In this propensity-matched cohort study, we linked nationwide Danish health registries to identify all patients with a first hospital diagnosis of unprovoked VTE who were new users of rivaroxaban or warfarin. Excluded patients included those who had not been residents in Denmark for at least 1 year before VTE diagnosis, patients with outpatient VTE diagnosis only, patients with other indications for oral anticoagulation treatment, patients with previous experience of oral anticoagulation, patients who did not have a prescription for rivaroxaban or warfarin within 7 days of VTE, and patients who redeemed prescriptions for both rivaroxaban and warfarin, or other oral anticoagulants. Primary effectiveness outcome was recurrent VTE and primary safety outcome was major bleeding. We used propensity matching and Cox regression to compare rates of the outcomes with rivaroxaban versus standard treatment.
From Dec 9, 2011, to Feb 28, 2016, we identified 29 963 patients with incident VTE. After exclusion, we identified 1734 propensity-matched patients given rivaroxaban (1751 before propensity matching) and 2945 propensity-matched patients given warfarin. The rate of recurrent VTE at 6 months' follow-up was 9·9 incidents per 100 person-years with rivaroxaban versus 13·1 incidents per 100 person-years with warfarin, yielding a hazard ratio (HR) of 0·74 (95% CI 0·56–0·96). The rate of major bleeding was 2·4 per 100 person-years at 6 months in rivaroxaban users versus 2·0 in warfarin users (HR 1·19, 95% CI 0·66–2·13).
In this clinical practice setting, rivaroxaban in patients with unprovoked VTE was associated with reduced risk of recurrent VTE compared with standard treatment, without compromising safety.
A related editorial pointed out the potential confounding effects of return visits for INR monitoring being interpreted as recurrent events.