Accounting for multiple births in systematic reviews of randomised trials: a methodological systematic review

OBJECTIVE: Multiple births are common in randomised trials targeting preterm populations. Clustering due to multiple births is often overlooked in individual trials and may impact the results of meta-analyses that pool their results. We aimed to assess how multiple births have been handled in the reporting and meta-analyses of recent systematic reviews. DESIGN: We conducted a methodological systematic review of Cochrane and non-Cochrane systematic reviews. The search was conducted on 10 September 2024 in the Cochrane Database of Systematic Reviews and PubMed for articles published in the previous 12 months. Reviews were eligible if they involved randomised trials of interventions delivered in pregnancy or infancy, included multiple births and reported results of at least one aggregate data meta-analysis for an infant outcome. RESULTS: After screening 222 articles, 39 had unclear eligibility due to making no mention of multiple births and nine met the eligibility criteria (five Cochrane and four non-Cochrane reviews). Multiple births were inconsistently handled across included reviews. The degree of clustering due to multiple births was poorly described and meta-analyses accounting for clustering were rarely reported (2/9 reviews; 22%). CIs around pooled treatment effect estimates were wider after accounting for clustering. CONCLUSIONS: Clustering due to multiple births is a poorly recognised issue in systematic reviews and meta-analyses. Given the potential for this clustering to alter conclusions about the effectiveness of interventions, we recommend accounting for clustering due to multiple births in future meta-analyses.