Characterization of the Risk of Bleeding with Novel Oral Anticoagulants and Warfarin: A Pilot Case-Control Study

Introduction: A recent meta-analysis has shown a reduced risk of intracranial hemorrhage (ICH) and an increased risk of gastrointestinal bleeding (GIB) in patients with atrial fibrillation (AF) treated with novel oral anticoagulants (NOACs–dabigatran, rivaroxaban and apixaban) as compared with warfarin [1]. Little is known about the safety of NOACs in the real-life clinical practice. Aim: To assess the risk of NOACs-related overall bleeding, and quantify the risk of ICH and GIB in patients with AF. Methods: A monocentric, retrospective, case–control study was carried out to compare patients with AF, discharged from June 2013 to March 2014 from the Hospital of Lucca with a diagnosis of bleeding (cases) and non-bleeding conditions (controls). Cases and controls were matched for gender and age (±5 years) with a 1:2 ratio. The risk estimates are given as odds ratios (OR). Results: The present analysis included 30 cases (13 males and 17 females; median age: 84.5 years; range 58–96) and 59 controls (26 males and 33 females; median age: 84 years; range 62–92). Both NOACs (OR = 14.25; 95 % CI 2.87–70.68; P = 0.001) and warfarin (OR = 2.71; 95 % CI 1.05–6.99; P = 0.04) were associated with an significant risk of overall bleeding. The risk of GIB was significant in patients treated with NOACs (OR = 21.00; 95 % CI 2.10–210.14; P = 0.01) and not significant in the warfarin group (OR = 0.53; 95 % CI 0.09–3.18; P = 0.49). By contrast, the risk of ICH was not significant in patients receiving NOACs (OR = 2.00; 95 % CI 0.12–34.82; P = 0.63) and significant in patients receiving warfarin (OR = 16.50; 95 % CI 3.06–89.06; P = 0.001). The univariate analysis showed a major distribution of risks factors for bleeding (treatment with medications that increase the risk of bleeding, etc) in the control group. However, owing to the small sample size, a logistic regression analysis to adjust ORs has not be performed. Conclusions: This exploratory analysis showed that NOACs are associated with a significant risk of overall bleeding and confirmed a significant risk for GIB. The high OR estimated for NOACs as compared with that of the warfarin group might depend on a selective prescription (channeling) of NOACs to more susceptible patients. The preliminary results obtained in this study will address the development of a future multicentric study on a larger population.