Case Control

Case-control studies are a nonexperimental form of medical research that informs cause-effect relationships. Their main purpose is the identification of risk factors for events of interest. Most famously, case-control studies provided the first evidence of a strong association between cigarette smoking and lung cancer. However, findings from a number of recent case-control studies have been subsequently contradicted or found to overestimate the strength of relationships compared with more robust epidemiological study designs. An example [Page 109]is the case-control finding that hormone replacement therapy (HRT) had a protective effect against coronary heart disease, following which randomized trial evidence identified a small increased risk associated with HRT.

Case-control studies identify cases as patients who already have a disease or condition of interest, and then attempt to identify characteristics of these patients that differ from those who do not have the condition of interest (the controls). For a defined exposure (e.g., walking alongside golf courses) and a defined outcome (e.g., experience of head injury), Table 1, a 2 × 2 table, represents hypothetical findings and informs analysis of an odds ratio.

The odds are expressed as the ratio of the probability that the event of interest occurs to the probability that it does not. In the example, the probability that a golf course walker experiences a head injury is 10/100 or .1, and the probability that he or she does not suffer such an injury is 90/100 or .9. The odds are therefore 10/90, or .11 (the number of events divided by the number of nonevents). The corresponding odds for non—golf course walkers are 5/150, or .03.

The odds ratio is estimated as the odds in the case group divided by the odds in the control group, that is, .11/.03 or 3.67 in the hypothetical golf course example. This is interpreted as golf course walkers being at more than 5 times the odds of suffering a head injury compared with non—golf course walkers.

Traditional case-control studies only inform estimates of the odds ratio between exposure states; they do not enable the estimation of absolute or relative risk because the full size of the population from which the cases (with and without the exposure(s) of interest) are drawn cannot be estimated in a straight case-control study.