Causal diagrams for epidemiologic research

Casual diagrams have a long history of informal use and, more recently, hav e undergone formal developmental for applications in expert systems and rob otics. We provide an introduction to these developments and their use in ep idemiologic research. Casual diagrams can provide a starting point for iden tifying variables that must be measured and controlled to obtain unconfound ed effect estimates. They also provide a method for critical evaluation of traditional epidemiologic criteria for confounding. In particular, they rev eal certain heretofore unnoticed shortcomings of those criteria when used i n considering multiple potential confounders. We show how to modify the tra ditional criteria to correct those shortcomings.