Introduction: Venous thromboembolism (VTE) is a common complication in hospitalized patients that may recur following treatment of the initial event. Patients at risk of recurrence are therefore often prescribed long-term (≥3 months) oral warfarin following resolution of the initial VTE with parenteral anticoagulants. However, recent studies have shown that many patients experience under- or over-anticoagulation while on warfarin. This study assesses patterns of warfarin use and discontinuation in US hospital patients, and analyzes the clinical outcomes associated with warfarin discontinuation.Methods: Patients from the ICHIS/Ingenix managed care database (Jan 2003–Sept 2007) were included in this retrospective analysis if they were aged ≥18 years, were hospitalized for VTE (deep-vein thrombosis [DVT] or pulmonary embolism [PE]) during the study period, and had at least 2 scripts or 60 days of warfarin within the study period. Subjects were also required to have health plan enrollment from a minimum of 180 days before to 365 days following the VTE hospitalization. Patients were excluded if they had non-continuous pharmacy benefit, were aged > 65 years and were not in the Medicare Risk group, or had a diagnosis of atrial fibrillation during the study period. Logistic regression analyses for predictors of clinical outcomes were conducted using explanatory variables including VTE type, plan and payer type, age, geographic region, medical condition, and anticoagulation characteristics.Results: This analysis included 8,380 patients with 4,056 (48.4%) experiencing DVT, 4,319 (51.5%) experiencing PE, and 5 (0.1%) experiencing both PE and DVT. A total of 5,315 (63.4%) patients discontinued warfarin, with a median time to discontinuation of 172 days. Surprisingly, International Normalized Ratio (INR) tests were not performed in 19.8% of patients, and only 39.8% of patients had a last INR that was within the therapeutic range (INR 2–3). When clinical outcomes in all patients were investigated, recurrent VTE was observed in 915 (10.9%) of patients and inpatient bleeding or outpatient visits associated with a bleeding concern were observed in 2,041 (24.4%) of patients. Following logistic regression analysis for predictors of VTE recurrence, warfarin discontinuation conferred a 4% increased risk compared to continued warfarin (Hazard ratio [HR] 1.04, 95% Confidence Intervals [CI] 1.03–1.06), time from initial VTE to warfarin therapy conferred a 1% increased risk per day without warfarin (HR 1.01, 95% CI 1.01–1.01), and increasing duration of index hospitalization conferred a 14% decreased risk per extra day of hospital stay (HR 0.86, 95% CI 0.83–0.88). The major predictor of inpatient or outpatient bleeding was a higher pre-index Charlson Comorbidity Index score (indicating increased patient comorbidity; HR 1.05, 95% CI 1.01–1.15). Patients undergoing abdominal surgery were less likely to have a bleed than patients with other conditions (HR 0.96, 95% CI 0.92–0.99).Conclusions: Patients on warfarin therapy for the prevention of recurrent VTE do not consistently receive therapy that is within the therapeutic range. Recurrent VTE is more likely to occur in patients that discontinue warfarin and in patients that do not receive warfarin in a timely manner following the initial VTE event.