Epidemiology is the study of disease patterns in human populations. Epidemiologic studies are purely statistical in nature and, by themselves, cannot prove cause-and-effect relationships. Biological plausibility, for example, is an essential element for proving causation.
There are four types of epidemiologic studies common to junk science. In order of least-to-most potentially credible, the four types are:
Because of inherent limitations, epidemiologic studies are typically only useful in the case of high risks of rare disease, as is typical of food poisoning occurrences. Epidemiologic studies typically are not suited for the study of low risks of common diseases.
- Gori G, Presentation: Epidemiology and public health: is a new paradigm needed or a new ethic?, J Clin Epidemiol. 51(8):637-41, August 1998.
- Abstract. "Public health militancy has been increasingly frustrated by what many perceive as the marginally fertile studies of risk factors operating at the individual level, whose causal underpinnings are often and inevitably weakened in multifactorial situations. As a remedy, leading advocates propose a refocusing of epidemiology and public health on socioeconomic, cultural, and political studies, and on broad interventions at population level. This new "paradigm" would be aided by a relaxation of evidentiary standards of causality, away from scientific criteria and more toward dialectic (rhetorical) precepts derived in a humanistic and sociologic tradition. It is countered here that such proposals would further reduce the objectivity and thus likely weaken rather than strengthen epidemiology and the justification of public health action. Instead, a realistic appraisal finds that multifactorial epidemiology raises warning signals of varying influence, and that the usefulness of epidemiology and public health could be enhanced by conceiving of methods to score the relative strength and priority of such warnings."
- Milloy S, Relax...You Might Not Be Doomed, Public Risk, February 1997.
- Wynder E and Stellman S, The "over-exposed" control group, Am J Epidemiol. 135(5):459-61, March 1992.
- Abstract. "Relative risk is determined as much by the level of exposure among controls as among cases. If cases and controls are drawn from a population in which the range of exposures is narrow, then a study may yield little information about potential health effects. This may be one reason why an association between dietary fat and cancer has not been consistently observed in Western populations. Since the fat intake as a percent of total calories in the US general population varies little, only very large relative risks can be detected in epidemiologic studies. Investigators of the dietary fat hypothesis need to select study groups from populations where the risk factor is not, essentially, narrowly distributed."
- Gori G, Epidemiology and the concept of causation in multifactorial diseases, Regul Toxicol Pharmacol. Jun;9(3):263-72. June 1989.
- Abstract. "Unlike infectious diseases of the past, diseases prevalent in modern industrialized societies have multifactorial origins whose complexity so far has defied an integrated scientific understanding. Their epidemiologic investigation suffers from the conceptual inability of formulating plausible causal hypotheses that mimic a complex reality, and from the practical difficulties of running elaborate studies controlled for multifactorial confounders. Until biomedical research provides a satisfactory understanding of the complex mechanistic determinants of such diseases, epidemiology can only field reductionist causal hypotheses, leading to results of uncertain significance. Consensual but rationally weak criteria devised to extract inferences of causality from such results confirm the generic inadequacy of epidemiology in this area, and are unable to provide definitive scientific support to the perceived mandate for public health action."