Venous thromboembolism (VTE) carries a considerable risk of recurrence and anticoagulants should be administered for at least 3 months, but the optimal duration of treatment remains uncertain for most patients. We aimed to explore the actual duration of anticoagulant treatment in clinical practice by using the database of an international, prospective registry on patients treated for VTE, RIETE. Information was collected on baseline characteristics, risk factors for VTE and bleeding, and on therapeutic strategies. Treatment duration was censored at >12 months and multivariate analysis using logistic regression was performed to identify predictors of treatment duration. A total of 8295 patients were enrolled; 29.9% had transient risk factors, 26.3% had cancer, and 43.7% had unprovoked VTE. Median duration of treatment was 390 days in patients with unprovoked events, 282 days in patients with transient risk factors and 181 days in cancer patients (P<0.001). After the exclusion of patients who died during the first year of observation, the rate of patients treated for >12 months was 62.4%, 46.8%, and 46.9%, respectively (P<0.001). After multivariate analysis, unprovoked VTE, prior VTE, pulmonary embolism at presentation, and age >65years were independently associated with treatment for >12 months; body weight <75kg, anemia, and occurrence of major bleeding were associated with treatment for <12months.

We conclude that the duration of VTE secondary prevention is heterogeneous in clinical practice and does not entirely follow current recommendations. A substantial proportion of patients with transient risk factors receives long-term anticoagulation and may be exposed to an unnecessary risk of bleeding.