APPENDIX IX: RISK ANALYSIS IN THE CODEX COMMITTEE ON RESIDUES OF VETERINARY DRUGS IN FOODS

1. Introduction

The ninth session of the Codex Committee on Residues of Veterinary Drugs in Foods (CCRVDF) endorsed the incorporation of a science-based approach to risk analysis into its work, and agreed that France should prepare a discussion paper, with the assistance of Australia, Canada, the United States, Norway, New Zealand and the Netherlands, for examination at its tenth session (ALINORM 97/31 - para. 14). France duly prepared a paper with the help of these countries and the United Kingdom, FAO and WHO. It was submitted to the tenth session of the CCRVDF in 1996 and gave rise to a number of observations which have been taken into account in this latest draft, in addition to input from the two expert consultations that have since been organized by FAO and WHO on risk management and communication.

Risk analysis has been described in several Codex documents: CL 1995/40 CAC, ALINORM 93/37, ALINORM 95/9, CX/RVDF 94/5, CX/EXEC 96/43/6, reports of the joint FAO/WHO expert consultations held in March 1995 (risk analysis limited in practice to risk assessment), January 1997 (risk management), February 1997 (food consumption and evaluation of exposure to chemical substances) and February 1998 (risk communication). Risk analysis is now recognized as a process comprising three stages: risk assessment, risk management and risk communication. This paper sets out to determine the extent to which each stage is taken into account in the Codex procedure for setting maximum residue limits (MRLs) of veterinary drugs in foods. It will then look into the respective roles of the Joint FAO/WHO Expert Committee on Food Additives (JECFA) and the CCRVDF in the risk analysis process. Lastly, it will put forward proposals for further integrating this risk analysis process into the setting of MRLs and the work of the JECFA and CCRVDF.

The definitions of the risk analysis components referred to in this paper are those adopted provisionally in July 1997 by the Codex Alimentarius Commission and given in the "definition" section of the Manual of Procedures.

The general aim of the Codex Alimentarius is to set standards to ensure that food is wholesome and safe. The veterinary use of chemical substances in the form of drugs can, however, have a prejudicial impact on health through food. The aim, therefore, is to assess both the direct toxic risks in food after the use of veterinary drugs and the secondary risks from possible changes in biological balances or husbandry practices that they may induce. Assessment of toxic risk comes under the general framework of dangerous substances likely to contaminate food, regardless of origin. The Codex Committee on General Principles should define a risk assessment policy. The analysis should lead to as broad an understanding as possible of the benefits and risks for public health of using these substances as veterinary drugs. On the toxicological level, it should result in Acceptable Daily Intakes (ADIs), maximum thresholds in animal-based foods and suggested methods of analysis.

This paper is open-ended in the sense that presently unknown or neglected aspects will eventually come to light and will have to be taken into consideration.

2. Mandates of the CCRVDF and JECFA

At its sixteenth session in 1985, the Codex Alimentarius Commission strongly supported the recommendation of the 1984 joint FAO/WHO Expert Consultation on Residues of Veterinary Drugs in Foods (JECFA) and decided to set up a Codex Committee on Residues of Veterinary Drugs in Foods (CCRVDF), giving it the following mandate:

- to identify priority veterinary drugs for analysis of residues in foods
- to recommend Maximum Residue Limits (MRLs) for these substances
- to establish codes of use, if necessary
- to determine criteria for selecting methods of analysis used to detect veterinary drug residues in foods.

The aim of the JECFA, a joint FAO/WHO committee of experts outside the structure of the Codex Alimentarius, is to help the CCRVDF in its work by evaluating scientific data on metabolism, pharmacokinetics and toxicity of drugs and their residues. On the basis of its scientific evaluation, the JECFA proposes ADIs and MRLs for the consideration of the CCRVDF.

3. Risk analysis

3.1. Risk assessment

Risk assessment is a science-based process involving four stages:

- hazard identification
- hazard characterization
- exposure assessment
- risk characterization

3.1.1. Hazard identification

The purpose of this stage is to identify drug residues capable of causing adverse effects on health and possibly present in a selected food.

The definition of veterinary drug residue adopted by the Codex Alimentarius includes both the parent substance administered to an animal for therapeutic purposes and all the chemical compounds produced by its biotransformation that may be present in food derived from the treated animal. The metabolic changes vary in magnitude depending on the substances and may in some cases be intense and rapid. In such cases, it is technically and hence economically difficult to identify all the residues resulting from the parent substance. Therefore, in the case of heavy metabolism of the substance under study, hazard identification is basically limited in practical terms to this substance and to the main residues resulting from its metabolism. Consequently, while for practical reasons the MRL values are usually expressed in substance equivalent, the calculations of consumer exposure consider the full range of residues from its metabolism.

There are however two exceptions to this general rule:

• When the substance with an adverse effect not relating to the digestive tract generates allied residues, take-up bioavailability studies make it possible to set aside the non-bioavailable compounds of all the residues covered by the MRL.

• When the risk assessment of a particular substance is based on some clearly defined pharmacological adverse effect, and above all if the substance under study is also used in human medicine, an appropriate model can be used to compare the pharmacological activity of the parent substance and that of its main metabolites. In this case, the MRL will only relate to the compounds expressing this pharmacological activity.

Once the adverse effects of the drug residues have been quantitatively evaluated in the hazard characterization stage, the toxic effects observed in the laboratory animal have to be extrapolated to humans. The question is whether the drug residues present in the food from treated animals are likely to have the same toxic effects on the consumer as those observed in the laboratory animal. This can only be answered by comparing the metabolic profiles of the substance in the laboratory animal, where the adverse effect was identified, and in the food-producing animal which, when treated, will be the source of consumer exposure to the drug residues. Analogy of metabolic profiles provides the scientific basis for the results of the toxicological evaluation of the laboratory animal to be extrapolated to humans. Such metabolic information is however incomplete and any extrapolation from animal to human is based more on assumption than analogy of metabolic profile.

3.1.2. Hazard characterization

In this stage, the nature of the adverse effects associated with veterinary drug residues that may be present in the food is evaluated qualitatively and/or quantitatively. This difficult task requires a methodology to evaluate the results of the necessary toxicological and pharmacological tests. In this connection, WHO published the methodology for evaluating the safety of food contaminants together with a list of toxicological tests in its 1987 compendium Environmental Health Criteria 70.

Hazard characterization can sometimes be based on observations in humans, but is more generally carried out by means of toxicological studies on laboratory animals. It can also be done with the help of in vitro experiments.

The epidemiological studies carried out on humans are very useful because a hazard (an adverse effect in humans from an intake of toxic drug residues) can be directly characterized without need for extrapolation. Unfortunately, the statistical power of this methodological tool is too weak to identify with the required accuracy the adverse effects of lower quantities of residues unlikely to produce acute toxic effects. The evidence of allergic effects in humans from penicillin residues is a fortunate exception. More frequently, useful information can be obtained for drugs that are also used in human medicine. In these cases it is possible to observe adverse effects caused by the higher doses used when treating humans. But it is still necessary to extrapolate the chronic risks at low dose. Therapeutic tests carried out on humans using drugs that are also employed in veterinary medicine can provide indications of doses associated with pharmacological effects. The difficulty however lies in the fact that the purpose of the exercise in human medicine is to determine an effective, optimal dose and only rarely a dose without effect, which is the whole point of evaluating the innocuousness of veterinary drug residues.

As public opinion is turning increasingly against animal experimentation, scientific research has sought to develop in vitro testing. However, despite progress made, the results are rarely comparable to in vivo tests because of their simplification, although they do provide invaluable information to enhance the qualitative characterization of the hazards.

The limitations of studies conducted in vitro and on humans make animal experimentation the best source of the toxicological and pharmacological information needed to evaluate the safety of veterinary drug residues. The JECFA uses a very complete battery of toxicological tests, most of them codified by OECD protocols, to detect general or specific toxic effects. This battery combines acute, sub-acute or chronic toxicity tests, toxic effects on reproduction, and teratogenic, mutagenic, carcinogenic and immunotoxic effects. The undesirable effects sought also include any pharmacological effects that might help characterize the hazards for antibiotic, tranquillizer, anti-inflammatory and other residues.

For ethical and economic reasons, this complex battery of toxicological pests is restricted to the parent substance and is not used to assess the toxicity of all the residues resulting from its metabolism. This neglect of the specific toxic potential of each residue has given rise to the premise whereby the parent substance and all its metabolites are jointly responsible for the observed toxic effects and where the toxicity of each metabolite is similar to that of the parent substance.

In each toxicological test, the laboratory animals are exposed to increasing doses of the substance, calculated, if necessary, to cause adverse effects to emerge. Identifying the correlation between dose and effect is an important component of hazard characterization. The objective is to determine any relationship that might exist between degree of exposure to a chemical agent and severity and/or frequency of adverse effect on health. The joint FAO/WHO expert consultation of March 1995 estimated that setting the ADI, the quantity of residue that can be absorbed daily without risk to consumer health, was the final stage of this hazard characterization process. It should therefore be inferred that, as far as veterinary drug residues are concerned, this stage concerns both:

- the dose-response relationship that must be established for the laboratory animal undergoing the toxicological tests and that helps determine a dose without observed toxic effect in the animal (NOEL)

- extrapolating to humans the conclusions of this toxicological test on the laboratory animal to set an ADI.

The JECFA procedure is therefore more pragmatic. It is based on determining a NOEL for the laboratory animal and a subsequent ADI for humans based on NOEL and safety factor. A NOEL is the highest dose in a toxicological test that caused no adverse effect in the laboratory animal.

The value of the safety factor used to calculate an ADI from a NOEL is normally 100 and itself comprises two factors:

- The first is designed to:

• offset the uncertainty of the NOEL value that arises from the necessarily restricted number of animals used in the toxicological study

• take into account the possibility that human beings might be more sensitive to the toxic effect than the most sensitive laboratory animal. If the NOEL has been determined on the basis of undesirable effects on humans, this factor is not used.

- The second factor is designed to take account of the genetic variability of consumers likely to absorb these drug residues, which is much wider than the genetic variability of the laboratory animals used in the toxicological study.

This ADI calculation process is based on the premise that humans are at least as sensitive as the most sensitive laboratory animal exposed to the most sensitive test. This concept is not based on any scientific evidence but is used as a precaution against the uncertainties inherent in the process of risk assessment. The ADI corresponds to the quantity of residue that consumers can absorb each day throughout their lives without incurring any appreciable risk to their health and, as such, expresses the intention to keep the risk to public health so low as to be insignificant. Under this perspective, the setting of this value is therefore strongly influenced by the concept of risk management.

But this approach has two drawbacks, one due to the need to have a NOEL, the other to the standard nature of the security factor.

If, for any reason, it is not possible to determine a NOEL for an animal then it is not possible to establish an ADI. In such a case, if it is still possible or desirable to set MRLs, the pragmatic approach used is an exercise in risk management.

The safety factor value of 100 that is often used does not consider the slope of the curve expressing the relationship between dose and frequency and/or severity of adverse health effect. It does not therefore always guarantee the same margin of safety in extrapolation from animal to human.

This hazard characterization stage is therefore an area requiring more research. It would be well to look further into the mechanisms that generate the observed toxic effects and thus to refine the modalities used to determine the NOELs and the value of the safety factors. Such an effort has been made with substances thought to be carcinogenic, seeing whether they are genotoxic using a battery of short mutagenic tests. At the same time attempts are made to find any pre-cancerous lesions that might be produced in the studies of sub-chronic toxicity. But the cost of these mechanistic studies too often precludes exhaustive investigation.

3.1.3. Exposure assessment

This expression is used to refer to the qualitative and quantitative evaluation of the likely intake of drug residues through food, as well as exposure from other sources, if applicable.

Estimating consumer exposure is based on the daily consumption of a particular food combined with its content of veterinary drug residues.

In view of the difficulty of assessing such exposure by scientific approach, the JECFA preferred for the time being and for purposes of simplification to reduce the risk to the consumer to the absolute minimum by deliberately overestimating exposure, by combining the worst-case scenario and a globally standardized consumer food intake.

The worst-case scenario is based on the assumption that all the food of animal origin from animals likely to have been treated with a veterinary drug is contaminated by its residues at a level at most equal to the value of the MRLs set for the drug. The scenario is not a realistic reflection, because very few veterinary drugs are administered on a massive scale to all the members, and throughout the lives, of any one animal species. Conversely, there are many seasonal and even occasional uses of veterinary drugs, or cases in which they are only administered to treat sick animals. Lastly, statistical methods for establishing withdrawal times used by national authorities responsible for registering veterinary drugs strengthen the highly protective character of this scenario in relation to public health. On the other hand, the possibility of using veterinary drugs incorrectly reduces this margin of safety.

Concern for international standardization translates as the adoption of the following daily food intake: 300 g of muscle, 100 g of liver, 50 g of kidney, 50 g of fat, 100 g of eggs, 1.5 l of milk and 20 g of honey. The value set for milk seems to be particularly high, but has been estimated as appropriate to ensure that infants do not consume veterinary drug residues at levels exceeding ADIs. The JECFA has considered that the potential error from using these intakes only accounts for a small proportion of the uncertainty inherent in the risk assessment procedure and that there is no need to specify these values any further.

The components of this diet should however be reconsidered on the basis of more relevant studies of intake if the exposure assessment stage is to use the scientific approach employed in the risk assessment procedure.

As the administration of veterinary drugs to an animal takes place under strictly controlled conditions, the values of maximum residue contents in foods can also be defined in particular by establishing appropriate withdrawal times. The MRL values are therefore established in such a way that maximum daily intake of residues is below that authorized by the corresponding ADI. The determination of MRL values therefore relates more to risk characterization than to exposure assessment.

3.1.4. Risk characterization

This stage sets out to provide a qualitative and/or quantitative estimate, given the uncertainties of assessment, of the probability of occurrence and severity of known or potential adverse health effects in a given population based on hazard identification, hazard characterization and exposure assessment.

The aim is to characterize the risks to the consumer from residues possibly present in animal products, on the basis of use of the substance and particularly the withdrawal time, given that the period of administration and the dose are predetermined by the objective of effectiveness.

The conditions under which the drug is used need to be estimated as do acceptable residues linked to the level of acceptable risk to the consumer. The acceptable level of risk, which is determined in theory at the risk management stage, has already been expressed in terms of residues by the ADI under hazard characterization. Moreover, the elements considered for hazard identification, hazard characterization and exposure assessment make it possible, for a given form of utilization of a particular substance, to establish a profile of residues in animal tissues and to associate this with a profile of consumer exposure. Comparison of this consumer profile and ADI indicates whether the mode of utilization of the substance is acceptable or not. Analysis of the different results of residue content in animal products then provides an indication of level of residues in one or several animal tissues, making it possible to distinguish between veterinary drug applications that do or do not permit compliance with the ADI.

As expressed by the 1995 joint FAO/WHO expert consultation, it is this risk characterization stage that leads to one or several proposed MRLs associated with sound veterinary drug practices which, on the basis of established food intake, can guarantee that ADI values will not be exceeded.

The JECFA does not use rigorous mathematical models to set MRLs from a particular ADI. The MRLs are set, using available metabolism and pharmacokinetic data, at the end of a procedure heavily dependent on trial and error and strongly influenced by risk management. The few examples below illustrate the close interaction between risk assessment and risk management in setting MRLs.

The MRLs express a ceiling for all residues of a drug likely to pose a risk to consumer health.

Since it is difficult in practical terms for a monitoring plan to measure analytically a series of residues with widely differing chemical structures, control exigencies require that MRL values be expressed in terms of a single chemical entity, know as the marker residue. It is important that the contents of this marker residue evolve in the different tissues of treated animals in proportion to all targeted residues, if it is to reflect them. But, for obvious practical reasons, this marker residue must also satisfy two requisites: it must permit a practical dose and must be commercially or otherwise available for the purposes of official controls.

The MRL values for the different tissues (muscle, liver, kidney, fat) are set in proportions that reflect the tissue distribution of the residues. But, to avoid producing a set of highly complex figures for different tissues and different animal species, the JECFA tries as far as possible to harmonize these values to keep their number down. Similarly, when it appears that the residue contents in a given tissue are likely to be too small for the feasible control after recommended withdrawal time of residue contents in other tissues, the JECFA cannot propose any MRL for that particular tissue.

When a veterinary drug is used for both meat and dairy animals, the ADI breakdown between meat and milk is done by trial and error. This is a decision pertaining to risk management. The CCRVDF should therefore give this matter serious attention. To help weigh up the various risk management options, the JECFA should provide precise indications on the conditions of use of the substance and the good veterinary practices that influence risk assessment, thus enabling Governments to gauge their margin of manoeuvre for the management options.

Lastly, the MRL values may be reduced to take account of the normal conditions under which a particular veterinary drug is used where these lower MRL values can always be controlled by a viable analytical method.

Even though the JECFA is not involved in setting withdrawal times, it has to refer to a practical withdrawal time in order to establish a consistent set of MRL values. If it emerges that compliance with the MRLs requires unrealistically long withdrawal times, the JECFA cannot recommend any MRL. This situation can arise in particular for milk and eggs.

Furthermore, the JECFA at the present limits its proposed MRLs to animal species for which the necessary information is already available. This strict approach raises the problem of controlling veterinary drug residues for the so-called minor animal species, for which the veterinary drug industry considers the economic market too small to justify the funding of the studies required. Thought needs to be given to defining a pragmatic approach that is compatible with reasonable risk management.

The whole pragmatic approach used in establishing MRLs indicates strong interlinkage between risk assessment and risk management. The particular relevance of scientific data from pharmacokinetics, metabolism and statistics suggests that the JECFA should retain its role of proposing MRLs for CCRVDF consideration. However, the CCRVDF is basically involved in risk management and, as such, should also assume greater responsibility in this connection when invited to consider JECFA-proposed MRLs that have been based on risk management decisions.

3.2. Risk management

Risk management is understood as the process whereby the policy options determined by risk assessment findings are weighed up, and where any necessary control and regulation measures are instituted and put into effect.

The joint FAO/WHO expert consultation that discussed this issue in January 1997 tried to configure this concept of risk management, but its conclusions were somewhat imprecise and further thought is needed on defining the components of risk management. The consultation divided risk management into four component parts: risk evaluation, assessment of management options, implementation of management options and monitoring and review.

3.2.1. Risk evaluation

This first stage of risk management includes:

- identification of a public health problem
- description of the problem
- classification of the identified danger in terms of risk assessment and management priorities
- establishment of a risk assessment policy
- appointment of a body to conduct the risk assessment
- consideration of risk assessment findings.

The 1997 FAO/WHO consultation considered that such policy should protect scientific integrity, coherence and transparency of risk assessment. More specifically, this component of risk management should deal with identification of populations at risk, criteria for ranking hazards and modalities for determining safety factors.

The protection of scientific integrity, coherence and transparency of risk assessment by the JECFA is crucial if confidence in the JECFA and its MRL proposals is to be total. As the JECFA is not strictly speaking a Codex structure, the CCRVDF and FAO/WHO should discuss how this objective of risk management can be achieved. They should focus on the management of JECFA meetings by FAO and WHO and look into the modalities of selection of the experts who should complete a declaration of interest.

The 1997 consultation addressed the topic of safety factors which is vitally important for the protection of public health. Setting MRLs is in fact based on a series of safety factors including:

- the assumption that humans are at least as sensitive as the most sensitive laboratory animal to a potentially toxic residue;

- the safety factor used to infer an ADI from an NOEL, including the additional safety factor, generally with a value of 2, to establish a provisional ADI until further information is available to convert this into a definite ADI;

- the over-estimate of consumer exposure to drug residues;

- the assumption that all the residues covered by the MRLs are as toxic as the parent substance;

- the assumption of total bioavailability of residues "free" from the human gastro-intestinal tract;

- the reduction of MRL values to take account of normal conditions under which the veterinary drugs are administered.

Establishing the value of these different safety factors would seem to be a basic component of public health policy as the exercise involves decisions on the magnitude of a socially acceptable risk. This needs to be assessed in the light of observed toxic effect, quality of information on residue toxicity and content, benefit-risk trade-off with assessment determined by the therapeutic or productive purpose for administering the substance in question. This is a central aspect of risk management that should be dealt with by the mandated parties. It is odd, to say the least, that the CCRVDF has never addressed this important matter and issued the necessary directives to the JECFA.

The JECFA is also involved in determining risk evaluation policy when it proposes guidelines to the CCRVDF, as in the case of evaluation of microbiological risk from antibiotic residues. The JECFA's scientific expertise is clearly to the benefit of Codex performance, but the CCRVDF should be more active in critically assessing the proposals it makes.

In contrast, the CCRVDF is involved in policy formulation when it drafts guidelines such as that for assessing the safety of veterinary drug residues at injection point.

The final component of risk evaluation, consideration of risk assessment findings, is clearly under the remit of the CCRVDF and corresponds to steps 4 and 7 of the Codex standards drafting procedure.

3.2.2. Assessment of management options

The joint FAO/WHO consultation divided this stage into three parts without giving details: identification of possible management options, selection of preferred option and final decision. So far, the CCRVDF has done very little in this area for which the States have the required competence.

The joint FAO/WHO consultation on risk management has insisted that decisions on acceptable levels of risk should be based on considerations of public health. It also accepted that other considerations such as economic costs, expected benefits, technical feasibility and social choices could be considered, where these could be objectively determined.

For its part, the JECFA has advised against using certain veterinary drugs with dairy cattle and laying hens when the withdrawal times needed to meet the MRLs seemed unrealistic in view of their normal conditions of use.

3.2.3. Implementation of management options, monitoring and review

These two last components of risk management are essentially under state responsibility. However, the JECFA advises states on appropriate methods of analysis to ensure compliance with the MRLs.

It is important to stress that risk management goes beyond straightforward analytical study of residues in animal products and must also include the control of good practices at, and prior to, the time of veterinary drug administration.

The JECFA can also make a contribution when it:

- studies the validity of analytical methods proposed to check MRLs

- specifies the statistical basis for establishing withdrawal times

- issues recommendations on the conditions of use of certain veterinary drugs in relation to MRLs set (tranquillizers for pigs) to reduce consumer exposure to veterinary drug residues.

A more recent joint FAO/WHO consultation in February 1998 sought to define this third component of risk analysis which was described in 1995 as an interactive exchange of information and opinion on risks among officials responsible for risk assessment and management, consumers and other interested parties. Although examination of this very complex subject is recent and needs further reflection, there would appear to be many parties potentially involved in such communication and the structures responsible for risk assessment and management have a duty to report on their respective areas of competence. This report only looks at the related responsibilities of the JECFA and CCRVDF via their secretariats.

3.3.1. Role of the JECFA

The JECFA provides satisfactory technical communication through:

- its summary reports of meetings
- its more detailed reports of meetings
- WHO and FAO monographs on evaluation of toxicological data and study of residue contents
- the publication of scientific information needed to assess the safety of veterinary drug residues.

It would be useful if, for each substance studied, the JECFA could clearly indicate the assumptions and choices made during the risk assessment process that relate to risk management, thus providing more information on its proposals. This would not be necessary for routine assumptions and decisions already announced in a general paper.

Greater involvement in JECFA activities by experts put forward by consumer associations and greater transparency in the nomination of experts would greatly enhance this interactive process of risk communication.

The formal publication of these technical documents under the authority of two international organizations of the stature of FAO and WHO is clearly a difficult and time-consuming task, given the obvious staffing shortages, but the time it takes to publish the detailed reports of JECFA meetings and the FAO and WHO monographs is far too long. This undermines the effectiveness of the CCRVDF which is thus deprived of the timely information it needs to critically assess the ADIs and MRLs proposed by the JECFA. This worsening state of affairs needs to be urgently redressed.

3.3.2. Role of the CCRVDF

The role of the CCRVDF in communication for risk management is extremely limited as it is reduced to reports of meetings, which, for budgetary reasons, are increasingly succinct to the point of having little substance to communicate. The important step of drawing up priority lists of substances, which is the point of departure of the JECFA and the CCRVDF work, provides no explanation for the choices made. Even the general criteria adopted in 1986 to determine priority lists have lost their transparency. It would be a good idea to see whether the amendments adopted in 1994 are applicable or not. It is also important to recall that the Codex procedure for establishing MRLs only considers substances for which the JECFA has been able to propose ADIs and MRLs. Other substances, whatever the reasons for the inexistence of ADI and MRL (too toxic, inadequate documentation) are cast aside and simply ignored. No relevant information is given to explain why these substances, some of which can be toxic, have been passed over by the Codex procedure. This is a matter where improvement is required.

4. Roles of the JECFA and CCRVDF

It should first be recalled that the JECFA and CCRVDF mainly consider risks to consumers from residues of a drug in animal products. They also consider the effect of the drug on the composition of the animal products (for example, IGF1 in the case of BST), but some aspects have been virtually ignored. It would be useful to decide at what point of the risk analysis process these should be taken into account and by which body:

- interactions between different uses of drug substances and their effects on residues in animal products: the use of a substance can modify the metabolism of another substance administered simultaneously, particularly when used continuously to modify animal physiology. It can also have an impact on modality of use of other substances (for example, BST leads to greater use of antibacterial drugs).

- risks of drug use to animal health: the question does not arise when the substances are used for therapeutic purposes and are therefore to the de facto benefit of animal health. It can arise, however, with they administered for production purposes that can generate risks to animal health.

- risks to human and animal health from the use of antimicrobial drugs and the resulting increase in microorganism resistance (zoonotic or not, pathogenic or not).

However, closer examination shows a slightly different picture and indicates that the respective roles of the CCRVDF and the JECFA in the risk analysis process need to be better defined. As the organization and division of work was decided before the introduction of the risk analysis concept, de facto systems have arisen that are perfectly logical in functional terms but that do not fulfil the recommendation of separate responsibilities for risk assessment and risk management. As a result, the JECFA includes elements of risk management in its risk assessment work. This can be acceptable for proper Codex functioning, particularly as it echoes a pragmatic observation made by the consultation of 1995 that there might be exceptions to any hard-and-fast separation of responsibilities. But when these aspects of risk management go to the very heart of public health protection, it would seem inappropriate for the CCRVDF not to assume its appointed risk management responsibilities. A clear example is establishing the values of the safety factors that used in the different stages of risk assessment.

The JECFA should nevertheless continue to provide the CCRVDF with technical assistance for risk management by proposing guidelines and protocols that will improve risk assessment policy.

5. Conclusions

This report shows that the procedure for setting veterinary drug MRLs incorporates the concept of risk analysis. The separation of risk assessment and risk management is a reality because of the way the work is divided between the JECFA and CCRVDF. The scientific approach needed for JECFA risk assessment needs to be reinforced by providing additional scientific information. Deficiencies need to be identified and research encouraged to provide the missing information. The CCRVDF is responsible for risk management and should focus more on risk management components that need to be used in risk assessment, thereby realizing the desirable separation of responsibilities for risk assessment and risk management.

6. Recommendations

Quality of work by the JECFA and CCRVDF is a requisite for consensual adoption of MRLs, without scope for contest. Such quality is firmly recognized.

However, a number of proposals can be made to further enhance the three components of risk analysis: assessment, management, communication.

· Risk assessment

- the establishment of veterinary drug MRLs should be based on objective analysis of available and relevant scientific data. The process must continually take account of new concepts emerging from an ever-evolving world of science.

- the mechanisms of toxic or pharmacological action to better substantiate the setting of doses without effect and safety factors;

- structure-activity relations which would help distinguish between residues that need to be taken into consideration and those that pose no risk to public health. The identification of marker residues would also be much more reliable;

- comparison of metabolisms of laboratory animals and animals destined for human consumption and humans to increase the pertinence of extrapolation of toxicity findings from laboratory animals to humans;

- this scientific evaluation must also take account of the aspirations of contemporary society, which is determined to reduce the number of animals used for experimental purposes. This should encourage the JECFA to integrate alternative evaluation tests that are more respectful of animal life. These new tests must however be properly validated beforehand.

The CCRVDF should revisit its action regarding components of risk management. Besides the improvements suggested in this report, two further proposals are drawn to its attention:

- it would be useful if, with the assistance of the Codex Committee on General Principles, the CCRVDF could identify reference factors other than impact of residues on consumer health when evaluating the JECFA's ADI and MRL proposals: health factors (animal health, public health, such as exposure to antibiotic resistant bacteria, etc.) and non-health factors (consumer expectations, organization and geographic distribution of production).

- The CCRVDF should review the procedure for establishing priority lists of substances to be evaluated by the JECFA. One criterion for including a candidate substance on the priority list is that all the necessary information be made available to the JECFA, but this condition can only be met by the veterinary pharmaceutical industry because of the growing complexity of the documentation. As a result, the JECFA works on the basis of CCRVDF priorities that are heavily influenced by industry decisions. Thought should be given to the ultimate aim of the work of the CCRVDF and JECFA and to the respective importance of public health and international trade. Without wishing to belittle the importance of evaluating new substances, which spearhead modern medicine and are a lifeline to the veterinary pharmaceutical industry, there is at the same time no reason to neglect older substances that are still in widespread use. The problem is that these substances are no longer protected by patent and therefore no longer represent an economic market sufficiently important to justify investment in the requisite studies. The unfortunate result is that the JECFA focuses especially on evaluation of new molecules which, under constant pressure from ever-tighter technical requirements, offer increasing guarantees of safety, and does not perhaps spend enough time addressing long-established substances, some of which, though prohibited here and there, can expose public health to considerable risk. There is an urgent need to draw up a list of these substances and to agree an appropriate methodology to identify their associated residue risks and/or provide interested parties with all relevant information.

Given the rapid advance of science, the JECFA should be able to regularly publish the scientific basis for its evaluation of safety of veterinary drug residues, with reference to good laboratory practices and internationally recognized procedures for validation of methods of analysis. Even more important, the JECFA secretariat should rapidly publish the reports of the JECFA sessions.

Finally, during CCRVDF evaluation of JECFA proposals, the inclusion in a so-called inactive list of substances for which ADIs and MRLs could not be established does not seem to come up to expectations on risk communication from the CCRVDF. There is an urgent need for Member Nations to be given the reasons for the inability to set ADIs or MRLs for these substances.