What is the difference between effect modification (interaction) and confounding?

• A. Effect modification is a differential effect of exposure across strata; confounding distorts the overall association; effect modification is a real difference, not bias.
• B. They are the same concept.
• C. Confounding is a real effect; effect modification is a bias.
• D. Neither affects study results.

Answer: (A)

Effect modification happens when the impact of an exposure on an outcome changes across levels of another variable. In other words, the exposure’s effect is not the same in every subgroup, so you see a different magnitude (or even direction) of effect in different strata. This is a real difference in the effect that matters for interpretation, not a bias.

Confounding, on the other hand, arises when a third variable is related to both the exposure and the outcome and creates a distorted, overall association. The observed link isn’t the true causal effect of the exposure alone; it’s biased by that third variable.

The best choice captures both ideas: effect modification is a differential effect across strata; confounding distorts the overall association; and effect modification represents a real difference, not bias. The other options mix up the concepts—for example, claiming they’re the same concept, or that confounding is a real effect and effect modification is bias, or that neither affects results. In practice, you’d look for different effects across subgroups to identify interaction, while adjusting for confounders to estimate the true association.