CASE-CONTROL STUDY

The case-control study, a widely used method of observational epidemiological study, is an application of medical history-taking that aims to identify the cause of disease among a group of people, or the cause-effect relationships of a condition of interest. The underlying concept is simple. The past medical history, or history of exposure to a suspected risk or protective factor, of a group of persons with the disease or condition of interest (the cases) is compared with the past history of another group of persons (the controls) who resemble them in as many relevant respects as possible, but who do not have the disease or condition of interest. Statistical analysis is used to determine whether there is a stronger association of past exposure to the suspected risk or protective factor with the condition of interest among the cases than among the controls. The method can be called a retrospective study because it is concerned with events in the past. However, the cases are often collected prospectively, with cases added as they occur, so there is possible confusion with what used to be called a prospective study but is now almost always called a cohort study. It has also been called case-compeer study and case-referent study, but case-control study is the most widely used term.

The method evolved out of analyses of series of cases. The concept was mentioned in the writings of the nineteenth-century French physician Pierre Charles Alexandre Louis and a simple form of it was used by the nineteenth-century English physician William Augustus Guy. In the 1930s, the English physician Janet Lane-Claypon used this method to study risk factors for breast cancer, and in 1939, just as war was breaking out in Europe, F. H. Muller, a German physician, used a case-control study design to demonstrate that a past history of cigarette smoking was strongly associated with lung cancer. Following World War II, several investigators in England and in the United States adopted Muller's methods for case-control studies of smoking and lung cancer, which had become a very common and lethal form of cancer. In 1950, Doll and Hill in the England and Wynder and Graham in the United States published large case-control studies of cigarette smoking and cancer of the lung almost simultaneously in the British Medical Journal and the Journal of the American Medical Association, respectively. Many more case-control studies of this and other kinds of cancer soon established the utility of the method.

Case-control studies have proved particularly useful in studying very rare conditions. During 1969 and 1970, eight case of adenocarcinoma of the vagina were seen in adolescent girls and young women in Boston, Massachusetts. This was, up till then, an extremely rare, almost nonexistent condition, and it was clear that these young women must have been exposed to some unusual cancer-causing agent. Each of the eight cases was matched with four otherwise similar but healthy females of the same age. Their, and their mothers', past histories of many kinds of exposure to medications, vaginal douches, and other substances, were compared. Seven of the eight cases had a history of their mothers having been given artificial estrogen to prevent miscarriage early in pregnancy (this had been a popular though unproven method of preventing threatened miscarriage since the 1950s; it has now been shown to be useless). None of the controls had a similar history. There was less than a 1 in 100,000 likelihood of this distribution occurring by chance. Adenocarcinoma of the vagina was caused by prenatal exposure of the developing female fetus to diethylstilbestrol, an artificial estrogen. Later studies showed that genital dysplasia in boys and young men was another consequence of prenatal exposure to artificial estrogen.

These examples, and many others, illustrate the value of the case-control study. It is a relatively cheap, rapid, and reliable method of establishing evidence of an association between an exposure to a risk (or protective) factor and an unfavorable (or favorable) outcome. It does not require study of large numbers. The concept is readily understandable, so members of the lay public, political decision makers, and the media can easily grasp the significance of the findings.

There are, however, some important shortcomings. The results can be biased in many ways— by flawed information about past exposure to risk, inappropriate selection of controls, and various confounding factors. The validity of results based on the use of controls who may have been exposed to similar or different combinations of risk, biases introduced by selective recall or recording of relevant past exposure to risk, and the most suitable way to analyze the results have generated endless debates in epidemiological journals.

The advantages of the case-control method are: (1) it is an excellent way to study rare diseases and diseases with long latency, (2) a relatively quick answer can be obtained, (3) it is relatively cheap, (4) it usually requires only a few cases, (5) it can often make use of existing records, and (6) it can study several possible causes or exposures to risk simultaneously.

The disadvantages of the method are: (1) it relies on subjects' recall and/or completeness of existing records, (2) it may be difficult or impossible to validate this information, (3) there is incomplete allowance for extraneous factors, (4) the selection of a suitable comparison (control) group may be difficult, (5) rates cannot be calculated, (6) the mechanism of disease cannot be studied, and(7) a proof of causation cannot be established.

Analysis of the results of a case-control study makes use of a simple approximation, the odds ratio. This ratio is based on the assumption that the condition under study is relatively rare. If it is, the total number exposed to risk is close to the number of healthy persons exposed, and the total number not exposed to risk is close to the number of healthy nonexposed persons.

No other epidemiologic method has been so much discussed. Several books and many learned articles have been written about it. Many of the shortcomings can be overcome by ingenious designs such as the use of a "nested" case-control study in which both cases and controls are drawn from the same large population that is being used in a cohort study. Case-control studies have undoubtedly been overused, and many spurious associations have been reported. Nonetheless, the value of the method in rapid and inexpensive assessment of a new or serious health problem has been proved beyond doubt.

John M. Last

(see also: Cohort Study; Cross-Sectional Study; Epidemiology )

Bibliography

Breslow, N. E., and Day, N. E. (1980). Statistical Methods in Cancer Research, Vol. 1: The Analysis of Case-Control Studies. Lyon: International Agency for Research on Cancer.

Rothman, K. J., and Greenland, S. (1998). "Case-Control Studies." In Modern Epidemiology, 2nd edition, eds. K. J. Rothman and S. Greenland. Philadelphia, PA: Lippincott-Raven.

Schlesselman, J. J. (1982). Case-Control Studies—Design, Conduct, Analysis. New York: Oxford University Press.