BACKGROUND: Direct oral anticoagulants (DOACs) are an alternative to warfarin for treatment of atrial fibrillation (AF). Evidence demonstrating the efficacy and safety of DOACs has primarily been from clinical trial settings. The real-world effectiveness of DOACs in specific nontrial populations that differ in age, comorbidity burden, and socioeconomic status is unclear. OBJECTIVE: To compare total downstream medical expenditure between AF patients treated with warfarin and DOACs dually enrolled in the Veterans Affairs (VA) Healthcare System and fee-for-service Medicare. METHODS: This was an exploratory treatment effectiveness study that analyzed VA administrative data and Medicare claims. We examined patients with an incident diagnosis for AF and initiated warfarin or DOAC treatment between 2012 and 2015. The primary outcome was total medical expenditure over 3 years following treatment initiation. To address potential informative censoring, we applied a multipart estimator that extends traditional 2-part models to separate differences between groups due to survival and cost accumulation effects. Inverse probability weighting was applied to address potential treatment selection bias. RESULTS: We identified 31,276 and 17,021 patients receiving warfarin and DOACs, respectively. Mean unadjusted (SD) expenditure was higher for warfarin ($56,265 [$96,666]) compared with DOAC patients ($32,736 [$52,470]). Compared with patients receiving DOACs, adjusted 3-year expenditure was $25,688 (P < 0.001) higher for patients receiving warfarin. CONCLUSIONS: VA patients with AF initiating warfarin incurred markedly higher downstream expenditure compared with similar patients receiving DOACs. The benefits of DOACs found in previous clinical trials were present in this population, suggesting that these DOACs may be the preferred option for treatment of AF in older VA patients. DISCLOSURES: This study was funded by a VA Health Services Research and Development Investigator Initiated Research Award (IIR 15-139). Support for VA/CMS data was provided by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, Health Services Research and Development, VA Information Resource Center (Project Numbers SDR 02-237 and 98-004). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The views expressed in this article are those of the authors and do not necessarily reflect the position or policy of the Department of Veterans Affairs, the University of Washington, Northeastern University, and Boston University. The authors declare no conflicts of interest. This research includes data obtained from the VHA Office of Performance Measurement (17API2), which resides within the Office of Analytics and Performance Integration (API), under the Office of Quality and Patient Safety (QPS; formerly known as RAPID). An oral presentation documenting a subset of the findings from this study was presented at the 2020 AcademyHealth Annual Research Meeting, delivered virtually on July 29, 2020.