Introduction to causal diagrams for confounder selection

Abstract

In respiratory health research, interest often lies in estimating the effect of an exposure on a health outcome. If randomization of the exposure of interest is not possible, estimating its effect is typically complicated by confounding bias. This can often be dealt with by controlling for the variables causing the confounding, if measured, in the statistical analysis. Common statistical methods used to achieve this include multivariable regression models adjusting for selected confounding variables or stratification on those variables. Therefore, a key question is which measured variables need to be controlled for in order to remove confounding. An approach to confounder-selection based on the use of causal diagrams (often called directed acyclic graphs) is discussed. A causal diagram is a visual representation of the causal relationships believed to exist between the variables of interest, including the exposure, outcome and potential confounding variables. After creating a causal diagram for the research question, an intuitive and easy-to-use set of rules can be applied, based on a foundation of rigorous mathematics, to decide which measured variables must be controlled for in the statistical analysis in order to remove confounding, to the extent that is possible using the available data. This approach is illustrated by constructing a causal diagram for the research question: 'Does personal smoking affect the risk of subsequent asthma?'. Using data taken from the Tasmanian Longitudinal Health Study, the statistical analysis suggested by the causal diagram approach was performed.

Assessing causal effects of early life‐course factors on early childhood caries in 5‐year‐old Ugandan children using directed acyclic graphs (DAGs): A prospective cohort study

“…[18][19][20] A causal model based on DAGs can be designed and used as an aid to check for the sufficiency of confounder adjustment instead of relying entirely on P-values derived from statistical models or stratification. 16 Using statistics to identify and adjust "for confounding" could induce bias unless built on well guided causal approach models. 15 So far, conventional analytic approaches in the ECC literature have included adjustment for several risk factors in multiple variable statistical models without assumptions of causal relationships.…”

“…[15][16][17] Using DAGs has been proposed as a tool for confounder selection. [18][19][20] A causal model based on DAGs can be designed and used as an aid to check for the sufficiency of confounder adjustment instead of relying entirely on P-values derived from statistical models or stratification.…”

A simulation study on matched case-control designs in the perspective of causal diagrams

“…In analytic epidemiology, various strategies could be adopted to remove confounding bias, such as Restriction, Adjustment, Stratification [6, 7], while strategy of matching on confounders C (e.g. matched case-control designs) mainly focuses on improving estimation precision of the effect of E on D, rather than removing confounding bias [8, 9].…”

Mortality effects of timing alternatives for hip fracture surgery

“…We obtained these risks by stratifying RESEARCH Mortality effects of timing alternatives for hip fracture surgery observations on confounders identified from an evidenceinformed causal diagram, 31,32 and weighting observations with the inverse propensity score of surgical timing for their respective strata. 33 We then combined the weighted observations across strata to con struct equalsized samples, each representing the same patient population treated on a certain day (Appendix 1, available at www.…”