Michel A. Ibrahim

The case control study: The problem and the prospect☆

Publisher Summary This chapter discusses the problem and the prospect of the case control study. The case-control study can be defined as that method of epidemiological investigation in which the frequency of an attribute or exposure to an environmental factor in cases is compared to that in non-cases or controls. The difference in the two frequencies is quantified statistically to rule out chance occurrence and to estimate the magnitude of the effect of the attribute or exposure on the disease. The case-control method of study differs from the former or the experimental method in that the investigator begins the cohort with the effect and then looks backward in time for the exposure. In the cohort and experimental modes, the investigator begins with the exposure and looks forward in time for the effect. The case-control and cohort methods depend upon observing. This is in contrast to the experiment in which the investigator is able to randomize the subjects to treatment or exposure and to manipulate one or more factors and then measure the consequences.

Psychological patterns and coronary heart disease: An appraisal of the determination of etiology by means of a stochastic process☆

Abstract The question of whether a selected psychological pattern precedes or follows the onset of coronary heart disease has served as a vehicle for a discussion of the fundamental issue, namely, the ability to draw causal inferences from observed associations in cross-sectional studies. The possibility of an independent contribution of the psychological pattern to the development of coronary disease, hence a possible etiological role, has been clarified and its implications have been illustrated The crucial distinctions between “predictive” and “etiologic” stochastic models have been reviewed. When such models are used to predict phenomena, the validity of their underlying assumptions is not mandatory. However, when stochastic models are used to explore etiologic relationships, their value depends on the validity of the underlying assumptions. In addition, the users of formal mathematical models must exercise caution to avoid the pitfall of excluding relevant knowledge available from epidemiologic observation. To do so may prevent selection of the most likely inferences to explain particular causal mechanisms.

Screening of the coronary prone by study of offspring.

All the authors are or were, at the time of the study, with the School of Medicine, State University of New York at Bufjalo. Dr. Ibrahim is Deputy Health Commissioner of Erie County, N.Y., and associate professor of epidemiology in the schools department of social and preventive medicine. Dr. Borden is now a resident in radiology, Boston City Hospital, and Dr. Pinsky is in private practice in Philadelphia, Pa. Dr. Kohn is clinical associate professor of medicine, in the school of medicine. Mr. Feldman is research associate in biostatistics in the schools department of social and preventive medicine. Dr. Winkelstein is professor of epidemiology, University of California School of Public Health, Berkeley. The study described in this paper was partly supported by New York State Health Department grant 67267 and grants from the Heart Association of Erie County. Tearsheet requests to Michel A. Ibrahim, M.D., 2211 Main Street, Bufialo, N.Y. 14214. THE INCREASED likelihood of coronary heart disease developing in persons with elevated levels of serum cholesterol, blood pressure, body weight, and certain electrocardiographic abnormalities has been repeatedly shown in retrospective and prospective studies (1). Other studies have revealed significant associations between fathers and their children for cholesterol and blood pressure levels and ponderal index values (2-4). These observations raise the question of whether it is possible to identify or screen coronary-prone middle-aged men by studying their children. To evalute this procedure, the so-called coronary risk factors should be measured on both fathers and children. The frequency of predicting the presence of coronary heart disease factors among fathers by examining their childrens values for these characteristics could then be quantified. Furthermore, if such a procedure were to be practical, a captive population of children whose fathers are at risk should be selected. Eleventh grade students seem to be ideal for this purpose as they are a captive, relatively stable population, and their fathers are mostly 35 to 55 years old. The feasibility of such a procedure was demonstrated in a pilot project involving 100 studentfather pairs (2). Briefly, the pilot study revealed significantly positive associations between the students and their fathers for serum cholesterol and blood pressure levels and ponderal index values. In addition, predominantly high levels of all variables combined (blood lipids, blood pressure, glucose, uric acid, skinfold thickness, and ponderal index) of the students predicted all the fathers with ST-T abnormalities. Also, three-fourths of the fathers with a history of coronary heart disease or with elevated levels of relative body mass and two-thirds of the