Beyond HAS-BLED in very elderly patients with atrial fibrillation treated with direct oral anticoagulants: predictors of major bleeding in a real-world cohort

Abstract

Bleeding risk assessment in very elderly patients with atrial fibrillation (AF) receiving direct oral anticoagulants (DOACs) remains challenging. The HAS-BLED score is widely used but may have limited applicability in this population. We aimed to evaluate its performance and to identify clinical predictors of major bleeding beyond conventional risk scores. We conducted a retrospective real-world cohort study including consecutive patients aged ≥ 90 years with AF treated with DOACs. The primary outcome was major bleeding. Discrimination was assessed using receiver-operating characteristic (ROC) analysis, and multivariable logistic regression was used to identify predictors and assess incremental value. A total of 487 patients (mean age 93.0 ± 2.6 years, 63.4% female) were included, with major bleeding occurring in 57 patients (11.7%). The HAS-BLED score was not associated with major bleeding (OR 0.83, 95% CI 0.55–1.23; p = 0.34) and showed poor discrimination (AUC 0.46). In multivariable analysis, increasing age (OR 1.14 per year, 95% CI 1.03–1.27; p = 0.014) and anticoagulant type were independently associated with bleeding. Compared with apixaban, rivaroxaban (OR 3.77, p = 0.002) and dabigatran (OR 3.34, p = 0.008) were associated with higher bleeding risk. A clinical model incorporating these factors modestly improved discrimination compared with HAS-BLED (AUC 0.69 vs. 0.46). In very elderly patients with AF receiving DOAC therapy, HAS-BLED has limited predictive value. Bleeding risk appears to be influenced by treatment-related factors, particularly anticoagulant type. Incorporating these factors may improve risk stratification and support a more individualized approach to anticoagulation management in nonagenarians.