Differential associations of aging attitude risk with depression, anxiety, and suicide-related risk among older Chinese adults: a network analysis and GGMNIRA-based simulated intervention study

Abstract

Attitudes toward aging reflect how individuals understand and appraise their own aging process and may be closely associated with mental health in later life. Previous studies have primarily examined associations of attitudes toward aging with depression, anxiety, and suicide-related risk at the total-score level, whereas symptom-level conditional associations remain insufficiently studied. This study aimed to examine the conditional associations of aging attitude risk (AAR) with specific depressive and anxiety symptoms and suicide-related risk, and to explore the projected importance of different nodes using GGMNIRA-based simulated intervention analysis. This cross-sectional study included 1,493 older Chinese adults. Attitudes toward aging, depressive symptoms, anxiety symptoms, and suicide-related risk were assessed using the Attitudes to Ageing Questionnaire (AAQ), PHQ-9, GAD-7, and the five-item Suicide-Related Risk Questionnaire (SRA-5), respectively. An 18-node Gaussian graphical model was estimated using EBICglasso, and strength centrality (Strength), expected influence (EI), direct cross-domain edges of AAR, and bootstrap accuracy and stability were evaluated. GGMNIRA-based simulated intervention analysis was then applied to the GGM by implementing equal-intensity simulated interventions on each node’s conditional mean. Kullback–Leibler (KL) divergence was used to quantify the model-implied change in the joint Gaussian distribution and to describe projected importance. AAR showed modest conditional associations with anhedonia (w = 0.096, 95% CI [0.051, 0.140]), SRA-5 (w = 0.080, 95% CI [0.028, 0.131]), depressed mood (w = 0.068, 95% CI [0.026, 0.110]), and guilt/worthlessness (w = 0.047, 95% CI [0.006, 0.089]); bootstrap edge-weight difference tests among the four edges did not support a statistically distinguishable ordering. Higher Strength values were observed mainly for anxiety activation symptoms and core depressive symptoms. In the GGMNIRA analysis, restlessness had the largest KL value (29.217), followed by nervousness (25.267), trouble relaxing (22.840), anhedonia (21.998), and depressed mood (21.210). AAR showed symptom-specific conditional associations with selected depressive symptoms and suicide-related risk rather than uniform associations with all symptoms. Anxiety activation symptoms and core depressive symptoms were more prominent in terms of network connectivity and GGMNIRA projected importance. These conditional associations and GGMNIRA findings provide testable research hypotheses for subsequent longitudinal and intervention studies.