Including self-reported race to improve cancer surveillance data for American Indians and Alaska Natives in Washington state.
BACKGROUND: American Indians and Alaska Natives (AI/AN) are frequently misclassified as another race in cancer surveillance systems, resulting in underestimated morbidity and mortality. Linkage methods with administrative records have been used to correct AI/AN misclassification, but AI/AN populations living in urban areas, and those who self-identify as AI/AN race, continue to be under-ascertained. The aim of this study was to evaluate racial misclassification in two cancer registries in Washington State using an urban AI/AN patient roster linked with a list of Indian Health Service (IHS) enrollees. METHODS: We conducted probabilistic record linkages to identify racial misclassification using a combined demographic dataset of self-identified AI/AN patients of a large, urban Indian health center, and administratively-identified AI/AN enrolled with the IHS. Age-adjusted incidence rates were calculated for 3 linkage populations: AI/ AN originally coded in each cancer registry, post-linkage AI/AN identified through the IHS roster alone, and post-linkage AI/AN identified through either the urban or IHS file. RESULTS: In the state and regional cancer registries, 11% and 18%, respectively, of matched cases were originally coded as a race other than AI/AN; approximately 35% of these were identified by the urban file alone. Incidence rate estimates increased after linkage with the IHS file, and further increased with the addition of urban records. Matches identified by the urban patient file resulted in the largest relative incidence change being demonstrated for King County (which includes Seattle); the all-site invasive cancer rate increased 8.8%, from 443 to 482 per 100,000. CONCLUSIONS: Inclusion of urban and self-identified AI/AN records can increase case ascertainment in cancer surveillance systems beyond linkage methods using only administrative sources