Objective: To identify risk factors for bleeding in atrial fibrillation (AF) patients treated with anti-coagulants such as warfarin, apixaban, edoxaban, dabigatran, rivaroxaban using a large claims database.
Methods: A claims database for 8926 AF patients from 2004 to 2016 was obtained from JMDC. Inc. We performed a retrospective cohort study in 2796 Japanese AF patients with 4-month screening and 12-month observation periods. Polypharmacy was defined as prescription of over six drugs. Logistic regression analysis was conducted after stratification based on the presence and absence of cerebrovascular diseases to detect the predictive factors for bleeding.
Results: Polypharmacy was observed in 815 of 2796 (29.1%) patients. A total of 371 AF patients (13.3%) experienced bleeding in the 12-month observation period. Bleeding risk assessment using multiple logistic regression analysis revealed that the odds ratio for the number of co-administered drugs in the elderly (age for ≥60, ≤74) was not significant in those without and with cerebrovascular diseases (1.05 [0.99-1.12], N.S. and 1.10 [0.96-1.27], N.S.). In contrast, in the young (age for <60), the number of co-administered drugs was a significant predictive factor in those without and with cerebrovascular diseases (1.09 [1.03-1.16], p = 0.0054 and 1.20 [1.05-1.36], p = 0.0059). Other observed predictors were"history of bleeding" in young and elderly, but "polypharmacy" and "start from warfarin" were observed in only young.
Conclusion: We determined the bleeding risk in the clinical setting using a large claims database. Physicians and pharmacists need to monitor patients for the initial bleeding signs, particularly in those with these predictive risk factors.