Performance of HAS-BLED and DOAC scores to predict major bleeding events in atrial fibrillation patients treated with direct oral anticoagulants: A report from a prospective European observational registry

Background: The DOAC score has been recently proposed for bleeding risk stratification of patients with atrial fibrillation treated with direct oral anticoagulants (DOAC). Objective: To compare the performance of HAS-BLED and DOAC score in predicting major bleeding events in a contemporary cohort of European AF patients treated with DOAC. Methods: We included patients derived from a prospective observational registry of European AF patients. HAS-BLED and DOAC scores were calculated as per the original schemes. Our primary endpoint was major bleeding events. Receiver operating characteristic (ROC) curves were used to compare the predictive ability of the scores. Results: A total of 2834 AF patients (median age [IQR] 69 [62–77] years; 39.6 % female) treated with DOAC were included in the analysis. According to the HAS-BLED score, 577 patients (20.4 %) were categorized as very low risk of bleeding, as compared to 1276 (45.0 %) according to DOAC score. A total of 55 major bleeding events occurred with an overall incidence of 1.04 per 100 patient-years. Both scores showed only a modest ability for the prediction of bleeding events (HAS-BLED area under the curve [AUC], 0.65, 95 % confidence interval [CI] 0.55–0.70; DOAC score AUC 0.62, 95 % CI 0.59–0.71, p for difference = 0.332]. At calibration analysis, the DOAC score showed modest calibration, especially for patients at high risk, when compared to HAS-BLED. Conclusion: In a contemporary cohort of DOAC-treated AF patients, both HAS-BLED and DOAC scores only modestly predicted the occurrence of major bleeding events. Our results do not support the preferential use of DOAC score over HAS-BLED.