Long-term stroke and bleeding risk in patients with atrial fibrillation treated with oral anticoagulants in contemporary practice: Providing evidence for shared decision-making

Abstract

Background: Oral anticoagulation is recommended as a lifelong therapy for most patients with atrial fibrillation (AF). However, data on long-term outcomes in clinical practice on these drugs are scarce, particularly for the recently approved agents. We aimed to describe differences in characteristics between patients in everyday practice and those enrolled in the pivotal trials, and to report long-term outcomes on oral anticoagulation in practice.

Methods: We performed a retrospective cohort analysis using a large U.S. administrative database to identify patients with AF initiating oral anticoagulation and examine incident stroke (effectiveness endpoint, including ischemic stroke and systemic embolism) and major bleeding (safety endpoint).

Results: We identified 107,373 patients with AF initiating anticoagulants 7/1/2006-6/30/2016. These patients were more likely to be elderly, female, or to have advanced kidney disease in comparison to those enrolled in the trials. The event rates for major bleeding (3.1%, 2.8%, 4.0% and 4.9%/year for in apixaban-, dabigatran-, rivaroxaban- and warfarin-treated patients, respectively) were higher than those observed in trials. The event rates for stroke 0.9%, 1.0%, 0.9% and 1.4%/year the four drug cohorts), were similar to the trials. The three-year risk of stroke was 2.3%, 2.1%, 2.3% and 3.5%, and the three year risk of major bleeding was 5.4%, 7.0%, 8.2%, and 11.7% in the four drug cohorts.

Conclusions: Clinical trials represent a narrow spectrum of the general AF population. The trials may underestimate the bleeding risk observed in practice. This study provides important data to help clinicians communicate expected outcomes to patient during shared decision-making.

Keywords: Anticoagulation; Atrial fibrillation; Major bleeding; Non-vitamin K antagonist oral anticoagulant; Stroke; Warfarin.