Ricky Schroder pulled his daughter out of school last week. No, he isn’t the latest parent to embrace homeschooling—he’s worried about cancer. Three teachers at the middle school were diagnosed with thyroid cancer in a span of six months. Parents at the Malibu school fear that contaminants in the soil or mold in the classrooms are responsible, and they have hired an independent environmental consultant to conduct testing. (His $5,000 fee isn’t much of a hardship in a town where the median household income is more than double the national average.) The former child star told TMZ he’ll homeschool his daughter until the school is proven safe.
On the face of it, the story makes Schroder seem like another nutty celebrity. Three teachers with thyroid cancer is hardly the kind of statistical anomaly that brings the Centers for Disease Control and Prevention to town. In addition, Schroder and his fellow parents could use a lesson on how cancer works. It’s not an infectious disease, although some carcinogens such as the human papilloma virus are contagious. (Which is why you and yours might want to get the HPV vaccine.) Most cancers occur spontaneously. And in those that are prompted by a carcinogen, symptoms do not immediately follow contact; the disease manifests years after exposure. Unless these three teachers have all been working together in the same school for a very long time, it’s unlikely the cases have anything to do with one another.
Nevertheless, Schroder’s panic is pretty typical. This scenario plays out nearly every day somewhere in the United States. Public health officials field between 1,000 and 2,000 calls annually from ordinary citizens concerned about a perceived cancer spike in their hometown. Call it the Erin Brockovich effect. Epidemiologists and oncologists can be forgiven for hating that movie, in which Julia Roberts plays a concerned citizen who susses out a public health emergency. In reality, it’s nearly impossible for ordinary people to identify a so-called cancer cluster.
Many of the calls can be dismissed out of hand. Most of the callers report 15 or fewer cancer cases, too few to detect a true cancer spike. Many of the complaints involve different types of cancer, which are very rarely caused by the same carcinogen. In plenty of cases, elevated rates of cancer in a community have more to do with human behavior than environmental exposure. If, for example, you live in a town full of smokers, lung cancer rates are likely to be high even if you have pristinely clean air.
But public health officials are understandably hesitant to ignore concerns about cancer, lest they find themselves cast as villains in a Julia Roberts film. So they launch investigations for many of the reports, resulting in a serious drain on public resources that rarely produces any useful public health findings.
In 2012, a group of researchers pored over 428 investigations of perceived cancer clusters that occurred since 1990. In 87 percent of those cases, the caller was simply overreacting. Cancer rates in the town or place of employment were well within the expected range. Among the 72 reports in which epidemiologists could verify a higher than expected incidence of a particular type of cancer, investigators found no evidence that the victims had been exposed to the same carcinogenic agent in 69 of the cases. These cases were likely statistical anomalies and nothing more. Only one of the three remaining cases with a plausible link to a known carcinogen turned out to be a verifiable cancer cluster. (Elevated rates of pleural cancer near Charleston, S.C. over a 20-year-period were likely the result of asbestos exposure among shipyard workers.) In other words, public health officials came up empty in 99.8 percent of the reported cancer cluster investigations.
If anything, those numbers overestimate the public’s ability to detect a cancer surge. Similar studies in the past have been even more damning: A review of 108 investigations conducted by the CDC between 1961 and 1982 found precisely zero cases of verifiable cancer clusters linked to a known carcinogen.
In addition, investigating a cancer cluster is biased toward finding a spike where there is none. Why is that? When epidemiologists are concerned about the levels of a particular carcinogen in a community, they set a boundary around the exposure area and assess cancer rates within that zone. Higher than expected cancer rates in that sample provide evidence that the carcinogen is affecting the population. Cancer cluster reports, on the other hand, start with a high concentration of cases and draw the boundary around them. The incidence of cancer within the area may, in fact, be elevated, but there’s no way of knowing if that rate is simply chance and balanced out by slightly lower rates in adjacent areas. The sample is biased from the beginning in favor of finding a statistical anomaly.
Despite decades of evidence proving that cancer cluster investigations are a drain on our resources and methodologically problematic, the public and our representatives can’t get enough of them. Congress considers bills nearly every year seeking to strengthen the federal response to reports of disease clusters. Sen. Barbara Boxer of California is among the most persistent advocates. The current iteration of her proposal, called the “Strengthening Protections for Children and Communities From Disease Clusters Act,” would establish regional response centers to deal with such reports. Many state legislatures have considered similar legislation. After Erin Brockovich, A Civil Action, and even lesser-known films like Eddie Murphy’s The Distinguished Gentleman, getting tough on cancer clusters is a cheap and easy way for a politician to appear to be protecting anxious constituents.
The truth is that our public health agencies are already doing too much in response to cancer cluster reports. Not even the defenders of strong government action can justify the investigations on public health grounds. In response to the 2012 study showing the low success rate of cancer cluster investigations, a trio of Massachusetts health officials wrote in defense of such studies. Their logic was that responding to calls about perceived cancer clusters gave the government an opportunity to talk to people about cancer, recommend healthy behaviors, and encourage them to get screened. Fair enough, but do we really need to launch a major epidemiological research study to make that point?
Last year, an Australian oncologist wrote in the Internal Medicine Journal, “The travesty in Erin Brockovich is not that ingested chromium VI causes breast and other tumor types; the travesty is the suggestion that the impact of any environmental carcinogen could be so marked as to be self-evident.” Perhaps it’s time for Ricky Schroder et al. to leave epidemiology to epidemiologists.