The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
Health Risks from Exposure to Low Levels of Ionizing Radiation: Beir VII Phase 2
the effect of confounders and the ecological fallacy2 in the evaluation of high-background-radiation areas and concluded that “these two problems alone are enough to make such studies essentially meaningless” (NRC 1990).
Another important consideration is the expected magnitude of the increase in health effect induced by excess background radiation. If one assumes a linear no-threshold response, a calculation can be made for expected cancers induced by excess radiation in a high-background-radiation area. As an example, consider the elevated levels of gamma radiation in Guodong Province, Peoples’ Republic of China (PRC). In this study, a population receiving 3–4 mGy per year was compared to an adjacent control population receiving 1 mGy per year. No difference in cancers was noted between the high-background area and the control area (NRC 1990). One can estimate the expected excess percentage of cancers resulting from the 2–3 mGy difference in exposure per year using a linear nonthreshold model and the lifetime risk estimates developed in this report. A calculation by this committee indicated that the expected percentage of cancers induced by the excess background radiation would be 1–2% above the cancers occurring from all other causes in a lifetime. Even if all confounding factors were accounted for, it is questionable whether one could detect an excess cancer rate of 1–2%. Excess cancers may indeed be induced by elevated radiation exposure in high-background areas, but the excess may not be detectable given the high lifetime occurrence of cancer from all causes.
Ordinarily, epidemiologists do not consider ecologic data such as this as being sufficient for causal interpretations. Since the data are based on populations, no information is available on the exposure and disease status of individuals. Such data cannot be controlled adequately for confounding factors or for selection bias. Although ecologic data may be consistent with an inverse association between radiation and cancer, they may not be used to make causal inferences.
A second type of epidemiologic study that has been used to evaluate the association between exposure to radiation and disease is the retrospective cohort study. Persons who have had past exposure to radiation are followed forward in time, and the rate of disease is compared between exposed and nonexposed subjects or between exposed subjects and the general population. Especially valuable are occupational studies that include both unexposed and exposed subjects, so that a dose-response evaluation can be made of the relation between radiation exposure and health outcome. Typically, study populations in retrospective cohort studies include persons who have worked with radiation in medical facilities or in the nuclear industry or patients with cancer or other disease who have been treated with radiation.
It is common in cohort studies of occupational populations to observe that the overall mortality rate is lower than that of the general population, commonly about 15%. This is not interpreted to mean that work per se reduces the risk of mortality, but rather that healthy persons start to work more often than unhealthy persons (Monson 1990). The term “healthy worker effect” (HWE) is commonly used to describe this observation. Diseases such as cancer that develop in later life ordinarily have less of an HWE than noncancerous diseases. The HWE is observed in most occupational studies, including those of radiation workers, and should not be interpreted to mean that low doses of radiation prevent death from cancer or other causes.
A third type of epidemiologic study that has been used to evaluate the association between exposure to radiation and disease is the case-control study. Persons with a specific disease are compared to a control group of persons without the disease with respect to their past exposure to radiation. This type of study is unusual in radiation epidemiology, in that most general populations have relatively low exposure to radiation.
While no phenomenon similar to the HWE is observed in case-control studies, the play of chance is always operative, as it is in cohort studies. Thus, if some exposure does not cause cancer and if a number of case-control studies are conducted, there will be a normal distribution observed in the odds ratios that describe the association between exposure and disease. Some studies will have an odds ratio that is less than 1.0; others will have an odds ratio greater than 1.0. In interpreting these studies, it is inappropriate to select only those that are consistent with an excess or deficit of disease. Rather, the entire distribution must be examined to assess the likely relationship between exposure and disease.
The studies discussed here illustrate the variability that is inherent in all epidemiologic studies and the need to evaluate the entire body of relevant literature in order to assess possible associations between radiation and disease, be they positive or negative. In its evaluation of the literature and in its discussions, the committee has found no consistent evidence in the epidemiologic literature that low doses of ionizing radiation lower the risk of disease or death. Some studies show isolated positive associations between radiation exposure and disease, and some show isolated negative associations. However, the weight of the evidence does not lead to the interpretation that low doses of radiation exert what in biological terms is called hormesis.
The committee concludes that the assumption that any stimulatory hormetic effects from low doses of ionizing radiation will have a significant health benefit to humans that exceeds potential detrimental effects from the radiation exposure is unwarranted at this time.
Ecological fallacy: two populations differ in many factors other than those being evaluated, and one or more of these may be the underlying reason for any difference noted in their morbidity or mortality experience (Lilienfeld and Stolley 1994).