Thirteen RCTs (n=1,177; 161 CBT versus control and 1,016 pharmacotherapy versus control) were included in the review. The sample sizes ranged from 5 to 105 participants, and the majority of studies were based on intention-to-treat analysis.
CBT (5 study arms) and pharmacotherapy (10 study arms) were significantly more effective than controls: ES 1.45 (95% CI: 0.68, 2.22, p=0.0002) for CBT and ES 0.48 (95% CI: 0.36, 0.61, p<0.00001) for pharmacotherapy. However, significant statistical heterogeneity was detected in studies comparing CBT with controls.
The subgroup analysis did not significantly alter the results, although the wait-list group showed a larger treatment effect than the placebo group. Sensitivity analyses involving the removal of outliers and subgroups of CBT studies also did not significantly alter the results, although subgroups of pharmacotherapy study arms showed a non significant pooled ES for completers.
The authors stated that Egger's test showed no evidence of publication bias.