Blue Cross Blue Shield of Minnesota, Eagan, Minnesota 55122, USA.
Popul Health Manag. 2009 Dec;12(6):325-31. doi: 10.1089/pop.2009.0006.
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.
健康计划和其他医疗机构可能会使用间接方法，如地理编码和姓氏分析，来估计种族、民族和社会经济地位，以衡量护理方面的差异或针对特定人群。本研究调查了通过地理编码进行种族和民族分层是否可以改善推断。从医疗补助计划登记记录和大型雇主进行的健康风险评估中报告的种族和民族，用于验证来自年龄分层模型和标准模型的推断结果。计算了敏感性、特异性和阳性预测值。这两种方法都成功地推断了白人、黑人、亚洲人和西班牙裔的种族和民族。分层方法比非分层方法识别出更多的黑人，并且正确地识别出更多的黑人和白人。这两种方法对于识别亚洲人和西班牙裔同样有效。年龄分层可能会提高推断方法的准确性，并帮助医疗保健组织更好地了解他们所服务人群的人口统计学特征。
