By Sue Rochman
An Unusual Collection of Cancers
Cancers can cluster within a community just by chance—so how do we tell if there’s actually a dangerous cause?
By Sue Rochman
When Kathleen Zemanek was diagnosed with breast cancer two years ago at age 47, she didn’t have to look far for support. Two of her neighbors had recently been diagnosed with breast cancer, as had the woman across the street. And they weren’t alone: There was the neighbor who had bladder cancer, and the one with leukemia, and at least two with melanoma.
Zemanek wasn’t the only person living on this bucolic San Francisco cul-de-sac of single-family homes who had begun to think the number of cancer cases was unusual. Half-jokingly, they had nicknamed themselves the “Sixth Avenue Cancer Cluster.” And though they didn’t talk much about it, many wondered: Was there something about their neighborhood that was responsible for all the cancer?
To date, no one who lives on Sixth Avenue has reported any concerns to the Environmental Health Investigations Branch of the California Department of Public Health. But if they did, they’d soon learn that theirs was one of more than 1,000 cancer cluster complaints made to state health departments nationwide every year. The vast majority do not meet the definition of a cluster and are never investigated. But even when the health department believes a situation looks suspicious, it’s far from easy to find a cause. In fact, more often than not, the investigators who study a cluster, and those affected by it, are left with more questions than answers.
Defining a cancer cluster is easy: It’s an unusually high number of cancers that occur during a specific period of time among people who live in the same geographic area, share the same workplace, or have the same occupation. But determining whether a group of cancers is not a random event but an actual cancer cluster—and if so, what might have caused it—is one of the more difficult tasks public health officials face.
“Clusters are mathematically going to occur all of the time,” explains Alfred I. Neugut, an epidemiologist at Columbia University in New York City. “If you flip a coin, you might get three heads in a row, or five. That’s clustering.” And this happens with cancer cases all the time. Imagine you’d expect to see 110 cases of a particular type of cancer in every 100,000 women, he continues. “Not every group of 100,000 women will have 110 cases. Some might have 90 and others might have 150. … And it has nothing to do with etiology or anything sinister under the surface.”
Except when it does.
“The overwhelming majority of cancer clusters are statistical fluctuations that don’t indicate anything at all,” says epidemiologist Wendel Brunner, the director of public health for the health services department of Contra Costa County, in California. “But the key here is ‘overwhelming majority.’ And that’s the rub: Every now and then a cluster is an indication of something that is important, so you can’t ignore them. But if you chase down every one, you will spend all of your time chasing statistical fluctuations.”
Identifying A Cluster
Public health officials go through a series of steps to separate clusters that are important from those that are not. They begin by determining if the cluster is just a random statistical fluctuation—or not. A clustering of cancers is generally considered “statistically significant” if its possibility of occurring by chance is less than 5 percent. “But 5 percent is one in 20,” says Brunner. “So if we have 20 neighborhoods and we look at rates of breast cancer in those neighborhoods, we would expect that one neighborhood would have a statistically significant elevated rate of cancer just by chance alone.”