Objective
To evaluate how post‐acute care (PAC) transitions affect minority older adults with Alzheimer's disease or related dementia (ADRD), and the extent to which dual Medicare‐Medicaid eligibility may attenuate or exacerbate disparities in PAC outcomes. We examined: (1) PAC referrals by race/ethnicity and dual status; (2) individual, hospital, and market‐level factors associated with PAC; (3) the association between PAC and outcomes.
Data Sources/Study Setting
We used the following secondary data: Master Beneficiary Summary File (MBSF), Medicare Provider Analysis and Review (MedPAR), Minimum Data Set (MDS), Area Health Resource File (AHRF), hospital Provider of Services (POS) file, and the area deprivation index (ADI).
Study Design
This observational study consisted of 619,262 community‐residing Medicare fee‐for‐service (FFS) beneficiaries with ADRD who had a hospital stay in 2017.
Data Collection/Extraction Methods
PAC discharge was to skilled nursing facilities (SNF), home health care (HHC) agencies or home without services. Outcomes were 30‐day readmission and death. Multinomial logistic regressions with hospital random effects (RE), stratified by dual eligibility, were fit.
Principal Findings
Dual‐related differences were significantly larger than race/ethnicity differences in PAC transitions. For example, the difference in the probability of SNF transitions between White and Black patients was 3.2% and 6.8%‐points for non‐duals and duals, respectively. The difference between non‐dual/dual White patients was 21.6% points, and among Black patients 18.0%‐points. The adjusted risk of 30‐day readmission was 5.6 percentage point higher among non‐duals discharged to SNF, compared to home, but such risk among duals was not statistically significantly different. The adjusted probabilities of 30‐day mortality were larger for duals and non‐duals who transitioned to SNF, compared to those discharged home.
Conclusions
PAC referrals and the resulting outcomes for Medicare beneficiaries with ADRD are associated with multi‐level variables that need to be incorporated in discharge decision making.