A book on the epidemiology of breast cancer (1) was published in Belgrade (Yugoslavia) more than a year ago and has been enthusiastically supported and promulgated by mass media ever since.
An amazing feature of this book is that the author misunderstood a vast majority (if not all) of the basic epidemiological terms, concepts and definitions (incidence, prevalence, mortality, relative risk, standardization, confounding, outbreak, epidemiology, sample, causality, hypothesis testing, analytical and experimental studies, etc.). The same holds true for many common facts on cancer epidemiology in general. None of these errors would call for major concern, except as anecdotal evidence on a discrepancy between one's ambitions and abilities to write a monograph. However, a major part of the book concerns a large case-control study, the results of which have been taken as an ultimate truth and aggressively conveyed to the general public in order to influence behaviour. At this point, the book indeed became dangerous, imposing the attempts to neutralize its messages as an ethical issue.
Neither the methods nor any other part of the study was presented in a consistent and understandable way. One of the few apparent facts seems to be that the author interviewed 2000 breast cancer cases, and that 1615 interviews 'were valid for data processing'. Five hundred and eighty-six cases were somehow selected and each of the cases was asked to find a control of the same age, occupation and place of residence. Though being matching criteria, these variables were