DR. YOUNGMAN: If we could start with question two now. In the same order, Lyle, do you want to comment on question two.

DR. VOGEL: Okay. Are the current sampling strategies for the retail mea arm of NARMS adequate, as currently conducted? And I guess I have questions about the sample sizes that we are testing. Even though they are increased, maybe even doubled since 2002, I think when we start looking at the rarer serotypes -- well, maybe not so much the rarer serotypes -- but at least in 2002, we are not getting very many isolates of Salmonella Newport; I think it was eight. And even Salmonella Typhimurium, I think we only had 15 isolates.

So we really can’t depend upon that data for giving us good trend data, much less resistance data for that single year. So I think we need a statistician, epidemiologist to really look at that sampling scheme and give good advice on what we need for sampling. But, again, that is related back to what is the objective of the retail sampling. So we need to examine that as well.

I also mentioned, and I have concerns about the E. coli sampling scheme. That is taking samples from only 4 of the 11 FoodNet sites. That if there is regional variation going on that is being missed by that sampling strategy, I would suggest looking at a systematic sampling from all sites. Maybe taking one out of every 20 samples from all sites. Or 1 out of 10, whatever is best epidemiologically.

I also would, like the previous question, I would advise take a look at Enterococci, and if something needs to be cut, that may be the organism that needs to be cut.

DR. YOUNGMAN: Okay, thank you very much. Awa.

DR. AIDARA-KANE: So I think that there is also a need to better define the objectives, and then to sample size estimate. According to the objective, may be helped by epidemiologist. And I think that we need to extend E. coli and Enterococcis testing to all the sites.

DR. YOUNGMAN: So expand from the four sites to all 10 FoodNet sites.

DR. AIDARA-KANE: Yes.

DR. YOUNGMAN: Okay, thank you very much. Scott.

DR. MCEWEN: Yes, just a couple of comments. I agree with what the previous people have said. I think the switch from a convenience to random sampling is a distinct improvement over the past. It enhances confidence and enables better interpretation of the results. I think the integration with FoodNet was an excellent move and strengthen, if possible. That makes a lot of sense in terms of integration with the other branches of NARMS and other activities.

The only other thing I would suggest is consider some pilot studies in other types of meat, like veal and lamb, and things like that. I understand the logic in targeting the high volume products, but in order to be more comprehensive, I think we have to look at some of the minor species and, hopefully, eventually include ready-to-eat products and so on. I think the whole spirit of the retail side is to try to get a closer picture of what consumers are exposed to.

DR. YOUNGMAN: So you are proposing looking at like luncheon meats, and things like that, for Listeria?

DR. MCEWEN: Well, nothing that specific. The only specific ones I would mention is to look at some of the named pathogens in the more sort of minor species of animals. So veal and lamb is two examples of that and, perhaps, others. But in terms of the ready-to-eat products, I think that needs a bit more thought and look for opportunities. But it should be on the list of things considered.

DR. YOUNGMAN: Okay, thank you very much. Marissa.

DR. MILLER: I agree with all the comments so far, so I won’t repeat them. The only thing I wanted to say in addition is I think there are a lot of opportunities that are starting to be taken for working together between the veterinary side and the human side in terms of the, I think it was monthly or quarterly, conference calls. And asking those next questions in terms of how you interpret the data in light of human disease in those areas. So I think there are great opportunities here.

DR. YOUNGMAN: Thank you very much. Sue.

DR. KOTARSKI: Yes, my compliments to the group in providing such a broad database and providing a solid database to go forward. I agree with the comments that have been put forward by the rest of the panel.

The addition that I might have to that is to, again, from the concept of focusing resources, is to consider -- you know, I just looked at the 2002 retain meat data. The number of isolates that were obtained for Campy from ground turkey was 4 out of the 300 and some isolates. The number of isolates of Campy for ground beef was 0 out of the 300 some isolates. And for pork, it was 5 out of the some 300 isolates.

So for the efforts that you expend in obtaining or conducting an isolation, you will get a lot of negatives. I wondered if you could work with a statistician, now having that previous database to say, okay, what sampling strategy do I need to have confidence that I will get “X” number of isolates. If I get those number of isolates, what confidence do I have in the percent resistance that I do measure from those “X” number of isolates? Because in this case, you are going to get from the database that you have generated for 2002, you would predict that you are going to actually get a very small number of isolates. So if one isolate has a resistance to whatever, you would have a 20 percent resistance rate. And so is that a matter of chance or is that truly representative of the area that you are sampling? And, of course, is it or is not representative of a national basis?

Again, we go back to the objectives. I think because of the broad database that you have already generated, by working with a statistician to say, here is the objectives that I wanted to accomplish in my sampling strategy, how many samples do I need, and with what confidence will I have in terms of interpreting the antibiotic resistance that I measure for those isolates.

It could be after thinking about it, you might decide I won’t expend the resources to get campylobacter isolated from ground beef because the likelihood that I will obtain a positive sample is going to be pretty low. So by extension, the confidence that I will have in identifying the resistance rate will be a wide-variability associated with that number. So you might want to rethink your strategy there.

Same comments apply to Salmonella. You had nine isolates in 2002 from ground beef, and you had 10 isolates from pork loins. So, again, it is the same thing. I would revisit that and revisit with the statistician about what is possible and the confidence you would have with the isolates.

The other thing that I might mention is, again, the concept of integration of the program and what is the primary and secondary objectives. If one of the primary objectives is to understand the implications of drug use in animals, and you are using this as an indicator, the retail meat is, of course, going to tell you what your product has as a result of the animal, plus the trucking, plus the commingling, plus the lehrage (phonetically spelled), plus the slaughter house, plus the processing that gets you to the retail meat product.

So it is even more indirect than the slaughter in terms of predicting or understanding what happens at the animal level. And also, when you think about the sampling strategy in using the FoodNet catchment areas which, I think, is an improvement, like has already been expressed by the panel members, to increase and enhance your understanding of exposure to the public at the retail level. It does increase the possibilities that you can do it for those catchment areas.

The implications for the national levels, I don’t have that expertise to comment. But at least that allows that intensive epidemiological sampling for the exposure retail to humans. The sampling strategy that you have for that, versus the information that you have from the food animal is such that we go back to the HACCP Program.

The HACCP Program is designed to get -- the national data level is not designed to reflect the catchment areas. So I don’t know how you interpret, or how you compare, HACCP samples to the HACCP sampling program, to what you have at the retail.

So, again, you come back to your objectives. What is your objectives for risk assessment? Is it for national prevalence levels, is it for epidemiology, and work from there in terms of refining your objectives. Or, excuse me, refining your sampling.

DR. YOUNGMAN: Thank you very much. Sean.

DR. ALTEKRUSE: I just wanted to note something. It looks like the actual prevalence, when the switch was made from a convenience sample to a randomized sample, it didn’t change very much. And I think that -- well, actually, what I am referring to is in the Iowa Study, it was a random sample and then, initially, in the FoodNet sites it was a convenience sample. And the percents were very similar.

So it is good to use a randomized sample approach, but the reason I bring this up is because I think the same sort of thing is likely to apply with the HACCP isolates. They are, in some senses, a convenience sample. But, I think, what we will find is that they correspond pretty closely with random sample baseline.

DR. YOUNGMAN: If I can just interject. That we haven’t shown data for the retail meats, subsequent to it being changed to a random sampling.

DR. ALTEKRUSE: Right. What I was saying is that in the Iowa Study, that was a random design.

DR. YOUNGMAN: Oh, okay.

DR. ALTEKRUSE: And then the convenience design had a very, very similar prevalence of resistance. And now we are going back to a convenience sample. And I suspect what we will find is that the two are very similar.

DR. YOUNGMAN: We are going to a random sample.

DR. ALTEKRUSE: I am sorry, it is Friday. It has been a tough week.

Now, let’s see. The only thing I would suggest in terms of the retail food samples is related to yield. And the skin on products, like chicken breasts, for example, may have a higher yield of pathogens. So if the objective is to look at the pathogens that are being transmitted via these various food producing animal commodities, it may be that, for example, sampling a turkey leg, or a turkey wing, could have some advantages for recovery of pathogens versus ground turkey. And that is, in particular, for campylobacter.

And then using primal cuts like pork chops, or is it pork loin?

DR. YOUNGMAN: Chops.

DR. ALTEKRUSE: But those are not cuts that are from near the surface of the animal. So they are, actually, sort of sterile site type cuts. And you may have a lower yield than if you took something like a picnic ham, or a picnic shoulder, or a ham sample with some skin on, for example, as your sampling unit.

The same thing with ground beef. There is meticulous attention being paid now to sanitary dressing because of O157. That seems to be the control. And ground beef trim is being sampled to make sure that it is free of even indicator organisms, to the extent possible.

So I don’t know if there might be some other product that would enhance the yield of the pathogens that are being transmitted. A skin-on type product, perhaps. Although, the skin is removed from beef, but something near the surface, like a flank. It is just a thought.

And, in fact, I think that the area where they are particular concerned is the back of the neck because that is where the incision is made for hide pulling. So I just throw that out there in terms of the retail product that is most likely to reveal the pathogens that are being transmitted.

But it is a trade-off, because you are getting away from the most frequently consumed product to the product that has the highest yield. It is just something to consider.

That is all I have.