Component-level meta-analysis finds social support and behavioural activation outperform specialist techniques in non-specialist-delivered mental health care

A systematic review and Bayesian individual participant data component network meta-analysis examined which specific components of task-shared psychosocial interventions — delivered by non-specialist providers rather than specialist psychological therapists — actually drive benefit for adults with common mental disorders such as depression and anxiety Meta-analysis. Of 34 eligible randomised controlled trials, 30 (89%) contributed individual-level data covering 10,612 participants.

Strengthening social support (incremental mean difference -9.48), behavioural activation (-4.15) and problem management (-4.08) were identified as the most beneficial components. By contrast, relaxation (7.97) and, less clearly, cognitive reframing (5.17) were associated with detrimental effects specifically when delivered within a non-specialist, task-shared model — components more closely associated with specialist-style technique. The authors also found component effects varied systematically by baseline symptom severity and other participant characteristics, and have made a personalisation tool available to support component selection.

This is a substantial, methodologically rigorous synthesis — a component-level network meta-analysis of pooled individual participant data is a stronger evidence base than trial-level aggregate comparisons — but it defines efficacy at the level of intervention components rather than testing a single named programme, and ethnicity data were not available across the pooled dataset, which the authors note as a limitation.

For services relying on task-shared or non-specialist-delivered psychological support to extend access beyond specialist capacity, this offers direct, actionable guidance on which components to prioritise in intervention design — and a caution that including specialist-style techniques such as cognitive reframing in a non-specialist delivery model may do more harm than good.