Additional file 1: Supplemental Materials. of Comparative effectiveness of rivaroxaban versus warfarin or dabigatran for the treatment of patients with non-valvular atrial fibrillation NorbyFaye BengtsonLindsay LutseyPamela ChenLin MacLehoseRichard ChamberlainAlanna RapsonIan AlonsoAlvaro 2017 Table S1. Characteristics of atrial fibrillation patients by anticoagulant use prior to final matching based on high-dimensional propensity score, MarketScan, 2010–2014. Table S2. Adjusted hazard ratios (95% confidence intervals) of selected outcomes comparing new rivaroxaban users (categorized by initial dose) to new warfarin users for the treatment of non-valvular atrial fibrillation, MarketScan, 2010–2014. Table S3. Adjusted hazard ratios (95% confidence intervals) of selected outcomes comparing patients who switched to rivaroxaban (categorized by initial dose) from warfarin to warfarin-only users for the treatment of non-valvular atrial fibrillation, MarketScan, 2010–2014. Table S4. Adjusted hazard ratios (95% confidence intervals) comparing new rivaroxaban users (categorized by initial dose) to new dabigatran users for the treatment of non-valvular atrial fibrillation, MarketScan, 2010–2014. Table S5. Adjusted hazard ratios (95% confidence intervals) of selected outcomes comparing new rivaroxaban users to new warfarin users for the treatment of non-valvular atrial fibrillation, MarketScan, 2010–2014. Restricted to 36,623 patients with at least 180 days of enrolment before first oral anticoagulation prescription. Table S6. Adjusted hazard ratios (95% confidence intervals) of selected outcomes comparing new rivaroxaban users to new warfarin users for the treatment of non-valvular atrial fibrillation, MarketScan, 2010–2014. Restricted to 68,927 patients with an enrolment date later than January 1st, 2011. Table S7. ICD-9-CM codes for outcomes. Table S8. ICD-9-CM codes used to define comorbidities. Figure S1. High-dimensional propensity score distribution by oral anticoagulant status for the outcome of stroke. These are the distributions prior to final matching based on high-dimensional propensity score. (DOCX 356 kb)