The review included 23 trials, sample sizes ranged from 16 to 337. Thirteen trials compared group-based CBT with usual care to usual care alone (1,247 participants), six compared group-based CBT to individually-delivered CBT (211 participants), four compared other therapies. There was evidence of risk of bias because no trials were blinded and randomisation procedures and allocation concealment were generally not reported.
Group-based CBT was found to have been more effective for treating depression than usual care alone immediately post-treatment (SMD -0.55, 95% CI -0.78 to -0.32, 14 trials) and in the medium to long-term (SMD -0.47, 95%CI -0.87 to -0.08, three trials), but was not quite statistically significant in the short-term (SMD -0.47, 95%CI: -1.06 to 0.12, three trials). Individually-delivered CBT was more effective than group-based CBT immediately post-treatment (SMD 0.38, 95%CI: 0.09 to 0.66, seven trials) but no difference was identified at longer follow-up times. For interventions other than CBT no conclusions were drawn because of the small number of studies and their methodological weaknesses.
There was generally substantial heterogeneity across trials, with five of the six meta-analyses performed with Ι² of 40% or more.