com.atlassian.confluence.content.render.xhtml.migration.exceptions.UnknownMacroMigrationException: The macro 'datalayer.push(arguments);' is unknown.

7.3.6.3.1 Confounding and confounders

Owned by Carolyn Phillips
Mar 26, 2024
2 min read

 

Confounding occurs when another factor other than primary factor of interest or being investigated, can directly influence the outcome being measured. To be classed as a confounding factor, it should not be a factor that appears in the casual pathway between and exposure and the outcome. Confounding bias is defined as “bias of the estimated effect of an exposure on an outcome due to the presence of common causes of the exposure and the outcome” (Miquel 2014) (p.55).  A confounder or confounding variable is a variable that can be used to decrease confounding bias when properly adjusted for (Miquel 2014) (p.55). 

Criteria for confounders are (Rothman, Greenland et al. 2008) (p.132-134):

  1. A confounding factor must be an extraneous risk factor for the disease; i.e. the confounder is a risk factor for the disease and the factor's association with disease arises from a causal pathway other than the one under study.

  2. A confounding factor must be associated with the exposure under study in the source population (the population at risk from which the cases are derived).

  3. A confounding factor must not be affected by the exposure or the disease. In particular, it cannot be an intermediate step in the causal path between the exposure and the disease. For example, in the case of increased risk of lung cancer from high levels of red meat consumption, the confounding factor could possibly be the ‘cooking method’ (Cancer Australia 2014).

Confounding can be controlled in the design and analysis phases in the case of observational studies. The two approaches used for the control of confounding in the analysis of data are stratification and statistical modelling. In stratification, study participants are split into strata that are different groups based on levels of the potential confounding variable, for example age. Although this approach is a simple method, this approach is limited by the fact that only a certain a number of potential factors could be stratified. Hence, it is not a common approach to control for confounding in observational studies in the analysis phase (Kahlert, Gribsholt et al. 2017). Statistical modelling (such as multiple logistic regression, conditional logistic regression, Cox proportional hazards regression, multivariable regression analysis) is used to estimate the strength of the relationship of interest while controlling for all of the potential confounders (Webb and Bain 2011).