LAUSR

OBJECTIVE The risk of severe uterine bleeding (SUB) reported in oral anticoagulant (OA) clinical trials1 is likely underestimated due to the low representation of women of reproductive age. In a previous letter, 2 we reported SUB incidence in women who received OA treatment using real-world data. In this study, we examined the incidence of SUB among women presenting uterine bleeding (UB), with and without ongoing OA (warfarin or non-vitamin K oral anticoagulants) exposure, to contextualize our previous findings. STUDY DESIGN Using the US Food and Drug Administration's Sentinel System, 3 we summarized the incidences as proportions of women receiving same-day transfusion (SUB-T) or gynecologic surgeries within 30 days (SUB-S) in all eligible women who had a first UB diagnosis (index date) between 10/19/2010 and 9/30/2015, preceded by no such diagnosis and continuous medical and prescription drug coverage in the recent 6 months. The Sentinel System encompasses administrative data, health insurance claims, and electronic health records routinely collected from inpatient, outpatient, and institutional care settings by 16 national and regional data partners.4 We differentiated OA users and nonusers by the presence of 1) OA exposure (evidenced by lasting days supply of outpatient pharmacy dispensing) on the index date and 2) diagnosis of atrial fibrillation/flutter or venous thromboembolism in the 183 days prior to or on the index date (baseline period). Women who experienced joint replacement at baseline were excluded. RESULTS We identified 80,967 OA users and 4,245,694 nonusers. OA nonusers were younger (mean age 44.1 vs 69.3 years) upon their UB diagnosis and more likely utilized intrauterine device (IUD, 1.7% vs 0.7%) or oral contraception (14.6% vs 2.2%) at baseline (Table). Overall, SUB-T incidence was higher in OA users than nonusers (2.6% vs 0.4%), but the SUB-S incidence was comparable (7.3% vs 7.2%). When stratified by age, OA users consistently had higher SUB-T and SUB-S incidences than nonusers in reproductive ages (Figure). The differences in SUB-T incidences were in fact significant across all age groups (standardized differences < -0.1). Within age group of 18-50 years, prior exposures to IUD and oral contraception were both associated with lower SUB incidence (SUB-T: IUD or oral contraception use 0.1% vs no use 0.4%; SUB-S: IUD 4.0% vs 7.0%, oral contraception 4.7% vs 7.4%) for OA nonusers; however, these trends attenuated among OA users. In contrast, history of gynecological disorders was associated with higher SUB incidence regardless of OA use in the same age group. Significantly higher SUB-T incidences in OA users persisted across all assessed strata. CONCLUSIONS Compared to nonusers, OA users had higher SUB incidences in reproductive ages. Despite this finding, the differences were significant only for transfusion and not for surgical outcomes. Future studies should confirm the relationships of lower SUB incidences with recent IUD and oral contraception use and examine whether OA use alters SUB risk by outcome type in women with history of gynecological disorders. Women need to be aware of SUB risk while on OA therapy. Our study obtained incidence estimates from a curated healthcare claims database with national coverage. Limitations include indirect linkage of transfusion or gynecological surgeries to the UB event, dispensing-based proxy for OA exposure, and absence of clinical presentation of UB diagnosis (i.e., quantity of bleeding). Incidence estimates may not be generalizable to women without UB diagnosis. Advanced confounding adjustment technique such as propensity score analysis is recommended to assess comparative SUB risk between OA users and nonusers in quasi-experiment setting.