Summary
The episode works through four questions that students often find difficult in cohort studies and randomized trials. It shows how differential loss to follow-up can be hidden by a small overall loss and how best-case and worst-case bounds reveal its effect, and it demonstrates how misallocating person-time creates immortal time bias. It then considers the healthy-worker effect and the choice between external and internal comparison groups, and explains why direct vaccine efficacy appears small where coverage is high. The hosts close by debating whether well-designed cohort studies can answer questions usually left to randomized trials, using the hormone therapy evidence as a case.