The cancer non-epidemic of county 20: case study of an epidemiological mistake

Two suburban communities in western Canada, with a combined population of 52,000, were affected by a false allegation of increased cancer risk. In 1986, a cancer research agency responded to community concern by conducting a study of cancer incidence (1979 to 1983) and reported elevation on the order of 125% of expected for most sites. Reanalysis of these data several months later revealed an error in the population figure used to calculate the rates. Correction brought the rates into line with Alberta as a whole and comparable to other communities surrounding Edmonton. National media reported the cancer excess but did not report the correction. A Joint Advisory Committee convened at the time by the Minister responsible was a valuable resource for public education. This case study may be useful in the instruction of students and as an example of clerical errors that can indirectly affect an otherwise useful study. Residents of two nearby communities in northern Alberta became concerned about an apparently elevated rate of cancer among adults in their area. Many speculated on an association with the concentration of refineries in eastern Edmonton and petrochemical facilities outside the city in one of the communities. In 1986, the responsible provincial agency conducted a preliminary study of cancer incidence in the area compared to Alberta as a whole. The findings were summarized in a draft document which became widely circulated, although it was never intended to be the definitive report. It was "leaked" from a high government office and became the basis for numerous news stories. A reanalysis of these data revealed that the population figure used to calculate the combined cancer rates for the two communities was valid only for one of them and not the two combined. Correction of this mistake brought the calculated rates into line with those for Alberta as a whole. The original error was shown to be one of interpreting a confusing set of tables that lacked specific instructions and is not likely to be repeated. The original study is now invalidated. Reference to its findings should be made only as it has affected recent local history and public concern. There is certainly no evidence for a serious chemical threat to the health of residents of the area today in the present study and no suggestion of an occupational or environmental factor at work in the past.