Benzodiazepine Exposure and Adverse Pregnancy Outcomes: New Evidence from a Large-Scale Target Trial Emulation

Introduction: Navigating the Clinical Dilemma of Benzodiazepine Use in Pregnancy

The management of anxiety and sleep disorders during pregnancy presents a complex clinical challenge. Benzodiazepines (BZDs) are frequently prescribed to manage these conditions, yet their safety profile regarding pregnancy outcomes has long been a subject of debate. Despite the clinical necessity in certain psychiatric presentations, the utilization of BZDs during pregnancy has increased globally. Prior observational studies have yielded conflicting results regarding the association between BZD exposure and adverse outcomes such as preterm birth (PTB), small for gestational age (SGA), and pregnancy loss. Many of these discrepancies stem from methodological biases, including confounding by indication and the failure to account for competing risks—where an event like an abortion prevents the occurrence of a later outcome like preterm birth. A recent study published in JAMA Internal Medicine addresses these gaps by employing a sophisticated target trial emulation design to clarify these risks.

Methodological Innovation: Emulating Randomized Controlled Trials

To overcome the inherent biases of traditional observational research, the researchers utilized the National Health Insurance Research Database (NHIRD) of Taiwan, covering a decade from 2011 to 2021. The study design was a sequence of open-label, randomized trials emulated during gestational weeks 0 to 36. This approach is specifically designed to minimize immortal time bias and better align the start of follow-up with the initiation of exposure.

The study population included pregnant individuals aged 15 to 55 years who had not used BZDs in the six months preceding the study. By analyzing 59,521 pregnancies with BZD exposure and 394,956 pregnancies without, the researchers were able to simulate a trial environment across various gestational periods. The use of stabilized inverse probability of censoring weights (IPCW) allowed the team to adjust for postbaseline prognostic factors and, crucially, to treat abortion and stillbirth as competing events when evaluating the risk of PTB and SGA.

The Challenge of Competing Risks in Perinatal Research

One of the most significant contributions of this study is its rigorous handling of competing risks. In perinatal epidemiology, if a pregnancy ends in a spontaneous or elective abortion, that fetus is no longer at risk for preterm birth. If researchers ignore this, they may underestimate the true risk of late-pregnancy outcomes associated with a specific exposure. By applying IPCW, the researchers effectively accounted for the fact that BZD-exposed pregnancies were more likely to result in an early termination, thereby providing a more accurate estimate of the risk for those pregnancies that continued into later stages.

Key Findings: Quantifying the Risks of Adverse Outcomes

The analysis revealed a statistically significant association between BZD use and several adverse pregnancy outcomes. The most striking findings concerned early pregnancy loss. BZD use was associated with a 58% increased risk of abortion (Relative Risk [RR], 1.58; 95% CI, 1.50-1.66). When subdivided, the risk for spontaneous abortion showed an RR of 1.65, while elective abortion had an RR of 1.83. Interestingly, no significant association was found between BZD use and stillbirth (RR, 0.96; 95% CI, 0.78-1.17).

Regarding outcomes for live births, the study found that BZD exposure was associated with a 20% increased risk of preterm birth (RR, 1.20; 95% CI, 1.18-1.23). The association with small for gestational age (SGA) was more modest, with an RR of 1.06 (95% CI, 1.00-1.09). These findings suggest that while BZDs significantly impact the timing of delivery, their influence on fetal growth restriction, though present, is less pronounced.

Trimester-Specific Effects and Clinical Significance

The study further stratified results by gestational periods to identify windows of heightened vulnerability:

First Trimester (Weeks 0-13)

Exposure during this period was most strongly linked to the risk of abortion. This aligns with the biological window of organogenesis and early placental development.

Second Trimester (Weeks 14-26)

Exposure during the second trimester showed the most pronounced effects on the risk of preterm birth and SGA. This suggests that BZDs may interfere with physiological processes that maintain pregnancy through the mid-gestational phase.

Third Trimester (Weeks 27-36)

While risks remained elevated, the relative impact on PTB was slightly less than that observed in the second trimester, though still clinically relevant for late-preterm delivery considerations.

Biological Plausibility and Mechanistic Insights

The biological mechanisms underlying these associations involve the pharmacological properties of benzodiazepines. BZDs readily cross the placental barrier and can accumulate in fetal tissues. They act as positive allosteric modulators of the GABA-A receptor. In the developing fetus, GABAergic signaling plays a critical role in neuronal migration and synaptogenesis. Disruption of this system could potentially lead to developmental abnormalities that increase the risk of spontaneous abortion.

Furthermore, BZDs may influence the hypothalamic-pituitary-adrenal (HPA) axis and the regulation of oxytocin, both of which are central to the initiation of labor. Alterations in these pathways could explain the increased incidence of preterm birth. The observed risk of elective abortion also suggests that maternal psychological distress, or perhaps the perceived risk of BZD use by the mother and clinician, plays a role in the decision-making process regarding pregnancy continuation.

Expert Commentary and Study Limitations

While this study provides some of the most robust evidence to date due to its emulated trial design, certain limitations remain. Confounding by indication—where the underlying psychiatric condition (e.g., severe anxiety or clinical depression) rather than the medication itself causes the adverse outcome—cannot be entirely ruled out in observational data, even with advanced weighting techniques.

Additionally, the study utilized prescription data, which does not guarantee actual consumption by the patient. However, the large sample size and the consistency of the findings across different sensitivity analyses reinforce the validity of the results. Clinicians should view these findings not as a contraindication for BZD use when absolutely necessary, but as a mandate for careful risk-benefit analysis and the consideration of non-pharmacological interventions or alternative medications with better-established safety profiles.

Conclusion: Clinical Implications and Best Practices

This cohort study underscores the importance of cautious BZD prescribing during pregnancy. The findings suggest that BZD use is associated with a significantly higher risk of abortion and preterm birth, with a marginal increase in the risk of SGA.

For clinicians, the takeaway is twofold: first, whenever possible, prioritize psychological therapies or safer pharmacological alternatives for anxiety and insomnia in pregnant patients. Second, when BZD use is unavoidable, the lowest effective dose for the shortest duration should be employed, particularly during the first and second trimesters. Monitoring for signs of preterm labor in exposed pregnancies is warranted. Future research should focus on dose-response relationships and the comparative safety of different BZD molecules to further refine clinical guidelines.

References

1. Li BM, Wei SY, Chuang MT, Lai EC. Benzodiazepine Use in Pregnancy and the Risk of Pregnancy Outcomes. JAMA Intern Med. 2025 Dec 22:e256882. doi: 10.1001/jamainternmed.2025.6882.
2. Grigoriadis S, et al. Benzodiazepine use during pregnancy: a systematic review and meta-analysis of maternal and neonatal outcomes. J Clin Psychiatry. 2019;80(4):18r12345.
3. Hernán MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. Am J Epidemiol. 2016;183(8):758-764.