scholarly journals Use of directed acyclic graphs (DAGs) in applied health research: review and recommendations

2019 ◽  
Author(s):  
Peter WG Tennant ◽  
Wendy J Harrison ◽  
Eleanor J Murray ◽  
Kellyn F Arnold ◽  
Laurie Berrie ◽  
...  
Keyword(s):  
Health Research ◽  
Web Of Science ◽  
Directed Acyclic Graphs ◽  
Causal Effects ◽  
Research Review ◽  
Future Research ◽  
Acyclic Graphs ◽  
Confounding Variables ◽  
Applied Health Research ◽  
Consistent Direction

ABSTRACTBackgroundDirected acyclic graphs (DAGs) are an increasingly popular approach for identifying confounding variables that require adjustment 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.MethodsOriginal 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.ResultsA 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. Diagrams varied in size but averaged 12 nodes (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 a 34% (n=49) were arranged so the constituent arcs flowed in a consistent direction.ConclusionsThere is substantial variation in the use and reporting of DAGs in applied health research. Although this partly reflects their flexibility, it also highlight some potential areas for improvement. This review hence offers several recommendations to improve the reporting and use of DAGs in future research.

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

2020 ◽  
Author(s):  
Peter W G Tennant ◽  
Eleanor J Murray ◽  
Kellyn F Arnold ◽  
Laurie Berrie ◽  
Matthew P Fox ◽  
...  
Keyword(s):  
Health Research ◽  
Web Of Science ◽  
Directed Acyclic Graphs ◽  
Causal Effects ◽  
Research Review ◽  
Future Research ◽  
Substantial Variation ◽  
Acyclic Graphs ◽  
Confounding Variables ◽  
Applied Health Research

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.

scholarly journals Adjustment for time-invariant and time-varying confounders in ‘unexplained residuals’ models for longitudinal data within a causal framework and associated challenges

2018 ◽  
Vol 28 (5) ◽  
pp. 1347-1364 ◽  
Author(s):  
KF Arnold ◽  
GTH Ellison ◽  
SC Gadd ◽  
J Textor ◽  
PWG Tennant ◽  
...  
Keyword(s):  
Directed Acyclic Graphs ◽  
Regression Methods ◽  
Modelling Framework ◽  
Acyclic Graphs ◽  
Confounding Variables ◽  
Confounder Adjustment ◽  
Causal Framework ◽  
Distal Outcome ◽  
Time Invariant ◽  
Insight Into

‘Unexplained residuals’ models have been used within lifecourse epidemiology to model an exposure measured longitudinally at several time points in relation to a distal outcome. It has been claimed that these models have several advantages, including: the ability to estimate multiple total causal effects in a single model, and additional insight into the effect on the outcome of greater-than-expected increases in the exposure compared to traditional regression methods. We evaluate these properties and prove mathematically how adjustment for confounding variables must be made within this modelling framework. Importantly, we explicitly place unexplained residual models in a causal framework using directed acyclic graphs. This allows for theoretical justification of appropriate confounder adjustment and provides a framework for extending our results to more complex scenarios than those examined in this paper. We also discuss several interpretational issues relating to unexplained residual models within a causal framework. We argue that unexplained residual models offer no additional insights compared to traditional regression methods, and, in fact, are more challenging to implement; moreover, they artificially reduce estimated standard errors. Consequently, we conclude that unexplained residual models, if used, must be implemented with great care.

Directed Acyclic Graphs: Alternative tool for causal inference in epidemiology and biostatistics research and teaching

MedPharmRes ◽  
2018 ◽  
Vol 2 (3) ◽  
pp. 12-16 ◽  
Author(s):  
Dang Tran ◽  
Long Khuong ◽  
Tram Huynh ◽  
Hong Le ◽  
Tuan Vo
Keyword(s):  
Traditional Approach ◽  
Univariate Analysis ◽  
Experimental Studies ◽  
Directed Acyclic Graphs ◽  
Causal Effects ◽  
P Value ◽  
Causal Inferences ◽  
Disease Patterns ◽  
Research And Teaching ◽  
Acyclic Graphs

The issue of causation is one of the major challenges for epidemiologists who aim to understand the association between an exposure and an outcome to explain disease patterns and potentially provide a basis for intervention. Suitably designed experimental studies can offer robust evidence of the causal relationships. The experimental studies, however, are not popular, difficult or even unethical and impossible to conduct; it would be desirable if there is a methodology for reducing bias or strengthening the causal inferences drawn from observational studies. The traditional approach of estimating causal effects in such studies is to adjust for a set of variables judged to be confounders by including them in a multiple regression. However, which variables should be adjusted for as confounders in a regression model has long been a controversial issue in epidemiology. From my observation, the adjustments using only "statistical artifacts" methods such as the p-value<0.2 in univariate analysis, stepwise (forward/backward) are widely used in research and teaching in Epidemiology and Statistics but without appropriated notice on the biological or clinical relationships between exposure and outcome which may induce the bias in estimating causal effects. In this mini-review, we introduce an interesting method, namely Directed Acyclic Graphs (DAGs), which can be used to reduce the bias in estimating causal effects; it is also a good application for Epidemiology and Biostatistics teaching.

scholarly journals Qualitative approximations to causality: Non-randomizable factors in clinical psychology

2021 ◽  
Vol 3 (2) ◽  
Author(s):  
Michael Höfler ◽  
Sebastian Trautmann ◽  
Philipp Kanske
Keyword(s):  
Clinical Psychology ◽  
Quantitative Methods ◽  
Quantitative Estimation ◽  
False Negative ◽  
Directed Acyclic Graphs ◽  
Causal Effects ◽  
Qualitative Approaches ◽  
Acyclic Graphs ◽  
Negative Conclusion ◽  
Causal Conclusion

Background Causal quests in non-randomized studies are unavoidable just because research questions are beyond doubt causal (e.g., aetiology). Large progress during the last decades has enriched the methodical toolbox. Aims Summary papers mainly focus on quantitative and highly formal methods. With examples from clinical psychology, we show how qualitative approaches can inform on the necessity and feasibility of quantitative analysis and may yet sometimes approximate causal answers. Results Qualitative use is hidden in some quantitative methods. For instance, it may yet suffice to know the direction of bias for a tentative causal conclusion. Counterfactuals clarify what causal effects of changeable factors are, unravel what is required for a causal answer, but do not cover immutable causes like gender. Directed acyclic graphs (DAGs) address causal effects in a broader sense, may give rise to quantitative estimation or indicate that this is premature. Conclusion No method is generally sufficient or necessary. Any causal analysis must ground on qualification and should balance the harms of a false positive and a false negative conclusion in a specific context.

scholarly journals Maternal smoking during pregnancy and early development of overweight and growth deficit in children: an analysis of survival

2018 ◽  
Vol 18 (2) ◽  
pp. 361-369
Author(s):  
Poliana Cristina de Almeida Fonseca ◽  
Carolina Abreu de Carvalho ◽  
Vitória Abreu de Carvalho ◽  
Andréia Queiroz Ribeiro ◽  
Silvia Eloiza Priore ◽  
...  
Keyword(s):  
Maternal Smoking ◽  
Cox Regression ◽  
Directed Acyclic Graphs ◽  
Z Score ◽  
Smoking During Pregnancy ◽  
Acyclic Graphs ◽  
Confounding Variables ◽  
Maternal Smoking During Pregnancy ◽  
Pregnancy Smoking ◽  
Growth Deficit

Abstract Objectives: to evaluate the association between smoking during pregnancy and nutritional status. Methods: cohort study with a sample of 460 children in the baseline. The children were assessed four times, being measured for weight and length to be converted in indexes length forage (L/A) and body mass index forage (BMI/A) in Z-score. The time until occurrence of growth deficit and overweight was calculated in days and compared to maternal smoking during pregnancy. To assess the association between smoking during pregnancy and the outcomes, a Hazard Ratio by Cox regression was obtained, adjusting by confounding variables selected from Directed Acyclic Graphs (DAG). Results: the time until occurrence of growth deficit and overweight was lower in children whose mothers smoked during pregnancy. Smoking during pregnancy was a risk factor for length deficit (HR = 2.84; CI95% = 1.42 to 5.70) and for overweight (HR = 1.96; CI95% = 1, 09 to 3.53), even after the adjustment. Conclusions: maternal smoking was a changeable factor associated with anthropometric outcomes, which demonstrates the need for actions to combat smoking during pregnancy in order to prevent early nutritional deviations.

scholarly journals Structural learning and estimation of joint causal effects among network-dependent variables

2021 ◽  
Author(s):  
Federico Castelletti ◽  
Alessandro Mascaro
Keyword(s):  
Network Structure ◽  
Causal Effect ◽  
Directed Acyclic Graphs ◽  
Causal Effects ◽  
Structural Learning ◽  
Acyclic Graphs ◽  
Dependent Variables ◽  
Underlying Network ◽  
Dag Model ◽  
Joint Posterior Distribution

AbstractBayesian networks in the form of Directed Acyclic Graphs (DAGs) represent an effective tool for modeling and inferring dependence relations among variables, a process known as structural learning. In addition, when equipped with the notion of intervention, a causal DAG model can be adopted to quantify the causal effect on a response due to a hypothetical intervention on some variable. Observational data cannot distinguish between DAGs encoding the same set of conditional independencies (Markov equivalent DAGs), which however can be different from a causal perspective. In addition, because causal effects depend on the underlying network structure, uncertainty around the DAG generating model crucially affects the causal estimation results. We propose a Bayesian methodology which combines structural learning of Gaussian DAG models and inference of causal effects as arising from simultaneous interventions on any given set of variables in the system. Our approach fully accounts for the uncertainty around both the network structure and causal relationships through a joint posterior distribution over DAGs, DAG parameters and then causal effects.

scholarly journals A formative evaluation of Collaboration for Leadership in Applied Health Research and Care (CLAHRC): institutional entrepreneurship for service innovation

2014 ◽  
Vol 2 (31) ◽  
pp. 1-124 ◽  
Author(s):  
Andy Lockett ◽  
Nellie El Enany ◽  
Graeme Currie ◽  
Eivor Oborn ◽  
Michael Barrett ◽  
...  
Keyword(s):  
Clinical Practice ◽  
Health Research ◽  
Institutional Context ◽  
Radical Change ◽  
Future Research ◽  
Institutional Entrepreneurship ◽  
Institutional Practices ◽  
Policy Makers ◽  
Ego Networks ◽  
Applied Health Research

BackgroundCollaborations for Leadership in Applied Health Research and Care (CLAHRCs) are a time-limited funded initiative to form new service and research collaboratives in the English health system. Their aim is to bring together NHS organisations and universities to accelerate the translation of evidence-based innovation into clinical practice. In doing so, CLAHRCs are positioned to help close the second translation gap (T2), which is described as the problem of introducing and implementing new research and products into clinical practice.ObjectivesIn this study, we draw on ideas from institutional theory and institutional entrepreneurship to examine how actors may engage in reshaping existing institutional practices in order to support, and help sustain efforts to close the T2. Our objective was to understand how the institutional context shapes actors’ attempts to close the T2 by focusing on the CLAHRC initiative.MethodsThe study employed a longitudinal mixed-methods approach. Qualitative case studies combined interview data (174 in total across all nine CLAHRCs and the four in-depth sites), archival data and field notes from observations, over a 4-year period (2009–13). Staff central to the initiatives were interviewed, including CLAHRC senior managers; theme leads; and other higher education institution and NHS staff involved in CLAHRCs. Quantitative social network analysis (SNA) employed a web-based sociometric approach to capture actors’ own individual (i.e. ego) networks of interaction across two points in time (2011 and 2013) in the four in-depth sites, and their personal characteristics and roles.ResultsWe developed a process-based model of institutional entrepreneurship that encompassed the different types of work undertaken. First, ‘envisaging’ was the work undertaken by actors in developing an ‘embryonic’ vision of change, based on the interplay between themselves and the context in which they were situated. Second, ‘engaging’ was the work through which actors signed up key stakeholders to the CLAHRC. Third, ‘embedding’ was the work through which actors sought to reshape existing institutional practices so that they were more aligned with the ideals of CLAHRC. ‘Reflecting’ involved actors reconsidering their initial decisions, and learning from the process of establishing CLAHRCs. Furthermore, we employed the qualitative data to develop five different archetype models for organising knowledge translation, and considered under what founding conditions they are more or less likely to emerge. The quantitative SNA results suggested that actors’ networks changed over time, but that important institutional influences continued to constrain patterns of interactions of actors across different groups.ConclusionThe development of CLAHRCs holds important lessons for policy-makers. Policy-makers need to consider whether or not they set out a defined template for such translational initiatives, since the existence of institutional antecedents and the social position of actors acted to ‘lock in’ many CLAHRCs. Although antecedent conditions and the presence of pre-existing organisational relationships are important for the mobilisation of CLAHRCs, these same conditions may constrain radical change, innovation and the translation of research into practice. Future research needs to take account of the effects of institutional context, which helps explain why many initiatives may not fully achieve their desired aims.FundingThe National Institute for Health Research Health Services and Delivery Research programme.

scholarly journals Heat-Related Illness Is Associated with Lack of Air Conditioning and Pre-Existing Health Problems in Detroit, Michigan, USA: A Community-Based Participatory Co-Analysis of Survey Data

2020 ◽  
Vol 17 (16) ◽  
pp. 5704
Author(s):  
Jacqueline E. Cardoza ◽  
Carina J. Gronlund ◽  
Justin Schott ◽  
Todd Ziegler ◽  
Brian Stone ◽  
...  
Keyword(s):  
Air Conditioning ◽  
Directed Acyclic Graphs ◽  
Academic Community ◽  
Extreme Heat ◽  
Future Research ◽  
Acyclic Graphs ◽  
Excellent Health ◽  
Central Air Conditioning ◽  
Future Research Directions ◽  
Fair Poor

The objective of the study was to investigate, using academic-community epidemiologic co-analysis, the odds of reported heat-related illness for people with (1) central air conditioning (AC) or window unit AC versus no AC, and (2) fair/poor vs. good/excellent reported health. From 2016 to 2017, 101 Detroit residents were surveyed once regarding extreme heat, housing and neighborhood features, and heat-related illness in the prior 5 years. Academic partners selected initial confounders and, after instruction on directed acyclic graphs, community partners proposed alternate directed acyclic graphs with additional confounders. Heat-related illness was regressed on AC type or health and co-selected confounders. The study found that heat-related illness was associated with no-AC (n = 96, odds ratio (OR) = 4.66, 95% confidence interval (CI) = 1.22, 17.72); living ≤5 years in present home (n = 57, OR = 10.39, 95% CI = 1.13, 95.88); and fair/poor vs. good/excellent health (n = 97, OR = 3.15, 95% CI = 1.33, 7.48). Co-analysis suggested multiple built-environment confounders. We conclude that Detroit residents with poorer health and no AC are at greater risk during extreme heat. Academic-community co-analysis using directed acyclic graphs enhances research on community-specific social and health vulnerabilities by identifying key confounders and future research directions for rigorous and impactful research.

scholarly journals Oropharyngeal Dysphagia: A Proposal for an Ecological Theoretical Model

2022 ◽  
pp. 7-16
Author(s):  
Rafaela Soares Rech ◽  
Bárbara Niegia Garcia de Goulart
Keyword(s):  
Data Analysis ◽  
Theoretical Model ◽  
Oropharyngeal Dysphagia ◽  
Theoretical Models ◽  
Epidemiological Studies ◽  
Directed Acyclic Graphs ◽  
Acyclic Graphs ◽  
Confounding Variables ◽  
Direct Causation ◽  
Definition Of

Background: The exponential growth in epidemiological studies has been reflected in an increase in analytical studies. Thus, theoretical models are required to guide the definition of data analysis, although so far, they are seldom used in Speech, Language, and Hearing Sciences. Objective: To propose a multicausal model for oropharyngeal dysphagia using directed acyclic graphs showing mediating variables, confounding variables, and variables connected by direct causation. Design: This integrative literature review. Setting: This was carried out until January 4, 2021, and searches were performed with the MEDLINE, EMBASE,and other bases.

Sign in / Sign up

Export Citation Format

Share Document