STATS at George Mason University

Dueling evidence - who’s right?

In determining its cancer guidelines, the EPA favor studies based on dose-response assessments, which means that the higher the dose of a substance administered, the greater the likely effect. Therefore, to assess the risk of a given substance in humans, it is necessary to know how much of it they have been exposed to, how often, and what the result was.

As the occupational studies on PCB exposure in humans cannot provide this level of precision (and as it would be unethical to conduct dose–response assessments on humans), the EPA dismisses much of the epidemiological evidence on PCBs as “inconclusive” — except, that is, when it appears to support its case.

Thus, the EPA cites three occupational studies for showing small but “statistically significant” increases in cancer among workers exposed to PCBs: Bertazzi et al. (1987), Brown (1987), and Sinks et al. (1992). There are compelling reasons to discount each of them.

In 1997, the Supreme Court reviewed a lawsuit brought by Robert Joiner, an electrician who claimed that exposure to PCBs, furans and dioxins in the course of working with materials manufactured by General Electric (GE) had “promoted” his small cell lung cancer. A review of “General Electric Co. et al. v. Joiner et ux. was occasioned by arguments over how much discretion lower courts could exercise in determining what counted as admissible scientific evidence. And both the District and the Supreme Courts agreed that the four epidemiological studies Joiner ‘s lawyers cited for a link between exposure to PCBs and cancer failed to show any such link and were therefore inadmissible.

Bertazzi et al. was one of those four studies. Even the EPA concedes that studies of chronic exposure to PCBs in humans “provide limited to inadequate evidence of carcinogenicity.” Though Bertazzi et al. found elevated rates of lung cancer deaths among former employees of an Italian capacitor manufacturing plant, the authors concluded that “there were apparently no grounds for associating lung-cancer deaths (although increased above expectations) and exposure in the plant.”

In delivering the opinion of the court, Chief Justice Rehnquist pointed out the obvious: “Given Bertazzi et al. were unwilling to say that PCB exposure had caused cancer among the workers they examined, their study did not support the experts’ conclusion that Joiner’s exposure to PCBs caused his cancer.” [The other three studies were dismissed for different scientific reasons]. The EPA’s description of Bertazzi et al. does not mention the authors’ reluctance to link elevated rates of cancer with exposure to PCBs.

Brown (1987) looked at rates of cancer mortality among workers at two capacitor-manufacturing plants in Massachusetts. According to the EPA “there were statistically significant increase in death from cancer of the liver, gall bladder, and bilary tract (5 observed, 1.9 expected).” But the U.S. Army’s Medical Research and Material Command’s Deployment Toxicology unit at Fort Dettrick notes that “a review of pathology reports indicated that two of the liver tumors counted in the follow-up study were not primary liver tumors. When these liver tumors are excluded, the elevation in incidence is not statistically significant. The result also may be confounded by population differences in alcohol consumption, dietary habits, and ethnic composition.” None of these caveats are noted by the EPA.

Finally, Sinks et al. (1992) compared the mortality rates of 3,588 capacitor manufacturing workers known to have been exposed to PCBs with national rates for those with a similar demographic profile. Mortality in general, and total mortality from cancer, turned out to be lower in the workers than in the general population. Again, this aspect of the study did not make the EPA report. Although there were elevated rates for malignant melanoma and cancer of the brain and nervous system, the authors stated that the risk of malignant melanoma “was not related to cumulative PCB exposure,” and that the increase in cases of brain cancer was too small to be statistically significant. In conclusion, they noted that “These results provide some evidence of an association between employment at this plant and malignant melanoma and cancer of the brain. The possibility that the results are due to chance, bias or confounding [factors] cannot be excluded.”

In sum, the scientific case for PCBs causing cancer in humans, based on chronic exposure in the workplace, remains unproven. In 2001, the German Federal Institute for Consumer Health Protection and Veterinary Medicine concluded in a consultation on behalf of the World Health Organization that the evidence from occupational studies amounted to “only a modest association for melanoma.” Even the EPA concedes that “Overall, the human studies have been considered to provide limited to inadequate evidence of carcinogenicity.”

about staff publications email contact search contribute
Salmon home
email this page to a friend

STATS home