Thirteen RCTs were included in the review (1,490 children, range 30 to 316, with 1,081 parents). Four trials reported allocation concealment, outcome assessors were blinded in three trials and an intention-to-treat analysis was performed in three trials. The largest trial had the highest quality level.
The interventions significantly decreased the risk of the children developing the same mental illness as the patient compared with control groups (RR 0.60, 95%CI 0.45 to 0.79; Ι²=0%; six RCTs). The number needed to treat to prevent one incident mental disorder in a child was 17 (95% CI 12 to 33).
The interventions significantly reduced the risk of children developing internalising symptoms compared with control groups (SMD -0.22, 95%CI -0.37 to -0.08; Ι²=0%; seven RCTs), but did not significantly reduce the risk of the offspring developing externalising symptoms (SMD -0.16, 95%CI -0.36 to 0.04; ; eight RCTs; Ι²=51.2%, with moderate heterogeneity).
Results were similar when mothers' reports were used rather than those of fathers. There was little evidence of funnel plot asymmetry for the first two meta-analyses but there was significant asymmetry for the externalising symptoms meta-analysis (p=0.01).
There was no strong evidence that interventions involving parents and children were more beneficial than those in parents alone (meta-regression p=0.09 for internalising symptoms and p=0.11 for externalising symptoms).