PubMed1996
BACKGROUND: Previous reports have suggested that American Indians are sometimes classified as other races on cancer registries. Also, cancer registries typically do not include data on tribal affiliation. This study determined the extent of racial misclassification of American Indians in the Washington State Cancer Registry (WSCR) and obtained tribal-specific cancer data for Washington State. METHODS: A computer file including persons registered for services with the Portland Area Indian Health Service (IHS) or who were enrolled members of 19 tribes in Washington, Oregon, and Idaho (n = 127,375) was linked with WSCR records of incident cases for 1992 and 1993 (n = 49,420). Linkage was conducted with probabilistic methods using the AutoMatch software package. RESULTS: Of 180 persons recorded as American Indian in the WSCR, 130 (72.2%) were identified in the IHS/tribal roll file. Of 259 American Indians included in the IHS/tribal file who were identified in the WSCR, 130 (50.2%) were classified as American Indian. The estimated age-adjusted cancer incidence among American Indians in Washington State increased from 153.5 per 100,000 population before record linkage to 267.5 per 100,000 after linkage. Of the 259 persons who were linked to the WSCR, 17 were not registered with IHS and appeared solely in the tribal rolls. Only two tribes had more than five identified cancer cases during the 2-year study period. CONCLUSIONS: The number of IHS-enrolled American Indians or tribal members included in the WSCR would be underestimated by one third in the absence of record linkages, and the estimated cancer incidence of 43.6% would be lower before linkage. It is feasible to obtain tribal-specific cancer rates by linking tribal rolls to cancer registries, although the small number of cases in most tribes is a significant limitation. Further efforts to improve racial classification of American Indians in cancer registries should be undertaken.