Common antimicrobials appear safest in pregnancy

Nitrofurantoin, pivmecillinam, and amoxicillin were not linked to higher miscarriage risk.
*Population-based cohort study; Level 2 (OCEBM).

Citation

Boissiere-O'Neill T, van Gelder MMHJ, Engjom HM, Nordeng HME. Exposure to systemic antimicrobials during pregnancy and risk of miscarriage. BJOG. 2026;133:1394-1404. doi:10.1111/1471-0528.18155

Background

Antimicrobials are commonly prescribed during pregnancy, but prior studies of miscarriage risk have been limited by timing bias and confounding from the infection being treated. This study used nationwide Norwegian registries and methods designed to reduce these biases.

Patients

704,082 pregnancies in Norway from 2009 to 2018. Ectopic and molar pregnancies, very early failed in vitro fertilisation pregnancies, and pregnancies exposed to known harmful medicines were excluded.

Intervention

Systemic antibacterial, antifungal, or antiprotozoal medicines dispensed in early pregnancy.

Control

No exposure to the specific antimicrobial during the same pregnancy risk period.

Outcome

Miscarriage before 20 weeks, identified through linked national health registries.

Follow-up Period

From last menstrual period through 20 gestational weeks.

Results

ExposureWeighted association with miscarriage
NitrofurantoinLower risk: hazard ratio 0.75 (95% confidence interval 0.66-0.84)
PivmecillinamLower risk: 0.91 (0.87-0.95)
MetronidazoleHigher risk: 2.00 (1.82-2.21)
CiprofloxacinHigher risk: 1.89 (1.62-2.20)
CephalexinHigher risk: 1.87 (1.57-2.22)
FluconazoleHigher risk: 1.61 (1.45-1.78)
Trimethoprim-sulfa medicinesHigher risk: 1.49 (1.36-1.63)

Amoxicillin was not associated with miscarriage. Analyses treated exposure as changing over time, used a 14-day lag to reduce reverse causation, and censored elective terminations. Bias analyses suggested several higher-risk findings may reflect the underlying infection rather than the medicine itself.

Limitations

Observational design limits causal conclusions. Infection type and severity were incompletely measured. Very early miscarriages without health care contact were missed. Hospital-administered drugs, over-the-counter fluconazole, adherence, body mass index, and socioeconomic status were not fully captured.

Funding

University of Oslo and Norwegian Research Council; no industry funding.

Clinical Application

Continue usual first-line choices for urinary infection in pregnancy; avoid assuming higher-risk antimicrobial associations are causal without considering infection severity.