Atrial fibrillation bleeding risk and prediction while treated with direct oral anticoagulants in warfarin‐naïve or warfarin‐experienced patients

BACKGROUND: In patients with atrial fibrillation (AF) treated with direct oral anticoagulants (DOAC), bleeding risk scores provide only modest discrimination for major or intracranial bleeding. However, warfarin experience may impact HAS‐BLED (Hypertension, Abnormal renal/liver function, Stroke, Bleeding history or predisposition, Labile international normalized ratio, Elderly (>65 years), Drugs/alcohol concomitantly) score performance in patients evaluated for DOACs, as HAS‐BLED was derived and validated in warfarin cohorts. METHODS: We performed a retrospective cohort study of patients prescribed DOAC for AF in the Veterans Health Administration between 2010 and 2017. We determined modified HAS‐BLED score discrimination and calibration for bleeding, for patients treated with DOAC, stratified by prior warfarin exposure. We also determined the association between DOAC–warfarin‐naïve status to bleeding (nonintracranial and intracranial) with DOAC–warfarin‐experienced patients as reference. RESULTS: The DOAC analysis cohort included 100, 492 patients with AF (age [mean ± SD]: 72.9 ± 9.6 years; 1.7% female; 90.1% White), of which 26, 760 patients (26.6%) and 73, 732 patients (73.4%) were warfarin experienced or naïve, respectively. HAS‐BLED discrimination for bleeds was modest for patients treated with DOAC, regardless of prior warfarin experience (concordance statistics: 0.53–0.59). For DOAC–warfarin‐naïve patients, as compared to DOAC–warfarin‐experienced patients, adjusted risk of intracranial bleeding was lower, while risk of nonintracranial bleeding was higher ( intracranial bleeding propensity adjusted with inverse probability of treatment weights [IPTWs]: hazard ratio [HR]: 0.86, 95% confidence interval [CI]: 0.78–0.95, p = .0040) (nonintracranial bleeding propensity adjusted with IPTW: HR: 1.15, 95% CI: 1.11–1.19, p < .0001). CONCLUSION: Patients’ modified HAS‐BLED score at the time of DOAC initiation, regardless of prior warfarin use, provided only modest discrimination for intracranial and nonintracranial bleeds. These data argue against maintaining DOAC eligible patients on warfarin therapy regardless of modified HAS‐BLED score.