ORAL COMMUNICATION | Comparison of HAS-BLED and DOAC score for bleeding risk prediction in patients with non-valvular atrial fibrillation treated with direct oral anticoagulants: a real-world cohort study.

Introduction. Bleeding risk assessment is recommended in patients with atrial fibrillation (AF) receiving oral anticoagulant therapy. HAS-BLED is the most commonly used score, but it was developed in vitamin K antagonist–treated populations and shows limited predictive ability. As direct oral anticoagulants (DOACs) are now first-line therapy in non-valvular AF, a DOAC-specific bleeding risk score was proposed in 2023, called the “DOAC score." This study aimed to compare the predictive accuracy of the two scores. Materials and Methods. We conducted a retrospective observational study including patients with non-valvular AF treated with DOACs and followed at the Anticoagulation Center of Pavullo Hospital (AUSL Modena). HAS-BLED and DOAC scores were calculated. Predictive performance for major bleeding (MB) and for the combined endpoint of MB and clinically relevant non-major bleeding (CRNMB) was assessed using ROC curve analysis. Results. The analysis included 228 patients (mean age 80.7 ± 8.6 years; 41.7% women). Major bleeding occurred in 16.2% of patients, while 25.0% experienced clinically significant bleeding. Both scores showed modest predictive ability for major bleeding (AUC 0.63 for the DOAC score and 0.59 for the HAS-BLED). The DOAC score demonstrated superior performance for both major bleeding and the combined endpoint of MB and CRNMB (AUC 0.66 vs 0.57; p = 0.02). Conclusions. In a real-world cohort of patients with non-valvular AF treated with DOACs, the DOAC score outperformed HAS-BLED in predicting bleeding events, despite overall modest discrimination.