Four RCTs (256 participants, range 30 to 95) were included in the meta-analysis. All of the included trials reported use of random assignment to treatment allocation. Sequence allocation was unclear in all trials. No trials reported adequate details for blinding of outcome assessors. Details of dealing with incomplete outcome data were unclear in three trials. Three studies were reported to be at risk of other potential threats to study bias.
There were no significant differences for post-treatment self-rated depression between behavioural therapy and a waiting list control (three RCTs, Ι²=78%), cognitive therapy (four RCTs, Ι²=43%) or brief psychodynamic therapy (two RCTs, Ι²=0%).
Behaviour therapy was significantly more effective than a waiting list control for post-treatment clinician-rated depression (weighted mean difference -5.68, 95% CI -7.71 to -3.66; three RCTs, Ι²=0%). But there were no significant differences between behavioural therapy and cognitive therapy (four RCTs, Ι²=21%) or brief psychodynamic therapy (two RCTs, Ι²=0%).
Sensitivity analyses (where there were sufficient studies) reported similar results to the overall analyses. Reported drop-out rates did not differ significantly between intervention and cognitive or psychodynamic therapy groups. No comparisons of drop-outs for intervention compared to waiting list control were reported.