[Clinical research XI. From the clinical judgment to the case and controls design]

The case-control design like the historical cohort carries a number of potential biases as a consequence of the reconstruction of events once the outcome has occurred, and as a consequence of the bias generated by the selection of the control group. This design is characterized by a number of cases (cases), for which we identify a comparison group (controls). It begins at the outcome in direction to the probable cause; therefore, it requires reconstructing events in the opposite direction as it occurs in the phenomenon of causality. However, we must always keep in mind the architectural design, and consider in each section--baseline, maneuver and outcome--characteristics that allow us to demonstrate the effect of the maneuver, avoiding improper assembly, susceptibility, performance and detection bias. The transfer bias can only be controlled with the provision of a defined population, whether it is a population based case-control study or a case-control study nested in a cohort. When a defined population is not possible, this design is only recommended in rare diseases.