Exposure to Systemic Antimicrobials During Pregnancy and Risk of Miscarriage: A Population-Based Registry Study

Abstract

Objective: To estimate miscarriage risk following gestational antimicrobial exposure while addressing biases that have affected previous studies. Design: Population-based cohort study. Setting: Linkage of four nationwide registries: Medical Birth Registry of Norway (MBRN), Norwegian Prescribed Drug Registry (NorPD), Norwegian Patient Registry (NPR) and Norwegian Control and Payment of Health Reimbursements Database (KUHR). Population or Sample: A total of 704 082 pregnancies (2009–2018), with 91 836 (13.0%) exposed to systemic antimicrobials in early pregnancy. Methods: Time-stratified Cox regression models with overlap weights were used, considering time-varying exposures and a 14-day lag to prevent reverse causation. Elective terminations were right-censored to address competing risks, with adjustment for common infections and probabilistic bias analysis for confounding by indication. Main Outcome Measures: Miscarriage and gestational age at miscarriage, captured from NPR, KUHR and MBRN, using the UiO pregnancy algorithm. Results: Nitrofurantoin, pivmecillinam and amoxicillin were not associated with increased miscarriage risk. Metronidazole (HR = 2.00; 95% CI: 1.82–2.21), ciprofloxacin (HR = 1.89; 95% CI: 1.62–2.20), cephalexin (HR = 1.87; 95% CI: 1.57–2.22), fluconazole (HR = 1.61; 95% CI: 1.45–1.78), trimethoprim-sulfas (HR = 1.49; 95% CI: 1.36–1.63) and others showed associations with miscarriage. Probabilistic bias analysis indicated that associations for common antimicrobials may be driven by the underlying infections. Conclusions: Nitrofurantoin, pivmecillinam and amoxicillin did not increase miscarriage risk, but other less commonly used antimicrobials may carry higher risks. By addressing key biases, this study provided a more reliable assessment of miscarriage risks associated with antimicrobial use in early pregnancy.