To identify a confounder, the variable must meet three criteria:
It must be associated with the exposure. It must be associated with the outcome. It should not be an intermediate step in the causal pathway between exposure and outcome.
Using statistical methods like stratification or multivariable analyses can help identify potential confounders.