Australia - Measures Affecting Importation of Salmon Report of the Panel (Continued) Dr. Wooldridge 103. OK, I have got one or two areas to talk about which I think all fall within the broad area of risk assessment and risk analysis methodology. I noted the comment that the type of risk assessment most appropriate depends on an examination of the data. I would suggest that actually, well, in my opinion that is not the appropriate way to view it. I would suggest that one normally undertakes the qualitative risk assessment first and then, if necessary, follows up with an attempt to undertake a quantitative risk assessment, and this does not depend upon an examination at the initial stage of the data to decide which type of assessment one wishes to undertake. I think a qualitative assessment gathers the information and categorizes it and orders it and comes up with a qualitative assessment. Leaving aside the meanings of the words, this may be things like high, low, negligible - and if everybody agrees that the risk is too high from this basis to be acceptable, we do not have a problem; if everybody agrees that the risk is negligible, we do not have a problem, and it does not matter what those words mean at that stage. 104. When you have a dispute, I would maintain again - there may be other occasions - but particularly when you have a dispute, I would maintain that the most appropriate way to then follow it up to try and get further on the matter is a quantitative assessment. I would here like to reiterate the point I made earlier in my written evidence that the basis of a quantitative risk assessment is the putting together of a model into which you feed data and the whether or not you have the data, you can still put together a model and put dummy-data to actually clarify the pathways by which you think your identified hazard may become the unwanted outcome that you are attempting to assess. I think the big value of this is that it does clarify the pathways required and the data that you need, and if you say "I do not think I have got the data to do a quantitative risk assessment" without actually attempting to make the model required for that quantitative risk assessment, I do not believe you can categorically state you do not have that data. That is my theoretical point. From a practical view point I would say that when you do try and put together a quantitative risk assessment you almost always do find that you have got a lot more data of appropriate type than you initially anticipated that you would have. So I would say that you cannot tell whether you have got the data to do a quantitative risk assessment until you try and do one. 105. I think, following on from that I would like to say I am afraid I did not find the Final Report transparent. I think there is a quote or a comment that the final version was transparent. I personally found it far less transparent than the draft and difficult to find the information for which I was looking, sometimes I found it, sometimes I did not. It did not appear to me to be set out in a way which made it easy to find the information and to differentiate between what is the assessed risk and the acceptable risk. 106. Having said that, I would like to reiterate and I agree with Australia's contention here that it is up to the country or a local area or the particular place in question to decide what is an acceptable risk for them, and therefore whatever came out of Australia's Draft or Final Report as far as the assessed risk is concerned and whether I believe that was done well or badly is immaterial in this instance, it is up to them to say that is an acceptable or not acceptable risk. However, I think international guidelines would suggest that if they come to the conclusion it is not acceptable, they have to be able to justify it at a meeting such as this. 107. A word about diseases themselves, not specific diseases but which diseases are appropriate to include in a risk assessment and how this interacts with Notifiable Diseases and standard lists of diseases and so forth. We had many discussions on which information it was valid to include in an international risk assessment, when we were redrafting the OIE Risk Analysis Code Guidelines. There were opinions that the only diseases which could be included from a given source were those which had been notified and there were opinions that said "no, you need to be able to include any disease for which you have reason to suspect, or knowledge to backup, that it might be in the product which you are considering". That was my opinion and it was actually the final opinion of the working party, that those diseases which one may legitimately look at are any that one has evidence could be in the product, and that might not simply be the information notified to the OIE, but it might be from a literature search or personal reports or communications. The rational for this was that one wishes when one is doing an import risk analysis and risk assessment to have the best, the most appropriate, the fullest data available and that is not necessarily only the data that has officially notified to OIE or any other body. I think those are my only comments at the moment. Thank you. Chairman 108. Thank you very much. Turning to the other table, can I ask Dr. Burmaster? You have the Floor. Dr. Burmaster 109. Well I have a couple comments in response to this overview statement from Australia. Risk assessments, at least the way I use the word, are always anticipatory of some future event. So by definition, if some group of people are thinking of doing, changing a policy or taking a new action in the future, by definition we never have enough data to describe it. A risk assessment is always lacking some data which someone, somewhere thinks are important. It is part of the process, we are looking ahead, and by definition we are missing some part of it. Nonetheless, I would agree with Dr. Wooldridge that by putting together a model has very beneficial results. It sharpens the thinking. By this I mean a quantitative model, a mathematical model, whereby we can put in numbers or put in distributions and make calculations about what if, what if this were true? what if that were true? what if we did this? what evidence do we have? But we will never have - I am not aware of any situation where risk assessment has ever been conducted in any field, at any time, for any purpose, where they had perfect data. The concept of perfect data does not exist in risk assessment or, for that matter, may not exist in all of science. 110. Nevertheless, there are ways working with the model to use a combination of data and a combination of expert opinion and expert evidence and evidence from other fields to come up with calculations which somehow give a numerical value to the proposed action. It can also highlight where data is missing, where to go looking for data. That would then show where the data are most missing and could most beneficially help (this is called the value of information approach). Now this is still in risk assessment, this is still trying to ascertain what risks may occur in the future. Then comes a separate step called risk management, is this risk that we have calculated acceptable or not? I strongly separate risk assessment from risk management. Let me stop here. Chairman 111. Thank you very much. Can I ask if Dr. Rodgers would like to respond to Australia? You have the floor. Dr. Rodgers 112. Thank you Mr. Chairman. One of the beauties of coming last is that you do not have much to say providing the other experts have said it for you and in this case to a certain extent that it is true. So I will be brief. At the last European Association of Fish Pathology International Biannual Meeting, in Edinburgh, in September of last year, we organized a workshop on risk assessment for aquatic animal health. In a room of 120 seats there was standing room only. It was freely available meeting for anyone that was present at the main conference. We addressed many of the issues which have been raised by both parties in their submissions. What came out of the meeting was not only a sense of relief because somebody had actually got people together to talk about this subject, but that there was a lack of data in certain important areas in aquatic animal health for use in either qualitative or quantitative risk assessment. These areas are particularly related to species susceptibility, diagnostic techniques, survival parameters of the pathogens themselves in the fish, particularly after harvest, and pathogen inactivation. 113. Now I would think that there is a real need for risk assessment in aquatic animal health but this is already leading to a parallel need for basic epidemiological studies to be carried out to support the data gaps, if you like. I personally have not known this situation before as a scientist because, what normally happens in this sort of situation, is you generate data through research projects which then lead to a need for something else, like risk assessment. This has been driven the other way around, it is coming down from the top and it is being led by agreements such as the SPS Agreement, in my opinion, which says you must use this technique. It is new, it is evolving okay, but at the end of the day it is pointing out where the data gaps are occurring and that is something which David Burmaster has just eluded to. I think that also one of the most important things at the moment for risk assessment is that it can drive research projects because it does show you where the missing information is. But at the same time sometimes we do seem to be standing on thin ice because we are looking for information, some of which is there, but most of the time it is not there. Where the information is not there, there is a need for expert opinion. I think this has not yet been addressed sufficiently in aquatic animal health. Nobody has actually got experts together other than in this sort of forum, where they can actually give their opinion about missing data. In risk assessment and statistical terms this is acceptable for supporting risk analysis. The beauty of the technique is that the experts do not need to agree either, in fact it is probably better if they do not agree, because you end up then with a range of probability, a range of distribution, a probability distribution - Dr. X would say "there is a one in a million chance of that happening" and Dr. Y will say "no it could happen tomorrow". That is useful and I do not believe that it has yet been done for aquatic animal health. I have no further comments to make on that. Chairman 114. Thank you very much. Can I ask ... any further questions or comments for the experts? Australia. Australia (Mr. Ric Wells) 115. Thank you Mr. Chairman. Yes we do have some questions for the experts, I wonder however, whether we could have the indulgence of five minutes among ourselves just before we ask those questions? Chairman 116. We will have five minutes in that case. [Break] Chairman 117. If we can now resume, perhaps I can put the question to Australia again if they have any further questions or comments for the experts please feel free to give them now. Australia you have the floor, Mr. Wells. Australia (Mr. Ric Wells) 118. Mr. Chairman, I will ask Dr. Gardner Murray, the Australian chief veterinary officer to ask questions. Thank you. Australia (Mr. Gardner Murray) 119. Thank you Mr. Chairman, I am Gardner Murray, Australia, despite the Scottish accent. We have a number of questions. I suppose the first question I would like to pose relates to consequences. I think only one of the experts talked about consequences this morning, the potential consequences of irreversibility of disease, should enter Australia. My question is this: how can one best resolve the issue of the low probability of establishment of a disease with the high socio-economic consequences that could result should such a disease enter a country? Because this is the point that is made in the Australian submission, low probability but very high consequence. Thank you. Chairman 120. Thank you. Is that addressed to all the experts or to anyone in particular? Dr. Burmaster. Dr. Burmaster 121. Well, perhaps I can shed some light on that, certainly that is a key issue in this dispute between the two countries, but it is a common feature of many, many, many risk assessments. For example, people do risk assessments of large earthquakes in Los Angeles, California, in the United States. Now if a low probability earthquake occurs in Los Angeles, California, there could be hundreds or thousands of deaths along with property damage valued in the hundreds of millions of dollars or more. So this is not the first time, I do not want to be deflected into thinking that this is the first time that anyone, anywhere has ever thought about low probability high consequence events. It is a very standard question that occurs many times in many risk assessments and I will stop there for the moment. Chairman 122. Thank you. Other responses? Dr. Rodgers. Dr. Rodgers 123. Yes, the only thing I would say in answer to that is that you can reduce the low probability to an even lower risk, if you like, by addressing certain risk reduction factors, as I mentioned this morning. Providing those risk reduction factors were agreed to by both parties, I do not see that there would be a problem. Obviously you do need to consider the high consequences of disease introduction although even that may not reduce the level of risk to an acceptable level for Australia in this case. Chairman 124. Dr. Burmaster. Dr. Burmaster 125. Yes, I guess I can go back to it now and say that it is precisely in the situations of low probability and high consequences where numbers help. If we thought that an even might recur in hydrology, with floods and so on, people talk about storms of a certain magnitude, and they talk about a storm that may recur every hundred year, a storm so large that a storm of comparable size might occur only once every hundred year on average, or once every five hundred years, or once every ten thousand years, different sizes of floods. It is important to begin to talk about those numerically, rather than just saying "well it is a big flood". Different people have completely different interpretations of what "big" means, there is no common definition. I think if you posed to people in this room "what is a low probability?" we would have no agreement here in this room, even amongst the Australian delegation, or just amongst the Canadian delegation. If I describe something as a low and each of you wrote down on a piece of paper what you think that might be, there would be orders of magnitude, differences of opinion here. I am using one word, I am saying "low probability", yet amongst the audience here today, there would be orders of magnitude, differences of opinion as to what that word meant in that context on the spoken day. So without numbers we are engaged in an exercise that cannot converge. The English language was not designed to handle this, but mathematics was designed to handle this kind of question. It gives us the tools that we need to try to resolve this kind of dispute and to find out what we know and to find out what we do not know. What we know is important, what we do not know is important. Both of those have to be brought up in a way where we can talk about them and understand what each other means. Thank you. Chairman 126. Thank you very much. Dr. Wooldridge. Dr. Wooldridge 127. Yes, I think I am more or less concurring with what Dave Burmaster says here but I will just go through it and put it the way I see it. I think what we need to do is decide first what we are talking about in terms of consequences and work through similarly to hazard identification listing each of the consequences about which we are worried. These can be simple consequences or very complex consequences. For example, we are talking about the probability of the import of disease, and that is perhaps the simplest consequence that we can envisage in this case. So importing disease is a consequence. We then might think about another consequence perhaps, whether one consequence is an epidemic of that disease. Now the import of disease may or may not lead to an epidemic of that disease. So there are various different consequences resultant upon the previous consequence, some of which may happen, some of which may not, and if you are talking about a quantitative model you can actually model that as the next part in your model pathway, so you can work out the probability of the import of disease, then you can work out the probability of an epidemic of that disease in a given species or in all species or whatever you are interested in. You may also have on your list of consequences that you are worried about and that you have categorized, the loss of a certain amount of economic good in some way. I am not an economist and I do not plan to try and describe this in any more detail - I have just written down loss of X pounds of money due to this epidemic occurring. You can in fact add that onto your model and work out the probability of that given loss, albeit all these probabilities will probably have very wide uncertainty limits on them. 128. So quantitatively, with each of your high impact consequences, you can work out your probability of it happening. You have assessed that risk, you have assessed the loss of that amount of money. What you then got, assuming everybody agrees with the methodology and that is the probability of loss of that amount of money or the probability of that epidemic occurring or whatever - you have then got a risk management decision, "is that acceptable?" This is quite different from your risk assessment. You have actually included your high impact consequences in your risk assessment by calculating this probability of each happening. So then, are we prepared to accept this probability of this epidemic happening? Are we prepared to accept these particular series of consequences? That is very difficult to make a decision on but one potential way is actually to do something similar to a risk assessment, but a cost benefit analysis. That can also be done quantitatively if you are actually in a position where you have something which has a low probability of happening, you can actually look at the benefits that might accrue by letting the process go ahead and then compare the two - as I say, a cost benefit analysis, which can be done in a similar methodological way to a risk assessment in that you can put uncertainties into your benefits as well as your potential for disease importation. So that is perhaps the most straightforward way I can see of actually judging this particular kind of problem. But I accept it is a difficult problem and of course different people from different areas with different backgrounds and different perceptions, and different cultures will view different risks in a different way with regard to their acceptability all over the world. So as I say, I would recommend something along the lines of a cost benefit analysis, which again is lengthy and difficult and time consuming but there we are. Chairman 129. Thank you very much. Dr. Winton? Dr. Winton 130. Just maybe a brief comment, Dr. Murray is certainly correct in that the consequences of certain introductions of certain diseases can be quite high in some cases but certainly not all cases. The range can be from nearly not effect or to a manageable level. For example when VHS virus was introduced by, presumably, uneviscerated marine fish into turbot culture in the UK, United Kingdom lost its VHS free status, but that disease was able to be quarantined and was eradicated and the UK has now regained its status as a VHS free zone. So, while there was potentially a very large impact, that impact was contained and managed. Certainly Australia has seen in wild fish, the impact of an introduction for example in the Pilchard mortality which I think by most agreement is probably the largest fish kill ever recorded on the planet occurred in a wild fish. Once a disease is introduced into wild stock it becomes much more difficult to eradicate. In the case of whirling disease in the United States, is an example, where introduced from Europe, this disease has proven virtually impossible to eradicate from large areas due to its presence in wild fish. 131. The consequences can also range from biological which may be losses of natural stocks of fish to economic, which are generally those in aquaculture. So I certainly support Dr. Wooldridge's contention that a risk assessment can also include, for example, those range of consequences and perhaps try to quantify those. 132. Earlier, Mr. Wells talked about Furunculosis as an example of a disease that is not on the OIE notifiable list presumably because it is broad spread. It is also because that disease is manageable in context of aquaculture now through the use of therapeutents or vaccines to an extent that makes it of less concern to the aquaculture industry. Australia also has a strain of Furunculosis that is certainly not virulent in its own right, and so Furunculosis is certainly an example of a disease which might be more concern in certain areas or less in others, but I do not think that its absence in Australia, and its presence in other parts of the world, argues that it could be used for example as a special case. I think there are a lot of cases of diseases that are either present in other areas in the world, or even strains of diseases which are already present in Australia that might be significant in some ways. So both the range of probability of import go from nearly zero to very high, but the consequences also can go from very minor to relatively high depending on where it is introduced and what pathogen into what species of fish. It is virtually impossible for the OIE Fish Diseases Commission or perhaps any group of experts to foresee all of these various permutations. That is where I think the strength of these risk assessments comes in, in that you can then begin to then estimate those probabilities and those consequences in a more quantitative manner that can be discussed more clearly. Thank you. Chairman 133. Thank you. Well that is the responses to the Australian question. It is now five minutes to one and this is probably a good moment to break for lunch. But before I actually adjourn, can I ask if there are further questions coming from the parties? Certainly from Australia, Canada too perhaps. Canada (Ms. Valery Hughes) 134. Not at this time. Chairman 135. So there will be further questions coming from Australia this afternoon. Well lets resume at three o'clock in this room, we have this room at - Mr. Wells. Australia (Dr. Gardner Murray) 136. Mr. Chairman, Gardner Murray, I do not have a question at this point but I would just like to make an observation on one of Dr. Winton's comments. That was in relation to Pilchard mortalities. It is our view and also the view of New Zealand that in fact the causual organism Herpesvirus is endemic to Oceania, and we could find no relationship between the importation of feed stuff and the event, in particular, as it started first in New Zealand. But that is an opinion that we have and others have different views of course. Thank you. Chairman 137. Dr. Winton. Dr. Winton 138. Yes, I appreciate that clarification and I think earlier I had mentioned that while scientifically Koch's Postulates had not been fulfilled and I do not think that epizootic is clearly understood, I appreciate that clarification and perhaps it is not a very good example of an imported disease in that case if that is in fact the case. I do not know, is there an explanation for the sudden appearance of such a large loss, is it an environmental issue in addition then? What is the other explanation? Australia (Dr. Gardner Murray) 139. The explanation, and they come in large part from New Zealand as well as Australia, is that the virus in question, Herpesvirus occurs in Pilchards but when stressed disease can manifest. The stressors in this case appear to have been a sudden drop in temperature, Pilchards are very sensitive to sudden temperature drops, plus probable physiological changes, hormonal changes at that time of the year. That is the explanation of stressors. Thank you. Chairman 140. Can I just clarify, is that the only remaining question from Australia or are there going to be others? There are going to be others. So lets resume at three o'clock in this room and we will take the further questions at that time. Before we leave the room, the Panel does have questions which will be putting in later to the experts. We will circulate those in writing right now, so before you depart, if you would like to pick up a copy of that. It will be available in about five minutes time, we are just going to finalize the questions. Chairman 141. Let me start by inviting Australia to continue. Australia you have the floor. Australia (Dr. Gardner Murray) 142. Thank you Mr. Chairman, Gardner Murray. Now this first question is to those experts who care to answer: To the best of your knowledge, is it the exception or the norm for countries to base their import policies on purely quantitative risk assessments? Chairman 143. Dr. Wooldridge. Dr. Wooldridge 144. I think at the moment it is probably still the exception but I think there are quite a lot of us who think that it is probably the way forward that things are going to go. Chairman 145. Any other, Dr. Burmaster. Dr. Burmaster 146. Well I guess I do not know the worldwide answer, but it certainly is the way for the future and this is how these kinds of disputes and disagreements will be settled, I think starting in the last year of two and proceeding into the future. Dr. Rodgers 147. If I can only say that I agree with Marion, since I do not think there are too many examples of any legislation in aquatic animal health being based on a quantitative risk assessment exercise or any risk assessment exercise. I am thinking in particular of the EU directives which now govern the movements of fish into and out of the Community and I think that was done simply by negotiation, based on expertise rather than a fully functioning risk assessment exercise. Chairman 148. Thank you. Next question. Australia (Dr. Gardner Murray) 149. Thank you Mr. Chairman. This again deals with quantitative risk assessment: Assuming one has conducted a quantitative risk assessment, is it not then still a matter of judgement as to how this relates to a country's acceptable level of protection? Chairman 150. Dr. Wooldridge. Dr. Wooldridge 151. Yes. Chairman 152. Dr. Burmaster. Dr. Burmaster 153. Well, I will say yes also, with the 'but'. I guess I will take my examples from another area where risk assessment is performed, and this might be with pesticides residues in foods - not the topic of discussion here. The notion of risk assessment is partly comparative in nature: you want to compare the risks of one activity compared to a different activity, and if a risk assessment finds, or as the fields develop and it is found, that there is great differentials in two different activities that result in two different levels of risk, that is an insight, that is what we are thinking about and trying to understand how that differential arose - is it really intended? Do people wish to maintain that differential policy? Or is there some desire upon viewing this differential to change policies and practices to bring the two risks to be more commensurate - by either raising one risk or lowering another risk? So the purpose of risk assessment is partly specific and partly general. Chairman 154. Thank you. I think that is the response unless anybody wants to say anything more to that. Thank you. Australia (Dr. Gardner Murray) 155. As far as Australia is concerned, we contend we have conducted a probability risk assessment. But having said that, how can one distinguish between a low probability and a possibility? Is not assessing risk as low or small a statement of probability? Chairman 156. Dr. Burmaster. Dr. Burmaster 157. Well it is a statement in the English language to say that something is a small probability. But again I would go back to the example I raised this morning, if we were to conduct a quiz in this room, and I said: I am thinking of a certain activity that has associated with a low probability; and then I asked each of you to write down on a card what you do you think Dr. Burmaster might be thinking of? What could possibly constitute a low probability? I think we would get back answers ranging over several orders of magnitude, such that in my view, let me just continue on that thing, I think that therefore the phrase, low probability is essentially meaningless, it has no meaning. It has no consistently agreed upon meaning even in this room, even in a ten minute interval. By the time you step outside of this room, and by the time you step into different times and locals and different places and speak with different people, many of whom may not have any education beyond high school, and so on, that the word "low probability" becomes a meaningless phrase. It is an absolutely meaningless phrase. So if we wish to proceed, if you wish to do something that is rational, if that is your goal, if your goal is to do something rational - maybe it is not - but if your goal is to do something rational, I think you have to attach numbers or ranges of numbers or probability distributions in order to make any kind of sensible discussion. Chairman 158. Thank you. Dr. Wooldridge. Dr. Wooldridge 159. I think I more or less agree with Dr. Burmaster. It is a problem and I think partly it is a pragmatic problem in that if you, theoretically I think, if you are talking about the pathway for a given disease to get into a particular country, you can always sort out a pathway which makes that possible. You can always make the most unlikely, and therefore improbable assumptions, about a given part of that pathway to say something is possible. But from the point of view of actually doing a risk assessment on whether something really will get in, and if you have come up with the answer that you say it is a low probability, obviously, you are also saying it is a possibility. If you say something is a possibility, you are not necessarily saying it is a low probability, you could also be saying it is high probability or any other probability. I mean, if you look up the words in the English language, they are different in meaning, but from the point of view of really being helpful to distinguish in terms of a risk analysis or risk assessment, I think it is a pragmatic thing; if you have come to the conclusion that you have something that you are calling a low probability, and you need to go on from there, I think the only way you can advance, if somebody is not happy with that, and is not prepared to accept that as is, is to try and make it into a quantitative assessment. So in the end, what one is saying I feel, in my own opinion, if you left in that position that you have a dispute over what "low probability" or "negligible probability" actually means, you have got to go down the quantitative road, that is my opinion. To continue with Australia - Measures Affecting Importation of Salmon: Annex 2, Section 160

Agreements  |  Disciplines | Trade Policy Developments | Countries | Trade & SMEs Home | What's New | SiteMap | Resources | Search | Disclaimer