By Alan Farago
This is a story about pediatric cancer clusters in Florida. It begins in Durham, N.C., at Duke University, where Dr. David Banks is a professor in the Department of Statistical Science.
In 2013 Dr. Banks was the new editor of Statistics and Public Policy, a journal of the American Statistical Association. In early February, Banks gave a speech to the Florida Chapter of the ASA at the University of West Florida in Pensacola. There, he met with Dr. Raid Amin, a distinguished statistics professor at the university.
Three years earlier, a team led by Dr. Amin had published a paper, “Epidemiological Mapping of Florida Childhood Cancer Clusters.”
Dr. Amin’s statistical analysis of pediatric cancers in Florida – from the years 2000 to 2007 – concluded that there are significant cancer clusters in two large areas of Florida: the southern region of Florida and in northeast Florida. That struck one of the most sensitive nerves in state government.
Its publication was lightly reported in the press, but to state officials charged with monitoring public health, it followed in the highly publicized wake of two claims of pediatric cancer clusters, one in Port St. Lucie in the 1990s and another in an unincorporated area of West Palm Beach called The Acreage in 2009. Even today, reading the plaintive cries for help from aggrieved parents is heart-wrenching.
Although the causes of pediatric cancer are poorly understood, the incidence of its most common forms – leukemia, cancers of the brain and nervous system, and lymphoma – are rising around the United States, according to the Centers for Disease Control and Prevention. As the incidence rises, so do public calls for some sort of accountability. In Florida, that particular responsibility falls to the State Department of Health.
In both Florida instances, Port St. Lucie and West Palm Beach, the Department of Health – the agency charged with registering data and investigating claims of cancer clusters – emphasized to the public that there was no identifiable cause and effect. As in the case of most claims about cancer clusters, the angst and public outcry generated media attention then blew out like a passing storm.
Cancer strikes like lightning. It doesn’t matter if it is adult or pediatric cancer, although the very mention of pediatric cancer makes me nauseous. According to the CDC, cancer is the leading cause of death by disease among children. For most people, when cancer strikes, it instantly becomes a full time occupation – from flights of hope, tumbles to despair and every step between where the mind can rest. There are no part-time victims of cancer.
In the midst of battling cancer, it rarely occurs – or seems irrelevant if it does – that one’s personal or family catastrophe ought to be weighed as 1) a deviation from a statistical norm or 2) that there is something in the external environmental that caused the cancer, unless one begins to hear about other rare cancers in the neighborhood. Then, alarm bells go off like no others. (A cancer cluster is where a greater than expected number of cancer malignancies occur within a group, a geographic area, or a period of time.)
Dr. Banks and Dr. Amin are not in the cause-and-effect end of cancer. Statistics is the practice or science of collecting and analyzing numerical data in large quantities. The idea is that you take a big soup of numbers, strain it, analyze what’s in the numbers and infer proportions from a representative sample. It is a practice that can be applied over any kind of dataset – from growing nuts to assembly of microchips.
In an interview, Dr. Banks told me that after visiting the University of West Florida, he was intrigued. Are there cancer clusters in Florida or not?
The 2010 report, covering data from 2000 – 2007, states that “… during this time, there were 4,591 cases of pediatric cancer diagnosed, of which 1,254 (27 percent) had leukemia, 839 (18 percent) had brain/central nervous system (CNS) cancer, and 252 (5.5 percent) had lymphoma.”
In south Florida, the cluster encompasses the southwest, south central and southeast regions where “… compared with the state, there is a statistically significant 36 percent increased risk of childhood cancer.” Also, “In the northeast Florida cluster, there were 466 and 375 observed and expected cases, respectively. This region appears to be smaller in size, although it may represent a more densely populated area. … In addition, a third overall childhood cancer cluster was identified in a small area of central Florida in which the observed number of cases was 31 as compared to 11 expected cases. The rates were statistically significantly higher in this area relative to the state … which implies that compared with the state of Florida, those in this area are almost three times as likely to be diagnosed with childhood cancer.”
“The relative risk values are not small, indicating cancer rates that are higher than what is found in other parts in Florida by 35 to 52 percent.”
Following its publication, in March 2010 Dr. Amin and his colleagues were invited to Tallahassee by the Surgeon General of Florida, state health officials and the CDC to discuss their findings. In an interview, Dr. Amin told me, “It was a chilly meeting.” At the time, a state health official told the Palm Beach Post that the report’s analysis methods were “relatively new and untested” and that “independent researchers will use this report to identify areas that require additional study using more traditional methods.”
A few weeks later, the Florida Department of Health told Dr. Amin that the agency’s data sets including pediatric cancers did not match the Florida researchers’. According to Amin, “They promised to share their data if we gave them ours. We gave them our data, and they never gave us theirs.”
For the ensuing three years, after the publication of their paper in Pediatric Blood Cancer, silence.
Fast forward to February 2013, when Duke professor David Banks met professor Raid Amin at the University of West Florida.
“Since I had just started as the publication editor, we were looking for interesting projects,” Dr. Banks told me in an interview. During the meeting Dr. Banks and Dr. Amin discussed a study Amin and colleagues had recently completed on clusters of adolescent and young adult thyroid cancers in Florida counties.
Banks said it would be an interesting idea to update the earlier 2010 study and have other statisticians review the data and conclusions. In other words, apply independent methodologies on the same data to ask the question: are there cancer clusters in Florida?
Amin said he would need to check if he could share the data, and a few weeks later informed Banks it would be okay to proceed. Banks “started to contact some of the statisticians (he) knows who are prominent and work in epidemiological statistics. Some were busy and declined, others were interested in the experiment of having multiple papers on the same data, and eventually I obtained five papers (one by Dr. Amin). These all went through peer review, and after revision were ultimately published. To the best of my knowledge, there were no previous peer-reviewed papers comparing different methods on the same data.”
On May 29, 2014, Dr. Amin and colleagues published their updated paper reviewing pediatric cancer data covering the 11-year period from 2000-2010. Dr. Amin’s report identifies two pediatric clusters – in the Miami metro area and an area nearby, west of the Everglades around the southern portion of Lake Okeechobee. The cluster analysis results for the three cancer types and the total cancer rates are similar to what Amin and colleagues had concluded in 2010. “The three most widely recorded pediatric cancer types all occur in a geographical area that is close to Miami and to Lake Okeechobee. The relative risk values are not small, indicating cancer rates that are higher than what is found in other parts in Florida by 35 percent –52 percent.”
A year later, on April 17, 2105, the American Statistical Association published its review of the same data. “The process followed the usual academic path. It takes a long time to do,” Dr. Banks said. “Each of the five teams that took on this project, submitted drafts to peer review. Each paper had two referees to review the work, and some of the papers went through a few rounds of revisions. It is the first time we had done something like this.”
Five separately conducted methodological assessments, summarized by Dr. Lance Waller, a biostatistics professor specializing in spatial epidemiology at Emory University in Atlanta, confirmed Dr. Amin and colleagues’ findings: “The five author groups find some consistent results: there seem to be local areas where the observed cancer rate is statistically significantly higher than we would expect and the different methods tend to identify a few common areas. As noted above, the clusters themselves are not identical, but they do overlap. …”
For Dr. Amin, five years have passed since he and his team alerted the state to the presence of cancer clusters in Florida, but the memory is still vivid.
Not as vivid, of course, as the memory of anyone who suffered through a child’s cancer.
In 2010, when the state of Florida contested Dr. Amin’s data, a state health officer recommended that Amin and his colleagues contact the journal in which the paper had been published. Amin was troubled. “We saw no errors (in our data), and we suggested to them that research is not done that way. If you (the state) think you found errors, then you write the editors and you provide the correct data and results. They never did.”
I recently submitted two questions to the Florida Department of Health: 1) In a recent journal of the American Statistical Association, Dr. Lance Waller summarized the results of five independent statistical analyses of cancer data in Florida. What is the DOH response to that report? 2) Pediatric Blood Cancer 2010 published an earlier study, “Epidemiological Mapping of Florida Childhood Cancer Clusters.” What was the DOH response to that article at the time, and has the position of DOH changed or is it the same?
In response, I received the following statement: “The Florida Department of Health cannot comment on the two studies mentioned as we have not had adequate time to review.”
This August in Seattle, at the upcoming meeting of the American Statistical Association, Dr. Amin will discuss his research in a special session titled, “Are there cancer clusters in Florida?”