Case-control studies in pharmacoeconomic research: an overview

The case-control approach has been a mainstay of epidemiological studies, particularly those investigating aetiology. Many articles addressing pharmacoeconomic topics have reported studies purported to be of the case-control type. However, on examination, these were actually standard cohort studies that were misnamed because they compared cases of a particular illness to 'controls' without the illness. The actual case-control design involves a series of cases with the outcome of interest. In pharmacoeconomic applications, the outcome of interest would typically be high cost, or hospitalization, or return to full quality of life. The illness does not define cases, but rather is actually the 'exposure'. The 'controls' must be a sample of the study base, not subjects without the illness. In this article, we review the features of a proper case-control study and contrast them with those of the more common cohort study. Confusing the control series of a cohort study with the 'controls' in a case-control study leads to serious problems with understanding the research, its strengths and drawbacks (e.g. confounding concerns), and interpretation of the findings. Although the case-control design has so far been used little to address pharmacoeconomic questions, it can be very efficient in certain situations, particularly when obtaining data on all subjects is burdensome or when conditions provide a ready case series but not the rest of the subjects.