• 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

Using Risk Assessment to Evaluate the Human Health Impact of Resistant Pathogens

Dr. Scott McEwen

DR. McEWEN: Well, thanks very much, ladies and gentlemen. It is very good to be here. I was sitting down here reflecting as Steve was talking that I am kind of doubly disadvantaged. One is I have to follow his act on stage and the other is that it is right after lunch. And I remember as a young faculty member, the first post they gave me was teaching vet. public health. And the lectures were right after lunch.


And I remember seeing a lot of yawning faces and people sleeping. And after one class, somebody came up to me afterwards and said, "Dr. McEwen, if it is my last hour on earth, I want it to be one of your lectures." And that really kind of boosted me up. I felt really terrific after that. And I went away and was thinking about how that could affect my pedagogic style and all that kind of thing. And I thought I better find out some more. So I went back and asked what exactly she was talking about. And she said, "Well, if it is my last hour on earth, I want it to seem as long as possible."


So I hope that is not the case with you. Well, this is not an easy talk to give after we have had so much excellent stuff on risk assessment already. But I have to say that I am really thrilled to be here. I think I kind of live for this stuff. As I say, I have been teaching vet. public health for years. And a lot of it seems kind of esoteric and hard to relate to.


But in terms of the role of veterinarians and the things that they do and effects on public health, this is really cutting edge. This is as good as it gets in terms of a controversial issue that has real importance to society, real importance to us as professionals. And veterinary students just love it.

It is something that -- you know, for those of you who have been around for a while, you have seen it kind of all before. We get these shifts in opinion about what the impact is and how important it is. And I guess that is a natural sort of event.

As we learn more, we sort of engage in more discussion. We go back and forth. And really, as Steve mentioned, he talked about the risk assessment as a process. And I think that is an extremely important concept. That's the way I look at it. I look at it as a process.

And, yes, we can talk about risk assessment as a tool for helping policy and all that sort of thing. But I prefer to think of it in sort of a larger risk analysis context, that is, we are sort of looking for policy decisions, how to manage risk. We are using assessment to fortify that sort of thing. We are engaging right now in communication as part of that.

And I think all of the activities that the FDA has been involved and you folks in the United States on this issue is really an example of risk analysis, risk assessment and process.


We can talk about policy. I am not a policy expert. I just kind of work at a vet. school. But there is a lot of kind generic things that come out about general principles for policy. You've got to focus resources on those things that really matter, those important questions, the primary issues.

We've got to try to make decisions, or at least policy-makers have to make decisions sometimes when the information is incomplete. And that can be very frustrating and especially when the consequences are not clear.

They also have to involve the greatest number of those who must be around to implement it. So they've got -- as I said before, this was driven home to me when I was in Berlin at the '97 meeting. And there the discussion was around risk assessment, risk management.

And I saw before my eyes this kind of notion that people that may not have felt totally franchised -- were unenfranchised I guess is the way of saying were -- created all kinds of problems. And it really drove home to me the notion that people have to be involved in the process if they are affected.

I have a quote by Henry Kissinger in his recent memoir. And he was talking about the principle for Presidents when they are deciding on foreign policy. But I think they are very much generic. They certainly apply to this risk assessment of drug use field, as well.


I think one of the take-home messages that I would like to deliver is that there are really different types of risk assessment. The term gets used a lot of different ways. The one that we have seen today is one in its purest form, I guess. But there are a lot of other different varieties.

People sometimes talk about epidemiological studies, hypothesis testing and observation as a type of risk assessment. Really they are looking for evaluating risk factors, trying to identify risk factors of disease. And it is a form of analysis when we are looking at risk. But it is a kind of a different thing.

We will also talk about results of outbreak investigations and trace-back studies. Those types of studies where we attempt to identify through, as said here, molecular fingerprinting, but other ways of -- like clones of bacteria might come through the food system. And those are valuable -- provide valuable bits of information for risk assessments that support the kind of quantitative stuff we've talked about today.

These studies are often descriptive in nature. That doesn't sort of diminish their importance or their value. It is just a different way of looking at things, at different information.

We've got the -- what I would call the FDA study is a type of ecologic or population level scenario analysis. We are looking at the U.S. population, the total U.S. poultry production, that sort of thing. It is an ecologic study. It is kind of a separate type.

There is also, as David Vose referred to, a type of mechanistic or systems analysis, process risk model, the type that we are seeing evolving in the microbial field which is, again, a different approach. And I think some that we should see more of than we have in the past is more theoretical studies involving population biology, population genetics and that sort of thing.

But I guess the bottom line for all of this is the approach that we take very much depends on the questions that are being addressed and the purpose of the assessment. And that is something that has been stated already, but it can't be over stated.


I guess in the past, we have seen a lot of sort of evidence or a lot of weight put on trace-back studies as a way of assessing the risk of antibiotic use in agriculture. And they still have an important role. And I guess I just need to fortify for you folks -- you probably don't need it -- that there is a lot of difficulties and challenges in that. But it still is a useful way of gathering information.

So, for example, treatment of cattle on farms. We are involved in some studies right now up in Canada with Doug Powell and Richard Reed Smith and some others trying to quantify and describe the types and extent of drug use that is going on in animal agriculture. So what kind of impact did that have on human health?


Well, there is lots of difficulties in following, obviously, the treatment information through this system and the impact in terms of drug resistance through the system. Animals go to slaughter. They may go to auction marts, go to different farms. In all of those different locations, they encounter other strains of bacteria that may have acquired resistance elsewhere or they may supply them to different animals.


Again, when they get to the packing plant, the slaughter plant in this particular beef example, again, we know that there are lots of changes that can be produced in the bacteria, both quantitatively and qualitatively. And we can introduction of new strains. We can have a growth of microorganisms and death due to various types of processes.


So what do all these kind of changes and dynamics have in terms of the impact of human health which is the main sort of index that we are interested in? And, again, we have had looking through the literature a variety of studies that have evaluated this and have provided good information.


Okay. What are some sort of broad applications to risk assessment in this domain of drug resistance from agriculture? I think something that hasn't been maybe emphasized enough is that there is a value here in pre-approval assessment. We have been focusing today in many places, we've focused on those drugs that are already out there in the market. And we have a resistance arising. And we are looking at the impact in that sense.

But now from a more sort of pragmatic purpose or theoretical basis, it would be great if we could attempt to anticipate the level of impact before drugs are actually approved. And there are difficulties in doing that, but there is also a lot of utility and possibly good value there.


In other fields, if you are involved at all in food microbiology, you know that risk assessment, as Dave mentioned this morning, is being used extensively in trying to better develop food standards. What is the allowable level of microorganisms in food at the time of consumption? The same thing for water microbiology.

We also can use them for hypothesis testing. What would happen if we do such and such? What are the effects of this intervention, for example, judicious use or prudent use in animal agriculture? What effects could that have on the level of resistance, the impact to public health.

And I don't think that we want to forget that another kind of application of these -- this approach is better understanding the biology of the process and better understanding leads to better decision-making as we said. And the thing that always gets left to the last and is hardest to do is trying to assess the economic and social impact.


Well, you are not expected to actually read this at the back. I was sitting there before and I know how hard it is to see the screen. Basically, this is one of those slides that shows the complexity of the interrelationship among all of the environmental and other factors on resistance.

And I guess I put it up here just to underscore the kind of difficulty and sort of shock that one would get when you try to think about developing models that can capture all of these things at once. And so we really do have to make choices because of the incredible complexity of this system.


And I think the choice the FDA has approached is a good one. It is similar I think conceptually to the -- to a risk assessment approach that was taken more than a decade ago by the National Academy of Sciences and, basically, again looking at the ecological impact of the sort of national level of resistance in animal agriculture and impact on human health.

We do have to remember that the ecological studies are very useful. But they also have some limitations. And some of this has been brought out this morning when we were talking about representativeness of samples.


Basically, these types of studies -- epidemiologists refer to ecological studies is where the unit of observation is really the group. It could be a community or a national sort of level. The exposure, in this case, the exposure to drug-resistant bacteria I guess is -- and the disease are sort of measured at the group level. And there is lots of reasons for doing that.

But one of the problems is that you kind of lose a lot of information that applies to the units of that group. And it is impossible basically to control for any confounding that may be happening at that sort of level. So in some cases, it is a problem. But that doesn't mean that the studies have no value.


I think we have to realize, again, that we do have a very hierarchical set of levels of organization and both in human society and in the way we have sort of managed farm animals. I could make this pyramid using a sort of farm structure. And it would have the national herd at the bottom, the states, sort of farm level, pens or different sights or pens and so on.

So we do have a very kind of hierarchical system which has major implications towards our sampling plans, for surveys. It has implications to dissemination of microorganisms. And it is very important eventually to try to capture these kinds of levels of organization if we are going to sort of better understand the process.


I am not going to get into any kind of technical details about how risk assessments are being done in other fields. It is better left to the experts.

But I think it is important to understand in addition to the sort of ecological approach, there is this sort of mechanistic or process approach where we tried to follow the animal, the food product through the system of slaughter and processing, and try to measure or at least anticipate the effects that the interventions or the treatment effects or the heat treatment or the cross-contamination is going to have so that we get a better understanding of the quantity of bacteria that people are being exposed to I think at any given point.

And has been mentioned by Dave and others this morning, that we attempt to model what effects that quantity of exposure will have on human health that takes into account the variability in human population and all those factors to try to, again, characterize risk in some quantitative way.


Okay. We'll skip that one. Now, in terms of just trying to schematically present the amount of information that we have, I think in a sort of rough way, highly unquantifiable sort of way, this is my impression. In terms of the four traditional categories of risk assessment, I would say that we have got proportionately a tremendous amount of information on the hazard identification step. What are the nature of the microorganisms; the genetic basis of resistance; the ways that the resistance are transferred between microorganisms, that type of thing. And I think we need more of that. We have got a tremendous cadre of microbiologists and other biologists out there doing that kind of research.

But I think comparatively, we have very little information on the other steps in risk assessment, the exposure assessment phase and dose response and ultimately the characterization step. So what the implications are is that I think risk analysts concentrate on the exposure and dose response part of the equation for very good reasons. But often people in other domains don't sort of recognize the importance for doing that. And we sometimes simplify the hazard ID phase and concentrate on the exposure and other phases. And we have to try to communicate to microbiologists and some others that aren't sort of working in the field why it is important to do that because that is where the uncertainties are, that is where the data gaps are. And that is why it is important to assessment.


Okay. We talked about in this particular assessment -- we will have comments later -- but the end point being exposure of people to fluoroquinolone-resistant bacteria. That's fine. Good reasons for that, regulatory reasons for it.

But, again, in the larger scheme of drug resistance from agriculture, there are other possible end points as Steve mentioned in his talk. We will look at that at some point. I guess one that keeps coming back to me and I am not really sure how big a deal it is is this notion that there is going to be disease in the community that arises because people are taking drugs for other reasons. And when they do that, then they are more susceptible to challenge from drug-resistant bacteria. We have got lots of examples in the literature from Salmonella. I don't know about Campylobacter. We can't I think forget that and try to measure it in some way.

And, again, people also talk about the pathogen load phenomenon, gene transfer and the possibility of increased virulence which has come out in different discussions. We have a variety of different hazards that need to be addressed.


I think another point I would like to make is that in some cases we can focus our efforts on certain components of the ecosystem if you want to call it that, the whole sort of domain of animals, the production systems and processing and so on. And the FDA approach which, again, is appropriate, we focus on the sort of ecological level.

It may be appropriate in some instances to go sort of back in the system and look at other aspects. For example, I am kind of most interested in the pre-harvest phase of animal production. And I think there is a lot that can be done at that level to try to sort of reinforce or further refine the exposure assessment phase of risk assessment.


And just as one example, one of my Ph.D. students, David Jordan, did some simulation studies looking at in this case not drug resistance, but E. coli 0157 in beef production in Ontario. And he was interested in how different management systems, different ways of trying to mitigate the risk of 0157 might transpire, might sort of feed through the slaughter system in terms of reduced exposure of positive animals in the feed lot.

This is an animal with a kind of heavy tag on his carcass. So basically, in this particular approach which is quite different than the other risk assessments you've heard of, basically it is one of devising a scenario which basically describes the system of beef production and collection and transport to the slaughter plant and says, okay, in an attempt to address what we would happen if we were able to, say, administer a vaccine.

We don't have a commercially available vaccine yet. And it would reduce the quantity of bacteria of 0157 being shed in feces or the prevalence of positive animals in the slaughter plant. What effect might that have on the eventual level of contamination of the slaughter plant? And then he also looked at other types of interventions.

But I guess the point here, as I was saying before, is these are hypothetical. We don't have them yet. But this is a way that we could attempt to identify the impact that they could have. And if it looks promising, invest more resources in research to get them.


And this is an example of output from Dave's model. And I think the main sort of thing to look at is that on the kind of left side of your screen if you can't read the words, the main thing is the shape of the curve, is that given what we sort of know about 0157, we can bet that just about on any day of the week, there is going to be bacteria coming into a beef packing plant.

But if we are able to test animals and positive lots were either excluded or in this particular case moved to the end of the slaughter queue, then we could shift back in the day the sort of first time that sort of positive animal comes into the packing plant. And this could conceivably reduce the level of exposure, the level of contamination.

It is not a public health measure per se. But it is a bit of information of exposure assessment that could be used in a public health risk assessment.


I think we also have to acknowledge that there is a lot of different types of scientific expertise that need to feed into this exercise. We have heard about the mathematical and statistical components and the microbiological ones, as well.

We have also got I think to look a little bit more broadly into some of the other areas of biology and so on that have an effect on resistance. We've actually got an expert here -- I haven't met him yet -- Mark Lipsitch I think who works in evolutionary biology with Bruce Levin's group, or had in the past at least. And there is some excellent work going on there in terms of the creation and the selection of resistant organisms in nature.


So I would just like to follow through on that particular theme. I think we have to sort of, again, look beyond our sort of obsession with real data. I think any of us who have had any kind of medical training or in other fields for that matter want to see some data.

Show us the results. Show us the information and we will believe what you say. But in other instances, it may be appropriate to be less reliant on this quest for data and actually look to theory, look to biology for some ways around these problems.


And as one particular example of this, Roy Anderson's group in Oxford, has been looking at the temporal changes in drug resistance in human populations and from a theoretical basis is showing that under a constant selective pressure of antibiotic, that we are going to have over time a sigmoid sort of relationship between acquisition of resistance.

And also using these approaches, his same group has shown that with intervention studies -- interventions in this case reducing the amount of antibiotic use in a human population, that we do see a decrease in pneumococcal resistance. But I think the important point from these theoretical studies is that the decrease is much slower to be realized and takes a lot longer and doesn't sort of tail out completely even in the absence of antibiotic treatment or in its reduced use.


I think in the interest of time, I will finish off with this point. We have got a lot to learn about antibiotic resistance. And I think that we have the same sort of sigmoid curve here where we have got a lot of uncertainty and some particular -- for some particular drugs, some particular pathogens. And for others, we know a lot about the system. We have less uncertainty.

And those of us in the room, the individuals can put ourselves on this curve in different places. But I think the point is for policy-makers, we have to realize that sometimes they are forced to make decisions along various points in this curve. And that is a challenge. So with that I will stop and entertain any questions.


DR. LONG: Any questions for Scott? Great. Okay. We will go on. The next thing on the agenda is to look at two other risk assessments that have looked at this antimicrobial resistance issue. And to help us to evaluate, to help us put this new risk assessment that we are here to talk about today into context with what others are thinking. And our first speaker is Steven Anderson. He is currently a AAAS science risk policy fellow in the Epi. and Risk Assessment Division at the Food Safety Inspection Service. And before this fellowship, he was research fellow at the Georgetown Center for Food and Nutrition Policy. While he was there, he got his masters in public policy and conducted risk assessments on antimicrobial resistance in cattle.