Bleeding is the main complication on vitamin K antagonist treatment (VKA), particularly in elderly patients. However, the bleeding risk prediction in geriatric patients remains difficult. We evaluated the predictive value of the HAS-BLED and ATRIA bleeding scores in VKA-treated patients aged 75 and over. Various clinical bleeding risk factors in elderly were also studied. 208 patients were included in a case-control study: 52 hemorrhages cases were compared to 156 hemorrhage-free cases (controls), mean age 83.1 years in cases and 82.6 in controls. This elderly subgroup was provided from the prospective SCORE cohort study (study designed to validate the use of bleeding scores in an ambulatory population). The patients were included during a VKA-therapeutic education between May 2009 and May 2010 in 4 French hospitals, and followed for 1 year. The primary endpoint, collected prospectively, was the occurrence of severe and clinically relevant bleeding events. According to the Receiver operating characteristics (ROC), the ATRIA score was as effective as HAS- BLED to predict all bleeding (c-statistic: 0.59 [95% CI 0.50-0.68] vs 0.56 [0.48-0.65]) including severe bleeding (c-statistic: 0.64 [95% CI 0.49-0.79] vs 0.62 [0.49-0.75]). Multivariate Cox regression analysis showed increasing bleeding risk with anemia (OR = 2.6 [95% CI 1.34-5.23], p = 0.005), serotonin reuptake inhibitors (2.8 [1.08-7.47], 0.034), and family-management of VKA-treatment (2.8 [1.28-6.15], 0.01). ATRIA hemorrhage predictive value can be improved by adding such parameters as family-management of VKA-treatment and serotonin reuptake inhibitors treatment. ATRIA appears as relevant as HAS-BLED in predicting all bleeding including major hemorrhages in elderly patients educated VKA-management. The ATRIA bleeding score is improved by including items of serotonin reuptake inhibitors treatment and family-management of VKA-treatment.
Keywords: ATRIA; HAS-BLED; bleeding; elderly; vitamin K antagonist.