Abstract
AIMS: The direct oral anticoagulant (DOAC) score was recently developed to predict bleeding risk in patients with atrial fibrillation (AF) receiving oral anticoagulants. However, limited data show inconsistent results comparing its performance to the conventional HAS-BLED score in Asian populations with non-valvular AF receiving DOACs. METHODS AND RESULTS: We enrolled 21 142 patients with non-valvular AF receiving DOACs from a multicentre database in Taiwan (June 2012-December 2021). The primary endpoint was major bleeding events. Major bleeding events were defined according to the ISTH criteria. Areas under receiver operating characteristic curves (AUCs) were calculated for each score, with differences assessed using DeLong test. A total of 21 142 AF patients (mean age 75.9 ± 11.0 years; 41% female) treated with DOAC were included in the analysis. Major bleeding events occurred in 681 patients in 1-year follow-up (3.66%/year). There were 82(0.43%/year) intracranial haemorrhage event occurred. Both the DOAC and HAS-BLED scores are associated with a significant risk of major bleeding event, with only modest predictive performance (AUC < 0.7). The DOAC score showed a slightly but statistically significantly higher AUC compared with the HAS-BLED score [AUC: 0.670 (95% CI: 0.650-0.689) vs. 0.642 (0.623-0.663); P < 0.001]. Results from several reclassification analyses favoured the DOAC score. Both the two scores showed a good calibration for the low to intermediate risk categories, while the two bleeding risk scores both overestimate the risk of major bleeding risk for the high risk categories. Subgroup analyses indicated that the superiority of DOAC score over HAS-BLED score is primarily driven by elderly patients (≥75 years) and prediction in risk of gastrointestinal bleeding. CONCLUSION: The DOAC score, which employs a more granular scoring system compared to the HAS-BLED score, may enable finer bleeding risk discrimination among Asian patients with non-valvular AF receiving DOAC therapy.