I read with great interest the paper by Ellis et al. [1], which is a retrospective cohort study to determine the incidence of bleeding in patients with atrial fibrillation (AF) beginning dabigatran, rivaroxaban or warfarin during 3-year. Bleeding incidences were calculated per 100 patient-years of treatment, and they were 3.9 (95% confidence interval (CI), 3.6–4.4) in warfarin patients, 4.2 (95% CI, 3.7–4.7) in dabigatran patients, and 4.1 (95% CI, 3.0–5.3) in rivaroxaban patients. In addition, intracranial hemorrhage (ICH) rates were 0.71 (95% CI, 0.56–0.90) in warfarin patients, 0.4 (95% CI, 0.18–0.87) in dabigatran patients, and 0.27 (95%CI, 0.10–0.80) in rivaroxaban patients.
