A Novel Method to Apply Race and Ethnicity Observation for Nursing Home Residents Using Multiple Medicare...

Nursing homes (NHs) in the United States serve about 1.2 million residents. Different NH resident groups, including by race and ethnicity (R/E), experience different health outcomes in this setting. Multiple R/E measures exist for this population, but all have drawbacks. It is known that some R/E groups are underrepresented by these measures. Missing and inaccurate R/E classification in national data hinders research efforts to understand health outcomes.

We developed a method to improve R/E classification of NH residents using multiple datasets for enhanced accuracy and completeness.

This was a retrospective, observational study using 2011-2022 data from the Minimum Data Set 3.0 (MDS) and the Master Beneficiary Summary File (MBSF). We assessed missingness and consistency of variables and created a nuanced, mutually exclusive R/E measure based on the MDS race variables and added information from the MBSF RTI race variable.

All Medicare-eligible residents in CMS-certified NHs are identifiable in the MDS and MBSF data (2011-2022).

The merged data included 19,491,681 individuals and 206,985,444 assessments. Individual residents in the MDS had low missingness (1.74%) and high consistency (97.29%) on R/E data. Integrating MBSF and MDS data reduced missingness (0.03%) and increased Hispanic resident representation (3.92%-6.12%). New categorization of White and Black, Hispanic and Black, and White and other races reduced the "other/multiracial" category from 2.14% to 0.67%.

Using multiple data sources for R/E classification enhances the identification of NH demographics, which is important for researchers to have more accurate data. The increase in Hispanic residents after imputing RTI race suggests potential correction for self-underreporting. This method retains self-reporting standards while improving data completeness.