One hundred and twenty-four studies were included: 53 studies of pharmacotherapy (n=1,811), 47 studies of psychosocial treatments (n=7,725) and 24 studies that used both treatments (n=1,747).
(C)BT versus no treatment control: (C)BT was more effective than no treatment in reducing anxiety (ES 0.87, 95% CI: 0.71, 1.03) and depression (ES 0.72, 95% CI: 0.54, 0.90) and in increasing quality of life (ES 0.85, 95% CI: 0.48, 1.21). No significant heterogeneity was found for any of these meta-analyses.
CBT versus BT: treatment had similar effects on anxiety, but CBT improved depression (ES 0.18, 95% CI: 0.01, 0.35) compared with BT in studies directly comparing these interventions. No significant heterogeneity was found for either of these meta-analyses.
Pharmacotherapy versus placebo: pharmacotherapy was more effective than placebo in reducing anxiety (ES 0.38, 95% CI: 0.31, 0.45) and depression (ES 0.34, 95% CI: 0.21, 0.47) and in improving quality of life (ES 0.35, 95% CI: 0.22, 0.48), but significant heterogeneity was detected for all of these meta-analyses. Further analysis showed no difference in efficacy between BZDs, TCAs and SSRIs. Smaller studies had larger ESs. There was evidence of publication bias.
(C)BT versus pharmacotherapy: (C)BT was as effective as pharmacotherapy for anxiety and depression in the 11 studies directly comparing these interventions, but significant heterogeneity was found for these meta-analyses.
(C)BT alone versus (C)BT plus pharmacotherapy: there was little difference between treatments for the main meta-analyses, but the results varied with the study characteristics included in the sensitivity analyses.
Attrition: the drop-out rate was 15.1% with (C)BT, 18.3% with BT, 12.7% with CBT and 20.4% with pharmacotherapy. There were significantly more drop-outs with pharmacotherapy in comparison with psychotherapy (P<0.05).