Improving the imputation of race: evaluating the benefits of stratifying by age

Abstract

Health plans and other health care institutions may use indirect methods such as geocoding and surname analysis to estimate race, ethnicity, and socioeconomic status in an effort to measure disparities in care or target specific demographics. This study investigated whether stratifying by age improved imputations of race and ethnicity made through geocoding. Self-reported race and ethnicity from Medicaid enrollment records and from a health risk assessment administered by a large employer were used to validate imputation results from both an age-stratified model and a standard model. Sensitivity, specificity, and positive predictive value were calculated. Both approaches successfully imputed race and ethnicity for whites, blacks, Asians, and Hispanics. The age-stratified approach identified more blacks than did the unstratified approach, and correctly identified more blacks and whites. The two approaches worked equally well for identifying Asians and Hispanics. Age stratification may improve the accuracy of imputation methods, and help health care organizations to better understand the demographics of the people they serve.