What is overmatching and how can it harm a cohort study?

• A. Overmatching reduces bias.
• B. Overmatching increases power.
• C. Overmatching occurs when matching on variables that are intermediates or associated with exposure/outcome, reducing ability to detect true associations and potentially biasing results.
• D. Overmatching is the same as randomization.

Answer: (C)

Overmatching happens when you choose matching factors that aren’t true confounders but are intermediates on the causal path between exposure and outcome or are closely tied to both. In a cohort study, matching on such variables can make the exposed and unexposed groups too similar with respect to parts of the causal chain, effectively blocking or dampening the effect of the exposure you want to study. That loss of variation reduces your ability to detect real associations (lower statistical power) and can bias the results because you’re conditioning on a variable that lies on the pathway from exposure to outcome or is influenced by exposure. In contrast, matching should target true confounders that are not affected by the exposure; matching on intermediates or exposure-related variables undermines the very goal of measuring how exposure affects outcome. Overmatching is not the same as randomization, which balances groups by design without conditioning on post-randomization variables.