Return to 2001 Abstacts List

55. Use of Self-reported Race vs. VA Administrative Data on Race: Implications for Health Services Research Findings

U Boehmer, Center for Health Quality, Outcomes, & Economic Research and Boston University School of Public Health; NR Kressin, Center for Health Quality, Outcomes, & Economic Research and Boston University School of Public Health; A Pitman, Center for Health Quality, Outcomes, & Economic Research and Boston University School of Public Health; D Berlowitz, Center for Health Quality, Outcomes, & Economic Research and Boston University School of Public Health; CL Christiansen, Center for Health Quality, Outcomes, & Economic Research and Boston University School of Public Health; JA Jones, Center for Health Quality, Outcomes, & Economic Research and Boston University School of Public Health

Objectives: Concerns about administrative data on race have led some researchers to consider self-reported race as superior. However, few studies have examined the differential impact of the type of race data on study outcomes. We investigated whether different sources of race data led to different results when examining the use of one therapeutic dental procedure, root canal therapy.

Methods: From VA dental outpatients who had either a root canal or tooth extraction procedure in FY 1998, we selected 9,425 on whom we had both self-reported race from the "1999 Large Health Survey of Veterans" and administrative data on race from VA outpatient clinic files (OPC). Using logistic regression, we estimated the probability of obtaining root canal therapy versus tooth extraction, for White, African American, and Hispanic patients, calculating two models, one using self-reported race and one using race extracted from OPC. All models were controlled for severity of dental disease, medical comorbidity (Charlson index), mental disorders (schizophrenia, bipolar, depression, and alcoholism), age, gender, income, and prior use of preventive dental services. For patients with disagreement on race between the two data sources, we used chi square and t-tests to examine if these patients differed with regard to number of dental visits, age, gender, or income from patients whose recorded race was in agreement.

Results: 98% of self-reported Whites, 94% of self-reported African Americans, and 82% of self-reported Hispanics were classified as such in OPC. 66 self-reported Whites whose OPC record indicated Hispanic or African American race did not significantly differ on number of dental visits, age, gender, or income, compared to 6847 White patients whose administrative and self-reported data were in agreement. 113 self-reported African Americans who were classified as White in OPC were not significantly different from the 1760 African Americans whose self-reported and administrative race was in agreement. Also there were no significant differences between 109 Hispanics whose OPC race was non-Hispanic compared to the 467 Hispanics whose race data agreed. 113 African Americans recorded as White and 59 whites recorded as African American in OPC were as likely to receive a root canal as 6847 whites whose race agreed in both data sources. African Americans were significantly less likely to receive root canal therapy than Whites, regardless of source of race data (95%CI 0.64 to 0.90 (self-report) and 0.63 to 0.88 (OPC)), while Hispanics’ probability of obtaining a root canal did not differ from Whites.

Conclusions: The source of race data had no significant influence on the outcome measure, receipt of root canal. Dental patients whose race was recorded correctly in OPC were not significantly different from patients whose race was incorrectly recorded on the dimensions we assessed.

Impact: The high level of agreement between administrative and self-reported race of dental patients ensures reliable estimates of racial differences in receipt of root canal.