Health Disparities
Decomposing Black/White inequities in hospitalization with Alzheimer’s Disease and related dementias in Medicare Lauren Mock* Lauren Mock Michelle Audirac Daniel Mork Danielle Braun Antonella Zanobetti Scott Delaney
Background
Alzheimer’s Disease and related dementias (ADRD) currently afflict more than six million Americans, with rates varying by location. Geographic variation in ADRD morbidity is caused by several factors, some of which are shared by Black and White Americans while others are not. Describing spatial variation that is shared across racial groups vs. race-specific is key to understanding effects of systemic racism, but prior studies mapping ADRD-related inequities often do not account for shared geographic variation.
Methods
Using 507,712,807 person-years of Medicare claims data from 60,879,784 non-Hispanic White and 6,497,198 non-Hispanic Black beneficiaries enrolled between 2000 and 2018, we fit a Bayesian shared component model to jointly estimate shared and race-specific variation in rates of hospital admissions with an ADRD diagnosis code. For each county, we obtained spatially smoothed relative risk estimates (based on observed vs. age- and sex-standardized expected rates) for both White and Black adults. We then calculated risk ratios (Black over White relative risk) to quantify the magnitude of racial inequities in each county after accounting for shared risk.
Results
Our data included 22,828,100 hospitalizations with an ADRD diagnosis code. Hospitalization risk varied substantially across space and race. Regarding relative risk, the median county-level estimate was 0.71 (range 0.25-2.05) for White adults vs. 0.99 (range 0.29-2.77) for Black adults. For Black/White risk ratios, the median ratio was 1.38 (range 0.72-2.93); 87% of counties had a risk ratio significantly greater than 1, indicating higher Black-specific risk, while white-specific risk was higher in 0.9% of counties. Inequities were greatest in the Midwest, Florida, Idaho, and parts of California.
Conclusion
ADRD racial inequities vary across the US. Estimating both shared and race-specific spatial risk can assist future efforts to disentangle contextual vs. sociodemographic risk factors.
