Abstract
BACKGROUND: The identification of modifiable bleeding risk factors may be of relevance. The aim is to evaluate if aortic stenosis (AS) provides additional information to bleeding risk scores for predicting major bleeding (MB) in non-valvular atrial fibrillation (AF). METHODS: We designed a retrospective multi-center study including 2880 consecutive non-valvular AF patients initiating oral anticoagulation between January 2013 and December 2016. AS was defined as moderate or severe according to European echocardiography guidelines criteria. HASBLED, ATRIA and ORBIT scores were used to evaluate the bleeding risk. MB was defined according to the International Society on Thrombosis and Haemostasia criteria and registered at 18 months of follow-up. RESULTS: 168 (5.8%) patients had AS. Patients with AS had higher risk for MB compared to those without AS (HR = 2.13, 95% CI: 1.40-3.23, P < 0.001). Patients without AS and low-intermediate bleeding risk (0 points) showed the lowest MB rate, whereas the MB rate observed among patients with AS and high bleeding risk (2 points) was the highest one. Discrimination and reclassification analyses showed that AS provided additional information to bleeding risk scores for predicting MB at 18 months of follow-up. CONCLUSIONS: In this population, AS was associated with an increased risk for MB at midterm follow-up. The three scoring systems showed a moderate discriminatory ability for MB. Moreover, the addition of AS was associated with a significant improvement in their predictive accuracy. We suggest that the presence of this valvulopathy should be taken into account for bleeding risk assessment.