[Use of causal diagrams in Epidemiology: application to a situation with confounding].