This letter originally appeared in the Northeaster on August 24, 2016, and is of interest again.
I read “Study finds correlation between GAF, cancer rates” with great interest, followed by dismay that preliminary, unscrutinized, and possibly quite flawed results had been released by the Bottineau Neighborhood Association and reported credulously by your newspaper. By way of background I am a Northeast resident, a cancer epidemiologist by training, and Professor at the University of Minnesota Medical School. Based on the reporting in the Northeaster and elsewhere, there are some serious problems in both the execution and interpretation of the study.
To start with, the study is based on mortality data rather than incident disease (i.e. both fatal and nonfatal cases). Epidemiologists strongly prefer to study incident disease when trying to draw connections between an allegedly causative exposure and disease. In fact, the Minnesota Department of Health already analyzed cancer incidence, rather than mortality, near the GAF plant, and found little evidence of an excess of cancer, a fact which was unmentioned in your story (report is at www.health.state.mn.us/divs/healthimprovement/data/reports/cancernnempls.html).
The choice to compare mortality in the area around the GAF plant to Mound, Minnesota was odd given that Mound is a wealthy, overwhelmingly white western suburb while the North/Northeast population under study is both ethnically and economically more diverse. The astute reader will have noticed that the number of deaths due to cancer in Mound are far below those of Minnesota or the United States as a whole. This gives the unfortunate impression that Mound was cherry-picked to exaggerate differences in cancer rates.
There are two further problems with the reporting of these rates. First, BNA appears to have released crude rates – meaning they have not had statistical adjustment for age, sex, or race. Apples to apples comparisons between populations are based on adjusted rates. Second, there is no indication of the statistical certainty of the findings. While the rate of cancer deaths in Hawthorne may look large at 325 per 100,000, the “confidence interval” around that number must be very large given the small population. To give some perspective, there are about 5,000 residents of Hawthorne, which means the reported rate of cancer deaths was based on only about 16 deaths per year.
Lastly, even if this study were conducted properly and still found elevated rates of cancer, that alone would not be enough information to attribute the excess to the GAF plant. One would have to exclude known risk factors as well. For instance, the cancers of special interest here are all smoking related. If the rate of smoking among residents near the GAF plant is higher than in a comparison population it would be unsurprising that lung cancer rates would be higher. You don’t need to be a scientist to understand this.
While I applaud BNA for its efforts to keep the neighborhood clean and safe, presenting numbers without scientific review or context while claiming they support their position is really irresponsible.
Logan Spector, Northeast Minneapolis