Theory 99/Selection & Bias/Hospital case-control
Hospital Controls Are Not Population
The claim
Hospital-based case-control studies produce spurious associations because both cases and controls are selected on admission, which depends on both exposure and disease.
01Mechanism
Admission is a collider; conditioning on it creates associations between exposure and the conditions that also cause admission.
02Prediction
Associations should differ from population-based controls.
03Refutation
Identical associations would remove the selection effect.