one of three people in the U.S. will develop some form of cancer; one in four will die of it (NCI, 1990). The expected rate of spontaneous abortion is estimated to be as much as one in four of all pregnancies (NRC, 1989). Therefore, clusters of cancer, spontaneous abortion, or other common diseases can easily arise by chance. Assessing whether a given cluster of these common health problems could be linked to environmental exposure requires either the study of very large numbers of persons or the finding of extraordinarily elevated rates. Further, most hazardous-waste sites involve potential or actual exposures of only small numbers of persons. Because many of them no longer live in the area when a cluster is identified, tracking down all who potentially could have contributed to a cluster is extremely difficult.
Another problem relates to the fact that analyses of health effects possibly linked with exposures from hazardous-waste sites usually involve making implicit multiple comparisons, which results in increased rates of disease due to chance alone. “When eager environmental epidemiologists check to see if cancer registry data suggest that a particular waste site has increased the incidence of any one of the 80 types of cancer with a p value of 0.01 or less, we know that there is a 0.99 probability of escaping an increase in all of these cancers. So there is a better-than-even chance that the risk of some kind of cancer will be elevated around the site” (Neutra, 1990, p. 5). Multiple comparisons are being made in an implicit manner, in that only the single type of cancer that is elevated becomes the subject of public concern and study, rather than each type of cancer separately or all cancers combined. In the state of California 55 percent of the 5,000 census tracts will have at least one type of cancer elevated because of chance alone. Hence, there are potentially 2750 false-positive clusters to investigate each decade.
Finally, for many hazardous-waste sites, there usually are no data on relevant exposures that could have occurred several decades earlier, given the long and indefinite period for development of many forms of cancer. This and the other considerations described in this section explain why the observed strength of an association between pathology and exposure to hazardous wastes can be weak, even though the link may be causal.
Specificity implies that a putative cause induces a specific disease. However, a one agent-one disease model has been shown not to apply for many diseases, such as lung cancer, in which multiple causes are involved. Further, many agents, such as cigarette smoke, pro-