Predictive accuracies of CHA2DS2-VASc and HAS-BLED, and anticoagulation quality in relation to thromboemblism and bleeding in patients with mechanical heart valves

Background: The variability of international normalized ratio (INR) is considered a risk factor in patients with mechanical heart valves (MHV)and INR target range still remains unestablished. CHA2DS2-VASc andHAS-BLED are risk stratification schemas designed for atrial fibrillation. Their ability to discriminate thromboembolism (TE) and bleeding for patients with MHV have never been investigated. Methods: We conducted a complete study of all patients with MHV followed in two centres. A total of 407 and 140 patients respectively were followed between 2008-2011. Data on INR, time in therapeutic range (TTR), variability, CHA2DS2-VASc and HAS-BLED were extracted. Results: The mean (±SD) age in centre 1 and 2 were 70 (14) and 61 (14) respectively. The target INR range for all MHV was 2-4 in centre 1 and 2-3 in centre 2 and mean INR was 2.9 (0.31) and 2.6 (0.19) respectively. The incidence of TE was 3.4 per 100 patient-years in centre 1 and 1.4 per 100 patient-years in centre 2, p=0.035, and for major bleeding 4.9 per 100 patient-years and 1.7 per 100 patient-years respectively, p=0.005. TheHAS-BLED score correlated well with bleeding, with area under the curveof 0.63 (95% confidence interval [CI]: 0.57-0.70). CHA2DS2-VASc had 0.56 (95% CI: 0.48-0.64) for TE. Adding atrial fibrillation as a risk factor did not improve the c statistic. INR variability (SD), comparing the 3rd tertile with the first had (Odds ratio [OR]: 4.05; 95% [CI]: 2.09-7.84) for major bleeding and (OR: 2.01; 95% [CI]: 1.0-3.99) for TE. INR SD was higher with a higher mean INR and target range 2-4 (p<0.001) andindependently predicted bleeding. Conclusion: HAS-BLED predicted bleeding with discriminatory ability similar to previous reports to atrial fibrillation whilst CHA2DS2-VAScpredictive ability for TE was modest. Higher INR intensity is associated with higher variability, which correlates to primarily bleeding, but also TE. Some of the difference between the groups considering TE can be accounted for more significant risk factors in the centre 1 cohort. A more narrow INR target range could be recommended to reduce variability ofanticoagulation.