Abstract 171: Warfarin Use And Stroke Risk Among Patients With Nonvalvular Atrial Fibrillation In A Large Managed Care Population

Abstract

Introduction Warfarin discontinuation among real world nonvalvular atrial fibrillation (NVAF) patients is common. Hypothesis We hypothesized that in a managed care population, warfarin discontinuation is associated with increased stroke risk. Methods Patients who initiated warfarin therapy between Jan 2005 and Jun 2009 and had a healthcare claim related to AF within 30 days prior to the first warfarin claim, but no evidence of valvular disease, were included. Warfarin discontinuation was defined as a supply gap of >60 days without evidence of International Normalized Ratio (INR) measurements. Follow-up was divided into periods of warfarin treatment and discontinuation. Stroke events were identified based on claims for inpatient stays with a primary diagnosis for stroke or transient ischemic attack (TIA). Cox proportional hazards models were constructed to assess the relationship between warfarin discontinuation and incident stroke while adjusting for baseline demographics, stroke and bleeding risk, and comorbidities, as well as time-dependent antiplatelet use, stroke, and bleeding events in the prior warfarin treatment period. Results The mean (SD) age of the study sample (N=16,253) was 67±12 years; 64.8% was male. Mean CHADS2 score was 1.84±1.30; mean HAS-BLED score was 2.00±1.18. Half (51.4%) of patients discontinued warfarin therapy at least once and the overall sample had a mean of 1.87 warfarin treatment periods during a mean of 668 days follow-up. Approximately 1186 patients (7%) had a stroke or TIA at any site of service during follow-up. Risk of stroke significantly increased during warfarin discontinuation periods compared with therapy periods (HR 1.60; 95%CI 1.35-1.90; P<.0001). Stroke risk was significantly increased for patients with baseline CHADS2 score ≥2 (HR 2.69; 95%CI 1.44-5.03; P=.002), HAS-BLED score ≥3 (HR 1.46; 95%CI 1.05-2.05; P=.027), or who had a bleeding (HR 1.29; 95%CI 1.06-1.57; P=.013) or stroke event (HR 3.04; 95%CI 2.47-3.75; P<.0001) in the prior treatment period. Conclusions In the real world, over half of patients on anticoagulation therapy had treatment gaps or permanently discontinued therapy. These usage patterns, as well as prior bleeding, were associated with increased stroke risk.