Comparing Total Medical Costs between Patients Receiving Direct Oral Anticoagulants Versus Warfarin for the Treatment of Atrial Fibrillation: Evidence from the VA

Abstract

Atrial fibrillation (AFib) is the second most common cardiovascular condition affecting up to 6.1 million individuals in the United States. Treatment of AFib focuses on preventing blood clots to reduce stroke risk. Traditionally, warfarin, a vitamin‐k antagonist, has been used to reduce stroke risk by two‐thirds. However, warfarin requires regular laboratory monitoring to ensure safety and efficacy, and may be cumbersome for patients. Newer direct oral anticoagulants (DOACs) are an alternative treatment and do not require laboratory monitoring due to their more predictable pharmacokinetics. DOACs are either noninferior or superior to warfarin on a variety of outcomes and may be cost‐effective for certain patient populations. Much of the evidence demonstrating benefits of DOACs over warfarin have been in clinical trial settings, which may have limited generalizability to real‐world practice. For example, patients enrolled in the Veterans Affairs (VA) Health care System have increased burden of illness and greater socioeconomic challenges relative to other populations. This study examined whether advantages of DOACs found in prior trials translated to the VA population by comparing total downstream costs between AFib patients treated with warfarin and DOACs.Using administrative data from VA and Medicare, we examined total medical costs (inpatient and outpatient costs incurred in VA and Medicare) over three years following initiation of pharmacological treatment for AFib. To address potential informative censoring, we applied a multipart estimator that extends traditional two‐part models to separate differences between groups due to effects on survival and cost accumulation. This estimator requires modeling the probability of surviving to each time period, hazard probability, expected patient‐level costs over period t conditional on death and expected costs over period t conditional on surviving the full period. We applied inverse probability weighting to address imbalance in characteristics between patients treated with warfarin and DOACs. Weights were estimated as a function of 99 baseline patient and clinic characteristics using boosted regression trees allowing for complex interactions between predictors.75 547 veterans dually enrolled in VA and fee‐for‐service Medicare who had a new diagnosis for AFib and initiated warfarin or DOAC treatment between 2011 and 2015The majority of patients initiated warfarin (64%) with the remainder initiating the DOACs apixaban (14%), dabigatran (12%), and rivaroxaban (9%). Mean annual costs in the first through third years of the follow‐up period were $33 364 (SD = $64 493), $18 379 (SD = $38 089), and $17 081 (SD = $38 725), respectively. In adjusted analysis, patients receiving warfarin had lower median survival time (−0.107 years, 95% confidence interval (CI) = −0.213 to −0.011) and higher annual cost conditional on death ($12 873, 95% CI = $8681 to $17 065) and surviving the full year ($9163, 95% CI = $8311 to $10 015). Total three‐year costs were higher among patients receiving warfarin ($22 718, 95% CI = $19 851 to $25 478).VA AFib patients initiating warfarin compared to DOACs incurred significantly higher downstream costs, even after accounting for a greater likelihood of death.The benefits of DOACs found in prior clinical trials appear to also be present among this patient population, suggesting these drugs may be the preferred option for treatment of AFib in elderly VA patients.Department of Veterans Affairs.