DOAC Score Development and Validation

Quick Takes

  • Current scores for predicting bleeding risk in patients treated with direct-acting oral anticoagulant (DOAC) medications are limited in their discrimination ability.
  • The novel DOAC score, derived from common demographic, comorbidity, and laboratory data, has modest predictive ability of DOAC-related bleeding.
  • The DOAC score performed better at predicting bleeding than the HAS-BLED score for patients with AF treated with DOAC medication.

Study Questions:

Can a novel score predict bleeding risk in patients with atrial fibrillation (AF) who are taking direct-acting oral anticoagulants (DOAC)?


The authors used the RE-LY trial (Randomized Evaluation of Long-Term Anticoagulation Therapy) comparing dabigatran to warfarin in patients with AF to derive a novel risk score for bleeding among patients randomized to dabigatran therapy. This novel DOAC score was subsequently refined among patients in the GARFIELD-AF registry (Global Anticoagulant Registry in the Field-Atrial Fibrillation), who were treated with any DOAC medication (apixaban, dabigatran, edoxaban, or rivaroxaban). Finally, the score was externally validated in the COMBINE-AF (A Collaboration Between Multiple Institutions to Better Investigate Non-Vitamin K Antagonist Oral Anticoagulant Use in Atrial Fibrillation) pooled clinical trial cohort and the RAMQ administrative dataset from Quebec, Canada. The primary outcome assessed in the derivation and validation was major bleeding. The score was compared to the HAS-BLED score.


From the derivation studies, the DOAC score contained points for age, creatinine clearance, underweight status, prior stroke or peripheral embolism history, diabetes, hypertension, antiplatelet use, nonsteroidal anti-inflammatory drug use, prior bleeding, and active liver disease. In the RE-LY study derivation and internal validation, the C-statistic was 0.73. The score has superior discriminatory ability as compared to HAS-BLED in the RE-LY cohort (C-statistic 0.73 vs. 0.60, p < 0.001) and in the GARFIELD-AF population (C-statistic 0.71 vs. 0.66, p = 0.025). Similar benefit was seen in the COMBINE-AF population (C-statistic 0.67 vs. 0.63, p < 0.001) and the RAMQ population (C-statistic 0.65 vs. 0.58, p < 0.001).


The authors conclude that patients with AF who are treated with DOAC therapy may benefit from use of the DOAC score to help stratify based on the risk of bleeding.


While the use of the CHA2DS2-VASc score is nearly ubiquitous to estimate stroke risk for patients with AF, there has not been widespread use of a bleeding risk score to identify when harm from anticoagulation therapy may outweigh the benefit of stroke risk reduction. This study used several large datasets, both randomized trials and registry/claims datasets, to derive and validate the novel DOAC score for predicting bleeding in patients with AF treated with DOAC medications. Thankfully, the risk score includes elements that can be objectively measured and automated into electronic health records. However, the utility of this novel score for clinical practice remains to be tested and validated. Importantly, the discriminatory ability of this score is only modest. However, this study lays the foundation for a prospective clinical management trial that helps guide stroke prevention strategies in patients with AF.

Clinical Topics: Anticoagulation Management, Arrhythmias and Clinical EP, Anticoagulation Management and Atrial Fibrillation, Atrial Fibrillation/Supraventricular Arrhythmias

Keywords: Anticoagulants, Atrial Fibrillation, Hemorrhage, Risk Assessment, Vascular Diseases

< Back to Listings