Exploring regional patterns of childhood cancer diversity in Alberta, Canada.

Abstract

Introduction: The study of cancer geographical patterns has helped to identify differences in the distribution of cancer risk factors. In this study, we propose the use of the Shannon index (H-index) for evaluating regional changes in childhood cancer diversity. The H-index identifies cancer diversity based on the number of cancer types plus their relative frequencies in a given area, adding an “evenness” dimension that may help identify regions with a particular cancer diversity where cancer drivers may differ. We used the H-index to evaluating the regional diversity of childhood cancer in Alberta, Canada, exploring the co-location of industrial carcinogens emissions to inform future research. Methods: We used data from 1994-2005 including all children cancer cases; reported IARC-1 carcinogens emitted to air by industry; percentage of cases living <3km to facilities emitting carcinogens (closeness); and children’s population. We aggregated data in14 regions and calculated the H-index for each region. Cluster analysis identified the geographical patterns. Results: Province-wide cancer diversity was high (H-index=2.03) composed of 13 cancer types, predominantly leukemias (31%), nervous system tumours (20%) and lymphomas (12%). Interestingly, only two regions (R1, R2) showed a significantly smaller cancer diversity (H-index 1.53 and 1.68, respectively; p<0.05). R1 had eight cancer types, largely dominated by leukemias (≈50%) and R2 had seven cancer types, mainly represented by lymphomas (39%). In R2, the H-index collocated with the highest carcinogens emissions in the province and 54% of closeness. In R1, the H-index collocated with no carcinogenic emissions. Currently, we are exploring collocation with regional agrochemicals. Conclusion: The study indicates that the Shannon-index is a useful exploratory tool for cancer diversity, opening the opportunity for evaluating associated childhood cancer risk factors patterns.