Indirect method to estimate specific Hispanic group cancer rates

Abstract

Several states with large Hispanic populations have historically served as the source for US Hispanic cancer incidence rates, with aggregation of data across all states limited by different methodologies to identify Hispanic persons. Now with data available for more than 85% of the US Hispanic population, state rates suggest regional diversity in their Hispanic cancer profiles. We tested an approach of using a surrogate indicator of county residential homogeneity for Hispanic groups based on the 2000 US Census. The indicator used the counts of specific Hispanic residents compared to the total Hispanic population in the county to define counties with homogenous Hispanic populations. From these data, we aggregated counties into homogeneity categories for each Hispanic group and defined thresholds and rules for allocating Hispanic persons to a specific Hispanic group. We found that it was possible to use county demographic data in many counties to meaningfully attribute a specific Hispanic ethnicity to incident cancer cases based on homogeneity thresholds. Cancer rates for the US Hispanic population describe a profile of high rates of cancers of the liver, gallbladder, cervix (in female), stomach, and lower rates of the cancers of the lung, female breast, and prostate compared with the non-Hispanic white population. In general, rates among US Mexicans are lower than the US Hispanic rates, while rates for Puerto Ricans and Cubans are higher than the US Hispanic rates. Additional variations among the three Hispanic groups were also evident. The approach yielded reasonable and useful information to explore etiologic differences among the populations, as well as to develop relevant cancer control interventions. However, direct identification of specific Hispanic ethnicity in medical records and annual Census estimates of these populations would be preferable if they ever became available.