Robust causal inference using directed acyclic graphs: the R package 'dagitty'.

Area of Science:

• Epidemiology
• Biostatistics
• Causal Inference

Background:

• Directed acyclic graphs (DAGs) are crucial for causal inference in epidemiology.
• They are used to identify covariate adjustment sets for minimizing confounding bias.
• The DAGitty web application is a popular tool for drawing and analyzing DAGs.

Purpose of the Study:

• Introduce the R package 'dagitty' for accessing DAGitty's capabilities within R.
• Provide new functions for evaluating DAG consistency, enumerating statistically equivalent DAGs, and identifying robust adjustment sets.
• Enhance causal inference by enabling detection of misspecifications and ensuring valid inferences.

Main Methods:

• Leveraging the R statistical computing platform.
• Implementing functions for DAG consistency evaluation against datasets.
• Developing algorithms for enumerating statistically equivalent DAGs.
• Identifying exposure-outcome adjustment sets valid across equivalent DAGs.

Main Results:

• The R package 'dagitty' offers comprehensive DAG analysis within R.
• Users can evaluate DAG-dataset consistency and identify equivalent DAGs.
• Robust adjustment sets can be identified, ensuring valid causal inferences.

Conclusions:

• The 'dagitty' R package empowers epidemiologists with advanced causal inference tools.
• It facilitates the detection of causal misspecifications in DAGs.
• This leads to more robust and reliable epidemiological findings.