We found five studies including 284 participants that met the inclusion criteria, all of which compared parent training with de facto treatment as usual (TAU). One study included a nondirective parent support group as a second control arm.
Four studies targeted children's behaviour problems and one assessed changes in parenting skills.
Of the four studies targeting children's behaviour, two focused on behaviour at home and two focused on behaviour at school. The two studies focusing on behaviour at home had different findings: one found no difference between parent training and treatment as usual, whilst the other reported statistically significant results for parent training versus control. The two studies of behaviour at school also had different findings: one study found no difference between groups, whilst the other reported positive results for parent training when ADHD was not comorbid with oppositional defiant disorder. In this latter study, outcomes were better for girls and for children on medication.
We assessed the risk of bias in most of the studies as unclear at best and often as high. Information on randomisation and allocation concealment did not appear in any study report. Inevitably, blinding of participants or personnel was impossible for this intervention; likewise, blinding of outcome assessors (who were most often the parents who had delivered the intervention) was impossible.
We were only able to conduct meta-analysis for two outcomes: child 'externalising' behaviour (a measure of rulebreaking, oppositional behaviour or aggression) and child 'internalising' behaviour (for example, withdrawal and anxiety). Meta-analysis of three studies (n = 190) providing data on externalising behaviour produced results that fell short of statistical significance (SMD -0.32; 95% CI -0.83 to 0.18, I2 = 60%). A meta-analysis of two studies (n = 142) for internalising behaviour gave significant results in the parent training groups (SMD -0.48; 95% CI -0.84 to -0.13, I2 = 9%). Data from a third study likely to have contributed to this outcome were missing, and we have some concerns about selective outcome reporting bias.
Individual study results for child behaviour outcomes were mixed. Positive results on an inventory of child behaviour problems were reported for one small study (n = 24) with the caveat that results were only positive when parent training was delivered to individuals and not groups. In another study (n = 62), positive effects (once results were adjusted for demographic and baseline data) were reported for the intervention group on a social skills measure.
The study (n = 48) that assessed parenting skill changes compared parent training with a nondirective parent support group. Statistically significant improvements were reported for the parent training group. Two studies (n = 142) provided data on parent stress indices that were suitable for combining in a meta-analysis. The results were significant for the 'child' domain (MD -10.52; 95% CI -20.55 to -0.48) but not the 'parent' domain (MD -7.54; 95% CI -24.38 to 9.30). Results for this outcome from a small study (n = 24) suggested a long-term benefit for mothers who received the intervention at an individual level; in contrast, fathers benefited from short-term group treatment. A fourth study reported change data for within group measures of parental stress and found significant benefits in only one of the two active parent training group arms (P ≤ 0.01).
No study reported data for academic achievement, adverse events or parental understanding of ADHD.
