It was unclear from the review how many studies were included. It appeared that there were 40 studies and some duplication. The included studies had small sample sizes (33 of the 40 included studies had less than 30 participants). The reported number of participants was 1,381, but this was not possible to verify. It appeared that 36 studies (44 comparisons) reported pre-post-treatment data, 22 studies (27 comparisons) reported treatment/control data and eight studies (12 comparisons) reported follow up data.
Overall compared to pre-treatment, group therapy improved post-treatment scores (effect size 0.42, 95% confidence interval: 0.33, 0.49, p=0.001); 1,700 non-significant studies would be required to reverse this outcome. In subgroup analyses, results remained significant in studies that compared a treatment and control (27 studies, effect size 0.24, 95% confidence interval: 0.06, 0.40, p<0.01), studies that provided follow up data (12 studies, effect size 0.42, 95% confidence interval: 0.28, 0.54, p<0.001) and in geriatric patients (30 studies effect size 0.36, p<0.01). Significant heterogeneity was observed for all outcomes.
Group treatments for both functional and organic problems resulted in post-treatment improvement. Group treatment was more effective for functional problems. CBT and reminiscence therapy both resulted in post-treatment improvement, but CBT was significantly more effective. The older the individuals, the less effective the group intervention. Further moderator analyses showed studies that treated patients under 70 years had the best outcomes (effect size 0.55, p<0.001), studies in patients aged between 70 and 76 had the next best outcomes (effect size 0.38, p<0.001) and studies in which the patients were aged over 76 had the weakest outcomes (effect size 0.26, p<0.001); in all groups there was evidence of significant improvement post-treatment.
Publication type (published or unpublished), gender, group characteristics (number of sessions and length of sessions) did not moderate the overall effect.