The Pooled Effect of Non-Pharmacological Interventions on Loneliness in Older Adults: A Meta-Analysis of Recent Meta-Analyses

Abstract

Published meta-analyses on loneliness interventions for older adults tend to agree that non-pharmacological programs help, yet the effect sizes they report differ considerably from one paper to the next. This study pooled six such meta-analyses, published between 2021 and 2026 and covering psychosocial, social-support, and technology-assisted interventions, to arrive at a single combined estimate rather than leaving readers to reconcile six competing figures on their own. Each source meta-analysis’s reported effect size was extracted, converted to a common sign convention so that negative values consistently reflect greater loneliness reduction, and combined through a random-effects (DerSimonian-Laird) inverse-variance model. The resulting pooled standardized mean difference was -0.52 (95% CI [-0.82, -0.22]), Z = -3.39, p < .001, a moderate and statistically reliable overall benefit. Between-study heterogeneity was substantial (I² = 73.2%), reflecting how differently the six source meta-analyses defined and measured “intervention.” A follow-up moderator analysis splitting the six sources by intervention type found that this heterogeneity traced largely to a single distinction: psychosocial and mixed-delivery programs produced a strong, consistent effect, while technology-only interventions did not reach significance on their own. Together the results support treating loneliness in older adults as a genuinely modifiable outcome while cautioning that not all intervention formats are equally effective. Limitations specific to pooling already-pooled estimates, including likely overlap in the underlying primary-study samples, are discussed.