2G-05 - Benzodiazepine Use in Pregnancy and the Risk of Abortion, Stillbirth, and Preterm Birth: Emulating a Sequence of Target Trials

Background: Maternal use of benzodiazepines has increased in recent years. Studies suggest benzodiazepines use may increase the risk of abortion, stillbirth, and preterm birth. However, these findings are limited by immortal time bias and competing events. To address these limitations, we explicitly emulated a target trial to better assess the risk of abortion, stillbirth, and preterm birth on benzodiazepines among pregnancies.

Objectives: To evaluate the risk of abortion, stillbirth, and preterm birth on benzodiazepines among pregnancies

Methods: We emulated a sequence of 37 target trials of benzodiazepine initiation during gestational week from 0 to 36 weeks, using Taiwan's National Health Insurance Research Database from 2011 to 2021. We identified pregnancies by antenatal care records and ICD codes. In each sequential trial, pregnancies without prior use of benzodiazepines were eligible. Treatment strategies were benzodiazepines use versus no-use at each gestational week. We used propensity score with IPTW to emulate randomization in the target trial. The outcome were abortion, stillbirth, and preterm birth. Follow-up began with treatment assignment and ended at outcome occurrence, delivery, or end of study period. The causal contrast of interest was intention-to-treat. We used pooled log-binomial regression models to estimate relative risk (RR) across the 37 sequential trials. Sensitivity analyses included restricting the cohort to pregnancies with underlying indications to mitigate confounding by indication and applying inverse probability of censoring weights to account for competing risks of pregnancy loss for stillbirth and preterm birth.

Results: Among 46,415 pregnancies initiating benzodiazepines and 368,176 not initiating, the mean age was 32.0 years (SD, 5.4) and 31.5 years (SD, 5.9), respectively. After propensity score weighting, benzodiazepines use was associated with an increased risk of abortion (RR: 1.74; 95% CI: 1.58 to 1.97), but not stillbirth (1.01; 0.92 to 1.18) and not preterm birth (0.95; 0.86 to 1.07). Findings remained consistent after restricting to pregnancies with underlying indications (abortion RR: 1.71; 1.52 to 1.99, stillbirth RR: 1.15; 1.02 to 1.30, preterm birth RR: 0.97; 0.82 to 1.12). After accounting for abortion as competing events, benzodiazepines use was associated with an increased risk of stillbirth (1.22; 1.09 to 1.37) and preterm birth (1.19; 1.05 to 1.32).

Conclusions: Benzodiazepines use in pregnancies was associated with an increased risk of abortion. After accounting for pregnancy losses, benzodiazepines use was associated with an increased risk of stillbirth and preterm birth.