Cancer Clusters Are Difficult to Nail down
State health departments now receive at least 5,000 reports of so-called cancer clusters a year, says Wendy Kaye, chief of epidemiology for the Agency of Toxic Substances and Disease Registry, an arm of the U.S. Department of Health and Human Services.
Reports are so numerous that experts have a name for false clusters: the Texas sharpshooter fallacy. The term refers to a hypothetical rifleman who fires several shots at a barn wall, then draws a bull's eye around the shots. "Clusters can look pretty dramatic, even when there's nothing there," Kaye says.
And usually, in the case of clusters that occur in the environment or in residential areas, there is nothing there, says Paul Schulte, an epidemiologist for the National Institute for Occupational Safety and Health. Cancers that appear to fit a pattern can and do occur by chance.
Schulte makes this comparison: In poker, "you can be dealt a hand and get four aces," he says. "That's a cluster, but it doesn't relate to anything. If something similar happens in the workplace or a neighborhood, though, we start to say, 'Whoa! That shouldn't be.' But random occurrence doesn't have to be spread out. It can occur in big clumps."
One method that epidemiologists use to help determine whether to be suspicious of a cluster of cancers is to actually calculate the probability that a cluster is happening by chance.
If, for instance, eight lung cancers are found in a neighborhood of 500 people, and researchers figure that such a cluster is likely to occur naturally 20 times in every 100 neighborhoods of 500 people, they're less worried than if the odds are that such a cluster will occur by chance only five times in every 100 neighborhoods. (Epidemiologists also get more worried if all the cancers are of the same type.)
Overall, about 90 percent of cluster reports are dismissed annually.
This article is excerpted from "Cancer clusters are difficult to nail down," USA Today, April 15, 1999, available at http://usatoday.com/life/science/general/lsx016.htm.