Fifty-seven controlled studies with a total of 1,956 treated patients were reported in the review. Most studies (87%) randomly assigned participants to study groups. The mean quality score was 3.47 (standard deviation 0.67) out of a possible maximum score of 4.
Depression: Overall, across the included studies self-rated depression improved on average by d=0.84 standard deviation units and clinician-rated depression by d=0.93. Large effect sizes were reported for cognitive behavioural therapy and reminiscence. Medium effect sizes were reported for psychodynamic therapy, psychoeducation, physical exercise and supportive interventions. Differences in treatment effects were not associated with participant age. Weaker effects were reported for studies with an active control group (as compared to placebo, β (regression coefficient)=-5.3, t (test of significance)=-4.97, p<0.001) and in studies of physically ill or cognitively impaired patients (β=0.22, t=-2.23, p<0.05). Studies where patients exclusively suffered from major depression (compared with other mood disorders) were associated with weaker intervention effects (β=-0.22, t=-2.21, p<0.05). Regression analyses reported that studies with smaller sample sizes had larger effect sizes (β=0.35, t=-2.25, p<0.05). R2 (explained variance)=0.34.
Dropouts: Rates were reported in 50 out of 57 studies with overall rates of 18.9% in intervention groups and 18% in control groups. Dropout rates were higher in studies of group interventions (compared with individual interventions, β=0.36, t=-2.45, p<0.05) and interventions with more sessions (β=0.35, t=1.82, p<0.05). R2=0.22.