by Dan Fagin
15. The Ames test works by using Salmonella bacteria that have mutated and lost their ability to produce the amino acid histidine, without which the bacteria cannot grow. The tester applies the mutated Salmonella to a medium that includes a small amount of histidine as well as the chemical being tested. The bacteria will grow for a short while until the histidine is consumed but then will stop—unless the test chemical causes the bacteria’s DNA to mutate back into a form that can produce its own histidine. If the chemical is highly mutagenic, the Salmonella will mutate and thrive even in a histidine-free medium. Some versions of the test involve using bacteria with defective outer coats, which makes them more vulnerable to the chemicals being tested. Sometimes liver enzymes are added to the culture medium to simulate the effects of human metabolism, which often alters the molecular structure of synthetic chemicals after ingestion.
16. The identification of mutagenic compounds is not always a clear-cut process and is heavily dependent on the assay being used. The compounds identified as mutagenic in this chapter are listed in two government sources: Walter W. Piegorsch and David G. Hoel, “Exploring Relationships between Mutagenic and Carcinogenic Potencies,” National Institute of Environmental Health Sciences, Mutation Research 196 (1988): 161–75, 167–69, table 2; and “2007 Right to Know Special Health Hazardous Substance List: Mutagens,” available online from the New Jersey Department of Health and Senior Services.
17. United States General Accounting Office, Superfund Program: Current Status and Future Fiscal Challenges, July 31, 2003.
18. The author was able to identify and interview Lisa Boornazian thanks to Ellen Tracy, the director of oncology nursing at the Children’s Hospital of Philadelphia. In 2007, the author contacted Tracy, who had worked in the oncology ward for many years, to ask if she knew the identity of the nurse who had anonymously initiated the Toms River investigation more than twelve years earlier. She did. Tracy agreed to find the nurse and to ask her if she was willing to consent to an interview. Lisa Boornazian gave her consent, having decided there was no longer any reason to remain anonymous because so much time had passed and she no longer worked at the hospital. Her sister-in-law, Laura Janson, who still works in the EPA’s Philadelphia office, felt the same way.
Chapter Fifteen
1. In America, the New York tycoon John Jacob Astor III built the country’s first cancer hospital in 1887 after his earlier effort to fund a cancer wing at an existing hospital was rejected on the grounds that it might threaten other patients with infection. See James T. Patterson, The Dread Disease: Cancer and Modern American Culture (Harvard University Press, 1987), 23.
2. The famous member of the Webb family was Thomas’s younger brother, Captain Matthew Webb, who in 1875 became the first person to swim the English Channel without floats or other artificial aids. Victorian-era matchboxes displayed his picture. Webb drowned in 1883 while trying to swim across the swirling rapids just below Niagara Falls in New York. His brother Thomas Law Webb, despite his contribution to the field of biostatistics, earns but one line in his brother’s two-page entry in the 2001 edition of the Dictionary of National Biography (60:104–5).
3. For more about the life and work of Karl Pearson, see Helen W. Walker, “The Contributions of Karl Pearson,” Journal of the American Statistical Association 53:281 (March 1958): 11–22; and M. Eileen Magnello, “Karl Pearson and the Origins of Modern Statistics: An Elastician Becomes a Statistician,” Rutherford Journal 1 (December 2005).
4. For a concise, if dated, recounting of the history of significance testing and the long-running argument over its importance, see Ronald N. Giere, “The Significance Controversy,” British Journal for the Philosophy of Science (May 1972): 170–81.
5. Karl Pearson, “On ‘Cancer Houses,’ from the Data of the late Th. Law Webb, M.D.,” Biometrika (January 1912): 430–35. Pearson cofounded Biometrika in 1901 after the Royal Society refused to publish articles on biostatistics, deeming the field insufficiently scientific. To test how likely it was that the 377 cancer deaths in Madeley were distributed randomly, Karl Pearson conceived of the following experiment: If a bag were filled with two thousand balls, numbered one through two thousand, and someone then drew a ball out of the bag 377 times, recording the ball’s number each time before returning it to the bag and picking again, at the end of the experiment how many numbers would be written down twice, or even three or more times? Writing numbers on two thousand balls would be a tedious task, so Pearson designed some less burdensome experiments to answer the question through the use of random numbers and the drawing of playing cards. He conducted five such experiments, and each time found that the distribution was very close to what he had predicted. In only one of the five experiments was a number selected more than twice, and it was never picked four times. But in Madeley, there were six houses with three cancer cases and one with four, an extremely unlikely result for a chance distribution.
6. Karl Pearson wrote, “Dr. Law Webb’s data provide sufficient evidence to justify a demand for a thorough investigation of the subject, such as is not feasible in the case of the individual medical man. They do not finally demonstrate that cancer is more frequent in one house than a second, but they do justify a complete inquiry into the possibility that ‘cancer-houses’ are not wholly a myth, in other words, that immediate environment is in the long run a factor of the frequency of cancer.” See Pearson, “On ‘Cancer Houses,’ ” 434.
7. In 1932, two French statisticians, Auguste Lumière and Paul Vigne, analyzed 6,703 cancer deaths in 5,027 homes in the city of Lyon over a twenty-year period, including one home in which there had been eight deaths. Instead of trying to use a statistical formula to estimate a normal distribution of cases, as Pearson had, Lumière and Vigne instead obtained 6,703 birth, marriage, and death certificates for the city and studied the distribution of addresses on the certificates. What they found was that births, marriages, and deaths in Lyon clustered just as much as cancer did. Their discovery strongly suggested, the Frenchmen concluded, that mere chance was at work, not a hidden cause. A description of Lumière and Vigne’s work in Lyon appears in Percy Stocks, “The Frequency of Cancer Deaths in the Same House and in Neighbouring Houses,” Journal of Hygiene (February 1935): 46–63. The Lyon study is also described on page 10 of the May 5, 1933, edition of Science entitled “Cancer House Disproved Statistically.”
8. Stocks, “Frequency of Cancer Deaths,” 46–63.
9. At the time, the CDC was known as the Communicable Disease Center. Since then, as its portfolio has expanded, its name has changed four times: National Communicable Disease Center in 1967, Center for Disease Control in 1970, Centers for Disease Control in 1980, and Centers for Disease Control and Prevention in 1992.
10. Clark W. Heath Jr. and Robert J. Hasterlik, “Leukemia Among Children in a Suburban Community,” American Journal of Medicine 34 (June 1963): 796–812.
11. Sister Mary Viva, the school principal in Niles, Illinois, did give investigator Clark Heath one bit of tantalizing information: During the same years the leukemia cases were diagnosed, several waves of a contagious fever had swept through the school. There were no reliable records of the outbreaks, but Heath’s conversations with the school nurses suggested that they coincided with the months in which several of the leukemia cases were diagnosed. The researchers also found antibodies that might be related to leukemia in the blood samples of several relatives of the victims. To Heath, this was a hint that the same virus that had spread the fever might also have triggered leukemia in a few genetically susceptible children. As Alfred Knudson would have put it, the infection might have provided the second “hit.”
Fifty years later, researchers are still searching for a virus associated with acute lymphocytic leukemia in children. Their efforts have been spurred by two closely related hypotheses suggested by studies of childhood leukemia clusters. The “population mixing” hypothesis was first proposed in 1988 by University of Oxford epidemiologist Leo Kinlen, who was seeking to
explain a cluster of childhood leukemia near a nuclear power plant. Kinlen suggested that ALL may arise from an extreme immune response to a viral infection. Though very rare, these extreme reactions would most likely occur in fast-growing communities in which infected populations mix with nonimmune children who are highly susceptible. Another eminent British researcher, biologist Melvyn Greaves of the Institute of Cancer Research in London, has expanded on Kinlen’s ideas, as well as Knudson’s, by proposing the “delayed infection” hypothesis. Greaves and his frequent collaborator, Joseph Wiemels of the University of California at San Francisco, believe that the first “hit” occurs in utero, while the second involves a communicable infection that most children easily suppress but may trigger leukemia between the ages of two and five in a very small number of highly susceptible children. (These rare children, Greaves suggests, are less likely to have been in day care, where exposure to other children would have built up their immune systems.)
In an interview with the author, Greaves said that it is possible that the initial hit during pregnancy could be from exposure to a carcinogenic compound but noted that the dose would have to be large enough to trigger genetic damage. (Greaves is especially interested in a translocation called TEL-AML1 often found in ALL patients but also present in many healthy children.) For a current assessment of the population mixing and delayed infection hypotheses, see Kevin Urayama et al., “A Meta-Analysis of the Association between Day-Care Attendance and Childhood Lymphoblastic Leukaemia,” International Journal of Epidemiology 39 (2010): 718–32.
12. Clark W. Heath, “Community Clusters of Childhood Leukemia and Lymphoma: Evidence of Infection?” American Journal of Epidemiology 162:9 (November 1, 2005): 817–22. After leaving the Centers for Disease Control, Heath was a vice president of the American Cancer Society, where his generally skeptical views about environmental causation of cancer would raise the ire of activists.
13. By the mid-1980s, thirty-seven states had established cancer registries or were in the process of doing so. See Stephanie C. Warner and Timothy E. Aldrich, “The Status of Cancer Cluster Investigations Undertaken by State Health Departments,” American Journal of Public Health 78:3 (March 1988): 306–7.
14. Michael Greenberg and Daniel Wartenberg, “Communicating to an Alarmed Community about Cancer Clusters: A Fifty State Survey,” Journal of Community Health 16:2 (April 1991): 71–82. Greenberg and Wartenberg, of Rutgers University, surveyed all fifty state health departments in 1989 and concluded that they were getting a total of 1,300 to 1,650 requests for cluster investigations per year.
15. For more about “occult multiple comparisons,” see Raymond Richard Neutra, “Counterpoint from a Cluster Buster,” American Journal of Epidemiology 132:1 (July 1990): 1–8.
16. This kind of post hoc hypothesis generation is sometimes called Texas sharpshooting, after a joke in which a cowboy wildly fires shots into the side of a barn, then looks to see where the bullet holes have randomly clustered and draws a target centered on the densest cluster, declaring himself a sharpshooter.
17. There is, arguably, at least one exception to the universal failure of residential cluster analysis in identifying new carcinogens. The mineral erionite was identified as a lung carcinogen in the 1980s based on studies of two villages in central Turkey in which there were very high rates of mesothelioma and also high concentrations of airborne erionite fibers. As Raymond Neutra points out in “Counterpoint from a Cluster Buster,” however, the erionite example is not a conventional time-limited cluster because lung cancer has apparently been endemic in those villages for centuries.
18. Glyn G. Caldwell, “Twenty-Two Years of Cancer Cluster Investigations at the Centers for Disease Control,” American Journal of Epidemiology 132, Supp. 1 (1990): S43–S47.
19. From 1978 to 1984, the National Institute for Occupational Safety and Health investigated sixty-one alleged workplace cancer clusters and identified a plausible cause in just five, none of which could be scientifically confirmed. In forty-six of those clusters, NIOSH found, rates were not actually higher than expected. See Paul A. Schulte et al., “Investigation of Occupational Cancer Clusters: Theory and Practice,” American Journal of Public Health 77:1 (January 1987): 52–56.
20. Phil Brown, “Popular Epidemiology and Toxic Waste Contamination: Lay and Professional Ways of Knowing,” Journal of Health and Social Behavior 33 (September 1993): 267–81.
21. Kenneth Rothman’s keynote address to the “Cluster Buster” conference in February of 1989 was later published in a special issue of the American Journal of Epidemiology that also included twenty-five other papers presented at the same conference. Rothman’s contribution, entitled “A Sobering Start for the Cluster Busters’ Conference” is on pages S6–S13 of the special issue, which was published in July of 1990 (vol. 132, Supp. 1). The entire issue serves as an excellent guide to the perils and promise of investigating disease clusters, including cancer clusters.
22. Atul Gawande, “The Cancer Cluster Myth,” The New Yorker, February 8, 1998, 34–37.
23. In addition to the 75 percent who were satisfied after an explanatory telephone conversation or letter, another 20 percent of cluster callers in Minnesota, according to Alan Bender, could be mollified by supplying them with preexisting data about cancer rates and environmental conditions. “Relatively modest” data collection and analysis took care of another 4 percent. Only 1 percent of the time did Minnesota get to step four: an in-depth investigation. Bender’s department conducted only five of those between 1981 and 1988, a period during which it had fielded 420 telephone calls or letters about clusters. See Alan P. Bender et al., “Appropriate Public Health Reponses to Clusters: The Art of Being Responsibly Responsive,” American Journal of Epidemiology 132, Supp. 1 (July 1990): S48–S52.
24. In its “Guidelines for Investigating Clusters of Health Events,” published in final form on July 27, 1990, the Centers for Disease Control declared: “The unofficial consensus among workers in public health is that most reports of clusters do not lead to a meaningful outcome.”
25. Craig W. Trumbo, “Public Requests for Cancer Cluster Investigations: A Survey of State Health Departments,” American Journal of Public Health 90:8 (August 2000): 1300–1302.
26. The 1994 report, entitled Childhood Cancers in New Jersey: 1980–1988, is no longer available on the state health department’s website. The revised report, Childhood Cancer in New Jersey: 1979–1995, was published in 1999, in the midst of the Toms River cancer furor, and does not include the county-to-county comparisons that incited local activists.
Chapter Sixteen
1. The terms of the 1992 legal settlement that resolved the criminal charges against Ciba-Geigy also required the company to put $50 million into a trust fund for environmental cleanup at the factory site—money it was already obligated to pay under the Superfund program.
2. Patricia A. Miller, “Ciba Guilty of Pollution,” Asbury Park Press, February 29, 1992.
3. From the river, the Toms River “core zone” extended north for two miles to State Road 571. Its eastern border was Vaughn Avenue, its western edge an irregular line that included parts of the Garden State Parkway and Old Freehold Road.
4. The cancer incidence rate for United States children under age twenty was fourteen cases per one hundred thousand per year in 1995, according to the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) program. By 2005, the rate had risen to sixteen cases per one hundred thousand.
5. This small-numbers problem was less severe in the township, where there were about twenty thousand children, and in Ocean County, where there were one hundred thousand.
6. The data tables in this chapter are based on tables produced by Michael Berry for his 1995 incidence analysis. They were later reproduced in Appendix 1 of Childhood Cancer Incidence Health Consultation: A Review and Analysis of Cancer Registry Data, 1979–1995 for Dover Township, New Jersey Department of Health and Senior Services (December 1997).
8. Fabio Barbone’s doctoral thesis, A Nested Case-Control Study of Lung Cancer and Central Nervous System Neoplasms Among Chemical Workers (1989), was later republished in the form of two journal articles: Fabio Barbone et al., “A Case-Control Study of Lung Cancer at a Dye and Resin Manufacturing Plant,” American Journal of Industrial Medicine 22 (1992): 835–49; and Fabio Barbone et al., “Exposure to Epichlorohydrin and Central Nervous System Neoplasms at a Resin and Dye Manufacturing Plant,” Archives of Environmental Health 49:5 (September 1, 1994): 355–58.