Time Trends in Patient Characteristics of New Rivaroxaban Users with Atrial Fibrillation in Germany and the Netherlands

Background: Use of the direct oral anticoagulant rivaroxaban has strongly increased in Europe since its market approval for non-valvular atrial fibrillation in 2011. Patients characteristics of rivaroxaban initiators may have changed over time but this has not been investigated so far. Objective: We aimed to describe time trends of patient baseline characteristics among new rivaroxaban users with non-valvular atrial fibrillation from 2011 to 2016/17 in two European countries. Methods: We used data from Germany (German Pharmacoepidemiological Research Database) and the Netherlands (PHARMO Database Network). We included new rivaroxaban users with (i) a first dispensing between 2011 and 2016/17, (ii) ≥ 2 years of age, and (iii) a diagnosis of non-valvular atrial fibrillation and described their baseline medication and comorbidity prior to starting rivaroxaban stratified by year of inclusion. Results: Overall, 130,652 new rivaroxaban users were included during the study period (Germany: N = 127,743, the Netherlands: N = 2909). The sex ratio and median age remained relatively stable over time. The proportion of patients without prior use of oral anticoagulants before initiation of rivaroxaban increased in both countries between 2011 and 2016/17 (Germany: from 51 to 76%, the Netherlands: from 57 to 85%). In Germany, we observed a relative decrease by 27% in the proportion of new rivaroxaban users with a history of ischemic stroke and by 18% in the proportion with a transient ischemic attack at baseline. No such a pattern was observed in the Netherlands. The proportion of patients with heart failure at baseline showed a three-fold increase in the Netherlands, while there was a relative decrease by 12% in Germany. Conclusions: Patient characteristics of new rivaroxaban users with non-valvular atrial fibrillation changed between 2011 and 2016/17, but changes differed between countries. These patterns have methodological implications. They have to be considered in the interpretation of observational studies comparing effectiveness and safety of oral anticoagulants, especially regarding potential bias due to unmeasured confounding.