• Decrease font size
  • Return font size to normal
  • Increase font size
U.S. Department of Health and Human Services

Animal & Veterinary

  • Print
  • Share
  • E-mail

Setting Thresholds and Next Steps

(Audio missing.)

DR. STERNER: --- small animal and exotic practice. She holds her bachelor's degree from Harvard University -- excuse me, Harvard University in 1983 and a D.V.M. degree from the Virginia-Maryland Regional College of Veterinary Medicine, 1987. Dr. Thompson, it is my distinct pleasure to turn the podium over to you.


DR. THOMPSON: Okay. Well, my big benefit was going to be that I was going to get us out of here early because I had budgeted extra time so that I could accomplish that. But you have done such an excellent job, Keith, that we -- I don't even have to work at it.

I wanted to just spend a minute commenting on some of the points that people made in terms of I guess their disappointment that we had not gotten more into the subject of thresholds at this particular meeting. As many of you know, initially in the planning of this meeting, we did plan to have a whole session to talk about the establishment thresholds.

But similar to you, we basically got -- CVM got the draft risk assessment model very late. And that was through no fault of anybody's. But that was the reality. And we just did not feel that we would be prepared to discuss not only the validity of the model, but exactly how it would be used in terms of the establishment of thresholds.

So just to explain to you that we certainly do think that the issue of thresholds is an important issue. I am going to provide you some very preliminary comments today. We do plan to look more at this issue in the future. And so just to give you a little bit of background on that.


First, I wanted to start by giving people some history in terms of how we talked about thresholds in the Framework Document and then to talk about how this could fit into the risk assessment model itself. In the Framework Document, the FDA talked about two different kinds of thresholds, a resistance threshold and a monitoring threshold.

The resistance threshold really was envisioned as being the upper limit of resistant bacteria that could be transferred from animals to humans and still be considered safe. And in the document, we basically had talked about this threshold being established in humans.

The monitoring threshold was viewed as being established either in humans or in animals. We didn't define which and actually asked for comment on that. But it was envisioned as being an early warning system so that when you were approaching the monitoring threshold, basically the monitoring threshold could either be loss of susceptibility or frank resistance in terms of a resistance prevalence.

And when you were to approach that, the sum action would be taken, basically mitigation action in terms of further investigation or potentially changes in terms of how the drug was used on the farm, changes in management practices. That was what was envisioned in terms of a mitigation action.

As I said earlier, basically the resistance threshold was defined in humans. And for Category 1 drugs -- I would like to focus on that today -- the Framework Document stated that the resistance threshold would need to be zero or very low. This doesn't necessarily mean that resistance in animals would also necessarily have to be zero if data was available to show that some level of resistance in animals would not result in crossing the resistance threshold in humans.

For each Category 1 drug, the Agency would need to define the end point of concern. And what I mean by that -- I will give you an example. The Framework Document discussed resistance in Salmonella as the end point of concern for quinolones. Therefore, resistance developing in other pathogens such as Campylobacter would not necessarily raise the same level of concern as resistance developing in Salmonella.

I don't mean to say that we would not be concerned about resistance in Campylobacter. Just in terms of the human health impact, we would be more concerned about resistance in Salmonella.

The end point is, obviously, very directly related to the risk standard. Now, as Linda earlier had mentioned, in terms of the Framework Document, that was defined as the loss of availability of safe and effective antimicrobial drugs to treat human disease. For Category 1 drugs, the end point was more specifically highlighted as the loss of significant human antimicrobial therapies when alternative drugs were limited. So there was a consideration of alternative therapies in terms of the categorization process.

Linda had also mentioned that we have put out an analysis of the comments that we received on the Framework Document. And in case anyone has not gotten that, it is out on the -- copies are out on the table outside.

But I wanted to briefly mention a few points with respect to the thresholds. Basically, we received many comments as were made earlier, as well, about the need for extensive public dialogue and stakeholder involvement as we move forward, and especially on the issue of threshold. FDA definitely agrees with that.

I think we -- and the reason I was late getting up here -- and I apologize -- was that I was following up with Dr. Sundberg in terms of what were his thoughts on how we could design a better process in the future in terms of interaction in a workshop. So I do think that that is very important.

We also mentioned in the comment analysis that because really of the difficulty that we envisioned in establishing thresholds, that we are considering limiting that requirement in terms of a formal threshold only to those products that would be classified as a Category 1 product.


Okay. In terms of setting thresholds, we really envision that there is two ways that it could be done. One way would be what we consider a technology-based method and the other would be more of a health-based method.

In terms of a technology-based method, what we mean by that is that a technology-based threshold typically would be established by measuring the amount of contaminant currently present. So, for example, HACCP limits for Salmonella contamination on carcasses were established by measuring the current level of carcass contamination.

And then if a qualitative risk assessment were to suggest that that amount represented an unacceptable risk, then further regulatory action could be taken. In the HACCP regulation, USDA concluded that the current food-borne disease burden due to Salmonella was too high and required the levels on carcasses to be lowered.

For antimicrobial resistance in animal food-borne pathogens, a technology-based threshold could be established by measuring the amount of resistance present in the food-borne pathogen for approved products or the amount projected to develop based on pre-approval studies. And if that level was viewed as representing an unacceptable public health risk, strategies could be developed to decrease the disease burden or the resistance level.

While technology-based thresholds have an advantage in terms of the ease of establishment, the values are not necessarily tied to public health outcomes.

The other method that is routinely talked about in the literature is health-based thresholds. And these are usually established based upon a safety assessment. Since public health risk is a product of hazard times exposure, health-based thresholds are generally established by performing a comprehensive evaluation of both the hazard and exposure.

Establishing health-based thresholds, however, on the basis of a quantitative risk assessment for all antimicrobials and all pathogens would be difficult and resource intensive due to the lack of quantitative data on public health outcomes related to the use of antimicrobials in food animals. And in some cases, these also may be difficult to directly relate to public health outcomes due to uncertainty and the quality of available data.

The risk assessment model does facilitate the establishment of thresholds because it builds a link between resistance levels of animals and resistance levels in humans and ties that to a human health impact. The ability to link these two can assist us a regulatory agency in setting thresholds.

But, however, as we heard during this meeting, we really must first define certain questions including a more clear definition of the risk standard. And then we must also talk about what is the regulatory end point of concern.

If we go with the definition of reasonable certainty of no harm as defined in the Framework Document where we look at loss of available therapy, then we would need to look potentially at the particular drug or class of drugs and say what are we most concerned about in terms of resistance development with this particular drug and the particular pathogen of concern.

So one approach could be to use a hybrid of a risk assessment and a safety factor approach to established thresholds. For example, the complete risk assessment would be conducted for the pathogen that develops resistance the soonest or what we could call the sentinel food-borne pathogen in the animal species associated with the most food-borne disease due to that pathogen. And we could call this the reference animal species.

For example, with quinolones, we could choose resistance developing in Campylobacter and this would be the sentinel food-borne pathogen, in chickens, which could be the reference animal species. The risk assessment model could then be used to determine when an unacceptable human health impact is reached for the resistance threshold established in humans.

And furthermore, to calculate the level of resistance permissible in the sentinel food-borne pathogen on the reference animal species at slaughter -- and this would be the monitoring threshold -- the monitoring threshold could then be applied to all other species and be protective of the public health because the food-borne disease burden from other species associated with that particular pathogen should be less than that of the reference species. And that is inherent in how you define what the reference species would be.


For food-borne pathogens with health impacts greater than that of the sentinel bacteria, it may not be possible to wait until resistance develops to assess the public health impact. And this point has been brought up during the meeting before, that you may not want to wait until you have enough data to a quantitative risk assessment and judge what is the human health impact because at that point, you know, you are already too far.

So, for example, specifically mentioning the issue that has been talked about, the Agency may not want to wait until fluoroquinolone resistance develops in Salmonella to support a full-blown risk assessment on this Salmonella-related human health impact.

In this case, a safety factor could be determined and applied to the monitoring threshold established for the sentinel bacteria to be protective of the public health. And mitigation action could be warranted when either the monitoring threshold in the sentinel bacteria or other food-borne pathogens would be reached.

So in this kind of approach, we would have more than one monitoring threshold for fluoroquinolone resistance that would trigger the need for mitigation. One might be in Campylobacter derived from a quantitative risk assessment and another might be in Salmonella, either reduced susceptibility or low level resistance depending on where we go, derived from a more qualitative risk assessment and the application of a safety factor.


I want to talk a little bit also about some other critical risk management tools. I mean, we are focusing here on thresholds. But I really do think it is very important that we believe it is critical to also look at judicious use of antimicrobials. I think we -- Dr. Sundlof mentioned this and others have mentioned how supportive we are of the work that is going on by the AVMI. And we really do believe that this is a critical piece of the equation.

The application of these principles are critical in managing the risk of antimicrobial resistance by limiting the use of important human antimicrobial in food-producing animals to only when it is really necessary and thereby reducing the selective pressure for the development of resistance.

In addition, another critical piece of the equation has also been mentioned during the meeting, the impact of HACCP and what impact that has in terms of overall food-borne disease. While we believe the risk assessment was appropriate designed to estimate risk to human health from resistance food-borne pathogens associated with the use of antimicrobials in food-producing animals, the current apparent effect of HACCP is to reduce human exposure to food-borne bacteria which could concurrently reduce illness of people.

So this is something you can't ignore in terms of the overall management of risk because if overall food-borne disease burden goes down, then concurrently hopefully resistant food-borne disease would go down. So I think -- we feel that although we think we have looked at the issue from our prospective appropriately, we also feel that this is a very critical piece of the equation.


Now, I just want to make a very few comments in terms of next steps, first, focusing really on the risk assessment itself and what we plan in terms of moving forward in terms of that. Basically, we do plan to review comments made both at this meeting, as well as comments made to the docket. And I have put here the docket number to submit comments to us on the risk assessment model.

We will consider whether the model, itself, should be revised. There were some comments made during the meeting in terms of suggestions that we should have looked at certain aspects. So we will certainly review those and make an assessment as to whether we believe the model should be revised.

We will also consider any additional data that is submitted to us as part of the comment process, either data that is submitted or referenced either at the meeting here or in the comments. We will look at that.

We will also consider suggestions to generate new information to refine the risk assessment. For instance, if an industry group has an idea of data that could be generated that they are interested in collecting that would perhaps answer some of the questions or address some of the data gaps that are identified in the risk assessment, we would certainly more than welcome conversations on that and suggestions on the data that would be most useful.


And then finally, we do plan to publish the final risk assessment after we consider the comments. We will try to address all the comments as much as possible in the final report. We will try to either clarify points or, obviously, modify things or include additional data. So we will try to improve the description.

There has been a number of people who have pointed out certain things in terms of -- either during the meeting or on the side about some need to clarify certain pieces of logic in the report. And we will certainly do that.

We are also aware -- it has been made aware to us the inconsistencies in the current draft. It has been pointed out that there are variations between some of the charts and formulas amongst the sections. And we do plan to try to correct those inconsistencies and put up a revised draft in the next few weeks. So -- and basically, I beg your indulgence.

We were more concerned with getting it out to the public so that you would at least have some time to review it before the meeting and rather than the report being a perfect draft. So we do plan to correct that and we will post a revised draft in the next few weeks.

In addition, we will be putting up the -- a spreadsheet on the web so that you can actually download that and look at the data yourself. So we will be putting that up and making that available to people. And that will be on the CVM home page.

Now, moving from the risk assessment in terms of the issue of the risk standard, we would certainly also appreciate additional comments submitted to us in terms of that particular issue. I think, at least I hope that you got a sense of the fact that this really is a very difficult issue that we are struggling with. And we do want stakeholder input on this issue. So we would look forward to additional comments.

We do also plan in terms of additional meetings, we plan to have a meeting on the design of pre-approval studies. That is currently scheduled for February 22nd through the 24th. And we hope to have an agenda, a draft agenda available soon. And that will be posted on our home page. So look for that, as well.

And we will also hold additional meetings as needed. Obviously, the issue of thresholds needs further discussions. So we do plan to engage the public on that issue, as well after we have looked further at the risk assessment and the comments that are submitted on that model and basically made some assessment of how we can use this.

So I think we do want to move forward as quickly as possible on that. But we felt it was important at this meeting to at least first get some validation and opportunity for people to give us comments on the validity of the model itself. So we do plan that. Also, monitoring that has been -- as some people have suggested, that we need to hold a meeting on that, as well.


In terms of future risk assessments, I think many of you are aware that we are also planning to do a risk assessment on enterococci. And I was listening very closely during this meeting in terms of public process. And I think that one message that I am certainly taking home from the meeting is the need to begin the dialogue early.

And so I think that is very important. And I fully agree with that. And so what I would like to do in terms of as we move forward into the next risk assessment is to develop a public process, at least have some sort of notice defining the scope of the risk assessment that we are looking at and a call for information in terms of any relevant information on the issue and suggestions for how potentially the model could be designed.

So I think at least I have heard that very clearly from people. And if people have any other suggestions to make to me in terms of how to deal with communication in the future on this, I would certainly appreciate that, those comments.

And finally, just in closing, I want to thank you for everybody's participation in the meeting, especially the speakers who I know in terms of organization, I pressured a lot of people to come on very short notice to the meeting. So I really do appreciate that. I appreciate everybody's input into the dialogue at the meeting.

I think that, at least from my perspective, it was a relatively balanced meeting and we had some good discussion, sharing of views but, as Dr. Sundlof said in terms of ground rules, no personal attacks. So I think that was excellent. So I will close there and answer any questions that I can.


DR. STERNER: Questions for Dr. Thompson Robert.

DR. CONDON: Could you clarify, you are going to put out another draft or something in the near future and would it be best to wait until that comes out and comment on it Because -- could you kind of maybe highlight a few of the things you are going to change like some of the arithmetic differences and some of that

DR. THOMPSON: Yes. And I mean mainly it has been -- and I may ask David Vose also to make a comment on this. But there were some -- we were working people in disparate places and people were out of the office for certain periods of time. And so there were some inconsistencies in some of the charts and the formulas from different parts of the draft.

In terms of really the discussion or the issues that were presented, that is not going to change. You know, we would clean up some of the typos, too. And that is -- if you would like to wait for that, we do plan to do that in the next couple of weeks.

But in terms of the issues that we are posing and how the model is constructed, none of that is going to change. It is just cleaning up some of the presentational issues like that. I don't know, David, if he is here or if he wants to comment on that either. But -- sure.

MR. : I would like to make a suggestion that you when you do the next draft, you put line numbers so when we are making comments.

DR. THOMPSON: Okay, we will try to do that.

DR. STERNER: Dr. Richard Carnival is recognized.

DR. CARNIVAL: Yes, I am Rich Carnival from the Animal Health Institute. And, Sharon, recognizing there is going to be continued discussions on this threshold it sounds like and further workshops, there are some questions that have been bothering me for a long time about thresholds.

And I just thought it may not be fair to ask you these now because you probably can't answer them. But I thought I would want to get them out there for the record just to have people think about the idea of thresholds and exactly where we go with them.

First of all, it has been troubling me for a while as to how FDA would, in fact, enforce thresholds. I think that is a big question on the industry. Now, one way I see that they could enforce thresholds is taking action against veterinarians and producers using the product. That is what happens with residues.

When tolerances are exceeded, there is usually investigation that occurs. And the FDA goes back and looks at who might be responsible for causing that residue so action is taken against the veterinarian and producer. Would it be envisioned that the FDA would put out some sort of general notice banning the use of this product or greatly prohibiting the product

Short of that, it sounds to me like the Agency might be considering taking action against the producer -- or against the manufacturer. So it would raise a question as to why would the action be taken against the manufacturer, what justification would there be for that when, in fact, the manufacturer is simply supplying the product.

They are not necessarily using it and causing the resistance that is occurring. So there, I mean, there is a question in my mind how these thresholds would be enforced. So you might want to answer that one.

The second part of the question is the current HACCP sampling is really not statistically-based at the moment. It is about the best the USDA can do because they are looking at Salmonella and they are testing Salmonella based on their program for trying to set standards for Salmonella plants.

But it is really not statistically-based. The kind of threshold monitoring you are talking about seems to me would entail a much larger program, one that is representative the nation's food supply as a whole with multiple species and multiple pathogens.

It sounds to me like a major increase in the NARMS type program. Is that envisioned and who would pay for that

Finally, it seems to me that the methods that are used in the detection and susceptibility testing would have to be validated just like methods that are validated for drug residues. I mean, there is a very elaborate process that goes into validating analytical methodology for drug residues.

And we all know how expensive and difficult that process is. And this, obviously, would involve the NCCLS and other agencies in trying to do that. So I guess it is fine to talk about thresholds. And we have been talking about them and listening to different concepts for the last year.

But I think these are some real, hard core questions that at some point in time we have really got to put on the table for the industry and say this is how it is going to be applied; this is how it is going to be enforced because, otherwise, I am afraid we could talk about this for the next ten years.

And as it stands right now, you know, the drug approval process is kind of being held hostage. So I just

-- if you could answer any of that today, that is great. If not, we will hold it for another time.

DR. THOMPSON: Well, I will make just a couple of comments, Rich. Obviously, it is a very difficult area which is why we need additional dialogue on it. So I can't answer your whole question.

But in terms of really the first question you posed in terms of whose responsibility, you know, focusing in really on the monitoring threshold, I think we have had some dialogue on that with the industry. And what we envision with that is for that to be, you know, the point where we would say some mitigation is needed.

And I think what we put out initially in the Framework Document, as you may remember, was kind of requiring drug sponsors to do on-farm monitoring programs from the onset.

And so some of the information that we would be looking for in terms of information potentially to aid us in mitigation in terms of intervention strategies, we would -- our idea was we would have some of that information from the very beginning.

And that would aid the Agency in encouraging the industry, both the drug industry as well as potentially the individual producer in the industry, in terms of the appropriate mitigation so that the product could stay on the market.

If you have looked at the comments -- comment analysis that we put out, basically we are saying now we don't believe that we would need to have -- or we are not -- we are moving away in terms of saying that we would require on-farm programs for all products. So we have moved away from that.

But in terms of when we do start to approach that monitoring threshold, I think we would still go back to the sponsor and say we need to do some investigation because, otherwise, we won't be able to tell, for instance, the producers, we will not be able to make those appropriate label changes to allow the product to continue to be used safely.

So I think what we have envisioned and I may -- Linda, if you certainly have any additional comments -- but think what we envision was at that point in time, we would need to really do some investigations to try to identify what are the risk factors that could be addressed in terms of mitigation so that, you know, resistance could be managed.

And in terms of action in terms of the Agency, what we may do may be certain changes in terms of how the drug is used, label restrictions, that sort of thing.

So I think it is a combined effort, at least that is how I would like to view it. But, obviously, at least from our perspective, the drug sponsor has a major role to play.

And in terms of the other more technical issues, in terms of laboratory validations, robustness of the NARMS program, statistical significance, I don't think we are there yet in terms of saying how we would define when we have reached a certain threshold level which I think is what you are getting at, what statistical basis there is in terms of saying that we have reached that. And so I think there is more discussion on those issues.

Linda, do you have any additional comments on --

DR. TOLLEFSON: No, I think you covered it well. I really do.

DR. STERNER: If you are going to talk, come to the microphone.

DR. TOLLEFSON: I am Linda Tollefson. I am Director of the Office of Surveillance and Compliance. And what Sharon said about the thresholds, I fully agree with it. And I thought pretty much how we laid it out in the Framework Document.

However, you raised a question in the beginning about would we go -- would we treat it like a residue and trace it back to the producer or the veterinarian. And, no, is the answer. We never envisioned doing that. We see no purpose in it. If you want to provide comments as to why or what rationale.

I don't understand what that would get us. I mean, what you are thinking of is individual misuse maybe. Right

(Away from microphone.)

DR. CARNIVAL: Well, I wasn't necessarily suggesting --- problem --- same kind of action taken ---.

DR. TOLLEFSON: Right. So you are thinking that it was being like a violative residue, we would be approaching the resistance or monitoring threshold. And, no, we never considered treating it as a result of an individual producer or veterinarian's actions.

(Away from microphone.)

DR. COPELAND: Linda, in that same regard --- resistance ---. And I think that needs to be ---.


DR. STERNER: Could you repeat the question for --

DR. TOLLEFSON: Right. Go ahead, Dennis.

DR. COPELAND: I'm sorry. I thought I could sneak it in without getting up to the microphone. I am Dennis Copeland with Bayer. And I just pointed out that I would envision that if resistance develops, you are going to have pockets of resistance where, you know, maybe in most parts of the country, there is no resistance. But there might be in one location. And I think somehow that has to be taken into consideration.

DR. THOMPSON: Actually, and I know Linda wants to say this, too, but I am dying, too, is that that is actually the exact reason that we said that we need more specific drug use information so that we could address that exact issue and look at things at more of a regional basis.

But --

DR. TOLLEFSON: Right, exactly. If we rely on NARMS to monitor those monitoring thresholds, we will not know any kind of regional variation or differences. And, in fact, we won't detect it. What will happen is it will just simply be wiped out and we won't see an increase or decrease in susceptibility or increase of resistance because it is not powerful enough.

Combining the drug use information with the trends in susceptibility would give us a better handle on that. But even so, it is pretty difficult.

DR. STERNER: Further comments or questions for various either panel members that participated today, speakers or CVM members Your opportunity to speak is rapidly disappearing because it may have to do with the lateness of the hour. I will point out to you, however, that we appear to be 45 minutes ahead of the scheduled departure time. And with that, we are now officially adjourned.

(Whereupon, at 3:40 p.m. on Friday, December 10, 1999, the Workshop on Risk Assessment and the Establishment of Thresholds was concluded.)