Abstract 13115: Association of Hospital Racial Composition and Anticoagulation for Atrial Fibrillation: A Nationwide Study

Introduction: Anticoagulation reduces stroke risk in atrial fibrillation (AF). Patient-level studies show significant Black-White disparities in any anticoagulant and direct oral anticoagulant (DOAC) prescribing. Little is known about whether such disparities exist at the hospital level, thus we examined the association of hospital racial composition and anticoagulant initiation.

Hypothesis: Hospitals with higher rates of Black patients would have lower anticoagulation.

Methods: We conducted a retrospective cohort study of patients with incident AF treated at a VA medical center (VAMC) from 2018-2021. Our independent variable was hospital racial composition, defined by the proportion of Black patients cared for at a VAMC, with Quartile 1 (Q1) being the lowest, and Quartile (Q4) being the highest. We used a multivariable logistic regression model, adjusting for clinical factors (e.g., stroke and bleeding risk), year of diagnosis, and region, to estimate the odds of any anticoagulant and/or DOAC initiation by hospital racial composition.

Results: We identified 86,436 patients with incident AF at 140 VAMCs, 10.5% Black, 89.5% White, mean age 73 years. Patients in Q4 were younger with similar stroke and bleeding risk to those in Q1. Overall, any anticoagulant use was 73.4% in Q1 and 71.4% in Q4 and lower for Black vs. White patients in both quartiles (Q1: 68.9% vs. 73.5%; Q4: 69.1 vs. 72.0%). The odds of initiating any anticoagulant overall were lower in Q1 hospitals compared to Q4: adjusted odds ratio (aOR): 0.85; 95% CI 0.73-0.99; p=0.04. DOAC use was similar in Q1 and Q4 (67.6% vs. 67.2%) though lower for Black vs. White patients in both quartiles (Q1: 57.4% vs. 67.7%; Q4: 63.3 vs. 68.5%). There was no significant difference overall in the odds of initiating DOAC in Q1 hospitals compared to Q4, aOR: 0.96; 95% CI 0.83-1.12; p=0.61.

Conclusions: In a national study of patients with AF in the VA, we observed a significant difference in any anticoagulant but not DOAC initiation by hospital racial composition. Racial differences in anticoagulant use existed across quartiles, as in prior work. These data suggest that racial disparities in anticoagulant use in the VA, a safety net health system, may require local rather than national-level solutions alone to eliminate them.