To analyze the impact of integrating randomized controlled trials (RCTs) and non-RCT designs in child psychiatry meta-analyses (MA), focusing on the influence of study design and PICO-similarity on effect size, and the effectiveness of a bias-corrected Bayesian model.

A systematic MEDLINE search identified child psychiatry MA published before March 2023 that included both RCT and non-RCT data. Initial qualitative assessments of confidence interval (CI) overlaps and conclusions were conducted, followed by a meta-epidemiological analysis to quantify design-related effect size relative to RCTs. Subgroup analyses based on PICO-similarity were also performed. The bias-corrected model's effectiveness was evaluated through CI overlaps and comparing effect size.

The search yielded 22 meta-analyses, encompassing 406 articles, of which 17 were suitable for our meta-epidemiological analysis. For this analysis, we focused on MAs that reported results in standardized mean differences, as continuous outcomes were more prevalent in our results. Consistent confidence intervals were found across study designs, but significant variations in conclusions and effect size emerged. Meta-epidemiological analysis revealed design-specific differences in effect size: 0.88±0.11 for all non-RCTs, 0.64±0.11 for non-randomized comparative trials, and 1.47±0.20 for comparative single-arm studies (Fig. 1). No significant difference was found for cohorts vs. RCTs. Subgroup analysis indicated that meta-analyses with lower PICO-similarity showed larger effect size differences. The bias-corrected model mitigated disparities from including various designs and appeared to enhance overall result precision.

The scarcity of RCTs in child psychiatry underscores the importance of integrating diverse study designs in meta-analyses to expand the knowledge base. Careful article selection and bias adjustment are crucial in this field to mitigate inaccuracies and inflated effect sizes. This approach enriches our understanding, supporting the development of evidence-based treatments in pediatric mental health.