Which components of talking therapy actually help? Large meta-analysis identifies the ones that work — and two that may not

A Bayesian component-network meta-analysis pooled individual participant data from 34 randomised controlled trials (10,612 participants) testing task-shared psychosocial interventions — delivered by trained non-specialist workers rather than specialist psychologists — for adults with depression or anxiety Meta-analysis (Papola et al.1). Strengthening social support, behavioural activation and problem management were identified as the most beneficial individual components. Relaxation, and less clearly cognitive reframing, were associated with a detrimental effect on outcomes.

The analysis also found component effects varied systematically by baseline symptom severity and other participant characteristics, though ethnicity data were not available for this analysis.

For services relying on task-shared delivery models to extend psychological care under specialist capacity constraints, this evidence suggests which components of a manualised programme are doing the work — informing which elements to prioritise, rather than assuming any such programme performs equally well regardless of its component mix.