How to develop causal directed acyclic graphs for observational health research: a scoping review

Louise Poppe, Johan Steen, Wen Wei Loh, Geert Crombez, Fien De Block, Noortje Jacobs, Peter W G Tennant, Jelle Van Cauwenberg, Annick L De Paepe*

*Corresponding author for this work

Research output: Contribution to journal(Systematic) Review article peer-review

Abstract

Causal directed acyclic graphs (DAGs) serve as intuitive tools to visually represent causal relationships between variables. While they find widespread use in guiding study design, data collection and statistical analysis, their adoption remains relatively rare in the domain of psychology. In this paper we describe the relevance of DAGs for health psychology, review guidelines for developing causal DAGs, and offer recommendations for their development. A scoping review searching for papers and resources describing guidelines for DAG development was conducted. Information extracted from the eligible papers and resources ( = 11) was categorised, and results were used to formulate recommendations. Most records focused on DAG development for data analysis, with similar steps outlined. However, we found notable variations on how to implement confounding variables (i.e., sequential inclusion versus exclusion). Also, how domain knowledge should be integrated in the development process was scarcely addressed. Only one paper described how to perform a literature search for DAG development. Key recommendations for causal DAG development are provided and discussed using an illustrative example.
Original languageEnglish
Pages (from-to)1-21
Number of pages21
JournalHealth Psychology Review
DOIs
Publication statusPublished - 27 Sept 2024

Keywords

  • Directed acyclic graph
  • causal diagram
  • causal inference
  • development
  • guidelines
  • recommendations

Fingerprint

Dive into the research topics of 'How to develop causal directed acyclic graphs for observational health research: a scoping review'. Together they form a unique fingerprint.

Cite this