Seventy-five studies were included in the review. Thirty-two studies (1,892 participants) were included in the analysis of BPT, while 45 studies (2,745 participants) were included in the analysis of CBT.
Two studies that met the inclusion criteria were excluded from the meta-analysis after a preliminary analysis of post-treatment ES showed their values to be outliers.
The meta-analysis showed that BPT and CBT had small to moderate beneficial effects.
The overall post-treatment ES was 0.4 (95% CI: 0.34, 0.47), based on 71 studies (30 BPT and 41 CBT). There was significant heterogeneity between the studies. The average quality score was 3.3 out of 6. Higher quality studies produced a significantly lower ES. After controlling for study quality, BPT had a significantly higher ES than CBT in the subset of studies with participants aged 6 to 12 years (7 BPT and 21 CBT). The differential effect between BPT and CBT was not seen when the intervention setting was included as a control variable in the regression. The overall follow-up ES was 0.22 (95% CI: 0.11, 0.34), based on 17 studies (4 BPT and 13 CBT).
For BPT, the overall post-treatment ES was 0.47 (95% CI: 0.34, 0.61); there was significant heterogeneity between the 30 studies. Sufficient data were available to conduct a regression analysis on age and intervention approach: after controlling for study quality neither showed a significant relationship with ES. There were insufficient data to calculate the follow-up ES for BPT.
For CBT; the overall post-treatment ES was 0.35 (95% CI: 0.25, 0.47); there was significant heterogeneity between the 41 studies. Sufficient data were available to conduct a regression analysis on age and ethnicity: after controlling for study quality there was a significant positive relationship between ES and age. The follow-up ES was 0.31 (95% CI: 0.13, 0.48).