Directed Acyclic Graphs in Surgical Research.

Area of Science:

• Surgical research methodology
• Causal inference in medicine

Background:

• Multivariable regression is common in surgical research for causal inference.
• Variable selection for control in regression lacks clear justification in surgical publications.
• Directed acyclic graphs (DAGs) are increasingly used in epidemiology but not surgical literature.

Purpose of the Study:

• Advocate for the adoption of DAGs in surgical research.
• Illustrate the importance of DAGs for defining and communicating variable relationships.
• Highlight the role of DAGs in mitigating bias in causal inference.

Main Methods:

• Review of the current literature on causal inference and DAGs.
• Explanation of DAGs as a formal logic tool for variable relationships.
• Discussion of bias introduction through inadequate control variable selection.

Main Results:

• DAGs offer a structured approach to identify confounders, mediators, and colliders.
• Lack of explicit DAG use can lead to unacknowledged bias in surgical studies.
• Epidemiology and other medical fields are increasingly adopting DAGs.

Conclusions:

• DAGs should be integrated into surgical research for robust study design and transparent reporting.
• Researchers should clearly communicate their rationale for control variable selection using DAGs.
• Adoption of DAGs will enhance the validity and interpretability of surgical research findings.