Directed Acyclic Graphs for Oral Disease Research.

Area of Science:

• Epidemiology
• Dental Research
• Causal Inference

Background:

• Directed acyclic graphs (DAGs) are graphical tools for causal inference.
• Their adoption in dental research remains limited despite advocacy since 2002.
• DAGs identify threats to causal inference like confounding and selection bias.

Purpose of the Study:

• To illustrate the application of DAGs in dental research using a temporomandibular disorder (TMD) study.
• To demonstrate how DAGs identify confounders, mediators, and colliders.
• To highlight DAGs' role in informing study design and data analysis strategies.

Main Methods:

• Utilized DAGs to model causal relationships in a temporomandibular disorder (TMD) study.
• Examined the causal effect of facial injury on TMD risk.
• Illustrated identification of confounders, mediators, colliders, and variables with dual roles.

Main Results:

• DAGs effectively identify potential confounders and mediators in exposure-outcome associations.
• Demonstrated how conditioning on colliders can introduce bias.
• Showcased that adjusting for certain variables (e.g., pressure pain threshold) can bias TMD risk estimation.

Conclusions:

• DAGs are valuable tools for enhancing causal inference in dental research.
• Their application can prevent bias and guide appropriate statistical analysis.
• DAGs aid in optimizing study design, subject selection, and variable measurement for TMD research.