Distinguishing Individual Risk from Population Attributable Risk in Toxic Tort Litigation. S. L. Schwartz and P. Witorsch, Dept. of Pharmacology and Medicine, Georgetown University Medical Center
Epidemiological and toxicological evidence in court can be misconstrued, if not distorted, by confusing general causation with individual causation. A particular logical error occurs when the population attributable risk for an exposure characteristic is considered a measure of causation probability in the individual exposed to the characteristic. Robins and Greenland (Biometrics 45:1125-1138, 1989) have shown by mathematical definition that probability of causation among exposed cases cannot be ascertained from epidemiological data. Evaluating the probability of causation in an individual case requires the evaluation of three factors: (1) the general background knowledge concerning the putative causative agent; (2) specific knowledge concerning the exposure to the putative causative agent, and (3) case specific knowledge concerning the individual being evaluated. Much has been written in the scientific and legal literature concerning the use of Bayes theorem to facilitate this evaluation. There is substantial reason to consider Bayes theorem as a useful heuristic device for examining relevant evidence. Bayes theorem is directed at a posterior probability, the probability than a hypothesis is true once all of the evidence has been assessed. A Bayesian assessment is sensitive to that evidence that is entailed by the hypothesis; is sensitive to the evidence that contradicts the hypothesis; and it is sensitive to alternative hypotheses that could result in the observations made. Though an actual mathematical Bayesian assessment could provide significant advantages, it is also not easily achieved. Further, there is controversy concerning the use of a purely mathematical approach at trial. It is feasible to use an algorithmic approach. Algorithms provide a useful tool for decision-making and they can incorporate structural components of Bayes theorem. We have devised and applied such methods in a wide variety of courtroom situations.
Go to . . .
2001 SRA Annual Meeting Table of Contents
2001 SRA Annual Meeting Author Index
Main Abstracts Menu Page
RiskWorld Home Page