- Risk Analysis In Relation To The Importation And Exportation Of AnimalProducts
|Last update: March 2, 2011 09:29:33 AM|
Risk Analysis In Relation To The Importation And Exportation Of Animal Products
Dr P. G Ma rais
Grootfontein Agricultural College, Middelburg, 5900 Cape Province, Republic of South Africa
Risk analysis is not a new discipline, the concept actually has a long record of application in areas of engineering and economic activity, but nonetheless its application in area of animal health is a recent phenomenon. Its origin goes back to the "Dunkel Proposal" under the General Agreement on Tariffs and Trade (GATT).
As from 1 January 1995, GATT is to set up the World Trade Organisation (WTO), and the Dunkel Proposal is to be incorporated in the text of the Agreement on the implementation of sanity and phytosanity measures, which inter alia lays down the principles of risk analysis, regionalisation, harmonisation, equivalency and transparency.
The increasing globalisation of trade in livestock and animal products increases the chances of the spread of diseases. In response to this prospect of trade deregulation there is now an imperative need to establish mechanisms which can be used to speed up international trade but at the same time protect the animal health situation of the countries involved.
Food borne disease is generally recognized as a major human health problem and an important cause of decreased economic productivity in both developed and less developed countries. Despite this, there is very little information available on the true level of exposure of specific populations to potential hazards, particularly in the case of bacterial diseases transmitted by consumption of meat and meat products. Attempts to quantify human health risks consequent to exposure to food borne hazards largely rely on extrapolation of information gained from individual disease outbreak investigations to the population at large.
In the case of meat hygiene, the qualitative recognition that unseen microbiological and chemical contamination rather than grossly apparent abnormalities are now the most important sources of hazards to human health has led to increasing demands for a more systematic regulatory approach to combat these hazards. In particular, dependence on traditional meat hygiene programs that focus the largely majority of resources on routine ante- and postmortem inspection is now recognised as being inadequate.
A wider recognition of the high level of complexity of food safety issues and increasing demands from consumers for maximum protection are other factors forcing regulatory authorities to adopt a more systematic and scientific approach to meat hygiene. Commensurate with these changes, protection of food from contamination, spoilage, and adulteration is no longer limited to being primarily a domestic issue; regulatory authorities must respond to increasing demands for facilitation of international trade.
According to the International Animal Health Code (IAHC) of the OlE (7):
'The principal aim of import risk analysis is to provide importing countries with an objective and defensible method of assessing the risk associated with the importation of animals, animal products, animal genetic material, feedstuffs, biological products, and pathological material The analysis should be transparent in order that the exporting country may be provided with a clear and documented decision on the conditions imposed for importation, or refusal of importation. Import risk analysis is preferable to a zero-risk approach because it provides a more objective decision, and enables Veterinary Administrations to discuss any differences in conclusion, which may arise concerning potential risks. "
Risk analysis. Risk analysis is based on an amalgam of scientific and technical information, and social and political policy decisions. Thus it is an applied rather than a theoretical science. The elements of risk analysis are risk assessment (RA), risk management (RM), and risk communication (RC).
RA is the primary scientific process and is regarded as the estimation of the likelihood (probability) and severity (magnitude) of harm or damage resulting from exposure to hazardous agent or situations. In the ideal situation, the RA process would be restricted to the "value-neutral" (non-normative) assessment of purely objective scientific data generated from alternative courses of action.
Scientific value judgements and policy choices are inevitably involved at some decision points in the RA process. Thus scientific experts assembling RA data should be guided by clear policy directives when any value judgements affecting the outcome of the RA are made. Determining this "RA policy" is an interactive process and examples of decision points where policy guidelines are necessary are: range of hazards included in the primary hazard identification, judging the scientific adequacy of the data set that is available, treatment of uncertainty, and deciding on the statistical basis for the standard of proof The treatment of uncertainty is an "RA policy" issue that has particularly important implications; should the "worst-case", "best-case", or mean of the range of uncertainty by chosen?
RM is concerned with development and selection of policy options for the purpose of decision-making, and the implementation of the regulatory programs that is developed from the RA. The options that are considered may be quantified solely in economic terms and RM decisions made according to some risk balancing standard, e.g., the risk-cost benefit analysis producing the highest benefit/cost ratio. However, RM decisions often have to be made in the face 0 f significant scientific uncertainty and the values that are considered may not be reduced to monetary values alone.
Risk managers must make a choice about what is an "acceptable" level of risk. The simplest technique is a risk-cost-benefit approach in those situations where all RM options can be reduced to, and quantified in, economic terms. Risk analyses involving human values will require use of other methodology such as threshold, comparative, "zero-risk", or as-low-as-reasonably achievable (ALARA) risk standards.
Recognition of RC is a vital part of the risk analysis process. The results of RA and RM need to be effectively communicated both within and between regulatory authorities and to the public. Additionally, RM decisions in animal health import risk analysis are likely to depend on the trustworthiness of information likely to depend on the trustworthiness of information gathered in another country on an on-going basis, focus on means of reducing risks associated with imported animals or animal products, and be limited to estimating the probability of the hazard being realised rather than including an estimation of the severity of such an outcome.
In discussing risk analysis, it should be recognised that while uniform principles of RA and RM can be pursued, uniform conventions (such as levels of statistical significance) are not necessarily advisable in deciding what level of proof is acceptable for policy purposes. As an example, it may be appropriate to rely on "most likely" estimates of risk when evaluating a chemical that is essential to a beneficial societal activity (e.g., use of radio nuclides in medicine), whereas a "worst-case" estimate may be appropriate when evaluating a nonessential chemical.
Why risk analysis in meat hygiene?
Meat hygiene programmes are primarily engaged to ensure that meat and meat products are "safe and wholesome". In the case of raw meat, this is only a qualitative measure of freedom form hazards to human (and animal) health (51). Postmortem meat inspection cannot guarantee freedom from all grossly detectable abnormalities, and sampling programmes have limited ability to detect randomly occurring violative levels of chemical residues and contaminants. More importantly, some degree of inadvertent microbiological contamination is inevitable in the slaughterhouse environment.
Risk assessment. In broad terms, the major "hazards" detectable at postmortem meat inspection are identified during observation of tissues. Following removal of the most important hazards, incremental benefits decrease as the level of inspection intensity increases. The optimum usage of postmortem inspection occurs when the incremental gain in benefits (in the broadest sense) equals the incremental increase in costs. Thus the optimal use of inspection resources does not eliminate all hazards, but removes all important hazards and ensures that any residual hazards are minor in nature and exist at a prevalence that constitutes a "negligible" risk to the consumer. A well-designed RA model can provide a quantitative basis for comparative evaluations, thus delivering scientifically appropriate information for RM decisions. The four analytical steps of the general health RA model can be suitably modified:
--Hazard identification: All hazards that could be present in the tissues of interest and that could be detected by organoleptic inspection procedures need to be identified. "Hazards" in meat hygiene include public health hazards, animal health hazards, and aesthetic defects that are unacceptable to the consumer;
--Hazard characterisation: The dose/response relationships that are developed from laboratory animal trials to assess chemical hazards are inappropriate for the characterisation of gross abnormalities detectable at postmortem meat inspection. Therefore all conditions that may possibly be present in the final product and that can be detected by postmortem inspection procedures are considered;
--Exposure characterisation: Exposure of the human population to "hazards" in meat that should have been detected by the procedures under investigation is very dependent on the particular processes and conditions that apply prior to human consumption. Despite this, the "worst-case" exposure characterisation must assume that the consumer will be exposed to all hazards that are capable of being detected by organoleptic meat inspection but which escape the inspection procedures in place. Thus the establishment of the performance attributes of individual procedures (sensitivity, specificity, nondetection rate) allow a quantitative characterisation of exposure. The nondetection rate can be defined as the proportion of abnormal tissues among those classed as normal by the individual procedure, and is equivalent to (1 minus the negative predictive value of the test);
--Risk characterisation: A consideration of the difference between nondetection rates for all identified hazards for each procedure, together with a scientific assessment of the consequences of each difference, provides the basis for the risk characterisation the investigator must consider the importance of all individual "hazards" that are missed during the application of particular procedures on a case-by-case basis. However, in the case of tissues that are not destined for human consumption, the only hazards of significance are those that serve as an indicator function for other tissues, or those which may have implications for animal health;
Risk management. Decision-making criteria for establishing an acceptable level of risk for postmortem meat inspection programmes may be complex. With the realisation that even high- intensity routine inspection procedures are neither 100% sensitive nor specific, RM decisions should focus on the comparative performance of the different procedures under test. It is neither technically feasible nor cost-effective for current meat inspection systems to eradicate all potential hazards from fresh meat produced for human consumption. Thus a "zero risk" approach to RM is inappropriate.
An important issue in RM is a consideration of scenario trees for meat and meat products. The scenario tree starts with an initial event and charts the functions that can affect the outcome of the initial event. Construction of a scenario tree collectively describes the risk model, and calculation of the likelihood of each of the risk scenarios (and their aggregation into an overall quantification of risk) can be achieved by several statistical methods. The use of probability density formats is gaining in acceptance (60) and in this case, quantification of each scenario parameter depends on expressing each numerical value as a probability curve against all possible values. PC software programmes such as @RISK (Palisad Corporation, New York) can use more elemental probability distributions, e.g., triangular distributions based on minimum most likely and maximum estimates, to carry out iterative calculations for RA.
Constructing scenario trees for risk analysis of raw meat and meat products is not a simple task, and should be undertaken on a case-by-case basis. Some process interventions remove public health hazards, e.g., freezing of meat to specified temperatures to kill parasitic cysts (59), whereas microbiological hazards may be potentiated by product abuse prior to consumption. In contrast, some aesthetic hazards may be rendered unrecognisable by further processing interventions. Difficulties in determining true levels of human exposure to microbiological hazards that may be associated with grossly detectable abnormalities (including amplification by cross-contamination), as well as the likely magnitude of effects (see later discussion), are compounded by the wide variation and unpredictability in events between processing and consumption for different meat products. In the absence of adequate dose-response data and the wide variability in outcomes in different individuals if transmission of an infectious agent occurs, RM will usually be limited to decisions based on the probability alone of the hazard(s) being realised under different operating systems.
It is likely that postmortem meat inspection activities that detect and remove grossly abnormal tissue contribute relatively little to safety in modem meat production systems (7, 15, 52, 54, 56). Unfortunately there is no RA model to assess the relative importance of these activities with those of process control systems that attempt to minimise inadvertent microbiological contamination during slaughter, dressing, and further processing. In the absence of any systematic data on the public health impact of current postmortem inspection procedures for poultry in the United States, a major study by the National Research Council concluded that the primary focus on organoleptic inspection should be shifted to quality assurance systems for microbiological and chemical surveillance (15). In a New Zealand context, an extensive bacteriological survey of grossly detectable abnormalities in lamb viscera revealed very few bacteria of public health importance; a subsequent literature review indicated that almost all potential meat borne zoonoses would be the result of unseen contamination with enteric pathogens, many of which have high asymptomatic carriage rates (54).
International activities. The need for systematic risk analysis of postmortem meat inspection programmes is now widely recognised internationally. Despite being addressed in principle, there are few international initiatives to translate this principle into generally agreed risk analysis methodology. National initiatives have led to some changes in domestic programmes but there is a concern that inadequate methodology may lead to criticism of such changes in the future (7, 15, 52). The widely recognised need to accept the equivalence of different national programmes where warranted, and harmonise international requirements for trade, also suffers from the lack of international initiatives in developing risk analysis models.
The recent redrafting of the codes of practice on all aspects of meat hygiene by the Codex Committee on Meat Hygiene (28) incorporates general principles for a risk analysis approach but it is beyond the purvey of this committee to initiate a working group to develop specific methodology. Notwithstanding this, the CAC has asked that all relevant Codex Committee describe the basis of any RA methods they utilise (23). International agreement on risk analysis methodology is also implicit in the GATT Uruguay Round SPS Decision (25).
Risk analysis of any postmortem meat inspection programmes would necessarily entail public recognition that some level of exposure to grossly detectable abnormalities is unavoidable. Regulatory authorities have avoided challenging the "zero-defect" concept of the consumer in the past, largely because they themselves have had very little scientific data with which to quantify their empirical knowledge. If regulatory authorities are to genuinely engage in a scientific and risk-based allocation of inspection resources, they will need to develop particular skills in communicating to the public the residual risks inherent in all components of any meat hygiene programme.
Routine monitoring and surveillance for chemicals, contaminants, and residues in meat and meat products (" chemical substances") constitutes a major element of meat hygiene programmes. Food additives will have been deliberately added during processing at levels up to those allowed by food standards whereas other chemical substances may have entered the food chain at any stage of production or processing.
Unlike the situation with on-line postmortem inspection, risk analysis in one form or another is relatively well established in the development of standards and guidelines for chemical substances in meat and meat products. This is largely consequential to the application of "risk analysis" in the general area of chemical contamination of foods, and the importance of meat as a dietary component in average daily food intake calculations. However, in some specialised applications in meat hygiene, e.g., residues of veterinary drugs, formal risk analysis is a relatively recent development.
Risk analysis of chemical hazards has primarily been used to establish maximum permitted limits in target tissues or meat products. In the international area, the CAC and its expert groups (the FAO/WHO Joint Expert Committee on Food Additives [JECFA] and the Joint Meeting of the FAO Panel of Experts on Pesticide Residues in Food and the Environment/WHO Expert Group on Pesticide Residues [JMPR]) have emerged as the most suitable bodies for facilitating the harmonising of food, health, and sanitary regulations for chemical substances worldwide (27,65). By comparison, the sampling plans for monitoring chemical substances in meat and meat products are rarely evaluated on a risk analysis basis. In the absence of 100% monitoring (c.f., postmortem meat inspection), the statistical adequacy of sampling plans is an essential component of a systematic risk-based approach. As a parallel activity, the scientific validity of regulatory responses when violative levels for specific substances are detected (domestically or at port-of-entry inspection) has rarely been subjected to risk analysis.
Risk assessment. Prior to toxicological evaluation, data is required on the chemical structure, stability, presence of impurities, and breakdown products of the compound to be assessed. This data facilitates identification of the chemical to be used in animal toxicity tests, and the type of studies to be performed.
In general terms, the safety assessment of chemical substances in meat and meat products should contain the elements of the four analytical steps described for all health RAs (6). Differing methodologies have been developed in different countries but the safety assessment of all chemical substances is largely based on the results of high-dose toxicological studies in laboratory animals, and precepts about what might occur at lower doses (14, 62, 69, 75). Detailed descriptions of the methodologies that are used are widely available and the dose-response part of the RA is based upon the scaling up of animal data to humans. Relevant biological data include biochemical tests; acute and chronic toxicity studies; special studies, e.g., testing possible neurotoxic, reproductive, or mutagenic effects; and observations in man. Metabolic studies may be used to complement extrapolation of laboratory animal findings to man.
The no-observed-effect-level (NOEL) determined from laboratory animal studies approximates a threshold level below which an adverse health effect will not ordinarily occur (14, 16, 17). (This assumes that a threshold dose actually exists and no toxic effect will occur below this). In most evaluations, the NOEL determined in the most sensitive species is divided by a safety factor so as to compensate for uncertainties in the scientific process. However, a shortcoming of this step is the inability to take the shape or slope of the dose-response curve into account.
The "safe" does is established as an acceptable daily intake (ADI) for a food and is expressed on a body weight basis. This dietary intake is not expected to result in any adverse health effects over the lifetime of an individual in the general population. In the case of contaminants inadvertently present in food, "provisional tolerable daily (or weekly) intakes" that denote permissibility rather than acceptability are calculated.
The total intake of a chemical determines exposure. for most safety evaluations of chemicals occurring in food, diet will constitute virtually the entire avenue of exposure. If a chemical is highly toxic but exposure is very low, the risk will be successfully ameliorated. In contrast, long-term dietary exposure to large amounts of a chemical of low toxicity may represent a significant level of risk. Information from national dietary intake studies (22) is used to evaluate whether a proposed maximum level for a chemical substance in a particular food is toxicologically supportable in terms of the cumulative intake in all foods.
Safety evaluations carried out in this manner cannot be regarded as a quantitative measure of risk. Although the approach contains some of the elements used in a formal health RA, the ADI end-point is derived by imposing a specific margin of safety. The use of safety factors has the advantage of preventing problems that may be associated with determining an acceptable level of risk against which a quantitative RA would have to be compared.
Evaluations of residues of veterinary drugs present a specific departure from the general methodology of safety evaluations of chemical hazards in food. Fixed assumptions are made about dietary intake and these are used to characterise the ADI (14). The number of tissues in which veterinary drug residues are found is limited, and intakes in the upper range limits for edible tissues (e.g., 300 g of muscle) are chosen. Maximum residue levels (MRLs) are calculated by using the ADI and the selected intake factors. Use of the drug according to good veterinary practice yields :MRLs that are compared with the potential :MRLs derived from the ADI.
True quantitative risk assessment (QRA) has a specific application in the safety evaluation of chemical substances that have carcinogenic potential. QRA begins with identification of those chemicals which may pose a human cancer or developmental hazard, and involves characterising the nature and strength of the evidence of causation (3, 70). Most methods use a "weight-of-evidence" approach and the classification will be of primary importance if it dictates the nature of the QRA to be carried out, e.g., chemicals placed in Group A and B by the United States Environmental Protection Agency (EPA) are always assessed using upper bound estimates of risk and worst-case default options (4). Chemicals placed in Group C are assessed on a case-by-case QRA basis.
QRA utilises a mathematical extrapolation to fit the observed data (usually derived at high-dose levels) to the expected dose-response at low-dose levels (2, 29, 30, 70). The available models evaluate the slope of the dose-response curve but are unable to take account of the biological factors that may modify the response at low levels, thereby having an influence on the calculation of excess lifetime cancer risks for humans. QRA also has been used in the United States as a method to develop cancer potency estimates (Q*) (16), but this application is now controversial (70). The outcome of QRA for carcinogenic chemicals is to estimate a "virtually safe dose" that correlates with an excess over background risk that is acceptable to society, e.g., 1 X 10-6 cases of cancer per lifetime of daily exposure to the calculated amount of the chemical in foods.
While QRA methodology offers considerable potential for improving RAs, the current inability to reliably model the underlying biological mechanisms of carcinogenicity suggest limited usefulness. In this respect, recent availability of human epidemiological data indicates that in some cases the QRA dose-response model may grossly overestimate the actual cancer risk (70), and in certain circumstances it has been suggested that application of the NOEL/safety factor approach would be a more rational approach to evaluation of non genotoxic carcinogens (38). Newer quantitative approaches include physiologically based pharmacokinetic models and biologically based cancer models that provide the prospect of more accurate scaling up of laboratory animal data to estimate human risk (70).
Risk assessment policy decisions. There are many RA policy decisions embodied in both the safety factor and the QRA approach to safety evaluations of chemical substances in food and some examples are given below. Scientific value judgements are implicit in the evaluation of nonuniform data sets and the determination of toxicological and end-points on a case-by-case basis. Technical concerns may also intrude at different decision steps.
The application of safety factors to the NOEL represents specific mechanisms to address uncertainty and create conservative margins of safety proportional to the development of a "no risk" level of exposure (83). The respective values of the safety factors that are used are arbitrary and have no measured biological significance; however, the value of the safety factor chosen in a particular evaluation has a marked effect on the ADI that is set. Notwithstanding this, their appropriateness is somewhat borne out of empirical experience.
In carrying out an exposure assessment in the safety evaluation for residues of veterinary drugs, the predicted dietary intake that is used is an upper limit value, an additional safety factor affecting the final MRL. The use of an upper-bound estimate of withdrawal in the animal also incorporates a safety factor, mean exposure levels being much less (14). Choice of these statistical parameters represents a particular RA convention, and contributes to characterising exposure as a "worst case" scenario.
Comparison of potential MRLs with MRLs established from use of the veterinary drug in field trials forms the basis for recommended MRLs. If concentrations of residues lower than the potential MRLs are found in field trials, the recommended MRLs are reduced accordingly. This is an RA policy option that is unique to the evaluation of veterinary drug residues (51). In the case of pesticide residues, the MRLs are usually established at the levels resulting from use of the pesticide in accordance with Good Agricultural Practice.
Risk Management. There are many situations where social benefits and economic need as well as human safety are taken into account when elaborating maximum permitted levels for chemical substances in food. It also will be apparent from the above discussion that separation of RA and RM can be difficult to achieve and the basis for RM decisions varies both between countries and within countries. In the US, cost-benefit analysis is required in risk analyses for pesticides whereas it is prohibited in risk analyses for standards for drinking water and occupational health (35). In Europe, risk analysis is the foundation of most health and safety standards but the EC Drinking Water Directive (80/778/EEC) sets virtually zero limits for individual pesticide residues (0.01 ppb) (81). These limits bear no relationship to a toxicological risk to consumers and RM essentially constituted a political decision.
RM of contaminants and natural toxicants often embodies consideration of the food's nutritional value as well as the chemical substance's toxicity and the extent to which it can be controlled. In setting an acceptable level in food, there may be explicit consideration of the consequences of this level on the quantity and price of the food supply (33, 45, 83).
The ability to set irreducible levels for chemical substances that are based on feasibility rather than safe numerical limits (14) is an RM option (51). For contaminants, the irreducible level usually represents the concentration of a substance that cannot be eliminated from a food without discarding that food altogether. Such is the case for mycotoxins where the data does not allow "safe" numerical limits to be set (46). However, it is difficult to achieve the consensus needed to develop international guidelines when maximum permitted levels are set in this manner.
The availability of analytical methods can be an important RM component in the setting of MRLs. If a practical analytical method to measure veterinary drug residues under the conditions of use is unavailable, the recommended :MRLs are raised so that compliance with them can be checked.
QRA models for carcinogens in food generate a numerical estimate of risk to be used in decision-making. Legal decisions over acceptable levels of risk inevitably require accommodation between law and science (34). In recent court decisions in the US, the legal interpretation of "safe" does not mean risk-free, and "acceptable risk" involves a judgemental determination based on three factors: the statutory basis, the scientific data, and the "risks that are acceptable in the world in which we live" (34).
In Europe, there is increasing use of "unacceptable," "tolerable," and "acceptable" levels of risk in regulatory decisions (1). "Unacceptable" represents exposure that is not acceptable on any reasonable basis whereas "tolerable" represents exposures that are not welcome but can be reasonably tolerated. "Acceptable" means that exposures can be accepted without further
improvement, i.e., when protection has been optimised. The Royal Society's view is that the annual fatal cancer risks per year that are represented by these terms are greater than 3 X l0-5, between 3 and 1 X 10-5, and I X 10-5 per year from a single source respectively (42).
In the Netherlands, the unacceptable level has been set at 1 X 10-6(1). Within the "tolerable" range considered conditionally acceptable, the question remains as to what extent hazards should be reduced in the light of social and economic factors. For pesticides evaluated in the United States, the judgment is made against a negligible risk standard of less than 1 X 10-6(1). Within the "tolerable" range considered conditionally acceptable, the question remains as to what extent hazards should be reduced in the light of social and economic factors. For pesticides evaluated in the United States, the judgement is made against a negligible risk standard of less than 1 X 1-10-6 additional cases of cancer over a 70-year lifetime (16). If dietary risks fall between 1 X 10-4 and 1 X 10-6, further studies are undertaken and there may be explicit consideration of benefits. If an RA is properly conducted, not only will exposures at or below the :MRL provide the level of safety desired, but exposures at levels higher than the :MRL will also provide some measure of safety (16).
Monitoring and surveillance. Establishment of maximum permitted limits for chemical substances in food involves only one aspect of a comprehensive risk analysis approach. Inclusion of monitoring and surveillance as an element of risk analysis provides a "risk profile" of potentially hazardous substances and a means of focussing limited analytical resources where they will have most benefit. In the case of international trade, "risk profiles" can only be fully utilised for risk analysis purposes if there is a complete and systematic exchange of information, coupled with continual updating of epidemiological data (26). The US Pesticide Monitoring Improvements Act of 1988 (18) mandates the Food and Drug Administration (FDA) to establish pesticide usage and other data in countries of origin of imported foods, thus adding to the information gained by analytical testing at port-of-entry.
The performance of sampling plans and analytical methods as they pertain to risk analysis of chemical substances in food is beyond the scope of this discussion. In general terms, sampling plans identify trends but are unlikely to be sufficient for prevention and are unable to cope with rapid change. Thus short-term ad hoc programmes may be needed to identify the presence of newly identified hazards or to establish whether small numbers of violative samples represent a significant risk. Sampling plans need to be linked to systematic traceback systems and establishment of quality assurance programmes at the point of entry of a hazard into the food chain. Rapid success in achieving a marked reduction in sulpha drug residues in bobby veal in New Zealand is a good example (13).
Accept/reject criteria. Following the detection of a violative level in an inspected "lot", the regulatory authority must decide on what is an appropriate action. There is a range of possible options, including outright rejection, intensified sampling of the same lot, intensified sampling of further lots or consignments of the same commodity, and recourse to monitoring data to gain more complete information on the extent of the problem (32). Decision making criteria should be based on systematic risk analysis, particularly in the case of chemical hazards in fresh meat and meat products, e.g., the heterogeneous origin of "lots" should engender a different approach to the regulatory response for a single violative test than would be the case for more homogeneous commodities.
Because of widely varying levels of dietary intakes of particular foods, it is probable that some individuals in the population will exceed the ADI for a chemical substance to some extent and for some limited length of time. Quantitative data on the health risk of these incursions is generally not available but the significance of any minor incursion above the ADI can be put into context by an understanding of the scientific basis upon which the standard was established, i.e., by reference back to the animal test data and the NOEL that gave rise to the ADI for the particular substance. In this respect, the "worst case scenario" for exposure that has been the recent historical basis for RAs for chemical substances is under challenge (70,81) and researcher developing new RA methods are already showing that in several cases the severity of human health hazards has been overestimated (70). A systematic risk analysis approach to decision-making in the event of detection of very low numbers of violative levels of chemical substances would include consideration of the precision of the NOEL (primarily for acute toxic effects), the steepness of the dose-response curve, the likelihood of acute toxicity, the likely extent of individual exposure consequential to the violative level, and the outcome of any subsequent, intensified sampling plan over a specific period of time.
Risk analysis is necessary because imposition of specific accept/reject criteria can have major economic (and political) importance, especially in international trading situation. An important question is: what effect will the regulatory decision have on reduction of the risk? A RM option in cases of violative levels of hazardous chemicals that are the consequence of illegal use or bad agricultural practice should include a punitive response, even though a low level of violations would be unlikely to have any effect whatsoever on human health.
International activities. Risk analysis of chemical substances in meat and meat products is inevitably tied to that in all foodstuffs, and a major responsibility of the CAC is to elaborate MRLs for chemical substances in all foods. Expert groups, primarily JECF A and JMPR, consider scientific data and make recommendations on food standards on a case-by-case basis to the relevant Codex committees. The committees consider a range of "equity and ethics" issues as well as the recommendations from the expert groups when elaborating draft Codex standards. The consensus modality governing decision-making at a committee level contains no formal elements of RA or RM.
A number of general recommendations on the future risk analysis activities of JECF A and JMPR were developed at the 1991 FAO/WHO Food Conference (27) and these included the need to harmonise the methodology used by different countries, establishment of internationally agreed-upon principles for the RA of substances that have been shown to be carcinogenic in animal studies, and ensuring transparency of the decision-making process. The Secretariat of the Joint FAO/WHO Food Standards Programme has followed up this initiative with the commissioning of a report on the use of risk analysis by JECF A, JMPR, and the relevant Codex committees (51).
Other wide-ranging initiatives are taking place under the auspices of the World Health Organisation, including those of the International Programme on Chemical Safety (IPCS) and the Intergovernmental Mechanism for Chemical Risk Assessment and Management that resulted from the United Nations Conference on the Environment and Development held in Rio De Janeiro in 1992 (83). The IPCS has recently begun a study on "Guiding principles and methodology for quantitative risk assessment in setting exposure limits" with the goal of harmonising risk analyses at the national and international level and work is also being undertaken by the Organisation for Economic Co-operation and Development to harmonise RA methodology for pesticides (58).
In the US, the EP A has sponsored a federal interagency working group to improve scientific interagency working group to improve scientific methods of RA so as to harmonise approaches, to reduce uncertainty, and to develop an inventory of existing databases and information needs(30).
International harmonisation of MRLs is a stated objective of the CAC and this depends on reducing national differences where those differences are not justified, and mutual recognition of comparable standards employed by different countries. Notwithstanding the above-mentioned international initiatives, work on risk analysis of accept/reject criteria for violative level of chemical substances detected in meat and meat products at port-of-entry testing is required and this is the purvey of the newly formed Codex Committee on Import inspection and Certification (32). Systematic exchange of information on monitoring and surveillance programmes in the country of origin will constitute an important element in this risk analysis activity.
Despite recognition of the problem, allocation of regulatory resources commensurate with the importance of microbiological contamination of meat and meat products is only just beginning to be addressed (7, 26, 43,45). Control of this source of hazards has generally depended on a traditional approach, i.e., ensuring that raw materials are as free of specific hazards as possible; keeping microbiological contamination during slaughter, dressing, and processing to the lowest practicable level possible; and preventing any subsequent growth during further processing or consumer activities. HACCP programmes are specifically designed to enhance achievement of these objectives and the CAC is encouraging their use as a means to assure food safety, to better utilise inspection resources, and to provide a more timely response to problems (27). However, there has been only limited uptake of HACCP in meat production systems (52) and this is largely a consequence of the specific adaptation that is required.
Quantitative evaluation of the microbiological safety of foods has primarily been dependent on the establishment of microbiological criteria as standards, guidelines, or specifications. There is considerable debate over the application of microbiological criteria to classify food as microbiologically acceptable or unacceptable and many standards and guidelines have proven to be impractical (5, 8, 9, 39).
The principles for the establishment and application of microbiological criteria for foods include: listing of hazardous micro-organisms, qualitative characterisation of likely exposure, evaluation of methods available for detection an quantification, and design of sampling plans (5,8,9). Any criteria that are elaborated must be effective and practical and a cost-benefit analysis should be a basic component in the development of a mandatory standard. It is apparent that application of these principles requires elements of a "risk analysis" approach in some form or another. Additionally, a direct and statistically based link is required between the adequacy of sampling plans and the severity of any microbiological criteria imposed (9), whereas such linkages are not currently employed with respect to MRLs for chemical substances in food. Decision criteria applied to a lot should be "administratively and economically feasible" and should take into account the heterogeneity of distribution of micro-organisms (9).
To date it has proved impractical to elaborate microbiological criteria that adequately define the safety and wholesomeness of raw meat (8, 9). WHO has stated that the "establishment of microbiological criteria for raw foods in general cannot serve the purpose of protecting the health of the consumer when the main source of the pathogenic micro-organisms is the raw food itself, and when processing does not include steps which will eliminate or substantially reduce the numbers of these micro-organisms" (39). This is clearly the case for raw meat and meat products. The US National Research Council has similarly reviewed data on the occurrence, potential for causing infection, and pathogenicity for humans of bacterial species known to be present on chicken and has reached a parallel conclusion (15). The research committee stated that "minimising microbial contaminants on chicken is a worthwhile objective, but it is premature to establish formal microbiological criteria for classifying raw products of poultry as microbiologically acceptable or unacceptable" and concluded that the data required to justify such formal regulatory standards do not currently exist.
At this time, genetically engineered micro-organisms are not regarded as fundamentally different to those that have been isolated from nature (and introduced into new environments) or that have been produced by breeding and selection (10); thus they have not engendered a specific RA approach.
Risk analysis. The problems associated with formal analysis of the risk of foodborne microbiological disease are very different to risk analysis of foodborne chemical hazards. Micro-organisms multiply and die and the biological interactions are complex. In the case of me at, the characteristics of contamination during slaughter and dressing dictate the character of the initial microflora but this can be markedly modified by subsequent events. Additionally, there are marked differences in the virulence and pathogenicity of animal and environmental strains for humans, and the interaction of host and microbiological systems that a prediction of exposure should not lead to an automatic assumption of risk (9,72,76).
If an uncritical assumption is made that a human health hazard exists because of the presence of particular contaminating microflora, it would be tempting to try and characterise that risk by exposure assessment. One approach would be to construct a numerical dose-response curve for each potential pathogen that may be present in the final product and attempt to characterise risk in these terms. However, experience in ecological RA would suggest that irrespective of the difficulty in gaining the quantitative data, developing this additive organism- by-organism approach would be very difficult (71, 76, 82). In the case of meat, the biological interactions in the microflora that occur after the product leaves the slaughter- or packhouse are not able to be quantified with any certainty and as stated above, prediction of exposure should not automatically lead to an assumption of a human health risk.
Notwithstanding the problems mentioned above, microbiological hazards in food can be subjected to a formal risk analysis process, with data generally being generated from clinical and epidemiological studies in humans, and surveillance. The best probability estimates would come from a "perfect" epidemiological study on the human population of interest at the range of doses or exposures of interest (64). Unfortunately, such studies rarely exist. Estimates of risk derived from epidemiological studies are therefore often quantified in terms of relative risk or attributable risk.
Development of quantitative microbiological RA's is in its infancy but will probably increase in the future; this may lead to establishment of more meaningful microbiological criteria in terms of human health risks. Despite the complex challenges of microbiological risk analysis, a quantitative RA model has been developed for waterborne disease in the United States (72). This work was based on a number of dose-response relationships derived from human experiments with a range of micro-organisms. A number of assumptions were made, including those of homogeneous distribution in water, average daily water intake, and an "acceptable" level of risk. Such work was only possible because of the low pathogenicity of the micro-organisms involved and is unlikely to be repeated for other microbiological risks. Thus future microbiological RAs are likely to have a more qualitative base.
The choice of a human health endpoint is very different for microbiological hazards compared with chemical ones and this would be an important RA policy decision in a theoretical quantitative microbiological RA. Possible outcomes of microbiological contamination are true exposure, infection, disease, or death. Positive diagnostic tests in epidemiological investigations are often indicative of infection rather than disease and undifferentiated RM decisions based on an outcome that had no consequence in terms of health would be wasteful of resources that could be better used elsewhere (44, 76).
A consideration of the DIE guidelines for import risk analysis (31) provides an interesting comparison with attempts at RAs for microbiological hazards in foods. A limited number of well-documented animal diseases make up the DIE List A and List B, and adequate data on the prevalence of a specific disease in the exporting country (country factor) is usually available.
Similarly, an adequate estimate of the probability of the specific disease agent being present at the time of import (country factor X animal import unit/animal product) can usually be determined. Estimation of the probability of exposure to animals or humans in the importing country and the likelihood of transmission is achieved by constructing scenario trees; this results in an unrestricted risk estimate.
Currently, RAs for microbiological hazards in fresh meat and fresh meat products are very unlikely to be able to draw on equivalent quantitative data to that described above for specific pathogens of animal health importance. The absence of detailed knowledge on the prevalence and specific zoonotic potential of the wide range of bacterial species commonly found as inadvertent contaminants on fresh meat, coupled with very limited dose-response data for those pathogenic strains known to be transmitted to humans by digestion, makes microbiological RA for public health hazards a difficult proposition. In addition, a microbiological RA for meat at a particular time of importation/distribution is only meaningful if the subsequent measures that are taken maintain the same microbiological quality.
International activities. Current risk analysis of microbiological hazards is primarily inhibited by lack of information and lack of a detailed conceptual framework. However, these problems are not intractable and national initiatives to address these issues in the general area of food safety are underway. Several. countries have recently embarked on studies to gather microbiological baseline data on dressed carcasses as a first step to provide quantitative input to a "risk analysis" approach to meat hygiene (43,85). Both pathogens and indicator organisms are being investigated but these research initiatives are not expected to provide enough comprehensive data to allow development of microbiological RA models in the short term. Nevertheless, it is important that the long-term goals of this research in different countries are strategically aligned. Food safety microbiological research is both research-intensive and time-consuming, and information-sharing in the initial stages of the above-mentioned research would be of marked benefit. Development of a substantial conceptual framework is a prerequisite for successful microbiological risk analysis. Development of HACCP systems for meat and meat products needs to be closely aligned with development of risk analysis methodology for microbiological hazards. A number of HACCP initiatives for raw meat appear to be based on a systematic application of traditional parameters of hygienic practice rather than a microbiological data base that validates the design of the HACCP plan. In addition, there is very little quantitative information available on the effectiveness of HACCP in reducing foodborne illness in the human population. It is assumed that operating procedures and process interventions that reduce overall microbial loads on carcasses will reduce overall microbial loads on carcasses will reduce public health risks; however, this assumption has been criticised by those who point to a general lack of correlation between total microbial counts and specific pathogens. Counter to this criticism, it should be realised that our current knowledge of micro-organisms capable of causing human disease is far from complete. As an example, only about half of the human cases of presumed infectious diarrhoea are of known aetiology (76) and thus it seems reasonable to assume that a higher level of microbial contamination of gastrointestinal origin on fresh meat will result in a higher level of human exposure to potential pathogens.
In the case of slaughter and dressing, initial research in New Zealand suggest that traditional organoleptic parameters may not be related to microbial loads on carcasses and monitoring of specific CCPs in these terms could be fallacious (36). If the additive marginal risks that microbiological contamination imposes at different CCPs are to be ranked and evaluated, together with an evaluation of the cost-benefit of reducing these risks, detailed microbiological data are required. Statistical limitations of organoleptic monitoring programmes for process control in raw meat production systems also impact on the scientific validity of the HACCP plan.
As science advances and the regulator's ability to detect risks improves, the opportunities for influencing risks have proliferated and a desirable goal for society is to develop systematic rules for decision-making across the entire spectrum of risks (86). Within the spectrum of risk associated with foodborne hazards, regulatory authorities need to meet the challenge with quantitative, unambiguous RAs that have a transparent and readily understandable methodology. International cooperation in food safety and harmonization of world food regulations are essential in today's environment, but this demands resources that are only now becoming available. However, there is an enormous diversity in possible food-related hazards and therefore it is unlikely that a single risk analysis approach can be developed that will suit all situations (44,74). In emerging as a "regulatory science", risk analysis offers a new opportunity to facilitate achievement of modern meat hygiene goals. International objectives in the future regulation of veterinary public health risks should include:
Alignment of national regulatory agendas;
Systematic exchange of information so as to establish conceptual frameworks for risk analysis;
Development of internationally accepted RA methodology, and the opportunity for peer review;
Development of international policy on standards of acceptable risk;
Mutual recognition of comparable standards and specifications employed by national agencies, and reduction in those differences where risk analysis shows that those differences are not justified;
Development of risk analysis-based principles to govern regulatory responses to violative levels of chemical substances detected in foods at port-of-entry inspection.
Design and implementation of HACCP systems that are correlated with new developments in the field of microbiological RA.
1. Ale, J. M. (1992). Dealing with risk in environmental policy in the Netherlands. In International Conference on Risk Assessment. London, 5-9 October 1992. United Kingdom Health and Safety Commission, 410-417.
2. Ames, B. N. & Gold, L. S. (1990). Too many rodent carcinogens: Mitogenesis increases mutagenesis. Science, 249, 970-971.
3. Ames B. N., Magaw R. & Gold, L. S. (1987). Ranking possible carcinogenic hazards. Science, 236,271-279.
4. Anderson, M. E. (1988). Incorporating pharmacokinetics and risk assessment into the setting of occupational exposure limits. Appl. Ind. Hygiene, 3, 267-273.
5. Anonymous (1980). General principles for the establishment and application of microbiological criteria for foods. Report of the 17th Session of the Committee on Food Hygiene. Codex Alimentarius Commission, Rome.
6. Anonymous (1983). Risk assessment in Federal Government: Managing the process. National Research Council. National Academy Press, Washington, DC.
7. Anonymous (1985). Meat and poultry inspection: The scientific basis of the nation's programme. National Research Council. National Academy Press, Washington, DC.
8. Anonymous (1985). An evaluation of the role of microbiological criteria for foods and food ingredients. National Research Council. National Academy Press, Washington, DC.
9. Anonymous (1985). Microorganisms in Foods. 2. Sampling for microbiological analysis: Principles and specific applications. (2nd Ed.). International Commission on Microbiological Specifications for Foods. University of Toronto Press, Toronto.
10. Anonymous (1986). Coordinated framework for the regulation of biotechnology. Office of Science and Technology Policy. Federal Register 51, 23301-233-50.
11. Anonymous (1986). Determining risks to health: Federal policy and practice. United States Department of Health and Human Services. Auburn House, Massachusetts.
12. Anonymous (1986). Prevention and control of foodborne salmonellosis through the application of the Hazard Analysis Criteria Control Point System. Document VPH 86/65. World Health Organization, Geneva.
13. Anonymous (1987). Chief Veterinary Officer's Annual Report. Surveillance, 15 (3), 11.
14. Anonymous (1987). Environmental Health Criteria 70. Principles for the safety assessment of food additives and contaminants in food. International Programme on Chemical Safety and Joint FAO/WHO Expert Committee on Food Additives. WHO, Geneva.
15. Anonymous (1987). Poultry inspection: The basis for a risk assessment approach. National Research Council. National Academy Press, Washington, DC.
16. Anonymous (1987). Regulating pesticide residues in food: the Delaney paradox. Committee on Scientific and Regulatory Issues Underlying Pesticide Use Patterns and Agricultural Innovation, National Research Council. National Academy Press, Washington, DC.
17. Anonymous (1987). 31st Report of the Joint FAO/WHO Expert Committee on Food Additives (1987): Evaluation of certain food additives and contaminants. WHO Technical Report Series Number 789. World Health Organisation, Geneva.
18. Anonymous (1989). Pesticides Monitoring Improvements Act. United States Congress, Washington, DC.
19. Anonymous (1989). A review of the NAS study of the scientific basis for the nation's meat and poultry inspection programme. Special Task Force, American Meat Institute, Arlington, Virginia. Presented as testimony before the Agricultural Research and General Legislation Subcommittee of the Committee on Agriculture, Nutrition and Forestry, Unites States Senate, July 20.
20. Anonymous (1989). HACCP principles for food production. National Advisory Committee on Microbiological Criteria for Foods. United States Food Safety and Inspection Service, Washington, DC.
21. Anonymous (1989). Report and resolutions. In Proceedings of the 10th Symposium of the World Association of Veterinary Food Hygienists, Stockholm, 1-2.
22. Anonymous (1989). Report of the 21st Session of the Codex Committee on Pesticide Residues. Codex Alimentarius Commission, Rome.
23. Anonymous (1991). Report of the 38th Session of the Codex Executive Committee, Codex Alimentarius Commission, Rome.
24. Anonymous (1991). Draft principles and application of the Hazard Analysis Critical Control Point (HACCP) system. Report of the 25th Session of the Codex Committee on Food Hygiene, Washington, DC, 75-80.
25. Anonymous (1991). Draft text on Sanitary and Phytosanitary Measures and Barriers. GATT Secretariat, 20th December.
26. Anonymous (1991). Emerging issues in food safety and quality for the next decade. FDA Contract No. 223-88-2124. Federation of American Societies for Experimental Biology, Bethesda, Maryland.
27. Anonymous (1991). FAO/WHO Conference on Food Standards, Chemicals in Food and Food Trade. Joint F AO/WHO Food Standards Programme, Rome.
28. Anonymous (1991). Report of the 6th Session of the Codex Committee on Meat Hygiene. Codex Alimentarius Commission, Rome.
29. Anonymous (1991). Report on the 23rd Session of the Codex Committee on Pesticide Residues. Codex Alimentarius Commission, Rome.
30. Anonymous (1991). Risk assessment practices in the Federal Government. Federal Register 56, (204), 54580-82.
31. Anonymous (1993). Report of the 61st General Session of the Office International des Epizooties, Paris, 24-28 May.
32. Anonymous (1993). Report of the 1st Session of the Codex Committee on Food Import and Export Inspection and Certification Systems, Canberra 21-25 September 1992. 20th Session of the Codex Alimentarius Commission. Geneva, 28 June- 7 July.
33. Ashford, N. A. (1988). Science and values in the regulatory process. Stat. Science, 3, 377
34. Barnard, R. C. (1990). Some regulatory definitions of risk; interaction of scientific and legal principles. Reg. Toxicol. Pharmacol., 11,201-211.
35. Belzer, R. B. (1992). The use of risk assessment and benefit-cost-analysis in U.S. risk-management decision making. In International Conference on Risk Assessment. London, 5-9 October 1992. United Kingdom Health and Safety Commission, 421-442.
36. Biss, M. & Hathaway, S. C. (1993). HACCP and raw food microbiology, In Proceedings of the Veterinary Public Health Branch of the Australian Veterinary Association. Gold Coast, 17-21 May, 12-16.
37. Brunk, C.G. (1992). Issues in the regulation of animal health risks. Report to the Animal Health Division. Agricultural Canada, Ottawa.
38. Butterworth, B.E. & Slaga, T. (1987). Non-genotoxic mechanisms in carcinogenesis. Banbury Report 25. Cold Spring Harbour Laboratory. Cold Spring Harbour, New York.
39. Christian, I.H. (1982). Microbiological Criteria for Foods: Summary of recommendations of FAO/WHO Expert Consultations and Working Groups 1975-1981. WHO, Geneva.
40. Ciolfi, E. (1985).USDA and EEC postmortem inspection. Food Safety and Inspection Service. United States Department of Agriculture, Washington, DC.
41. Clarke, G.C. (1989). The hazard analysis critical control point (HACCP) approach to slaughter quality control in red meat abattoirs. In Proceedings of the 10th Symposium of the World Association of Veterinary Food Hygienists, Stockholm, 194-198.
42. Clarke, R.H. (1992). Radiation protection standards: a practical exercise in risk assessment. In International Conference on Risk Assessment. London, 5-9 October 1992. United Kingdom Health and Safety Commission, 173-182.
43. Cross, H.R. (1993). Statement before the Subcommittee on Livestock, House Committee on Agriculture, United States House of Representatives. Washington, DC, 16 March.
44. Crouch, E.A. & Wilson, R. (1982). Risk/benefit analysis. Ballinger Publishing Company, Cambridge, Massachusetts.
45. Denner, W.H.B. (1992) Food and drinking water safety: can risk assessment help us to get our priorities right? In International Conference on Risk Assessment. London, 5-9 October 1992. United Kingdom Health and Safety Commission, 190-203.
46. Dichter, C.R. (1987). Cost-effectiveness analysis of aflatoxin control programmes. Joint FAO/WHO/UNEP Second International Conference on Mycotoxins, Bangkok.
47. DiGiacomo, R.F. & Koepsell, T.D. (1986). Sampling for detection of infection or disease in animal populations. J: Ameri. Vet. Med. Assoc., 1989,22-23.
48. Drucker, P.F. (1973) Management tasks, responsibilities and practise. Heinemann, New York.
49. Harbers, A.H., Smeets, J.F. & Snijders, J.M. (1991). Predictability of postmortem abnormalities in shipments of slaughter pigs as an aid to meat inspection. Vet Quarterly, 13,74-80.
50. Hathaway, S.C. (1991). The application of risk assessment methods in making veterinary public health and animal health decisions. Rev. sci. tech. Off Int. Epiz., 10 (1) 215-231.
51. Hathaway, S.C. (1993). Risk assessment procedures used by the Codex A1irnentarius Commission, and its subsidiary and advisory bodies. Joint FAO/WHO Food Standards. Programme. 20th Session of the Codex Alimentarius Commission, Geneva, 28 June- 7 July.
52. Hathaway, S.C. & Bullians, J. (1992). The application of HACCP in a red meat slaughter and dressing operation. In Proceedings of the World Congress of Food borne Infections and Intoxications, Berlin, 895-898.
53. Hathaway, S.C. & McKenzie, A.I. (1989). Evaluation of postmortem meat inspection procedures for sheep slaughtered in New Zealand. Volume I. The viscera of lambs. New Zealand Ministry of Agriculture and Fisheries, Wellington.
54. Hathaway, S.C. & McKenzie, A.I. (1990). Ovine meat inspection and public health. In Proceedings of the 26th Meat Industry Conference, Hamilton, New Zealand, 52-56.
55. Hathaway, S.C. & McKenzie, A.I. (1991). Postmortem meat inspection programs: separating science and tradition. J: Fd. Protect., 54,471-475.
56. Hathaway, S.C. & McKenzie, A.I. (1988). The impact of ovine meat inspection programmes on processing and production cost. Vet. Rec., 124, 189-93.
56. Hathaway, S. C. & Pullen, M. M. (1990). A risk-assessed evaluation of postmortem meat inspection procedures for ovine thysanosomiasis. J: A mer. vet. med. Assoc., 196, 860-864.
57. Hathaway, S. C. & Richards, M. S. (1993). Determination of the performance attributes of postmortem meat inspection procedures. Prevo vet. Med. (in press).
58. Hill, R. N. (1992). Pesticide risk assessment in the United States. In International Conference on Risk Assessment. London, 5-9 October 1992. United Kingdom Health and Safety Commission, 162-165.
59. Hilwig, R. W., Cramer, J. D. & Forsyth, K. S. (1978). Freezing times and temperatures required to kill cysticerci of Taenia saginata in beef Vet. Parasitol., 4, 215-19.
60. Kaplan, S. (1992). The general theory of quantitative risk assessment--its role in the regulation of agricultural pests. Risk Assessment Workshop, Fort Collins, 20-24 July. United States Department of Agriculture Animal and Plant Health Inspection Service.
61. Kellar, J. A. (1992). Analysis and management of health risks in the international trade in animals and animal products. In Report of the 60th General Session of the Office International des Epizooties, Paris.
62. Kokoski, C. J. & Flamm, W. G. (1984). Establishment of acceptable limits of intake. In Proceedings of the 2nd National Conference for Food Protection, Washington, DC. Department of Health and Human Sciences.
63. Lathrop, J. & Linnerooth, J. (1983). The role of risk assessment in a political decision process. In Analysing and aiding decision processes. (P. Hupphreys, P. O. Svenson & A. Vari, eds.). North Holland, Amsterdam.
64. Lave, L. B. (1987). Health and safety risk analysis: Information for better decisions. Science 236,291-295.
65. Lyng, R. E. (1988). Food and agriculture in a global economy, Fd. Technol., 42 (9),115-116.
66. McMahon, J., Kahn S., Batey R., Murray, J. G., Moo, D. & Sloan, C. (1987). Revised postmortem inspection procedures for cattle and pigs slaughtered at Australian abbatoirs. Aust. vet../, 64, 183-187.
67. Murray, G. (1986). Ante-mortem and post-mortem meat inspection: an Australian Inspection Service perspective. Aust. vet. .J, 63, 211-215.
68. Ogunrinade, A. F. & Oyekole, O. D. (1990). Evaluation of the efficiency of beef inspection procedures for tuberculosis, fascioliasis and cysticercosis in a Nigerian abattoir. Prevo vet. Med., 8,71-75.
69. Paustenbach, D. J. (1989). The risk assessment of environmental and human health hazards: a textbook of case studies. John Wiley & Sons, New York.
70. Paustenbach, D. J. (1989). Important recent advances in the practice of health risk assessment: implications for the 1990s. Reg. Toxicol. Pharmacol., 10,204-243.
71. Payne, J. H. & Medley, T. L. (1992). Risk assessment for Federal regulatory decisions on organisms produced through biotechnology. In International Conference on Risk Assessment. London, 5-9 October 1992. United Kingdom Health and Safety Commission, 227-237.
72. Regli, S., Rose, B., Haas, C. N. & Gerba, C. P. (1991). Modelling the risk from Giardia and viruses in drinking water. J Amer. Water Works Assoc., 83,76-84.
73. Roberts, T. (1983). Benefit analysis of selected slaughterhouse meat inspection practices. Economic Research Service, United States Department of Agriculture, Washington, DC.
74. Rowe, W. D. (1977).--An anatomy of risk. John Wiley and Sons, London.
75. Russell, M. R. & Gruber A. (1987). Risk assessment in environmental policy-making. Science 236, 286-290.
76. Skinner, R. (1992). Microbiological risk assessment and public health. In International Conference on Risk Assessment. London, 5-9 October 1992. United Kingdom Health and Safety Commission, 204-209.
77. Skovgaard, N. (1981). Meat hygiene: Preslaughter problems and trends. In Proceedings of the 8th Symposium of the World Association of Veterinary Food Hygienists, Dublin, 3-8.
78. Slovic, P. (1987).--Perception of risk. Science 236, 280-285.
79. Snijders, J. M. A., Smeets, J. F. M., Harbers, A. H. M. & van Logtestijn, J. G. (1989). Towards an improved meat inspection procedure for slaughter pigs. In Proceedings of the 10th Symposium of the World Association of Veterinary Food Hygienists: Satellite Symposium, Stockholm, 22-25.
80. Stenholm, C. W. & Waggoner, D. B. (1989). A congressional view on food safety. J Amer. vet. med. Assoc., 195,916-21.
81. Thomas, B. (1992). An industry approach to the risk assessment of pesticides. In International Conference on Risk Assessment. London, 5-9 October 1992. United Kingdom Health and Safety Commission, 166-172.
82. Tiedje, J. M., Colwell, R. K., Grossman, Y. L., Hodson, R., Lenski, R. E., Mack, N., & Regal, P. J. (1989). The planned introduction of genetically-engineered organisms: ecological consideration and recommendations. Ecology 70,298-315.
83. van der Heijden, K. A. & Stem, R. M. (1992). The role of risk assessment in the work of the World Health Organisation in Europe. In International Conference on Risk Assessment. London, 5-9 October 1992. United Kingdom Health and Safety Commission, 78-87.
84. Wesson, K. M. (1983). A study of the new cattle postmortem procedure for steers and heifers. Food Safety and Inspection Service, United States Department of Agriculture, Washington, DC.
85. Wigg. A., Miller,P. &Ng, G. (1993). International regulatory perspectives on fresh red meat microbiology. In Proceedings of the Veterinary Public Health Branch of the Australian Veterinary Association, Gold Coast, 17-21, May, 17-22.
86. Zeckhauser, R. J. and Viscusi, W. K. (1990). Risk within reason. Science 248,559-564.
Veterinary Congress 1997