Confounding in Health Research Explainer
Confounding occurs when the observed relationship between an exposure and an outcome is distorted by another factor. A confounder is associated with the exposure and independently affects the outcome, while not simply being a step in the causal pathway being studied.
Suppose a study finds that people who carry lighters have more lung cancer. Carrying a lighter is not the likely cause; smoking is linked to carrying a lighter and also increases lung-cancer risk. If smoking is not measured and addressed, the lighter-cancer association can be misleading.
Researchers can reduce confounding through randomization, restriction, matching, stratification, and statistical adjustment. Randomization helps distribute known and unknown factors across groups, while observational studies depend more heavily on measuring likely confounders accurately. No method can correct a factor that was never measured without additional assumptions or data.
Adjustment must be guided by a causal question, not by automatically controlling every available variable. Controlling for a mediator or a variable caused by both exposure and outcome can introduce new bias. Readers should look for the study’s causal model, which factors were considered, how they were measured, and whether sensitivity analyses tested residual confounding.
A confounder creates a competing explanation for an association because it is connected to both the exposure and the outcome.
It can make a true effect look larger, smaller, reversed, or present when there is no causal effect.
Randomization, restriction, and matching address confounding during study design. Stratification and modeling address measured confounders during analysis.
Each approach has tradeoffs, and none guarantees that every relevant difference has been removed.
Residual confounding remains when a confounder is unmeasured or measured too crudely. Statistical adjustment cannot create information that is absent.
Adjusting for the wrong variable can open a biased pathway, which is why subject knowledge and causal diagrams matter.
Explore more "Explainers"
Discover additional explainers across politics, science, business, technology, and other fields. Each explainer breaks down a complex idea into clear, everyday language—helping you better understand how major concepts, systems, and debates shape the world around us.
