Doxycycline exposure during the first trimester of pregnancy may be associated with a higher adjusted relative risk of pregnancy loss compared with alternative antibiotics among patients treated for respiratory tract infections. However, findings among patients treated for acne differed between databases.
In a retrospective, active-comparator cohort study, investigators used 2005 to 2021 Merative MarketScan commercial insurance claims and 1999 to 2019 Medicaid claims to analyze 218,234 pregnancies among patients with respiratory tract infections and 6,636 among those with acne. Eligible patients were aged 12 to 55 years and had filled prescriptions for doxycycline or comparator antibiotics during the first trimester. The investigators matched pregnancies by gestational age and propensity scores to account for measured differences between treatment groups, including demographic characteristics, obstetric history, comorbidities, and health care utilization.
The primary outcome was pregnancy loss, defined as spontaneous abortion before 20 weeks of gestation or stillbirth at 20 weeks or later. The investigators compared doxycycline with amoxicillin or macrolides among patients with respiratory tract infections and with topical clindamycin among those with acne. Additional analyses examined the components of pregnancy loss separately, gestational timing, potentially unrecognized pregnancies, and differences in infection severity. Follow-up began 7 days following the initial prescription to reduce potential reverse causation.
Among patients with respiratory tract infections, doxycycline exposure was associated with a 54% higher adjusted relative risk of pregnancy loss vs. amoxicillin in MarketScan and a 40% higher risk in Medicaid. Pregnancy loss occurred in 32% of doxycycline-exposed pregnancies vs. 20% of amoxicillin-exposed pregnancies in MarketScan and 27% vs. 19% in Medicaid.
Comparisons with macrolides yielded similar findings. Doxycycline exposure was associated with a 39% higher adjusted relative risk of pregnancy loss in MarketScan and a 36% higher risk in Medicaid. Pregnancy loss occurred in 32% vs. 22% of pregnancies in MarketScan and 27% vs. 20% in Medicaid.
Among patients treated for acne, results were less consistent. In MarketScan, doxycycline exposure was associated with a 49% higher adjusted relative risk of pregnancy loss compared with topical clindamycin. Pregnancy loss occurred in 27% of doxycycline-exposed pregnancies vs. 18% of comparator pregnancies. However, the Medicaid analysis showed no statistically significant association, with pregnancy loss occurring in about 20% of pregnancies in both groups.
Sensitivity analyses generally supported the primary findings. Associations persisted when the investigators delayed follow-up, restricted analyses to potentially unrecognized pregnancies, or limited the respiratory tract infection cohorts to patients with upper respiratory tract infections. Analyses restricted to antibiotic exposure at 6 weeks of gestation or later showed stronger associations in the respiratory tract infection cohorts. Stillbirth-specific findings were inconclusive because of small numbers of events.
In a post hoc quantitative bias analysis, adjusted risk estimates remained elevated even under assumptions of higher rates of unintended pregnancy among doxycycline recipients. Prescribed doxycycline doses differed by indication, with 200 mg daily predominating among patients with respiratory tract infections and lower doses more common among those treated for acne. The investigators could not determine whether differences in medication adherence, systemic exposure, population characteristics, or residual confounding explained the inconsistent acne findings.
The study had several limitations. Prescription claims did not confirm medication use, and inaccuracies in estimated gestational age could have resulted in exposure misclassification. Early pregnancy losses not resulting in medical encounters may have been missed. Although matching and statistical adjustment addressed numerous measured characteristics, residual confounding related to infection severity, acne severity, pregnancy intention, and other unmeasured factors remained possible. The study did not evaluate congenital malformations, and its observational design prevented causal conclusions.
Physicians should consider the observed association with pregnancy loss when weighing the potential benefits and risks of doxycycline during early pregnancy. Doxycycline may remain appropriate for certain life-threatening conditions, including anthrax, when maternal benefits outweigh potential pregnancy risks.
“This study provides additional data to inform clinical recommendations for the treatment and postexposure prophylaxis of infections during pregnancy and helps bridge the evidence gap for prescribers when evaluating the benefits and risks of doxycycline for their pregnant patients,” wrote lead study author Almut G. Winterstein, PhD, of the University of Florida, and colleagues.
Full disclosures of the study authors can be found in the study.
Source: Annals of Internal Medicine
