Thirteen controlled studies (n=812) and 12 uncontrolled studies (n=261) were included in the review. Eight of the controlled studies were randomised controlled trials (RCTs). No RCTs used intention-to-treat analyses. Reporting of drop-outs varied substantially across controlled studies.
When pooling all controlled studies, interventions were associated with a significant improvement in symptoms for bereavement and grief reactions in children and adolescents (Hedges’s g 0.35, 95% CI 0.15 to 0.57; 13 studies). Significant heterogeneity was observed for this outcome (I2=92.84%).
When pooling all uncontrolled studies, interventions were associated with a significant improvement in symptoms for bereavement and grief reactions in children and adolescents (Hedges’s g 0.49, 95% CI not reported; 12 studies). Significant heterogeneity was observed for this outcome (I2=62.17%).
Subgroup analyses showed that in both controlled and uncontrolled studies there were larger effect sizes of interventions for symptomatic or impaired children and adolescents compared with those without symptoms. Results of other subgroup analyses were reported.
Sensitivity analyses did not significantly alter the results. The fail-safe N showed little evidence of publication bias. Regression analyses showed possible moderators of treatment effects: symptom severity, age, amount of treatment, time since bereavement, confrontation and publication bias.