Fourteen studies comprising 799 participants were included in the review.
Twelve studies used random assignment and eight of these included a control group (waiting-list, placebo or no treatment). The others compared two or more interventions.
The mean ES was 0.77 (95% confidence interval, CI: 0.55, 0.98). Heterogeneity was not found. 'Orwin's fail-safe N' was 40.
Several studies did not use a control group. Hence, the authors calculated the ES on the basis of the improvement from pre- to post- test. The mean ES was 1.07 (95% CI: 0.92, 1.22). Heterogeneity was not found. 'Orwin's fail-safe N' was 104.
A test was performed to examine if the ES in the controlled studies differed from that in the studies without a control group. There was a small, non significant difference between the two ESs (1.14 and 1.06). The ES in the 6 studies in which participants were selected on the presence of a formally diagnosed major depression, was larger (ES 1.25) than that in the studies without such a selection criteria (ES 0.90); this difference was not significant.
Four studies compared the effects of CBT with 7 other therapies. There was a non significant trend that CBT was more effective than the other therapies (ES 0.20). Five studies compared behaviour therapy with 8 other therapies, but it was not significantly more effective than any of them (ES 0.05).
Data were presented at 1, 3 and 6 months follow-up in several studies. For each of these periods, the ESs from post-test to follow-up were calculated. The results indicated that the effects remained stable for 1 to 6 months.
A multiple regression analysis was performed in order to identify those characteristics of the participants and outreach programmes that were significantly related to the outcome. Only participation in a cognitive behavioural intervention reached a significant level, indicating that the effects of cognitive behavioural interventions are larger.
The mean drop-out rate of the interventions was 0.23 (standard deviation 0.13; range: 0 to 0.47). The multiple regression analysis found four significant predictors of drop-out: participation in a group; participation in a cognitive behavioural intervention; the percentage of female participants; and the number of sessions. These four variables accounted for as much as 94% of the variance (F=28.47, p=0.0002).