Comparative effects of activity-based interventions on quality of life among community-dwelling older adults: A systematic review and Bayesian network meta-analysis.

BACKGROUND: Meaningful activity participation may support healthy aging and quality of life (QoL) in community-dwelling older adults. We compared interaction-based social, exercise-based physical, and self-directed individual activities. METHODS: Thirteen English- and Korean-language databases were searched through May 2026 for RCTs of community-dwelling older adults with a study-sample mean or median age ≥65 years. Interventions were classified using a theoretically informed operational framework. Risk of bias was assessed with RoB 2. Pairwise and Bayesian network meta-analyses estimated standardized mean differences (SMDs); certainty was assessed using GRADE, with SUCRA rankings considered secondary. RESULTS: Seventy-one RCTs (12,159 participants) were included. The overall pairwise effect favored activity-based interventions (SMD=0.59, 95% CI 0.45-0.73), but heterogeneity was substantial and the prediction interval crossed the null; certainty was very low. Category-specific pairwise estimates were 0.68 (0.27-1.08) for self-directed, 0.61 (0.43-0.78) for physical, and 0.46 (0.21-0.72) for social activities, all with very-low-certainty evidence. In the network meta-analysis, low-certainty evidence suggested that all three categories may improve QoL versus no treatment: self-directed SMD=0.78 (95% CrI 0.25-1.32; SUCRA 82.5%), physical 0.73 (0.46-1.00; 79.6%), and social 0.40 (0.11-0.69; 37.6%). Excluding high-risk studies yielded an overall pairwise estimate of 0.53 (0.37-0.69); the estimate for physical activities remained 0.61, whereas the estimates for social and self-directed activities were 0.43 and 0.45. CONCLUSIONS: Evidence does not establish superiority of one activity category. Findings remain limited by heterogeneity, risk of bias, potential publication bias, and sparse direct active comparisons. Comparative rankings should be considered hypothesis-generating, and higher-quality head-to-head trials are needed.