Confounding

Abstract

In confounding, the effect of the exposure of interest is mixed with the effect of another variable. It is important to identify relevant confounders and remove the confounding effect as much as possible. There are three criteria that need to be fulfilled to determine whether a variable could be considered a potential confounder. The first criterion is that the variable needs to be associated with the exposure. The second criterion is that the variable needs to be associated with the outcome or disease. The third criterion is that the variable should not be an intermediate variable in the causal pathway between exposure and outcome. Only if all the criteria are fulfilled is the variable under question a confounder. If one incorrectly adjusts for a variable that is not a confounder, one risks overadjustment or adjustment for spurious associations. Confounders can be prevented from entering the study, during the design of a study, or if this is not possible, one can try to remove it during the analysis phase.