Comparing Cancer Registry Abstracted and Self-Reported Data on Race and Ethnicity

Data on racial and ethnic subgroups from the National Cancer Institute’s Surveillance, Epidemiology, End Results (SEER) program and Census Bureau population estimates are used to estimate cancer incidence rates. A SEER-National Longitudinal Mortality Study (NLMS) linkage of cancer cases diagnosed during 1973–2001 revealed mismatches in race classification from these sources affecting race-specific cancer incidence and mortality rates, particularly for minorities such as American Indians and Alaskan Natives (AIANs). Cancer registries obtain demographic data from various sources, including patient intake and provider records, administrative databases, and imputation algorithms. The primary Census Bureau source for racial/ethnic population denominators is self-reported survey data. We examined 7,970 SEER-NLMS cases diagnosed during 2003–2011 to update the comparison of patient race/ethnicity in cancer registry and population data sets. SEER and self-reported data did not agree for 5% of cases. The sensitivity of SEER data was better for whites (99%) and non-Hispanics (98%) than for multiracial individuals (23%) and all AIANs (40%). Intermediate sensitivities were seen for blacks as well as AIANs in Indian Health Service contract health service delivery areas (91%), Asians and Pacific Islanders (90%), and Hispanics (84%). As the United States becomes more diverse, a need exists to align race and ethnicity data from central cancer registries with population data, particularly for minority and multiracial groups. High-quality registry data on race and ethnicity, collected in a similar way as population estimates, will enhance cancer surveillance.