Campbell (1980) suggested the thesis that the risk (R) of some agent or event can be estimated as a function of the product of the probability (P) of the event and the severity of the harmfulness of the event or agent (H):
R = P × H
From the simplest viewpoint, the risk-taker may accept harm of greater severity (high value of H) only if the probability of occurrence (P) is very low. Conversely, events that are only mod- estly harmful (low value of H) may be acceptable at higher levels of frequency or probability. From this argument, safety may be taken as a measure of acceptability of some degree of risk.
Table 13.9, taken from the work of Oser (1978) and Upton (1980), lists the risk of death classified in relation to specific activities. From this table, all of the activities listed exhibit some degree of risk or probability (P) of death or harm (H). The important point to note is that the probability of risk per million persons per year ranges from 0.1 for lightning striking to 20,000 in the case of motorcycling. A careful person presumably compares the risks of any event to his or her health with the benefits that will potentially accrue before making a decision. Relatively few people may actually do this, and even when they do, precisely what index is chosen as the indicator of relative safety is a function of the value judgment of each individual.
As to the risk of cancer, the harm (H) is considered by most lay persons to be extremely great. In view of this concern by the public, the U.S. government through its regulatory agencies, such as the Environmental Protection Agency (EPA), the Food and Drug Administration (FDA), and others, has assumed a major role in practical considerations of human risk from environ- mental agents. Two theorems are the basis for the estimations of human risk from carcinogenic agents in the environment.
1. A threshold (no-effect) level for a carcinogenic agent cannot be determined with any degree of accuracy.
2. All carcinogenic agents produce their effects in an irreversible manner, so that the actions of small amounts of carcinogenic agents in our environment are additive— producing a “carcinogenic burden” for the average individual during his or her lifetime.
These bases may be considered as default assumptions that are utilized if there is not sufficient evidence to alter these assumptions. Recent guidelines by the EPA have indicated that at least one regulatory agency is beginning to consider and even include in their final disposition of the regulation of a chemical data that may alter these default assumptions (Page et al., 1997). Since the presence of thresholds of promoting agents as well as their reversibility has already been noted (Chapter 7), such data may become useful in consideration of regulation of chemicals in the future. However, the gold standard chronic 2-year bioassay that is utilized as the mainstay in regulation of both industrial and pharmaceutical chemicals does not distinguish between initiat- ing, promoting, and progressor agents; it will require substantial additional studies to give cause to alter the default assumptions. As noted above, scientific risk estimation should be carried out with the full knowledge of the action of the carcinogenic agent as a complete carcinogen, or as having a major action at one or more of the stages of carcinogenesis.
The extrapolation of bioassay data to human risk estimation is one of the most difficult problems that has faced society and will face us for years to come as numerous new chemicals enter the environment. In attempting to predict the behavior of a chemical in the human from data obtained from bioassays, a number of factors should be considered in extrapolation of bio- assay data to human risk (Kraybill, 1978). These include:
• Reproducibility of experimental data
• Tumor incidence in experimental animals on a dose-dependent basis
• Relative approximation of experimental dose to that of human exposure
• Acceptable design and statistical evaluation of bioassay
• Consensus on interpretation of histopathological changes
• Availability of biochemical, metabolic, and pharmacokinetic data to be considered in final decision making
Not included in these factors proposed more than two decades ago is a knowledge of the action of the agent as an initiating, promoting, or progressor agent. Unfortunately, not all of these fac- tors are taken into account when regulatory decisions are made at the governmental level con- cerning specific compounds in our environment. Newer requirements for more extensive studies of compounds that would satisfy these factors are a goal to be achieved but as yet not attained.
Another consideration in determination of human risk is whether or not the estimation is qualitative or quantitative. Qualitative risk estimation is much easier to develop based on qualita- tive analyses of the variety of bioassay procedures utilized. As noted in Chapter 11, the IARC as well as regulatory agencies throughout the world take very seriously the qualitative finding of induction of neoplasia in one or two species of animals as a qualitative indication of risk of car- cinogenicity to the human. However, quantitative risk analysis is much more difficult. In fact, a number of epidemiologists have refused to make such quantitative relationships on the basis of animal data and would only use data in the human to carry out such estimates. Still, as we have seen from the utilization of various “safe” doses of carcinogenic agents and a variety of other factors, quantitative risk assessment has been and is being applied to human risk situations of specific chemicals and mixtures. Paramount in such considerations are the use of mathematical models in which, making a variety of assumptions, one may develop quantitative risk estimates for the human. Some of these models are considered below.