Seven RCTs (n=700) were included. The number of participants per study ranged from 24 to 216.
For depressive symptoms (6 studies), the pooled effect size for post-treatment outcomes was 0.42 (95% confidence interval, CI: 0.23, 0.60; no evidence of heterogeneity, I2=0%), indicating a statistically significant improvement in depressive symptoms for the intervention compared with control. Only 2 studies provided results for 6-month follow-up assessments, and four for 12 month assessments: no significant effect was seen at either time point.
For incidence of major depressive disorder (4 studies), the pooled IRR was 0.70 (95% CI: 0.47, 1.03; no evidence of heterogeneity, I2=0%), indicating a reduced incidence of major depressive disorder in the intervention groups
There was evidence of publication bias. Adjusted effect sizes and IRRs did not differ significantly from unadjusted results.