Use of directed acyclic graphs (DAGs) to identify confounders in applied health research: review and recommendations

Peter W G Tennant; Eleanor J Murray; Kellyn F Arnold; Laurie Berrie; Matthew P Fox; Sarah C Gadd; Wendy J Harrison; Claire Keeble; Lynsie R Ranker; Johannes Textor; Georgia D Tomova; Mark S Gilthorpe; George T H Ellison

doi:10.1093/ije/dyaa213

Use of directed acyclic graphs (DAGs) to identify confounders in applied health research: review and recommendations

Int J Epidemiol. 2021 May 17;50(2):620-632. doi: 10.1093/ije/dyaa213.

Authors

Peter W G Tennant^{1

2

3}, Eleanor J Murray⁴, Kellyn F Arnold^{1

2}, Laurie Berrie^{1

5

6}, Matthew P Fox^{4

7}, Sarah C Gadd^{1

5}, Wendy J Harrison^{1

2}, Claire Keeble¹, Lynsie R Ranker⁴, Johannes Textor⁸, Georgia D Tomova^{1

2

3}, Mark S Gilthorpe^{1

2

3}, George T H Ellison^{1

2

9}

Affiliations

¹ Leeds Institute for Data Analytics, University of Leeds, Leeds, UK.
² Faculty of Medicine and Health, University of Leeds, Leeds, UK.
³ Alan Turing Institute, British Library, London, UK.
⁴ Department of Epidemiology, School of Public Health, Boston University, Boston, MA, USA.
⁵ School of Geography, University of Leeds, Leeds, UK.
⁶ School of GeoSciences, University of Edinburgh, Edinburgh, UK.
⁷ Department of Global Health, Boston University, Boston, MA, USA.
⁸ Department of Tumour Immunology, Radboud University Medical Center, Nijmegen, The Netherlands.
⁹ Centre for Data Innovation, Faculty of Science and Technology, University of Central Lancashire, Preston, UK.

Abstract

Background: Directed acyclic graphs (DAGs) are an increasingly popular approach for identifying confounding variables that require conditioning when estimating causal effects. This review examined the use of DAGs in applied health research to inform recommendations for improving their transparency and utility in future research.

Methods: Original health research articles published during 1999-2017 mentioning 'directed acyclic graphs' (or similar) or citing DAGitty were identified from Scopus, Web of Science, Medline and Embase. Data were extracted on the reporting of: estimands, DAGs and adjustment sets, alongside the characteristics of each article's largest DAG.

Results: A total of 234 articles were identified that reported using DAGs. A fifth (n = 48, 21%) reported their target estimand(s) and half (n = 115, 48%) reported the adjustment set(s) implied by their DAG(s). Two-thirds of the articles (n = 144, 62%) made at least one DAG available. DAGs varied in size but averaged 12 nodes [interquartile range (IQR): 9-16, range: 3-28] and 29 arcs (IQR: 19-42, range: 3-99). The median saturation (i.e. percentage of total possible arcs) was 46% (IQR: 31-67, range: 12-100). 37% (n = 53) of the DAGs included unobserved variables, 17% (n = 25) included 'super-nodes' (i.e. nodes containing more than one variable) and 34% (n = 49) were visually arranged so that the constituent arcs flowed in the same direction (e.g. top-to-bottom).

Conclusion: There is substantial variation in the use and reporting of DAGs in applied health research. Although this partly reflects their flexibility, it also highlights some potential areas for improvement. This review hence offers several recommendations to improve the reporting and use of DAGs in future research.

Keywords: Directed acyclic graphs; causal diagrams; causal inference; confounding; covariate adjustment; graphical model theory; observational studies; reporting practices.

Publication types

Research Support, Non-U.S. Gov't
Review

MeSH terms

Bias
Causality
Confounding Factors, Epidemiologic
Data Interpretation, Statistical
Humans
Research*

Grants and funding

MR/K501402/1/MRC_/Medical Research Council/United Kingdom