Abstract
IMPORTANCE: Previous studies have suggested that social participation helps prevent depression among older adults. However, evidence is lacking about whether the preventive benefits vary among individuals and who would benefit most. OBJECTIVE: To examine the sociodemographic, behavioral, and health-related heterogeneity in the association between social participation and depressive symptoms among older adults and to identify the individual characteristics among older adults expected to benefit the most from social participation. DESIGN, SETTING, AND PARTICIPANTS: This cohort study included participants aged 65 years or older in the 2016, 2019, and 2022 waves of the Japan Gerontological Evaluation Study (JAGES), a nationwide survey examining social determinants of health and social environment in Japan. All participants were functionally and cognitively independent. Analyses were performed from February 2024 to February 2025. EXPOSURES: Social participation recorded in the 2019 JAGES. Participation included involvement in hobby groups, sports groups or clubs, volunteer clubs, senior citizen clubs, and neighborhood communities more often than once a month. MAIN OUTCOMES AND MEASURES: The main outcome was depressive symptoms defined as a score of 5 or higher on the Geriatric Depression Scale (GDS; score range: 0-15, with higher scores indicating more severe depressive symptoms) in the 2022 JAGES. After 1:1 propensity score matching on the status of social participation in the 2019 JAGES to adjust for characteristics in the 2016 JAGES, a machine learning-based causal forest algorithm was used to assess the heterogeneity in the association between social participation and depressive symptoms. Conditional average treatment effect (CATE) was estimated for each individual, and the median of the CATE distribution was used as a threshold to categorize participants by their characteristics into the high-benefit group or low-benefit group. RESULTS: Among the 11 146 propensity score-matched older individuals (mean [SD] age, 72.40 (5.15) years; 5762 men [51.7%]; 15.7% developed depressive symptoms), social participation was associated with reduction in the risk of depressive symptoms (2.2 [95% CI, 0.9-3.5] percentage points). The causal forest model showed heterogeneity across individual characteristics. Compared with the low-benefit group, the high-benefit group were more likely to be older and have lower socioeconomic status (lower educational level and household income). The reduction in risk of depressive symptoms associated with social participation was observed in the high-benefit group (4.1 [95% CI, 1.3-6.9] percentage points) but not in the low-benefit group (0.4 [95% CI, -2.0 to 2.8] percentage points). CONCLUSIONS AND RELEVANCE: In this cohort study, the association between social participation and depressive symptoms varied across individual characteristics. Promoting social participation, particularly for older adults with lower socioeconomic status, would play a role in not only improved mental health but also decreased mental health disparities during the global population aging.