When meta-analyses mislead: Promises and pitfalls in teratology counselling

Meta-analysis has become a central tool for evidence synthesis in clinical teratology, offering the promise of increased statistical power, earlier signal detection and consolidated guidance for teratology counselling. When conducted rigorously and interpreted carefully, pooled analyses can provide clinicians with a structured basis for risk communication that no single observational study could achieve alone. However, the methodological assumptions underlying meta-analysis are frequently underappreciated, and their misapplication in teratological research carries real clinical consequences. This review critically examines both the promises and pitfalls of meta-analysis as applied to clinical teratology with particular attention to the implications for clinical risk communication and decision-making in teratology counselling. Drawing on selected examples from both the authors' work and independent research groups, we address key methodological concerns including study quality, heterogeneity, choice of statistical model, outcome classification, publication bias, confounding by indication, protopathic bias, the disproportionate influence of single large studies and the grouping of pharmacologically distinct agents within the same class. We also examine the conditions under which pooling is inappropriate and discuss the risk of outdated evidence shaping current practice. Uncritical reliance on meta-analytic findings may lead to unnecessary discontinuation of beneficial medications, unnecessary termination of otherwise wanted pregnancies or reassurance that exceeds what the available evidence can legitimately support. Translating pooled estimates into meaningful counselling requires communication of absolute rather than relative risk, honest acknowledgement of uncertainty and shared decision-making that incorporates the patient's circumstances, values and reproductive goals. Meta-analysis should inform clinical judgement, not replace it.