Hazard identification is the first step of a risk assessment, addressing the question of whether the agent or factor poses a risk to human health. This step is inherently integrative, as it may draw evidence from structure-activity relationships for chemical agents, in vitro evidence of toxicity, animal bio-assays, and epidemiologic data (NRC 1983). Epidemiologic data indicative of an adverse effect, when available, are strongly weighted in the evaluation of the weight of evidence to determine if an agent presents a hazard. Human data provide direct evidence of a hazard without the need to extrapolate from knowledge of toxicity in analogous agents or from another species. In fact, as we have gained a further understanding of the complexity of cross-species extrapolation from animal to man, such extrapolations are viewed with less certainty, unless buttressed by an understanding of human and animal pathways of absorption and metabolism and of mechanisms of action. Further, epidemiologic studies evaluate the impact of exposures received by the population, including complex mixtures which may not be readily replicable in the laboratory. Epidemiologic research captures the consequences of interactions among agents, and investigations in populations may capture the full range of susceptibility. However, given the numbers of agents of concern, epidemiologic data have been available on only a small number of environmental contaminants, and there is more often reliance on toxicologic evidence in identifying a hazard. In addressing gaps in the database on an agent's toxicity, the Blue Book gives toxicologic data collection higher priority than epidemiologic data collection, justifying this ordering by the cost of epidemiologic data and the ambiguity of the findings of some observational studies.
In using epidemiologic data for the step of hazard identification, researchers' interpretation of the evidence is fully parallel to the assessment of the causality of an association between an exposure and an adverse health effect. There are no specific guidelines for interpretation of epidemiologic data in risk assessments that go beyond the conventionally applied criteria for causality. For cancer, guidelines for interpreting the strength of evidence have been published, for example, by the International Agency for Research on Cancer (WHO I ARC 1972). However, these guidelines are not rigid criteria and, as with the widely applied criteria for causality, there may be disagreement on the proper classification of epidemiologic evidence for the purpose of hazard identification. For example, the interpretation of negative epidemiologic findings in the hazard identification process is a major source of disagreement.
Nonetheless, epidemiologic data have played a central role in some risk assessments, e.g., lung cancer/environmental tobacco smoke (US EPA 1992a) and lung cancer/indoor radon (US EPA 1992b). In its 1992 risk assessment, the Environmental Protection Agency classified environmental tobacco smoke as a class A carcinogen (i.e., a carcinogen for which definitive human carcinogenicity was available). The report noted that the data from active smokers, in combination with an understanding of mechanisms and dose-response relations, were sufficient for this classification. However, a principal basis for the classification was the result of a meta-analysis of the epidemiologic studies on lung cancer risk in never-smoking women married to smokers. The agency's analysis was also careful to consider potential sources of bias affecting the findings of the epidemiologic studies. In the example of radon, there is convincing evidence on human carcinogenicity from epidemiologic studies of radon-exposed miners (NRC 1988). All of the studies show strongly increased overall risks of lung cancer, approximately increased by three- to sixfold, and radon exposure alone causes lung cancer in laboratory models. Approximately 20 studies have shown excess lung cancer occurrence; the excess cannot be explained by confounding by smoking or other factors and the effect is strong. In assessing the risk of indoor radon, the agency has considered that the findings from epidemiologic studies of underground miners are sufficient for the hazard identification step.
Was this article helpful?