Evaluation of African-American and white racial classification in a surveillance, epidemiology, and end results cancer registry

Objectives: This study evaluated the validity of registry-reported race for individuals who participated in research studies conducted since 1980 through the Metropolitan Detroit Cancer Surveillance System (MDCSS), a Surveillance, Epidemiology, and End Results (SEER) Program registry. Methods: 5329 individuals who self-identified as African American or White and were classified in the MDCSS registry as African American or White were included. Self-identified and registry-reported race were compared, and associations between demographics and racial misclassification were examined. Results: Most self-identified African Americans and Whites were correctly classified (sensitivity=98.5%, specificity= 99.7%). Males were two times more likely to be misclassified than females [odds ratio (OR)=2.13, 95% confidence interval (Cl): 1.06-4.29]. Individuals diagnosed with cancer after 1990 were two times more likely to be misclassified than those diagnosed before 1990 (OR= 2.17, 95% Cl: 1.07-4.42). African Americans were four times more likely to be misclassified than Whites (OR=4.39, 95% CI: 2.24-8.60). Conclusions: Misclassification in the MDCSS registry of African Americans as Whites, and vice versa, is relatively low. Additional studies should evaluate misclassification of African Americans and Whites as other races and/or ethnicities in the SEER registry.