Risk analysis is commonly represented as a process involving three phases: risk assessment, risk management and risk communication. In the risk assessment phase, scientific and quantitative tools are most intensively applied. The risk management phase involves integrating broad social goals with science and developing and choosing strategies for addressing risk. Risk communication recognizes the need for broad input from diverse segments of the public and for outreach to affected parties. These phases are interactive, iterative and often take place at the same time, making it difficult to categorize specific activities conducted in risk analysis as uniquely assigned to that one phase. Bearing this qualification in mind, it is useful to review the role of value judgements and ethical considerations implicit in each of the three phases.
While we use the conventional categories of risk assessment, risk management and risk communication in this document, we emphasize that these distinctions are far less than they are often perceived. Indeed, in practice, risk analysis operates as a non-linear process.
First, a given hazard is identified and a risk profile is developed to guide decisions on how to proceed, giving particular consideration to whether resources should be allocated to a detailed risk assessment.
Second, a risk assessment is undertaken, according to guidelines established during the preceding step.
Third, the results of the risk assessment are used, along with other information, to select risk management options; the chosen options are then implemented and the results are monitored. While all this takes place, risk communication is occurring.
All three phases involve making judgements that are not unequivocally based on factual evidence. Many, if not most, of these judgements are ethical in nature, or have implicit ethical dimensions. Such judgements fall under both the "sound scientific analysis" and the "other legitimate factors" clauses in the Codex Alimentarius Statements of Principle Concerning the Role of Science, as well as in similar documents used in national systems of risk assessment and risk management. Taken together, these elements include a variety of economic, legal, cultural, social, environmental and scientific values needed to undertake an adequate risk analysis.
In this section, we discuss some of the sources of these value judgements and clarify how the role of values in the three phases of risk analysis can be made more explicit, and how value judgements are informed by ethical considerations.
Risk assessment was defined by FAO/WHO in 1995 as:
"The scientific evaluation of known or potential adverse health effects resulting from food-borne hazards. The process consists of the following steps: (i) hazard identification; (ii) hazard characterization; (iii) exposure assessment; and (iv) risk characterization."
In an ideal situation risk assessment involves quantifying the probability of occurrence of adverse health effects. In many situations, however, problems arise in the interpretation of available data. Furthermore, only a few countries currently provide most of the data for international risk assessments, especially on dietary intakes and exposure, making it difficult to apply the results widely. The Codex system and its parent agencies have recognized the need to include data from a broader range of countries in order to develop a universal data set. The problem, however, is that in most cases there are no data that allow such a quantification to be made.
These fundamental considerations led to a recommendation of the Joint FAO/WHO Expert Consultation:
"Many sources of both uncertainty and variability exist in the process of risk assessment of food-borne hazards to human health. Explicit consideration should be given to uncertainty and variability in the risk assessment process so that these may be taken into account in the formulation of risk management policies."
Value judgements and policy choices in risk assessment may be divided into three types.
· Internal dynamics. Carrying out a scientific risk assessment requires a host of value judgements. For example, hazard identification in food safety may be confined to well-characterized outcomes such as mortality and morbidity associated with known mechanisms of toxicity or disease processes. Or, the scope of hazards considered can be expanded to include less well-characterized, or even unknown and unforeseen, outcomes. If people exposed to even a small chance of harm see little personal benefit from the circumstances of their exposure, they may prefer a relatively expansive interpretation of hazards.
Even when hazards are well characterized, the decision to use a given data set reflects an implicit assumption that extrapolation from the population base from which data are collected can be made to the population actually exposed to the hazard. The problems that can arise in making inferences from, for example, animal studies to human populations, or in extending observed segments of a dose-response curve to a lower level of exposure are well documented in the food safety literature. In the international context, exposure estimates from certain (usually developed) countries often must be extrapolated to populations with very different food intake patterns in countries where residue data are scarce. The response to such problems often reflects an ethical judgement about the level of precaution that should be incorporated into a risk assessment. The general philosophical approaches of optimization and informed consent can underlie contrasting attitudes about the way such issues should be addressed. If, for example, one thinks that it is particularly important not to subject the rights of vulnerable groups - such as child-bearing women or young children - to a trade-off against benefits for the majority, at least two stances may be adopted, based on optimization and informed consent. On the one hand, one may be more likely to incorporate assumptions and scenarios or rely on data that suggest higher levels of risk. On the other hand, one may use less conservative scientific parameters, and plan instead to recognize the differing segments of consumers by labelling the final product so as to warn those at greater risk.
For some key areas of food safety, ethically difficult questions must be addressed before one can begin to undertake a risk assessment. For many hazards, one must estimate exposure based on assumptions about whether best practice and intended product use represent a realistic approximation of actual exposure. Sometimes an ambiguity in the way a risk is conceptualized becomes apparent only when fairly technical decisions in risk assessment are formulated. For example, with respect to genetically modified (GM) foods, the probability of harm might be assessed relative to all transformation events, even though many events are never developed into commercial crops or food products. Alternatively, it might be assessed with respect to only those transformation events intended to be introduced as foods, even though there is a non-zero probability that transformations intended for animal feeds or production of non-food items (such as drugs, biologics or industrial products) will enter the food system. One's final estimate of the aggregate risks associated with genetic modification can vary considerably based on such choices in characterizing the reference population for GM crops.
The international model of risk analysis includes an element called risk assessment policy which is defined as the "guidelines for value judgement and policy choices which may need to be applied at specific decision points in the risk assessment process"(FAO/WHO, 1997). Risk assessment policy, generally and in specific cases, is developed by risk managers in consultation with risk assessors and with input from stakeholders, and it is needed before a detailed risk assessment is undertaken. This step of risk analysis offers an important context in which value choices in risk assessment can be identified, clarified and examined.
It is not practical to recount here all of the value judgements, assumptions and policy choices that arise in constructing a risk assessment in the food safety context. Risk assessors are generally aware of the nature of these judgements, and there is an ongoing debate as to the ethical standards that should be used in making them. One school of thought is that there should be standard and consistent ways of making such judgements that reflect a consensus view of professionals undertaking risk analyses. An alternative position suggests that the burden of proof sought in a technical risk assessment varies according to the risk management context. In some circumstances, risk assessment policies may be fixed by statute or by the internal policies of an organization. In other circumstances, the need to address specific public concerns may militate in favour of assumptions specific to the context at hand. If, for example, there is a great deal of public distrust of organizations carrying out the risk assessment, it may be appropriate to be more conservative than usual when calculating exposure, while in a more typical case, it may be appropriate to adopt more empirically based assumptions.
· External pressures and demands. Because value judgements about which data to use, how to extrapolate beyond observed data and how to merge findings using different study methodologies can result in wide variations in the estimated degree of risk associated with an activity or situation, risk assessors can be subjected to pressure and demands to adopt assumptions or rely on data that interested parties believe will favour their case. A specific problem is the availability of dietary exposure data needed to estimate risks. Too often, data are available from only a few countries and it is not clear how the exposure estimates should be adjusted to assess risks in other cultures and geographic regions.
External demands to adjust their assumptions and selection of data place risk assessors in an interesting ethical dilemma. On the one hand, they feel that they must resist such pressures in order to preserve the objectivity and scientific integrity of the risk assessment. On the other hand, external groups can be a source of important information about how to characterize hazards and how to develop studies that will ultimately be of the most use in the risk management process. This is an area of risk assessment that has not been subjected to particularly explicit or thoughtful ethical debate, much less case study analysis.
· Transparency. Because of the technical complexity of most scientific risk assessments, value judgements employed in developing the analysis can be very difficult for outsiders to recognize and discern, much less to evaluate in light of their own values and interests. As a result, key elements in the ethics of risk assessment can be functionally hidden from view, and groups who lack the technical expertise to contribute data to or participate in such analyses can be effectively (if perhaps unintentionally) excluded from this critical decision-making process. For example, many developing nations lack scientific expertise to participate or cannot devote resources to the data collection and analysis involved in scientific risk assessment. This situation can, in turn, create distrust in the results of a risk assessment. There is thus an ongoing need to make transparent the judgements and choices in a scientific risk assessment, and to increase the technical capacity of parties (which are typically excluded) to participate actively in risk assessment.
Risk management is, according to the 1997 FAO/WHO Consultation:
"The process of weighing policy alternatives in the light of the results of risk assessment and, if required, selecting and implementing appropriate control options, including regulatory measures."
Consensus international food safety regulations have important implications for trade:
"The food standards, guidelines and other recommendations of the Codex Alimentarius Commission are considered by the WTO to reflect international consensus regarding the requirements for protecting human health from food-borne risk. (...) While the adoption and application of Codex standards remain technically non-mandatory, failure to apply Codex standards creates the potential for dispute if a Member applies standards that are more restrictive of trade than necessary to achieve required levels of protection."
Risk management may be examined with regard to several value dimensions. Risk management is iterative and overlaps with elements of risk assessment and risk communication. Nevertheless, it is useful to distinguish value judgements in risk management that tend to be made before a technical risk assessment is carried out -or at least in conjunction with the early stages of the risk assessment - from those that tend to be made once a more scientifically complete understanding of risk is available.
Starting the risk analysis
Value judgements play a central role in triggering the decision to engage in risk analysis. The fact that a risk analysis is being conducted reflects a judgement that the risk arising from some sort of hazard is non-routine and needs to be assessed. The triggering circumstance for a specific risk analysis is potentially of great ethical significance. The fact that a risk analysis is being done often reflects a prior judgement that some person or group should take action to reduce the likelihood or severity of harm. It reflects the judgement that someone should provide potentially affected parties with an opportunity to give or withhold consent, or that there is someone who could be held responsible for harms that might eventually ensue. Each of these "shoulds" and "coulds" can be understood to have an ethical, and sometimes legal, connotation. In this context, the ethically laden question of how much precaution is necessary comes up repeatedly and often plays a decisive role.
These possibilities mean that the decision to conduct a risk analysis is not undertaken out of idle scientific curiosity. It is one in which prior ethical understandings of the roles and responsibilities of both private sector stakeholders and government agencies are deeply embedded. Furthermore, in almost all situations, some groups will want the risk analysis carried out with the expectation that it may lead to new risk management activities, while others' interests may be served by continuing the status quo.
The triggering of a risk analysis is thus one of the most deeply value-laden dimensions of risk management. The triggering circumstance must be judged to be important and to require a response. For example, in advance of any deliberate characterization of hazards or attempt at quantification, the situation must be deemed serious enough to warrant the cost of a risk analysis, given limited resources to pursue such analyses. It may be useful to bear in mind that there is a difference between the level of resources needed for a quick and qualitative risk assessment versus a detailed risk assessment. Public expectations can also influence what triggers the need for risk analysis. If the status quo is perceived to be one in which the food system is plagued by hazards - as was the case in the early years of the twentieth century - new or novel technologies may have less tendency to trigger risk analysis than in circumstances such as prevailed at the end of the twentieth century, when the food safety system was generally presumed to be working well.
Once the decision to engage in risk analysis has been reached, decisions must be made as to:
whether a known hazard can be characterized and, if so, what potential harms and unwanted outcomes are associated with it;
who will participate in identifying and characterizing the hazard; and
whether a preliminary characterization of alternative risk management options might form the basis for a comparison of risks.
It will then be necessary to determine whether conducting a detailed scientific risk assessment is the best way to proceed in dealing with the problem. But the iteration between risk management and the hazard identification stages of technical risk assessment can blur the distinction between the two. Just as factors such as novelty and unfamiliarity influence the judgement that risk analysis should be undertaken in the first place, previous studies, existing data and good scientific models can influence judgement about whether launching a comprehensive risk assessment is likely to improve significantly the knowledge base for coping with the situation. Finally, both ethically and politically driven considerations can add urgency to the need for a systematic response. Each of these parameters requires informed judgements and the ultimate choices are often value laden.
Managing assessed risks
Risk management includes policies, actions and choices to reduce risk and protect public health that might be undertaken by a variety of groups, including government regulatory agencies, international standard-setting bodies, private sector firms, NGOs and individuals in their capacity as consumers or citizens.
Avoiding or mitigating a given hazard is not necessarily the sole goal of risk management. Risk management may also be undertaken to secure human rights, to rectify inequities or to secure outcomes not directly related to food safety, such as environmental quality, economic opportunity or protecting the interests of future generations.
The goals of risk management and the manner in which risk management choices are based on a scientific risk assessment are particularly influenced by the tension between optimization and informed consent.
With the optimization approach, risk management often involves balancing the scientifically estimated likelihood and degree of harm associated with the food safety hazard (along with other costs or unwanted consequences) against the benefits expected to follow from activities that reduce exposure to food safety hazards. Simply put, it may be said that the risks are acceptable whenever benefits outweigh risk, although in actual practice, optimization involves complex comparison of expected costs and benefits associated with a number of possible options.
However, the informed consent approach dictates that the scientific risk assessment is used not to determine the proper course of action in mitigating or accepting food safety risk, but to provide a basis for activities intended to empower those who would bear the risks in question. Such activities include, at a minimum, identifying populations at risk, informing risk bearers about the circumstances and offering them an opportunity to opt out of the circumstance for which their informed consent is being sought. In such circumstances, allowing individuals or groups to choose which risks to bear may be the appropriate risk management option (e.g. via informative labelling).
Risk management activities undertaken in response to problems of trust are particularly sensitive to this tension. From the standpoint of optimization, once governments or private sector stakeholders have lost the public's trust, the behaviour of consumers, activists and media sources is much more likely to produce suboptimal results than otherwise may have been achieved. From the standpoint of informed consent, trust is a signal that groups and individuals have delegated to others their authority to give and withhold consent. Once these decisions have been delegated, maintenance of trust is critical to the entire legitimacy of food safety policy decision-making, as people who cannot trust experts to exercise their rights of exit and consent on their behalf have effectively been coerced into risk-bearing situations.
A number of specific activities in risk management might contribute equally well to either of these ethical goals, optimization or informed consent. Involving affected parties in decision-making can help experts ascertain the value that consumers and citizens place on food safety risk reduction relative to other social outcomes, such as economic opportunity and environmental quality. Thus risk management decisions that are open to stakeholder involvement may be more capable of achieving optimal trade-offs. The same opportunities for stakeholder involvement may contribute directly to the empowerment of affected parties, which is a goal more consistent with the framework articulated by informed consent. As such, the divergence between these approaches should not be overstated. Nevertheless, the tension between optimization and informed consent approaches can produce radically divergent perceptions of the fundamental goals of risk management.
In practical application, a number of contingencies other than these overarching principles may lead stakeholders towards actions and responses of a particular type. The capacity of various governments, firms and individuals to carry out a proposed programme of activities may sharply limit the potential for implementing idealized versions of either optimization or informed consent. The degree and distribution of costs from both administering and complying with policy may influence the framing of a general strategy for risk management, as well as particular tactical responses.
A closer consideration of the difference between "strategy" and "tactics", in this sense, follows.
Strategy. The strategy chosen to cope with a given risk once it has been identified and assessed raises a number of ethical issues, irrespective of the overall approach. In devising a risk-management strategy, a risk manager needs to consider all of the following: What constitutes an appropriate level of protection or an adequate signal of consent? What is the opportunity cost of a given risk management strategy? Are there particularly vulnerable populations and, if so, should the strategy aim to provide them with additional protection? Will these strategies hold up in court if challenged? How much of an error margin is acceptable? How will non-tariff trade barriers be avoided?
Tactics. Once a general strategy is adopted, specific tactical questions inevitably arise. For example, labelling might be chosen as a means either to mitigate risk exposures (a goal consistent with optimization) or as a mechanism of information and exit (a goal more consistent with informed consent). In either case, however, the specific design and content of the label can be subjected to additional ethical questions about effectiveness, truthfulness and enforcement. Tactical questions also include choices about technical procedures and equipment to measure conformity to standards arising from risk analysis. Choices about equipment have distributive economic consequences; they raise questions about who decides and whether alternative (including low cost) means are available for meeting a given standard, as well as other costs of compliance. For example, pH may be measured for a small sum with a portable test kit or at great cost with highly sophisticated machinery. If the latter is required in order to meet standards, then those individuals or firms unable to purchase the sophisticated equipment may be put out of business in the name of food safety.
Again, the overarching tension between optimization and informed consent can strongly influence the way that these questions are addressed. On the one hand, considerations reflecting the desire to optimize the trade-off between costs and benefits may lead one to evaluate such questions in light of the efficiency with which alternative management practices deliver the desired level of exposure to a food-borne hazard. On the other hand, since a goal of informed consent is to empower those who are exposed to risk, questions about "who and how" might be evaluated more in terms of the way that risk bearers themselves occupy key decision-making roles.
Risk management must also be sensitive to ethical issues that arise in connection with equity. While equity issues are not uniquely associated with food safety, these considerations cannot be ignored by policy-makers. Equity considerations include the following.
The balance of power among the various stakeholders in the risk analysis and what is at stake for each group in a given risk analysis. The relevant parties include, for example, members of various disciplines, delegations from various nations, nongovernmental sectors and firms.
Participation (or lack thereof) of developing nations in risk management decisions at the international level.
Distribution of costs and benefits across nations, within nations, among various ages/genders/incomes/ethnicities, etc.
The impact of a decision on future generations. Care must be taken so that future generations inherit a better, safer world and are not burdened with an endless need to mitigate the negative impacts of our actions.
Equity is crucial in the consideration of ethical issues related to food aid and the right to food (discussed later in this report). One significant element of equity in food safety policy concerns the potential consequnces of standards derived from risk analysis, such as the creation of non-tariff trade barriers. Some food safety standards can have the practical effect of denying access to world markets, particularly by food producers from poor countries or from regions lacking technical expertise. If the effect of a food safety standard is to deny access to world markets, particularly by food producers from poor countries or from regions lacking technical expertise, the need for such inequitable standards should be clearly demonstrated by an appropriate risk assessment.
The FAO/WHO Expert Consultation on Risk Communication recommends the following definition of risk communication (FAO/WHO, 1998):
"Risk communication is the exchange of information and opinions concerning risk and risk-related factors among risk assessors, risk managers, consumers and other interested parties. (...) Effective risk communication should have goals that build and maintain trust and confidence. It should facilitate a higher degree of consensus and support by all interested parties for the risk management option(s) being proposed. (...) The Consultation considered that the goals of risk communication are to (...) foster public trust and confidence in the safety of the food supply; (...) exchange information, attitudes, values, practices and perceptions of interested parties concerning risks associated with food and related topics."
As suggested by the quoted passage, there are many forms of and contexts for food safety risk communication. The need for interactive communication among parties in risk analysis and the value of such communication in establishing the basis for decisions is addressed in some detail in the FAO/WHO report.
Risk communication about food hazards, risk assessments and risk management should take as its primary point of reference the perspective of those affected by the risk. Risk communication should address the question of "Is this food, with these known risks, safe?". The concept of safe in popular language differs from the concept of safe among food scientists. In day-to-day language, "safe" is defined (Shorter Oxford English Dictionary) as:
"Safe: Free from danger; secure; free from risk, not involving danger; not likely to cause harm or injury."
This definition of "safe" highlights the value of "trust" in relation to our perception of food safety. When we have trust in the safety of food, we do not think of the potential risks involved in consuming it. From the perspective of the food safety system, however, it is important to remember that food, although it is considered "safe", is never totally risk-free.
The goals of communication
In both ethical approaches discussed in this report - optimization and informed consent - communication is critical in establishing a better understanding of scientific processes involved in risk analysis and facilitating better interactions among all interested parties. From an optimization perspective, risk communication is a means to an end. Under the informed consent approach, successful non-directive communication is the primary goal of a strategy that seeks not to influence, but to inform and empower.
Since the most difficult issues in food safety decisions generally involve competing values and interests, it follows that much of the risk communication in food safety policy-making should focus on the value and ethical dimensions of the decision. Parties (food safety regulators, food industry, food trade, consumers) interested in food safety policy decisions may hold two views of associated risk communication concurrently. They may believe there is a lack of openness and transparency in decision-making, which may pose barriers to their effective participation in the process or to their acceptance of its results. Also, it may often seem that there is a surfeit of information that is not always easy to understand.
The information an individual needs about food safety may vary over time and circumstance. For example, when there is a change in lifestyle, when there are extraordinary events such as a crisis or food-borne illness outbreak or when a matter becomes a cause célèbre, an individual's information needs may change. A range of risk communication strategies should be developed anticipating unforeseen events and should be sensitive to the changing circumstances of the recipients of the information.
The interactive nature of risk communication between and among decision-makers and stakeholders emphasizes the need for communication to be effective and embedded in all the steps of risk analysis - not just added at the end to offer a retrospective of past events. It also requires all interested and involved parties with a role in any stage of the risk analysis to understand and apply an ethical approach to their actions in communicating food risks. Some of those ethical dimensions are elaborated below.
Individuals. Individuals can participate in the risk communication process either as individual citizens or as members of consumer organizations. Individual citizens require that:
any perceived information asymmetry be redressed to facilitate their participation in the debate on food safety;
their right to information be respected so that they are able to make informed choices;
information be available currently if it can affect decisions that will have an impact many years later; and
information be available if, when and where it is needed to enable them to influence important decisions being made at all stages of the risk analysis.
While there is a need to democratize the communication process, not all consumers may wish to be encumbered at all times with detailed risk analysis information, nor do all consumers have the capacity to interpret and use masses of information.
The ethical issues that arise for consumer organizations include the need to be transparent in matters relating to their funding and to ensure that it is clear whom they represent when they act as spokespersons on food safety and related ethical concerns. In particular, the voices of consumers from the developing countries have been under-represented in international food safety governance, and the need to build capacity for consumer participation at both national and international levels is widely recognized.
While everyone in the food safety system has a role in managing risks, we refer here to the professional risk manager within a public food safety authority. Risk managers have an interactive and iterative partnership with risk assessors, to ensure that complete and appropriate scientific information is available to support risk management decisions. Risk managers also are responsible for ensuring that affected stakeholders have opportunities to participate at appropriate points in the risk analysis process, and that the valuable information provided by these participants is effectively weighed in choosing among risk management options. The risk manager also communicates with individual consumers by participating - as part of a public authority - in national food information programmes. The manner of delivery and the effectiveness of such information programmes should be reviewed and evaluated periodically. Effective management of each of these aspects of risk communication is crucial for maintaining trust in the food safety system and achieving optimal results from risk management strategies.
Risk assessors must recognize the challenge that scientific terminology and language can pose to non-scientists involved in risk analysis. Risk assessors also face particular problems in attempting to communicate explicitly about the value judgements used in scientific enquiry, the uncertainties associated with the risk assessment process and the presentation of data. Iterative communication by risk assessors with risk managers and other interested stakeholders - and in particular the development of risk assessment policy as an essential step in risk analysis - are important ways of clarifying and reaching agreement on the scientific and policy judgements that guide and are reflected in a risk assessment.
Modern science is based on research. Risks associated with new products of applied research may trigger the need for a food safety risk analysis, and knowledge generated by research is essential for effectively completing a risk analysis. In the latter case, researchers, particularly in the fields of health and food sciences, can contribute throughout the process. Ethical concerns for the researcher arise in interpreting and presenting research data, and may also arise with respect to the setting of national research agendas and the extent of pro bono research. Research into the actual communication processes employed throughout risk analysis is also relevant.
The agrifood industry
The agrifood industry (including input suppliers, producers of agricultural products, food processors, food service organizations, producer and industry organizations and food retailers) engages in risk communication through keeping quality assurance records, interactions with government regulatory agencies, and communication with the consumer through product labels, among other processes. It also is an important stakeholder and participant in all stages of risk analysis and in other activities, such as lobbying for particular food laws and advertising its products and services. In so doing, the food industry needs to address a range of ethical values including trust, truthfulness, fairness and transparency.
The media (radio, television and printed and electronic media), as the industry of communication, are significant, if not primary, sources of public information on food safety risks. The media are highly effective at delivering information to large numbers of people in easily understood forms and at overcoming barriers of language, level of education and cultural appropriateness. However, by their very nature, the media may sometimes oversimplify or sensationalize information to attract an audience, and may intentionally or accidentally mislead the public about food-related hazards. Mechanisms are needed to enhance media communicators' understanding of the science and values of a risk assessment and of the risk analysis framework. Also needed are training programmes for regulatory officials and other participants in risk analysis, to enhance their ability to communicate effectively with the media.
As the international food safety system comes to terms with its need and responsibilities for risk communication, several considerations should be kept in mind. First, communication should be structured to ensure that ethical components of food safety decisions are clearly identified as early as possible. Second, the system should function so that value-laden choices made by risk managers are made in an open, participatory process that respects the rights and roles of all stakeholders. Following such a strategy will not necessarily make food safety risk analysis more efficient, as dealing with all difficult questions may be time-consuming. But a strategy more sensitive to ethical issues should make food safety risk analysis more effective, by making decisions sounder, more transparent, more democratic and better understood. This, in turn, should make risk analysis decisions more acceptable to, and useful for, the governments and citizens of all nations.
 FAO/WHO. 1995. Report of an
Expert Consultation on Application of Risk Analysis to Food Standards Issues.
Geneva, 13-17 March 1995. WHO/FNU/FOS/95.3. Geneva.|
 FAO/WHO. 1997. Report of a Joint FAO/WHO Consultation on Risk Management and Food Safety. Rome, 27-31 January 1997. FAO Food and Nutrition Paper No. 65. Rome.
 FAO/WHO. 1998. Report of a Joint FAO/WHO Expert Consultation on Application of Risk Communication to Food Standards and Safety Matters. Rome, 2-6 February 1998. Geneva, WHO.