Evidence synthesis for constructing directed acyclic graphs (ESC-DAGs): a novel and systematic method for building...

Area of Science:

• Epidemiology
• Causal Inference
• Systematic Reviews

Background:

• Directed acyclic graphs (DAGs) are crucial for identifying adjustment strategies in epidemiological analysis.
• Current methods lack clear direction for constructing DAGs, posing a challenge for researchers.
• Integrating DAG construction into research review stages is needed.

Purpose of the Study:

• To propose a novel protocol, Evidence Synthesis for Constructing Directed Acyclic Graphs (ESC-DAGs), for building DAGs.
• To combine evidence synthesis strategies with causal inference principles for DAG development.
• To demonstrate the ESC-DAGs protocol using literature on parental influences on adolescent alcohol use.

Main Methods:

• ESC-DAGs utilizes empirical studies from systematic reviews or reviews of systematic reviews.
• Key stages include mapping study conclusions to DAGs, assessing and correcting causal structures using causal inference principles, and synthesizing DAGs into integrated versions.
• The protocol was applied to literature concerning parental influences on adolescent alcohol use.

Main Results:

• ESC-DAGs provides a practical, systematic, and transparent method for DAG development.
• The generated DAGs can guide primary data analysis and DAG-based sensitivity analyses.
• The protocol's modular design accommodates both experienced and novice DAG users.

Conclusions:

• ESC-DAGs offers a structured approach to DAG construction from background knowledge.
• The protocol enhances the reliability of DAGs used in epidemiological research.
• ESC-DAGs is adaptable and accessible, promoting wider adoption in research.