The geographic variation of cancer incidence in Ontario

OBJECTIVES: Our objectives were (1) to describe an analysis of the spatial pattern of cancer incidence in Ontario and (2) to discuss the quality of data in the Ontario Cancer Registry with respect to the accuracy of local cancer rates. METHODS: Cancer incidence rates were calculated for 22 cancer sites in 49 counties of Ontario during 1976 to 1986. Capture-recapture methods were used to estimate completeness of case registration, and completeness of residence information was also assessed. Spatial autocorrelation was used in measuring the geographic pattern of incidence rates. Comparisons were also made between sexes and with earlier data from 1966 to 1975. RESULTS: The quality of the geographic data in the registry appeared good, and corrections for incomplete or inaccurate registration had little impact. About one third of the sex-site combinations showed some evidence of spatial patterning in the cancer rate. Particularly strong regional variation was noted for cancers of the stomach, lung, uterus, and prostate. CONCLUSIONS: The analysis revealed a number of cancers with significant spatial patterning of risk. Further work is needed to relate the cancer data to other information on potential life-style and environmental factors.