In other words, individuals who subsequently developed AD (cases) and those who did not (controls) should be representative of those groups in the clinical population of interest and this is made possible by selecting them randomly after classifying all subjects in the relevant population according to the outcome.