The use of epidemiologic data in the courts

It is difficult to make generalizations concerning the type of epidemiologic data currently preferred in the courts because trial judges' decisions are infrequently published, appellate judges' decisions are not always accompanied by an opinion, and there is often disparity between the opinions and standards of different courts. In this paper I have chosen cases either for their legal significance or their illustrative nature. The cases essentially represent numerator data, and therefore, I cannot determine definite trends in the courtroom use of epidemiologic data. When trying to determine if a causal relationship exists between an exposure and illness, epidemiologists attempt to examine the statistical strength, consistency, specificity, temporal sequence, dose-response characteristics, and biologic plausibility of the association. In contrast, a widely used legal standard of proof of association is "reasonable probability." However, the interpretation of "reasonable probability" is evolving. And just as epidemiologists may be pushed to the limit of their methods in trying to determine if and at what level a health risk exists, so the courts have had to forge new legal paths in considering whether epidemiologic evidence is sufficient to establish a causal relation between exposure to a substance and ill health. Furthermore, it is not certain what standard of significance the courts will require in the future. In 1975 in Reserve Mining Company vs. EPA, and Ethyl Corporation vs. EPA, the theoretical existence of risk was adequate to move the Appeals Courts to take precautionary action. But when considering a standard for safe levels of an occupational toxin, the Supreme Court has rejected a theoretical risk as significant (in Industrial Union vs. American Petroleum Institute) and instead, has required factual evidence of a dose-response relationship (as in American Textile Manufacturer's Institute vs. Donovan) because it did not want to extrapolate the risk of low levels of exposure from data based on high levels of exposure. Since demonstration of a dose-response relationship, although rarely obtained easily, is one of the criteria used by epidemiologists to infer a causal relationship, there can be no doubt that such evidence will strengthen any attempt to prove future damages from low level exposure to toxins. In addition, I predict evidence of attributable risk or attributable fraction will be useful in determining the extent of potential losses and, therefore, the value of future damages.(ABSTRACT TRUNCATED AT 400 WORDS)