Congenital malformation risk following prenatal antipsychotic exposure: a systematic safety surveillance approach

Background Medication safety studies in pregnancy typically focus on selected or composite outcomes (eg, any malformation) to test specific hypotheses or verify suspected safety signals, potentially overlooking or masking other clinically significant associations (eg, individual malformations or malformation types). Objective To conduct a comprehensive, systematic screening for potential teratogenic safety alerts associated with first-trimester exposure to individual antipsychotics, using a tree-based scan statistic (TBSS) approach for simultaneous evaluation of a broad range of specific malformations and malformation groupings. Methods Using a US-nationwide cohort of >4.2 million mother–child dyads (2000–2020), pregnancies with ≥1 first-trimester antipsychotic dispensing were compared with antipsychotic-unexposed pregnancies. Individual congenital malformations were identified via International Classification of Diseases codes, grouped into increasingly aggregated higher-level clinically related categories. Leveraging this hierarchical classification tree , TBSS was used to scan for associations with individual malformation codes and code categories while tightly controlling type 1 error. Confounding was adjusted for via propensity score fine-stratification, and relative risks (RRs) were estimated using an unconditional Poisson scan statistic. The p-values were used to prioritise alerts for further investigation, and follow-up analyses were conducted to refine the understanding of statistical alerts. Findings Exposed pregnancies ranged from 68 (fluphenazine) to 18 366 (prochlorperazine). Antipsychotic-exposed versus antipsychotic-unexposed women tended to be older and to have a higher comorbidity burden and more healthcare encounters. Alerts for an increased risk (with p<0.1) were observed for skin anomaly after haloperidol (RR=1.88) and polydactyly after ziprasidone (RR=3.06) exposure. Results were consistent in sensitivity analyses. The data in the two external data sources were too sparse to confirm a safety alert. Conclusions TBSS identified two potential alerts previously unreported for prenatal antipsychotic exposure. Importantly, no alerts for severe or life-threatening malformations were detected. Findings from this screening-based approach, therefore, align with existing evidence suggesting that antipsychotics are unlikely to be major teratogens. Clinical implications The potential risk increase observed for some less-severe malformations and some antipsychotics needs to be weighed against the high potential for relapse and mental health deterioration following treatment discontinuation. While no consistent patterns suggesting a strong teratogenic effect have emerged thus far for newer antipsychotics, continued monitoring of these medications is important.