Performance of the HEMORR2HAGES, ATRIA, and HAS-BLED Bleeding Risk–Prediction Scores in Nonwarfarin Anticoagulated Atrial Fibrillation Patients
What is the relative performance of the HEMORR2HAGES, ATRIA, and HAS-BLED bleeding risk–prediction scores in patients anticoagulated for stroke prevention in atrial fibrillation with nonwarfarin anticoagulants?
The authors used data from the idraparinux arm of the AMADEUS trial (Evaluating the Use of SR34006 Compared to Warfarin or Acenocoumarol in Patients With Atrial Fibrillation), to compare the effectiveness of three published bleeding risk–prediction scores in patients receiving idraparinux for stroke prevention in atrial fibrillation. They looked at c-indexes and receiver-operating characteristic curve analyses to compare the discriminative ability for the three tests for outcomes of major bleeding, any clinically relevant bleeding, and death. The authors performed Cox regression analysis to assess the association of an abnormal result on the HAS-BLED score, the best performing of the three tests, with the outcomes of major bleeding, any clinically relevant bleeding, and death.
The idraparinux arm of the AMADEUS study included 2,283 subjects (67% men; mean age 70.1 ± years), among whom there were 74 major bleeding events, 46 clinically relevant bleeding events, and 62 deaths. All three scores demonstrated only modest discriminative ability, with an ATRIA score c-index of 0.61 (95% confidence interval [CI], 0.54-0.67) for major bleeding, 0.56 (95% CI, 0.53-0.59) for any clinically relevant bleeding, and 0.65 (95% CI, 0.58-0.73) for death. The HAS-BLED score c-indexes were 0.60 (95% CI, 0.54-0.66), 0.61 (95% CI, 0.58-0.65), and 0.62 (95% CI, 0.55-0.69) for major bleeding, any clinically relevant bleeding, and death, respectively. Finally, the HEMORR2HAGES score demonstrated c-indexes of 0.60 (95% CI, 0.53-0.66), 0.60 (95% CI, 0.56-0.63), and 0.64 (95% CI, 0.57-0.71) for the outcomes of major bleeding, any clinically relevant bleeding, and death, respectively. These were not significantly different. The HAS-BLED and HEMORR2HAGES scores were both superior to the ATRIA score. The HAS-BLED score correctly reclassified 11.6% of the population (95% CI, 3.6-19.7; p = 0.005) compared with the ATRIA score, and the HEMORR2HAGES score correctly reclassified 4.7% (95% CI, 1.8-11.2; p = 0.152) of the population compared with the ATRIA score. A HAS-BLED score of ≥3 predicted bleeding endpoints with hazard ratios of 2.3 (95% CI, 1.1-5.0; p = 0.028), 2.7 (95% CI, 1.9-3.8; p < 0.001), and 2.8 (95% CI, 1.2-6.5; p = 0.013) for major bleeding, any clinically relevant bleeding, and death, respectively.
In this research correspondence letter to the editor, the authors concluded that all three bleeding risk–prediction schemes had similar, modest discriminative performance for the outcome of major bleeding and death, but that the HAS-BLED and HEMORR2HAGES scores were superior to the ATRIA score for the outcome of any clinically relevant bleeding. These results were in accordance with the author’s prior published observations of the relative predictive power of these three bleeding risk–prediction scores in the warfarin arm of the AMADEUS study. The authors further opined that these data suggest that the three leading risk–prediction scores evaluated retain their modest predictive performance in patients receiving nonwarfarin forms of anticoagulation, in spite of the fact that the scores were initially validated in warfarin-treated populations. The authors opine that the data suggest HAS-BLED and HEMORR2HAGES scores were clearly superior to the ATRIA score for the outcome of any clinically relevant bleeding.
Although published as a research correspondence to the editor, this observation provides an important piece of clinical information: Bleeding risk–prediction scores derived in populations of patients treated with warfarin seem to retain their effectiveness in populations treated with nonwarfarin, novel anticoagulants. Furthermore, these data again suggest that the HAS-BLED score has superior predictive efficacy. I hope these data also serve to remind clinicians that, when weighing the clinical propriety of prescribing anticoagulant therapy for stroke prevention in patients with atrial fibrillation, it is just as important to consider the patient’s bleeding risk on anticoagulation as it is to consider the patient’s stroke risk without anticoagulation. Risk scoring tools are available for estimating both sides of this clinical decision making challenge. Atrial fibrillation treatment guidelines should emphasize the importance of estimating bleeding risk in clinical decision making surrounding anticoagulation for atrial fibrillation.
Keywords: Oligosaccharides, Stroke, Hemorrhage
< Back to Listings