Author + information
- Stavros Apostolakis, MD, PhD,
- Deirdre A. Lane, PhD,
- Yutao Guo, MD,
- Harry Buller, MD, PhD and
- Gregory Y.H. Lip, MD⁎ ()
- ↵⁎University of Birmingham Centre for Cardiovascular Sciences, City Hospital NHS Trust, University Department of Medicine, Dudley Road, Birmingham B18 7QH, United Kingdom
To the Editor:
Estimation of bleeding risk is a crucial step in the management of patients with atrial fibrillation (AF). Three bleeding risk–prediction schemes have been derived and validated exclusively in AF populations: HEMORR2HAGES, HAS-BLED, and ATRIA (1,2). In the present analysis, the performance of these 3 schemes was tested in the idraparinux arm of the AMADEUS trial (Evaluating the Use of SR34006 Compared to Warfarin or Acenocoumarol in Patients With Atrial Fibrillation) (3).
A total of 2,283 patients (67% men; age 70.1 ± 9 years) were randomized to the idraparinux arm. Overall, 74 major bleeding events, 346 any clinically relevant bleeding events, and 62 deaths occurred over 311 ± 161 days of follow-up. Specific data for each risk score are shown in Table 1.
The 3 scores demonstrated only modest discriminative ability for all outcomes as reflected by the c-indexes in receiver-operating characteristic curve analysis. The ATRIA score presented c-indexes of 0.61 (95% confidence interval [CI]: 0.54 to 0.68), 0.56 (95% CI: 0.53 to 0. 59), and 0.65 (95% CI: 0.58 to 0.73) for the outcomes of major bleeding, any clinically relevant bleeding, and death, respectively. The HAS-BLED score demonstrated c-indexes of 0.60 (95% CI: 0.54 to 0.66), 0.61 (95% CI: 0.58 to 0.65), and 0.62 (95% CI: 0.55 to 0.69) for the outcome of major bleeding, any clinically relevant bleeding, and death. Finally, the HEMORR2HAGES score demonstrated c-indexes of 0.60 (95% CI: 0.53 to 0.66), 0.60 (95% CI: 0.56 to 0.63), and 0.64 (95% CI: 0.57 to 0.71) for the outcome of major bleeding, any clinically relevant bleeding, and death. Comparison of c-indexes revealed no statistically significant differences in the discriminative ability of the 3 tested scores for the outcomes of major bleeding and death.
The HAS-BLED and HEMORR2HAGES scores were both superior to the ATRIA score for the outcome of any clinically relevant bleeding (HAS-BLED vs. ATRIA, c-index difference 0.054, z-score = 3, p = 0.002; HEMORR2HAGES vs. ATRIA, c-index difference 0.036, z-score = 2.3, p = 0.02). For the outcome of any clinically relevant bleeding, using the HAS-BLED score compared with the ATRIA score correctly (and significantly) reclassified 11.6% of the population (95% CI: 3.6 to 19.7; p = 0.005), whereas using the HEMORR2HAGES score compared with the ATRIA score correctly reclassified 4.7% (95% CI: 1.8 to 11.2; p = 0.152) of the population, although this finding was nonsignificant.
In a Cox regression analysis, a HAS-BLED score ≥3 was a predictor of major bleeding, any clinically relevant bleeding, and death, with hazard ratios of 2.3 (95% CI: 1.1 to 5; p = 0.028), 2.7 (95% CI: 1.9 to 3.8; p < 0.001), and 2.8 (95% CI: 1.2 to 6.5; p = 0.013), respectively (vs. low-risk category as baseline risk).
This is one of the first comparisons of bleeding risk–prediction schemes in a cohort of nonwarfarin anticoagulated patients with AF. All 3 bleeding risk–prediction schemes demonstrated similar modest discriminative performance for the outcome of major bleeding and death. HAS-BLED and HEMORR2HAGES demonstrated superior discriminative performance compared with ATRIA for the outcome of any clinically relevant bleeding. Clinically relevant bleeding was the primary safety endpoint in AMADEUS (which was centrally and blindly adjudicated) and would be clinically meaningful and highly relevant to patients as well as to clinicians who ultimately wish to assess, in everyday clinical practice, those patients who are at risk of important bleeding events. The modest predictive performance of the scores should be interpreted in light of the low bleeding risk population in this clinical trial setting, with the highest categorization into low risk (89.7%) seen for the ATRIA score.
These results are in accordance with our observations in the warfarin arm of the AMADEUS cohort (2), suggesting that despite being initially validated in warfarin-treated populations, the ATRIA, HAS-BLED, and HEMORR2HAGES schemes retain their modest predictive performance in patients receiving other forms of (nonwarfarin) anticoagulation (1), and that the HAS-BLED and HEMORR2HAGES scores were clearly superior to the ATRIA score for the outcome of any clinically relevant bleeding.
Please note: The AMADEUS study was funded by Sanofi-Aventis, but the present analysis received no funding. Dr. Lane has received research funding and/or honoraria for educational symposia from Boehringer-Ingelheim, Bayer HealthCare, and Bristol-Myers Squibb/Pfizer in relation to atrial fibrillation. Professor Buller has served as a consultant to Sanofi-Aventis, Bayer, Pfizer, GlaxoSmithKline, Astellas, Boehringer-Ingelheim, and Daiichi Sankyo. Professor Lip has served as a consultant for Bayer, Astellas, Merck, AstraZeneca, Sanofi-Aventis, ARYx, Portola, Biotronic, and Boehringer-Ingelheim and has been on the speakers' bureau for Bayer, Boehringer-Ingelheim, and Sanofi-Aventis. Drs. Apostolakis and Gao have reported that they have no relationships relevant to the contents of this paper to disclose.
- American College of Cardiology Foundation
- Lip G.Y.,
- Andreotti F.,
- Fauchier L.,
- et al.,
- European Heart Rhythm Association
- Apostolakis S.,
- Lane D.A.,
- Guo Y.,
- Buller H.,
- Lip G.Y.