Medicaid Payments and Racial and Ethnic Disparities in Alzheimer Disease Special Care Units

Abstract

Importance: Alzheimer disease special care units (ASCUs) are associated with improved outcomes for residents with dementia, yet they are unavailable in most nursing homes.

Objectives: To examine racial and ethnic disparities in the availability of ASCUs and whether more generous Medicaid payments are associated with reduced disparities.

Design, setting, and participants: This cohort study used 2009-2019 Certification and Survey Provider Enhanced Reporting data and resident assessments from Medicare- and Medicaid-certified nursing homes in the US, as well as state Medicaid payment-to-cost ratios for 2019. Statistical analysis was performed from September to December 2024.

Exposure: The percentages of Black residents and Hispanic residents in a facility and the state mean nursing homes' ratio of Medicaid payment to estimated Medicaid cost of care.

Main outcomes and measures: The main outcome was whether a nursing home had an ASCU. Multivariable logistic regression was conducted on ASCUs, and then separate logistic regressions were performed for states with different quartiles of Medicaid payment-to-cost ratios.

Results: Most of the 13 229 nursing homes in the study were for profit (9561 [72.3%]) and were part of a chain (7775 [58.8%]). The overall mean (SD) Medicaid payment-to-cost ratio among all states was 0.87 (0.13) (range, 0.58-1.29). Each 1% increase in the percentage of Black residents was associated with a 0.1% decrease in the probability of having an ASCU. Compared with facilities with 0% to 0.8% of Black residents, the odds of having an ASCU were 37% lower in nursing homes with 4.3% to 15.2% Black residents (odds ratio [OR], 0.63; 95% CI, 0.53-0.74), and 45% lower in nursing homes with 15.2% or more of Black residents (OR, 0.55; 95% CI, 0.46-0.65). Compared with facilities with no Hispanic residents, the odds of having an ASCU were 27% lower in those with 3.7% or more of Hispanic residents (OR, 0.73; 95% CI, 0.62-0.86). In states with Medicaid payment-to-cost ratios between 0.58 and 0.81, nursing homes with 15.2% or more of Black residents were 68% less likely to have an ASCU (OR, 0.32; 95% CI, 0.21-0.50). This difference decreased to 45% in states with Medicaid payment-to-cost ratios between 0.82 and 0.94 (OR, 0.55; 95% CI, 0.44-0.69) and almost disappeared in states with Medicaid payment-to-cost ratios greater than 0.94 (OR, 0.86; 95% CI, 0.53-1.40). Higher Medicaid payment-to-cost ratios were not associated with reduced disparities among Hispanic residents.

Conclusions and relevance: This cohort study of nursing homes suggests that racial and ethnic disparities in ASCU availability narrowed in states where Medicaid payment rates cover a greater share of costs. Racial disparities in specialized dementia care may be mitigated and even eliminated by more generous Medicaid payments.

Figure 1.. State Variation in the Availability of Alzheimer Disease Special Care Units (ASCUs) in Nursing Homes, 2019

The sample included 14 058 nursing homes in 50 states from the 2019 Certification and Survey Provider Enhanced Reporting. States were ranked based on the percentage of nursing homes with an ASCU.

Figure 2.. Association Between the Percentage of Black Residents in a Nursing Home and the Presence of an Alzheimer Disease Special Care Unit (ASCU), 2019

The sample included 13 229 nursing homes in 47 states from the 2019 Certification and Survey Provider Enhanced Reporting merged with Minimum Data Set 3.0 assessments. The percentage of Black residents was calculated as the number of Black residents in 2019 in a facility divided by the total number of residents. Nursing homes with the same percentage of Black residents were grouped together. The number of nursing homes grouped at each percentile from 0% to 60% are 20 or more. For percentiles greater than 60%, the number of facilities were usually less than 20. The dotted line indicates a simple linear regression between percentage of Black residents and ASCU. The slope of the regression line was −0.05% (P = .07).