(272) Incidence of pregnancy among publicly and commercially insured childbearing-age women with warfarin treatment

Background: Due to potential teratogenicity, warfarin is contraindicated in pregnancy except in patients with mechanical heart valves. Currently, there are no Risk Evaluation and Mitigation Strategies from regulatory agencies to prevent pregnancy among childbearing-age women on warfarin treatment. Since 45% of pregnancies in the US are unplanned, warfarin exposed pregnancies are likely to occur.

Objectives: To estimate the incidence of pregnancy among publicly and commercially insured childbearing-age female warfarin users and to examine pregnancy outcomes among those who become pregnant.

Methods: Using nationwide MarketScan (2003-2020) and Medicaid (2000-2018) databases, we identified a cohort of 12- to 55-year-old women who initiated warfarin. Warfarin initiation was defined as filled prescription of warfarin and we used 14-day gap periods to define the discontinuation of warfarin. The period from warfarin initiation until warfarin discontinuation was regarded as a treatment episode and the same individual could have multiple treatment episodes.

We followed patients until warfarin discontinuation, pregnancy start date, end of enrollment or end of study period, whichever occurred first. We matched two cohorts by age and estimated (1) the incidence rate of pregnancy among warfarin users stratified by indications, and (2) the proportion of non-live births out of all pregnancies. The 95% confidence intervals (CI) were estimated using bootstrapping.

Results: We identified 467,034 and 697,551 warfarin treatment episodes in MarketScan and Medicaid, respectively (mean age: 45.3 in MarketScan and 41.7 in Medicaid). The incidence rates of pregnancy among warfarin-exposed and 8.8 (95% CI: 8.4-9.3) per 1,000 person-years in MarketScan, and 18.6 (18.1-19.2) per 1,000 person-years in Medicaid. The incidence rate among warfarin-exposed decreased to 10.9 (10.4-11.4) in Medicaid after matching age distribution to that among warfarin-exposed in MarketScan. Stratified by indication, the incidence rate was 12.1 in MarketScan and 17.8 in Medicaid among patients with venous thromboembolism (VTE), and 5.4 in MarketScan and 7.1 in Medicaid among patients with valve disorders. In both cohorts, about 45% of warfarin-exposed pregnancies ended with non-live births (spontaneous abortion: 27.3% in Medicaid, 30.3% in MarketScan, termination: 14.2% in Medicaid, 12.5% in MarketScan, and stillbirth: 1.4% in Medicaid, 0.8% in MarketScan).

Conclusions: It is not uncommon for childbearing-age women to become pregnant while on warfarin treatment, with a higher incidence among patients with VTE than patients with valve disorders. The risk of non-live birth was higher among warfarin-exposed pregnancies compared with the risk in the general population previously reported.