Definitions of key realist terminology.

Group psychosocial interventions can be a scalable treatment for depression, anxiety, and post-traumatic stress disorder (PTSD) in low- and middle-income countries (LMICs) but effects are heterogeneous. Little is known about how intervention mechanisms and context interact to generate different outcomes in different populations. We conducted a realist systematic review, combining traditional systematic review methods with analysis of intervention mechanisms and contextual factors to further understanding of complex interventions. This involved: (i) a scoping review to build initial theory and inform data extraction and analysis; (ii) systematically searching six databases for randomised controlled trials (RCTs) of group psychosocial interventions for participants aged 0–19 years in LMICs (21 November 2022, updated 26 April 2024); (iii) extracting data on outcomes, contextual factors, and intervention mechanisms to build hypotheses about how context interacts with mechanisms to generate outcomes (context-mechanism-outcome configurations; CMOCs); (iv) narratively synthesising CMOCs into wider programme theories about how and why psychosocial interventions work; (v) meta-analyses and meta-regressions to assess trends and test CMOCs. We included 38 RCTs with data for 6,086 participants (52% female, mean age 13). These data informed 14 programme theories including theories that interventions work best when adapted for participant cognitive level, incorporating age-appropriate activities, accounting for local gender-specific issues, and being culturally adapted. Pooled post-intervention effect sizes were -0.72 (95% CI -1.01 to -0.42, p