave 2) demonstrated the highest strength centrality in both cross-sectional networks. Node strength exhibited strong stability (correlation stability [CS]-coefficient = 0.75 for both waves). The presence of edges (φ = 0.802; p < 0.001) and edge weights (ρ = 0.921, p < 0.001) across two cross-sectional networks showed high reproducibility. In the longitudinal network, "lack of happiness" showed the highest out-expected influence (out-EI; r = 1.404), followed by "felt depressed" (r = 0.994). Both in-expected influence (in-EI) and out-EI showed acceptable stability (CS-coefficient = 0.594). Conclusions: Targeting core symptoms, such as "felt depressed" and "lack of happiness" may disrupt depressive symptom networks and reduce overall depression severity, informing precision interventions in older adults. Clinicians could prioritize these symptoms in screening and treatment. Future research should explore whether symptom-targeted interventions can reshape network structures over time.