Data Quality: Race and Ethnicity Information in Medical SAS Datasets

Data Quality     Data Quality Alert addresses VA race/ethnicity and mortality data     Lab Specimen Source in DSS Laboratory Data     Race and Ethnicity Information in Medical SAS Datasets

Issues regarding VHA race and ethnicity data quality include missing values and inconsistency across time due to changes in data collection method and in allowable response categories and format.

Responding to the revision of OMB Directive 15, the VHA adopted a new standard for race and ethnicity data collection in FY2003 (VHA Directive 2003-027). Under this policy, self-identification of race and ethnicity is preferred, multiple race reporting is allowed, and ethnicity data must be collected separately from race.

Prior to FY2003, race and ethnicity information was most frequently extracted from clinical documentation and/or observation of administrative staff. In the databases, there was just one race field and there was no indicator of the method of data collection. Since FY2003, self-identification of race and ethnicity is the preferred method of data collection. Up to seven races can be recorded at a time and there is a separate field for ethnicity. In addition, the race categories have changed and the values of the race variables are coded to incorporate the means of data collection (e.g., white-observer, white-proxy, white-self-identification).

Missing values on race variables are not uncommon. Following the implementation of the new standard, the frequency of missing values has increased (Sohn, et al., 2006). Researchers are advised to supplement data on race in the Medical SAS Datasets starting in FY2003 with data on race from VA datasets prior to FY 2003 and/or from Medicare data. VIReC is currently conducting an investigation of methods to fill in missing race information, including back filling race and ethnicity values from prior years and obtaining race values from other data, such as Medicare.  For more information on supplementing race data, please see the VIReC Data Issues Brief for March 2004.

References:
Sohn, M.W., Zhang, H., Arnold, N., Stroupe, K., Taylor, B.C., Wilt, T.J., Hynes, D.M. (2006). Transition to the new race/ethnicity data collection standards in the Department of Veterans Affairs. Population Health Metrics, 4(7). Available at: http://www.pophealthmetrics.com/content/4/1/7.