Identifying confounders is a challenge that statisticians encounter all the time. Confounding determines whether or not a causal association exists between an exposure and an outcome. A (rather silly) example is the notion that carrying matchboxes causes lung cancer. The factor – confounder – here is the smoking status. Smokers are likely to carry matchboxes; smokers have a higher chance of getting lung cancer. If this confounder is not identified, one may conclude that having matchboxes is the exposure that caused the outcome of lung cancer.
As per Jager et al., a confounding variable must satisfy three criteria: 1) it must have an association with the exposure of interest, (2) it must be associated with the outcome of interest, and (3) it must not be an outcome of the exposure.
