Directed acyclic graphs: A tool to identify confounders in orthodontic research, Part II
Am J Orthod Dentofacial Orthop 2017;151:619-21
In the previous article, we discussed the problem of confounding and presented 3 fundamental methods for assessing and adjusting for confounders: the traditional approach, the noncollapsibility approach, and the directed acyclic graphs (DAGs) or causal diagrams approach. DAGs are nonparametric structural methods to identify potential confounders through the presentation of variables and the relationship between them in the form of a graph. A DAG depicts the relationship between the exposure (E) or intervention and the disease (D) or outcome in addition to any other variables associated with E and D. The first step in building a causal diagram is to determine the effect of E on D independent of all other associated variables or paths. The second step is to statistically assess conditional independence in the causal diagram.