Recent large randomized controlled trials have shown that direct oral anticoagulants (DOACs) are at least as effective as vitamin K antagonists (VKAs) for prevention of stroke or systemic embolism in patients with non-valvular atrial fibrillation (AF) and are associated with similar or lower rates of bleeding. The results for bleeding seen in Phase 3 trials might not be applicable to real world practice. Population-based studies suggest that the bleeding risk for DOACs and VKA is similar however neither the risk of bleeding associated with different doses of DOACs nor that associated with apixaban in routine clinical practice is well established. We performed a large population-based study to determine the incidence of bleeding in patients with AF beginning treatment with different doses of dabigatran, rivaroxaban, apixaban or a VKA.


From the computerized database of the 4.5 million member Israeli Clalit Health Services health care provider, consecutive patients initiating a VKA or DOAC for AF between January 1, 2011 and December 31, 2014 were studied. Bleeding patients who required hospitalization were identified and key clinical and laboratory data were recorded. Incidence of bleeding was calculated during the first 20 months of treatment which was the minimum duration of treatment for all of the drugs. Adjusted hazard ratios for overall bleeding, intracranial hemorrhage (ICH) and gastrointestinal (GI) bleeding and all-cause mortality within 30 days were recorded and case fatality rates were calculated as the proportions of bleeding patients who died within 30 days.


26184 patients initiated anticoagulants for AF: 14258 received VKA, 214 -received dabigatran 75 mg, 3563 received dabigatran 110 mg , 1410 received dabigatran 150 mg, 2570 received rivaroxaban 15 mg, 2140 received rivaroxaban 20 mg, 1227 received apixaban 2.5 mg and 802 received apixaban 5 mg. Key patient demographics and the overall and site-specific bleeding rates are shown in table 1. Hazard ratios for any bleeding, ICH and GI bleeding adjusted for age, renal failure, CHADS2 score, aspirin use and Charlson comorbidity score favored dabigatran 150 mg versus VKA (P<0.05). The case fatality rate for VKA bleeding was 11,4%, dabigatran 110mg-10.5%, dabigatran 150 mg- 6.25%. rivaroxaban 15mg- 15.5%, rivaroxaban 20 mg- 10%, apixaban 2.5 mg- 11.4% and for apixaban 5mg-7.14%


The results of our population-based non-randomized study of unselected AF patients demonstrate a decreased bleeding rate for dabigatran 150mg and an increased bleeding rate for apixaban 2.5 mg compared to VKA. There was a consistent tendency for increased bleeding in patients receiving lower versus higher doses of the NOACs which probably reflects physician tendency to select lower doses of DOACs for patients at greater risk for bleeding.


Ellis:Boehringer Ingelheim: Speakers Bureau; Bayer: Speakers Bureau; Pfizer: Speakers Bureau. Eikelboom:Pfizer: Honoraria, Research Funding; Eli Lilly: Honoraria, Research Funding; Boehringer Ingelheim: Honoraria, Research Funding; Daiichi-Sankyo: Honoraria, Research Funding; Bayer: Honoraria, Research Funding; Astra Zeneca: Honoraria, Research Funding; Bristol Myer Squibb: Honoraria, Research Funding; Sanofi-Aventis: Honoraria, Research Funding; Janssen: Honoraria, Research Funding.

Author notes


Asterisk with author names denotes non-ASH members.