Association of Race and Ethnicity and Anticoagulation in Patients With Atrial Fibrillation Dually Enrolled in Veterans Health Administration and Medicare: Effects of Medicare Part D on Prescribing Disparities

Abstract

Background: Racial and ethnic disparities in anticoagulation exist in atrial fibrillation (AF) management in Medicare and the Veterans Health Administration (VA), but the influence of dual VA and Medicare enrollment is unclear. We compared anticoagulant initiation by race and ethnicity in dually enrolled patients and assessed the role of Medicare Part D enrollment on anticoagulation disparities. Methods: We identified patients with incident AF (2014-2018) dually enrolled in VA and Medicare. We assessed any anticoagulant initiation (warfarin or direct-acting oral anticoagulants, DOACs) within 90 days of AF diagnosis and DOAC use among anticoagulant initiators. We modeled anticoagulant initiation, adjusting for patient, provider, and facility factors, including main effects for race and ethnicity and Medicare Part D enrollment and an interaction term for these variables. Results: In 43,789 patients, 8.9% were Black, 3.6% Hispanic, and 87.5% White; 10.9% participated in Medicare Part D. Overall, 29,680 (67.8%) patients initiated any anticoagulant, of which 17,568 (59.2%) initiated DOACs. Lower proportions of Black (65.2%) than Hispanic (67.6%) or White (68.0%) patients initiated any anticoagulant (p= 0.001), and lower proportions of Black (56.3%) and Hispanic (55.9%) than White (59.6%) patients (p=0.001) initiated DOACs. Compared to White patients, Black patients had significantly lower initiation of any anticoagulant, adjusted odds ratio (aOR) 0.89; 95% CI 0.82-0.97. The aORs for DOAC initiation were significantly lower for Black (0.72; 95% CI, 0.65-0.81) and Hispanic (0.84; 95% CI, 0.70-1.00) than White patients.The interaction between race and ethnicity and Medicare Part D enrollment was non-significant for any anticoagulant (p=0.99) and DOAC (p=0.27) therapies. Conclusions: In dually enrolled VA and Medicare patients with AF, Black patients were less likely to initiate any anticoagulant and Black and Hispanic patients were less likely to initiate DOACs. Medicare Part D enrollment did not moderate the associations between race and ethnicity and anticoagulant therapies.

Association of Race and Ethnicity With Oral Anticoagulation and Associated Outcomes in Patients With Atrial Fibrillation

“…We examined several patient sociodemographic and clinical as well as hospital characteristics considered determinants of the association between our primary independent variables (ie, race and ethnicity) and our study outcomes. Race and ethnicity were identified by self-report as a required multiple-choice data element in the GWTG-AFIB registry . This variable was defined in mutually exclusive categories as non-Hispanic Asian, non-Hispanic Black, Hispanic (any race), and non-Hispanic White (the referent group).…”

Sociodemographic determinants of oral anticoagulant prescription in patients with atrial fibrillations: findings from the PINNACLE registry using machine learning

“… 26 In a study by Llorca and colleagues, 25 those living in more socioeconomically deprived and rural areas had lower OAC prescription rates. Moreover, previous studies by Essien and colleagues 27 , 28 showed lower initiation of OAC for Black patients and lower DOAC use for Black and Hispanic patients. This was also evident in our descriptive results; however, these did not emerge as high-ranking features in our ML models, which may be due to the low sample sizes of these populations in our database.…”

Beyond Detecting and Understanding Disparities in Novel Diabetes Treatment

“…Second, the study did not account for dual enrollment in Medicare insurance, Medicaid insurance, or both, which occurs in up to 60% of VHA enrollees and could influence receipt of guidelinerecommended medications from outside the VHA. 11 Third, given the association of clinician characteristics with racial and ethnic health disparities, understanding the proportion of variance in SGLT2i and GLP-1 RA prescribing at the clinician level, and having a wider range of clinician-level covariates (eg, clinical practice behaviors, demographics, and personal beliefs) to potentially explain prescribing differences would add depth to future analyses. Similarly, given the observed 16-to 18-fold variation in SGLT2i and GLP-1 RA prescribing across the 130 VHA facilities defined at an aggregated VA health care network level, future work should consider more finely classifying the facilities where patients receive their prescriptions (eg, VA community-based outpatient clinics) and assess a broader array of facility characteristics as study covariates.…”

“…First, even though cost-sharing is lower in VHA than non-VHA health plans, it is not negligible and more precisely understanding how the magnitude of prescription co-payment influences use of newer, evidence-based therapies is important. Second, the study did not account for dual enrollment in Medicare insurance, Medicaid insurance, or both, which occurs in up to 60% of VHA enrollees and could influence receipt of guideline-recommended medications from outside the VHA . Third, given the association of clinician characteristics with racial and ethnic health disparities, understanding the proportion of variance in SGLT2i and GLP-1 RA prescribing at the clinician level, and having a wider range of clinician-level covariates (eg, clinical practice behaviors, demographics, and personal beliefs) to potentially explain prescribing differences would add depth to future analyses.…”