Low bleeding acceptance is associated with increased death risk in patients with atrial fibrillation on oral anticoagulation

Bleeding is the most feared complication of anticoagulation. We sought to investigate whether the bleeding risk acceptance has a prognostic value during long-term follow-up in the era of direct oral anticoagulants (DOACs) among patients with atrial fibrillation (AF). We studied 167 consecutive AF outpatients [aged 68.8 SD 10.6 years; 141 (84.4%) on DOACs]. The bleeding acceptance was assessed based on the Bleeding Ratio defined as the declared maximum number of major bleedings that a patient would be willing to accept to prevent one major stroke. We recorded cerebrovascular ischemic events, major or clinically relevant non-major bleeds (CRNMB), and mortality. A median Bleeding Ratio was 4 (IQR 2–5). During follow-up of 946 patient-years, cerebrovascular ischemic events and/or death were observed in 28 patients (3.3%/ year) and major bleeding or CRNMB in 33 (4.6%/ year). The Bleeding Ratio was lower in patients who experienced cerebrovascular events or death (p = 0.004), but not bleeding. Patients with the Bleeding Ratio 0–3 were more often non-persistent to the OAC therapy, and more likely to have cerebrovascular event or die than those with higher bleeding acceptance (odds ratio 2.55; 0.95% CI 1.08–6.02) which was driven by the impact on mortality. The multiple Cox proportional hazards model showed that lower Bleeding Ratio, higher CHA$_{2}$DS$_{2}$-VASc score, and older age predicted cerebrovascular events or death during follow-up. AF patients who are willing to accept fewer serious bleedings to avoid major stroke during anticoagulation are more likely to experience death and/or cerebrovascular ischemic events, but not bleeding, what might be related to non-persistence.