Joseph V. Rodricks, Ph.D., is a Principal of Environ International Corporation,
Life Sciences Group (Arlington, Virginia, United States) and a Visiting Professor
at The Johns Hopkins University School of Public Health.
An understanding of the potential risks to human health posed by exposures to food contaminants is a central component of the process involved in establishing health protective standards. Thus, risk managers - those with the responsibility for establishing and implementing such standards - are required to engage in a dialogue with risk assessors, and to extract from that dialogue a firm grasp of what scientific evaluation can and cannot reveal about the risks associated with specific exposures. Because most risk assessments involving chemical toxicity and microbial pathogenicity are beset with scientific uncertainties, and because the nature of these uncertainties varies among different chemical and pathogenic agents, a general understanding of the risk assessment process, while important, is not adequate to support informed decision-making in specific cases. A dialogue between risk assessors and managers regarding the nature of and justification for the scientific analysis supporting specific risk results is essential if risk information is to be used in ways that are consistent with scientific understanding and its limits.
Risk managers are burdened with many other difficult tasks. When dealing with food contaminants, they need to consider factors other than health risk - the availability of technical means to reduce risk, the consequences for food supply of any actions taken, public perceptions and desires, and even the possibility of unintended and offsetting reductions in health benefits. Each of these considerations can and should be supported by additional technical ana-lyses, and a systematic and transparent methodology should be applied to integrate these considerations and the risk information to arrive at risk management decisions.
Unfortunately, the difficult task of risk management does not end with a decision, but usually involves the responsibility for explaining the basis for the decision to various affected parties (many of which will have contributed to the decision-making process). At the least this involves an exposition of how adequate safety criteria have been met.
Substantive, open communication is essential to the successful completion of every phase of the risk management process, and decisions emerging from this process will gain creditability in direct proportion to the care with which communication has been encouraged and managed. In this brief discussion of the topic, communications between risk assessors and managers, and between technical analysts and the lay public, will be emphasized, but it should be kept in mind that many other types of dialogues are involved in risk management de-cisions (Slovic et al., 1980).
All parties involved in the process of developing risk-based decisions are in some sense concerned about the ultimate consequences of these decisions; they are thus tempted to influence communi-cation in ways that favour their preferred outcomes. That risk communication can be impeded because of the introduction of extraneous concerns (sometimes in the guise of legitimate concerns) is well recognized, and probably unavoidable. Conscious awareness of the existence of such potentially impeding factors in ourselves and others, and constant efforts to engage in clarifying dialogue, are perhaps the only feasible means to avoid the undue influence of such factors. Figure 1 lists some typical impediments to risk communication.
Scientists prefer to communicate with other scientists in their own disciplines, and are often hesitant to explain their work to non-scientists, or even to scientists in other disciplines. Explanations of the methodology and results of risk assessments in ways that do not distort scientific understanding and that are also useful to an audience of non-specialists are especially problematic, because they require explicit discussion of the bases for choosing among different sets of toxicity and dose-response data. Even more difficult are explanations for dealing with extrapolations across species and with variability
FIGURE 1
Some impediments to risk communication
Risk assessors |
Risk managers |
The public and various |
|
|
|
And everyone is influenced by perceptions that do not match the technical understanding of risk (see section on Risk perception and risk communication in the text). |
FIGURE 2
The relative roles of science and policy in protecting people
from risks of chemical toxicity
Research |
Risk assessment |
Risk management |
Science |
Science |
Science |
Policy |
Policy* |
Policy* |
* The National Research Council (1994) distinguished “risk assessment policy” from “risk management policy”. The former involves the selection of various assumptions and models to deal with uncertainties in the risk assessment process (these are the “default” choices). Risk management policies are different in nature. - Source: Rodricks, 2001. |
It is useful for managers to acquire a grasp of what might be called the "anatomy" of specific risk assessments. Some of the useful anatomical characteristics of risk assessments for the effects of non-carcinogenic agents, generally assumed to act only when exposures exceed a threshold dose, are listed in Box 1 and illustrated in Figure 3. For the effects of carcinogenic agents, which may not obey threshold rules, see Box 2. Indeed, managers would be advised to seek out this type of information, on a substance-specific basis, through direct dialogue with risk assessors.
One aspect of the information shown in Boxes 1 and 2 that has been insufficiently discussed concerns the appropriate methods for characterizing human exposures to food contaminants. The risk assessment process for threshold agents leads to an estimate of a lifetime daily intake that is likely to be protective (i.e. to represent a threshold dose) for the most sensitive members of the human popu-lation (see Figure 3). To assess risk for such an agent it is necessary to understand whether, and to what extent, exposures to such an agent exceed the estimated threshold dose. The best approach to the latter evaluation requires the development of sufficient data to estimate the distribution of such exposures within the population. Having this type of infor-mation allows: 1) estimation of the fraction of the population with exposures that exceed the threshold; and 2) estimation of the effects on the exposure distribution that would result from the imposition of various maximum limits (MLs) for the agent in food. Similar analyses may be made for carcinogens, based not upon the criteria of exceeding the threshold, but rather on the criteria of the magnitude of the risk reduction achieved.
Much could be gained were it possible to estimate accurately the probability (which, in most cases, would be extremely small) that individuals at the "high end" of contaminant exposure were also at the "high end" of susceptibility to toxic effects. Such certitude could assist in defining protective tolerance levels that did not place undue weight on extremely unlikely exposure/risk scenarios. Scientific methodologies for gaining such knowledge are under study, but are perhaps not ready for routine application.
BOX 1 What do risk assessors need to tell risk CONCERNING TOXICITY HAZARD/DOSE-RESPONSE
CONCERNING HUMAN EXPOSURES:
|
Explicit recognition of risk is a feature of low-dose risk assessment for carcinogens, together with the concept that the only no-risk condition is when there is no exposure (Box 2). Because exposure avoidance is possible only in the case of intentionally introduced ingredients in food, decision-making for unavoidable contaminants requires both acknowledgement of and some tolerance for risk. Most decision-makers would prefer not to be placed in such circumstances, for understandable reasons. They are offered some comfort when told by risk assessors that low-dose risk estimates are "upper bounds" on actual risk (because actual risks cannot be predicted), and that "actual risk may be zero". But they are nevertheless required to admit to the possibility that some small risk may exist at contaminant intake levels which they have selected as acceptable.
BOX 2 What do risk assessors need to tell risk managers about carcinogenic risks predicted through low-dose extrapolation?
|
Analysis of the following characteristics of risk-based decision-making may assist risk managers in dealing with this difficult problem.
FIGURE 3
Towards conceptual clarity regarding the meaning of
acceptable daily intake (ADI) and tolerable daily intake (TDI)
- Typical derivation of TDI/ADI from animal data: factor of 10 for animal-human extrapolation, and factor of 10 for "average" to "sensitive" extrapolation within human population (note: factors other than 10-fold often used).
-The fraction of the human population that might be at risk below a derived ADI/TDI, while almost certainly very small, cannot be proven to be zero. Thus, it is not possible to claim that ADI/TDI is "absolutely safe". Development of a contaminant intake distribution would reveal the fraction of the exposed population that falls within the zone of potential risk (i.e. near or above the ADI/TDI).
- The effectiveness of various MLs in shifting intake distributions toward safer zones can be evaluated for purposes of risk management decision-making (note: the exposure distribution averaging time must match the averaging time upon which the ADI/TDI is established).
Notwithstanding the advantages of risk-based decision-making, it is clear from much research in social psychology that most people do not think of risk in terms of probabilities, but focus on the other attributes of a risk. Thus, it is possible, with some simplification, to categorize risks as "tolerable" or "non-tolerable" to people, irrespective of any objective measure of probability (Table 1). These other attributes appear to be far more important to most people (possibly including risk managers) than do the measures of risk provided by risk assessors. These varying degrees of risk toleration need to be considered by risk managers (and most managers sense their existence intuitively) when their decisions are offered for public consumption. Note that food contaminants, to which people are exposed involuntarily and from which they derive no benefit, especially those that are carcinogenic and are of industrial origin, generally pose risks of the highly "non-tolerated" type, so that communication of risk decisions regarding them is exceedingly troublesome.
It is interesting to note that many of the very large risks people face - from smoking, excessive caloric intake and inadequate caloric expenditures, other unhealthy dietary habits, excessive alcohol consumption, automobile accidents - tend to have attributes of the "tolerable" type, and public health officials find themselves seriously challenged when they seek ways to reduce these large risks. Those required to manage the many thousands of small risks of the "non-tolerable type" face a completely different challenge. It is to be recommended that several of the available guidance documents on this topic, which are listed in the bibliography, be given study by anyone involved in risk communication (Rodricks, 2001).
Pesticides being applied to tomato plants. Pesticides are used in accordance with good agricultural practice to reduce risk - FAO/19436/A. Conti
Although an understanding of risk perception is important in the communication process, it should nonetheless be only a minor part of formal risk management decision-making. Such decision-making will be most successful if it is based to the maximum extent possible on technical analyses, and if it avoids significant reliance upon subjective factors, such as those listed in Table 1, because their relative influences vary widely among individuals. Only when decisions rest primarily on technical analyses, communicated openly among all parties involved, is it possible for risk managers to offer credible explanations for them; without that possibility decisions will not easily find broad acceptance.
TABLE 1
Subjective attributes of risks that influence public perceptions
WHEN RISKS ARE TOLERABLE,* |
WHEN RISKS ARE INTOLERABLE,* |
voluntarily assumed |
imposed by others |
of high personal benefit |
of no perceived personal benefit |
scientists agree |
scientists disagree |
not catastrophic |
catastrophic |
natural in origin |
industrial in origin |
not fearsome |
highly dreaded |
common event |
rare event |
equitably distributed |
not equitably distributed |
* “Tolerable” risks are those that people tend to find less threatening than intolerable types, irrespective of the magnitude of the risk (Slovic et al., 1980). |
1 This article is based on a presentation at the Joint FAO/WHO Seminar on Risk Analysis of Contaminants, held in Rotterdam, the Netherlands,10 March 2002.
2 Variability among species and individuals within species is a fact of nature, and must be dealt with in every risk assessment. The fact of variability is scientifically certain; its magnitude in specific circumstances (e.g. the variability in response among humans to a specific level of methylmercury exposure) is highly uncertain. Risk assessors have adopted so-called "default assumptions" to deal with uncertainties; these are based in part upon scientific knowledge but also include an element of caution.
3 Only in the trivial cases in which it is possible to ban a substance completely is absolute safety assured.
4 Adoption and enforcement of MLs can do much to reduce the "high end" of cancer risk distributions for individuals (i.e. MLs can be set so that no individual is expected to incur risks from carcinogen exposure greater than, say, 10-5/lifetime). But an analysis of how exposure distributions shift by imposition of an ML will generally reveal that such an imposition will do very little to the exposures incurred by the vast majority of a population. Thus, population risks will be affected very little by imposition of an ML. Only contamination prevention will significantly affect population risk.
Bibliography
Finkel, A.M. 1990. Confronting uncertainty in risk management. A guide for decision-makers. Washington, DC, Resources for the Future.
National Research Council. 1989. Improving risk communication. Washington, DC, National Academy Press.
National Research Council. 1994. Science and judgment in risk assessment. Washington, DC, National Academy Press.
Rodricks, J.V. 2001. Some attributes of risk influencing decision-making by public health and regulatory officials. Amer. J. Epi., Suppl. 154(12): S7-S12.
Slovic, P., Fischhoff, B. & Lichtenstein, S. 1980. Facts vs. fears: understanding perceived risk. In R. Schwing & W.A. Albers Jr, eds., Societal risk assessment: how safe is safe enough? New York, USA, Plenum Press. 374 pp.