Anaes SAQs · Measurement & monitoring physics

Bias and confounding in clinical research

Primary — bias as systematic error distinct from random error/chance; five bias types (selection, performance, detection, attrition, reporting) each with its defence; confounder definition (associated with both exposure and outcome, not on the causal pathway); confounding versus effect modification; design-phase control (randomisation the only method covering unknown confounders, restriction, matching, stratification); analysis-phase control (multivariable regression, propensity-score matching); residual confounding and why RCTs sit atop the hierarchy.