[Attributable risk percent under an effect modifier and legal causality].

Among civil trials there are cases which are influenced by evidence derived from epidemiologic studies. In such cases, causality of a factor (X) illegitimately introduced by defendants is considered to be measured with an epidemiologic measure, attributable risk percent (AR%) expressing the level of risk. This paper aims to discuss calculations and interpretations of AR% in complicated cases where plaintiffs themselves introduced an additional risk factor (Y) for the alleged health injury. When X is not an effect modifier, AR% of X adjusted for Y can be simply adopted for arriving at a judgement of causality. Where Y is an effect modifier, and is not an indispensable item in the plaintiffs' daily life (e.g., smoking), the AR% of X for those not exposed to Y and also the AR% for those exposed to Y may both need to exceed a legally determined threshold, in order for X to be legally acknowledged as being causal to the alleged injury. The role of epidemiologists for such trials is to determine the AR% of X for each level of Y, and to advise the court on a realistic range of the AR%s.