Bias remaining after adjustment because confounders were mismeasured, unmeasured, or imperfectly controlled.
Residual confounding is the distortion in an observed exposure-outcome association that persists even after statistical adjustment for known confounders, typically because those confounders were measured imprecisely, categorised too coarsely, or because relevant confounders were never measured at all. It means an association (or lack of one) in an adjusted analysis may still not reflect the true causal effect. Because it can rarely be fully excluded in observational research, clinicians should treat adjusted findings as suggestive rather than definitive, particularly when plausible unmeasured confounders exist.
This guide was auto-drafted and is pending editorial review.