Maternal antipsychotic use during pregnancy and congenital malformations

Abstract

Existing data may underestimate the potential teratogenic effects of prenatal antipsychotic exposure because of lacking data on miscarriages and induced abortions. This study aimed to present a comprehensive analysis based on information on pregnancies ending in termination, miscarriage, stillbirth, and live birth. We conducted a population-based cohort study in Denmark of clinically recognized singleton pregnancies with the first-trimester scan performed from 2008 to 2017. We compared the risk of major malformations between pregnancies exposed to antipsychotics in the first trimester and unexposed pregnancies. In secondary analyses, the comparison was made with pregnancies of women who used antipsychotics before but not during pregnancy (discontinuers). We used weighted log-binomial regression to estimate adjusted prevalence ratios and propensity score fine stratifications for confounding control. We performed 4 sensitivity analyses, including a sibling-controlled analysis. Of the 503,158 pregnancies, 1252 (0.2%) were of women who filled an antipsychotic prescription in the first trimester. Major malformations were present in 7.3% of antipsychotic-exposed pregnancies, 5.1% of unexposed pregnancies, and 6.0% of discontinuers' pregnancies. The adjusted prevalence ratio was 1.23 (95% confidence interval, 1.01-1.50) among exposed pregnancies compared with unexposed pregnancies. The prevalence ratio was attenuated to 1.14 (95% confidence interval, 0.88-1.48) compared with discontinuers and 1.08 (95% confidence interval, 0.47-2.49) in the sibling analysis. Similar findings were observed with cardiac malformations. Results were consistent for classes and individual antipsychotics, and remained robust across the 4 sensitivity analyses. Our findings suggest limited or no overall teratogenic effect of first-trimester antipsychotic exposure. For individual antipsychotics, with estimations based on very few cases, further studies with sufficient sample sizes are warranted.