Abstract 14037: Nonspecific Tools Performed Better in Discriminating Major Bleeding From Direct Oral Anticoagulants Than Specific Tools in a Real-World Population With Atrial Fibrillation

Abstract

Background. Direct oral anticoagulants (DOACs) are the first-line guideline-recommended therapy for preventing stroke in non-valvular atrial fibrillation (AF) patients. DOACs are increasingly being prescribed in the US compared to warfarin because of lower bleeding risk overall, but the proportion of Emergency Department visits involving DOACs-related bleeding increased from 2.3% in 2011 to 37.9% in 2017. Therefore, tools that accurately predict bleeding risk from DOACs are needed. Aim. This study compared the performance of six tools for predicting major bleeding risk in non-valvular AF patients on DOACs. Methods. A retrospective cohort study was carried out with 2,364 Caucasians diagnosed with non-valvular AF and treated with rivaroxaban or apixaban at the University of Michigan Health System. Bleeding risk was calculated using six existing tools: CHA 2 DS 2 -VASc, Anticoagulation-specific Bleeding Score (ABS), HAS-BLED, ATRIA, HEMORR2HAGES, and ORBIT. Major bleeding events were retrieved from medical records and classified according to International Society of Thrombosis and Haemostasis criteria. Discrimination was estimated using the c-statistic, and the areas under the curves (AUC) were compared using the Hanley-McNeil test. RGui was used for all statistical analyses and p-values <0.05 were considered statistically significant. Results. A total of 97 (4.1%) patients experienced major bleeding on DOACs over 2.27 ± 2.03 years of follow-up. HAS-BLED and ABS are tools for specifically predicting major bleeding risk, but both had significantly lower performance than ATRIA, HEMORR2HAGES, and ORBIT ( Figure 1 ). There was no significant difference in the discriminative performance between HAS-BLED and ABS, as well as between ATRIA, HEMORR2HAGES, and ORBIT. Conclusion. ATRIA, HEMORR2HAGES, and ORBIT were the tools that performed better in discriminating major bleeding in AF patients if compared to HAS-BLED and ABS.