Summary cards for 3 teaching tips on understanding, interpreting and quantifying confounding variables in studies of therapy or harm

摘要

Scenario: Consider a randomized trial, in contrast to an observational study, on an issue of therapy or harm. Pose the issue: "Why do we randomize?" Lead the learners to understand the answer in successively more precise terms.Randomization helps ensure that groups have similar characteristics.Contrast characteristics such as eye color and shoe size to age and disease severity.Lead the learners to understand the relevant characteristics to be those that affect outcome.Lead the learners to distinguish between any outcomes and outcomes of interest to the question being studied.Label such characteristics "prognostic variables" and clarify that they are appropriately called "confounding variables" when they are unequally distributed between study groups.Lead the learners to understand that randomization minimizes the likelihood of both known and unknown confounding variables, whereas observational studies are only able to control for known confounding variables.