Causal graphs such as directed acyclic graphs (DAGs) are a novel approach in epidemiology to conceptualize confounding and other sources of bias. DAGs visually encode the causal relations based on a priori knowledge among the exposure of interest and the outcome while considering several covariates. The application of formal rules on these diagrams enables the identification of the causal and non-causal structures in the DAG. The causal effects are of interest and require no adjustment. Whereas the non-causal effects have to be checked for confounding and for which covariates adjustment is necessary. The identification of the adjustment set depends on the causal relations among the variables. The consideration of these relations is valuable because adjusting for more variables increases the risk of introducing bias. Considering every single path of a DAG allows the systematic identification of the causal structures in the DAG, and the determination of minimally sufficient adjustment sets for estimating the causal effect of the exposure on the outcome based on the underlying DAG. The aim of this paper is to provide an introduction to the basic assumptions as well as the steps for drawing and applying a DAG.
© Georg Thieme Verlag KG Stuttgart · New York.