Association between maternal benzodiazepine or Z-hypnotic use in early pregnancy and the risk of stillbirth, preterm birth, and small for gestational age: a nationwide, population-based cohort study in Taiwan

Benzodiazepines and Z-hypnotics are commonly prescribed for anxiety and insomnia during pregnancy, but the evidence regarding potential adverse neonatal outcomes is insufficient because of poor control for confounding factors in previous studies. We therefore aimed to evaluate the association between the use of benzodiazepines or Z-hypnotics during early pregnancy and adverse neonatal outcomes (stillbirth, preterm birth, and small for gestational age).We did a nationwide, population-based cohort study in Taiwan using three data sources: Taiwan's National Birth Certificate Application database, the National Health Insurance database, and the Maternal and Child Health Database. The study cohort included all singleton pregnancies of females aged 15-50 years who gave birth between Jan 1, 2004, and Dec 31, 2018. Pregnancies without valid information were excluded. Benzodiazepine and Z-hypnotic use was defined as at least one benzodiazepine or Z-hypnotic prescription during early pregnancy (the first 20 weeks of pregnancy). The primary outcomes were stillbirth (fetal death at or after 20 weeks' gestation), preterm birth (<37 weeks' gestation), and small for gestational age (birthweight below the 10th percentile for gestational age by sex). Logistic regression models with propensity score fine stratification weighting were used to control for potential confounders and examine the association between benzodiazepines or Z-hypnotics use during early pregnancy and the risk of adverse neonatal outcomes. Odds ratios (ORs) and 95% CIs were reported. We used confounding by indication control analyses, a sibling control study, and a paternal negative control design to account for unmeasured confounders. The risk associated with exposure during late pregnancy was also assessed.Between Oct 7, 2021, and June 10, 2022, we analysed the study data. The cohort included 2 882 292 singleton pregnancies; of which, 75 655 (2·6%) of the mothers were dispensed one or more benzodiazepines or Z-hypnotics during early pregnancy. Women exposed during pregnancy were older (mean age at delivery was 31·0 years [SD 5·3] for exposed women vs 30·6 years [4·9] for unexposed women), had a higher prevalence of psychiatric disorders, and were more likely to have unhealthy lifestyle behaviours than unexposed women. Information about ethnicity was not available. Early pregnancy exposure was associated with adverse neonatal outcomes compared with non-exposure. The propensity score-weighted OR was 1·19 (95% CI 1·10-1·28) for stillbirth, 1·19 (1·16-1·23) for preterm birth, and 1·16 (1·13-1·19) for small for gestational age. After controlling for confounding by indication, there was no significant association between drug exposure and stillbirth risk; however, this attenuation was not observed for preterm birth and small for gestational age. In models with sibling controls that accounted for familial confounding and genetic factors, early exposure to benzodiazepines or Z-hypnotics was not associated with an increased risk of stillbirth and preterm birth, but it remained significantly associated with small for gestational age. The paternal negative control analyses with point estimates close to the null indicated no strong evidence of unmeasured confounding shared by the mother and the father. Substantially increased risks of stillbirth and preterm birth were observed for late pregnancy exposure.Benzodiazepine or Z-hypnotic use in early pregnancy is not associated with a substantial increase in the risk of stillbirth and preterm birth after accounting for unmeasured confounding factors. Clinicians should be aware of the increased risk of small for gestational age and caution should be taken when prescribing these medications during late pregnancy.National Science and Technology Council, Taiwan.For the Taiwanese translation of the abstract see Supplementary Materials section.