More accurate racial and ethnic codes for Medicare administrative data

Area of Science:

• Health Services Research
• Health Disparities
• Biostatistics

Background:

• Medicare administrative data analyses of health care disparities are often limited to Black and White beneficiaries.
• Existing race and ethnicity codes have inaccuracies and small sample sizes for other groups, potentially biasing results.

Purpose of the Study:

• To develop and validate an imputation algorithm to significantly improve the accuracy of race and ethnicity coding for Hispanic and Asian or Pacific Islander beneficiaries in Medicare data.

Main Methods:

• An imputation algorithm was developed to enhance the accuracy of race and ethnicity coding.
• The algorithm's performance was validated by comparing its recoded data against self-reported race and ethnicity.
• Key metrics included sensitivity, specificity, and Kappa coefficients.

Main Results:

• Sensitivity for Hispanic beneficiaries increased from 29.5% to 76.6%.
• Sensitivity for Asian and Pacific Islander beneficiaries increased from 54.7% to 79.2%.
• Specificity remained high, and Kappa coefficients reached 0.80, indicating substantial agreement.

Conclusions:

• The developed imputation algorithm dramatically improves the accuracy of race and ethnicity coding for underserved populations in Medicare.
• This advancement enables more comprehensive and less biased analyses of health care disparities for Hispanic and Asian or Pacific Islander beneficiaries.
• Over 2.5 million beneficiaries were accurately recoded, enhancing the utility of Medicare data for health equity research.