by Dr. David White

DR. WHITE: I am going to take, hopefully, the next 20 to 25 minutes to talk about my real job. I am a team leader for the NARMS retail part, and I would like to share with you how it started, the history of it and some pre-work we did before we went into the NARMS to become the third arm and end with a little of data.

(Slide)

Of course, as background, the food supply, including meat and poultry, is an important source of enteric bacteria, including Salmonella, Campylobacter, E. coli and possibly enterococci. I think the debate is still on enterococci coming from foods as pathogens.

Antimicrobial resistance among these foodborne bacteria is not uncommon and is often associated with the use of antimicrobial agents in food animals. Retail food represents a point of exposure close to the consumer, and when combined with data from slaughter plants and on-farm studies, may provide an indication of the prevalence of resistance in foodborne pathogens.
(Slide)

So, I would like to start with the history of microbiological surveys, focusing on animal derived meats from CVM’s perspective, and our primary objective has always been to characterize antimicrobial resistance among isolated from retail meats of animal origin in meat and poultry.

I will talk about three studies in particular. The first one was our collaborations with the University of Maryland. In particular, Dr. Jianghong Meng, looking at the greater Washington, D.C. area as our sampling areas.

We will talk a little bit about the Iowa project, which was our pilot project for NARMS retail, and then how we expanding into retail meats as part of the NARMS program.

(Slide)

So, we started off with back in 1998/1999, partnering with Dr. Jianghong Meng, and we initiated a study looking at retail raw meats in the greater Washington, D.C. area. We tested 825 samples of retail raw meats for E. coli and Salmonella, 719 for Campy and these were random samples taken from 59 stores, published in the Applied Environmental Microbiology in 2001.

Of course, one of the things that catches your eye, in terms of poultry, we had 71 percent positive for Campylobacter. So, it was one of the first indications that Campylobacter was highly prevalent in chicken breast. Now, this is pre-HACCP. Okay? These were samples pre-HACCP.

(Slide)

We followed up with another study with Dr. Jianghong Meng that we were fortunate enough to publish in the New England Journal of Medicine looking at ground meat samples from the same area, and we looked at chicken, beef, poultry and pork, and these were ground samples. We were able to go back and find out that there were four poultry processes, one pork processor and beef was ground in the store.

Again, we found Salmonella. In this case from 20 percent of the ground samples. We found a variety of serotypes, most frequently isolated from poultry meats, and we found eight isolates of Typhimurium. The majority of those were DT104.

(Slide)

And just quickly. I don’t know if we can see this. This is what caught our interest. When we looked at those Salmonella isolates we did recover from the ground meats, we found the multi drug resistant to several antimicrobials. In particular, S. agona. We saw a lot of beta lactam resistance, strep-sul-tet and trimethoprim. We did find some DT208s which were resistant to about nine to 12 drugs. Very interesting.
These appear to be all plasmid mediated as well.

(Slide)

So, these two studies led us to think that we needed to investigate further the role of retail meats as reservoirs of resistance, and we initiated a study called the Iowa Retail Meat Pilot Study in 2001 and 2002, and then we segued it into NARMS, the FoodNet -- I will go over the FoodNet Retail Meat Surveillance. The first year was 2002.

(Slide)

In terms of the Iowa Study, this was started in 2001 and goes into 2002. We went to 24 samples a week, and Dr. Terry Proescholdt will talk about sampling later on and how we developed the study.

But we picked four meats, much like we had done in the other studies in D.C. Ground turkey, pork chops, ground beef and chicken breasts; 24 samples a week. They went to six grocery stores and they picked four meats of each for 24 samples per week.

And ideally, what we were trying to determine was the prevalence and susceptibility patterns of Enterococcus, Salmonella, Campylobacter and suspect extended spectrum beta lactamase producing gram negative bacteria. This goes back to Dan’s questions.

As part of our study we are trying to determine if we could find ESBLs on foods, and we did a whole screen with ceftazidime and cefotaxime with one microgram per ml. The Salmonella and enterococcal plates were -- we used the NARMS plates to determine susceptibility, and for Campylobacter we used agar dilution and E-test methods.

(Slide)

So, I will just give you some brief data. The studies I mentioned went on for 15 months. We had 24 samples per week. We ended up sampling about 263 grocery stores; 209 for Campylobacter. It took us several months to optimize the Campylobacter methods. Overall though we had almost 1,000 meats analyzed, and as I mentioned, we looked for Salmonella, Campy, Enterococcus and ESBL-producing gram negative bacteria.

Susceptibility testing was done using the NARMS panel or agar dilution. All Campylobacter and Salmonella were subjected to Pulsed Field Gel Electrophoresis as well using two enzymes, and we just finished and it has taken us two years to do that. For those of you that do pulsed field, doing one enzyme is hard enough. When you have to do two enzymes, you double your amount of work.

We were also able to get the human isolates from that same time from the State of Iowa. So we were able to have comparisons from what we recovered on food to what was submitted to the clinical diagnostic lab at the University of Iowa as well.

(Slide)

So, some quick data on that. Of the 981 samples, 126 were positive for Salmonella. So, about 13 percent overall. Ground turkey and chicken breast accounted for the majority of those. Forty-one percent of ground turkey samples were positive for Salmonella and 12 percent on chicken breasts.

Overall, 21 percent of samples were contaminated with Campylobacter, the majority being chicken with 75 percent. So, if you remember back to our original study in Washington, D.C., it was about 71.7 or pretty similar. We are talking almost three quarters.

(Slide)

And in terms of susceptibility, one thing I didn’t point out is that we did not recover Salmonella from any pork chop sampled. They were completely free of Salmonella. And all of our Salmonella from beef, there were five of them, were completely susceptible to all antimicrobials tested.

We saw the majority of resistance to isolates from either poultry, ground turkey or chicken breasts, and all isolates were susceptible to ciprofloxacin, amikacin and imipenem.

(Slide)

In terms of pulsed fields, we just want to show you a couple of things where we did do Pulsed Field Gel Electrophoresis. This was on Salmonella Heidelbergs that we found. We did get several hits in terms of indistinguishable pulsed field patterns between those recovered from ground meats in Iowa and also those recovered from human illness.

(Slide)

In terms of Campylobacter, this is the breakdown of what we found. We found primarily the Campylobacter in the chicken breast, and we also had about 181 isolates that we had from humans that we were able to compare as well.

(Slide)

This is just a -- I will give you quickly some ciprofloxacin resistance. This is based on agar dilution in an interpretative resistance breakpoint of four µg/mL. As you can see, that is supposed to be a micro.

But just to show you overall, 19 percent of the C. jejuni from chicken breast were cipro resistant at four microgram per mil or greater, 22 percent of C. coli. None of the other C. jejuni were resistant to cipro from any other meats. And overall, Campylobacter recovered from humans were about 12 percent resistant. So, we are talking in the teens for ciprofloxacin resistance.

(Slide)

Again, we did pulsed field on all Campylobacter jejuni and on several occasions we got identical pattern matches between those Campylobacter jejuni from retail meats and from human illness in Iowa.
(Slide)

Now, during the time of this study the WHO recommended a tripartite approach to include isolates from human clinical cases, food animals and retail meats when conducting antimicrobial resistance surveillance and monitoring for foodborne pathogens. So they recommend this three-prong system of surveillance, if you are looking at foodborne pathogens of on-farm, slaughter, retail, slaughter/retail and human isolates.

(Slide)

So therefore, in 2002 NARMS teamed up with FoodNet out of the Centers for Disease Control and Prevention in developing a program aimed at determining susceptibility of bacteria isolated from retail foods of animal origins, and this has become the third part of NARMS and the most recent. It is located at the Office of Research in Laurel, Maryland.

(Slide)

So, the overview. We started in 2002 with six of the 10 FoodNet sites. We had initially Connecticut, Georgia, Maryland, Georgia, Maryland, Minnesota and Tennessee started in January 2002. Oregon joined us much later in the year, in September. In January of 2003 we added New York and California, and in January of 2004 we added Colorado and New Mexico.

And currently we have 10 of the 11 sites. I think the 11th site is Texas. Correct? And they have yet to join our surveillance system yet, but they are still in their first year of FoodNet as well. So, we have the 10 major players.

Each of the 10 sites has a similar sampling scheme where the sites visit at least one grocery store per month. In the past, between 2002 and 2004, they did this. In 2005 we introduced a random sampling scheme. Again, Terry Proescholdt will talk about that later on.

Each site purchases 40 meats per month. Ten packages each of chicken breast, pork chops, ground turkey and ground beef. All 10 sites culture the rinsates for Salmonella and Campylobacter, and due to the large numbers we recovered of Enterococcus and E. coli, we only have four sites culture the rinsates for E. coli and Enterococcus.

And I may have a slide to show you that. The reason why is our Enterococcus were close to 95% contamination. So, if we had all 10 sites doing that, we are talking 5,000 Enterococcus isolates per year. It would quickly overwhelm us.

(Slide)

This just gives you a picture of the laboratories; where they are disbursed throughout the United States. You can see we have Oregon and California on the West Coast, Colorado and New Mexico, Minnesota, Tennessee, Georgia, Maryland, Connecticut and New York. So, these are the sites that are participating with us currently in the sampling.

(Slide)

As I mentioned, we used similar standardized methods at each of the sites to isolate the respective bacteria. They are adapted from the FDA BAM Manual. All of the meats are initially rinsed in peptone broth and then the peptone broth is then put into the selective broth, depending on what isolate we are trying to recover.

And for Salmonella there are several broths. Lactose broth, RVR10 and the M Broth, and we use a Tekra EIA Kit. For Campylobacter we use Bolton broth. For E. coli, MacConkey broth. Enterococcus is enterococcosel broth. All of these broths have been incubated at appropriate incubation temperatures for an appropriate amount of time and then plated onto selective agars to isolate the respective bacteria. XLD for Salmonella, CCA Agar for Campy, EMB for E. coli and Enterococcosel agar for Enterococcus.

The first two years we provided training where, we had the sites come out to our facility, and we had a week-long training to show them how we were doing it, because we wanted to make sure that all of the sites followed the same methods. It is critical.

(Slide)

So, the sites are the ones that isolate the bacteria, and once they isolate it, they presumptively identify it. Then they send it to us at the Office of Research for confirmation of identification and antimicrobial susceptibility testing.

We confirm all bacterial species by biochemical identifications, primarily using the Vitek, and we also have API if we have some disparate results coming from the Vitek. We have PCR capabilities for Campylobacter speciation using several genes that are specific for lari*, jejuni and coli.

We have our own serotyping laboratory that serotypes all of the Salmonella isolates that come again. And again, we do susceptibility testing, which is our bread and butter. We use broth microdilution/agar dilution, when appropriate, and as I mentioned before, we follow CLSI/NCCLS standards where available.

(Slide)

And just to give you an idea what happens in this whole scheme of collection and reporting, the sites go out and buy the meat. They isolate the bacterium from the different meats. Again, on the selected medias. There is a log sheet. Dr. Elvira Hall-Robinson will talk about this later on. It goes into our database. It is an access database. Again, that will be talked about later on by Dr. Hall-Robinson.

(Slide)

As part of the retail arm, all Salmonella and Campylobacter isolates are subjected to PFGE, which is a big burden on us, but we do them all. I think it is very important information.

We follow CDC guidelines for PFGE analysis and CVM-OR is a PulseNet certified laboratory. Our PulseNet unit is headed up by Dr. Shaohua Zhao, who will talk about that later on.

All of our data goes into the PulseNet program, and our isolates can be used for future research projects for anyone who is interested in these isolates as well. We have been involved in biosource tracking experiments, virulence studies and, of course, resistance studies when we try to find particular genotypes associated with particular phenotypes.

Enterococcus and E. coli. Right now a lot of these should have question marks next to it because we have these isolates. We are trying to determine what to do with them in terms of follow up studies.

Some questions that have come up are what virulence factors are present in Enterococcus isolates and are they present on the ones we see in foods? Streptogramin resistance, of course, is a big interest to CVM, and we have initiated some studies to determine what streptogramin resistance genes are there.

We find that in a lot of our Enterococcus isolates that show high levels of streptogramin resistance do not have the typical genes that people have found in the literature. So, there are genes yet to be identified, I believe, in these isolates.

E. coli. We are going after generic E. coli. We are not going after 0157 or shiga-toxin producing E. coli. So, it is important to serotype these isolates? Jim Johnson’s work out of Minnesota is showing that some of these E. coli from foods are potentially urinary tract pathogens. We don’t know that. And we do see fluoroquinolone resistance and ESBL phenotypes in our E. coli, and we are working those up via PCRs for mutation analysis and gyrase* and so forth and what type of beta lactamases are present.

(Slide)

Just to give you some data on the three years, what I wanted to point out here is the number of meats that have increased from 2002 to 2004. Our first year we had approximately 2,500 meats under analysis. As of 2004 we were almost up to 5,000 meats. So we are talking 4,600 and 4,700 meats in terms of our preliminary data, and the numbers are pretty stable across the board for Salmonella.

So, chicken breast was about 10 percent to 13 percent over the three years. For ground turkey it is about 11 percent to 13 percent again down to 12 percent. Our ground beef is almost identical over the years in terms of percent positive, and our pork chop is pretty similar as well.

But it is interesting. You can see it is mostly coming from poultry products.

(Slide)

This is to compare two years of data for Salmonella, 2002 versus 2003, and I tried to group the beta lactams together on the left, then your aminoglycosides and then our other drugs. You can see where we see a little more resistance in 2003 to several drugs, with the exception of TET where we did see a decrease. So, we don’t have 2004 data yet, but I would be curious to see what we get.

(Slide)

Again, I just wanted to show you that we are doing pulsed field, and Dr. Zhao will talk a little bit about this. But we do see some hits between clinical isolates and what is getting through into the ground products. Where it says turkey, they are actually clinical isolates from the diagnostic lab.

Again, we see another hit here. We see three indistinguishable clusters here, and this big cluster is mostly ground turkey. You can see from multiple states. Tennessee, Connecticut, Minnesota and different sampling times as well. So, we are seeing the same clonal isolates out there in the food supply.

(Slide)

In terms of Campy for the three years, it does appear like we have an increased rate of isolation of Campylobacter. I think a lot of this is due to the laboratories improving their methods of isolation. Our first year we had a lot of isolates sent to us from the laboratories that were indeed not Campylobacter.

Remember, we confirm everything at OR. So we had to throw a lot out, because our initial rates were pretty similar to what we see here in 2003 and 2004. But again, you can see we are not seeing much Campy coming from any other of the three meats sampled. Primarily, when we recover it, it is coming from chicken breasts.

(Slide)

And here is two years of data on resistance. This is agar dilution data. In 2002 we had approximately 300 Campy isolates. In 2003, 464. Our ciprofloxacin is pretty stable. About 14 percent to 15 percent. Erythromycin went down from six percent to about three percent. Doxycycline is fairly high, about 27 percent to 28 percent. Gentamicin we -- and remember, these are interpretive criteria or breakpoints that we have determined. They are not CLSI/NCCLS interpretative criteria.

We did see our first Gentamicin resistant Campylobacter this year, and it is at an MIC of 32 microgram per ml. And interestingly, in erythromycin here these are all C. coli’s. We don’t have a single C. jejuni that is erythromycin resistant.

(Slide)

On E. coli, I will quickly go over this. This is, again, three years worth of data. The numbers, remember, are lower because we only have four sites doing E. coli, and enterococci versus the 10 sites for Salmonella and Campylobacter. But your numbers are pretty steady for the three years.

We do seem to see this reproducible phenomenon with pork chops. They seem to be less contaminated with E. coli than the other three meat types.

(Slide)

And again, this is comparing two years of data. You can see almost the E. coli from 2002 and 2003; they are almost like mirror images across the board in terms of percent resistance to the drugs under surveillance of NARMS.

(Slide)

One thing we are trying to do now is to pull out to see if we see certain phenotypes associated with certain meat products, and this is an example of what we can do. If you look at gentamicin and nalidixic Acid, among those isolates that were GEN resistant and NAL resistant, the majority of them are coming from poultry products.

We see very little GEN or NAL resistance in other ground beef or pork chops, and they are primarily coming from chicken breast and ground turkey. So, we can do this for all of the drugs under surveillance at NARMS. At least for the retail arm.

(Slide)

In summary, since its inception in 2002 NARMS retail has increased the number of sampling sites and retail samples analyzed and improved its sampling scheme. For example, six sites in 2002 to 10 sites now in 2005. We have almost doubled the number of meat and poultry samples under surveillance, about 2,500 to almost 5,000, and we introduced random sampling strategy in January 2005 that will be discussed later on.

Its primary purpose is to determine the prevalence of antimicrobial resistance among foodborne pathogens and commensal organisms, and the commensals that we are looking at, of course, are Enterococcus and E. coli as potential reservoirs of resistance from retail foods.

We hope the results will generate baseline data and establish a reference point for analyzing trends of resistance among these foodborne bacteria at a point closets to the consumer. As I mentioned earlier, when combined with data from the rest of the NARMS program, it can give us an idea of what is happening and where we can target intervention and mitigation strategies.

(Slide)

With that, I would like to thank -- there are a lot of people involved in this project and there are certainly many more players, but this is a good collaboration between CVM and primarily CDC, because without their help we can’t work with the FoodNet sites as well. So, thanks. I will take any questions.

DR. VOGEL: I have got several questions. I have always been curious why you chose ground turkey as one of your products to sample. It doesn’t seem to me that per capita consumption is very high of ground turkey.

DR. WHITE: That is a great question, Lyle. During the Iowa Pilot Project we tried to determine what would be the most appropriate and the best handled in the laboratories, and we found out that buying the big carcasses was not the best way to go, because they were Fed Ex’ing them to us. So we tried to go with a product that was easily obtainable.

I think Terry will talk about it later. Ground turkey is actually the fastest growing commodity in the turkey industry being sold in grocery stores. So, that is why we went with ground turkey. But we could have easily gone with turkey legs and turkey thighs. Absolutely.

The thing to remember too is that USDA has performance standards for ground turkey. So that is one other thing we were looking at as well; to try to use that data to maybe help some other agencies when we get some data.

DR. SAHM: Thanks, Dan Sahm. Just in terms of trying to keep questions on target for the discussion points, with regard to sampling, now that what you have learned this data is sort of what we talked about at break. In looking at this data, if I only had a dime to spend I would put it on poultry.

How much information are you getting out of the other meat sections and would you not concentrate on poultry as being the highest risk of exposure to antibiotic resistant organisms for this program?

DR. WHITE: I think for the overt pathogens. Yes. They are primarily coming from poultry products. We are not getting that many isolates from either pork chops or ground beef. Absolutely. So in terms of bang for the buck, yes. We are not getting that much for those meats.

DR. SAHM: What would the risk be of going forward and not pursuing those other meats at this point if you wanted to concentrate more thoroughly on -- if people would agree that is where the risk is, to put your efforts in that area.

DR. WHITE: That is a great question. I think the major risk would be we would want to have information from those particular food commodities, and as we approve drugs in numerous animal classes we would only have information then from poultry. We would not have that information from say swine and from the resulting meats that are obtained from the swine, as well as cattle as well. So, we wouldn’t have that information, and we are trying to look at drug use in certain animals with what might be happening.

We wouldn’t be able to do that connect between say approval -- not to point out Sue’s drug, but ceftiofur is approved in numerous animals species. Without looking at the different foods, we would not be able to see if there is anything going on in terms of decreased susceptibility or what have you.

DR. SAHM: But in our surveillance efforts -- I just want to pursue this a little more to think through it. You learn from your baseline data and you redesign. It is analogous to your decision not to include imipenem anymore.

I just put that out as a point to consider based on the progress of surveillance of learning what you know and where you put your efforts going forward.

DR. WHITE: I think it is a good point. Thank you. I agree.

DR. ALTEKRUSE: I would just like to say from the FSIS standpoint we believe that also the greatest risk is for poultry and Salmonella. Our programs are being sort of reconfigured to address that. So we are going to put greater emphasis on the testing of poultry products. That will have some implications for sampling issues later related to the FSIS isolates.

But I am also very intrigued in the finding of matches with the PFG patterns between human and food isolates, and I would like to hear your opinion on the potential of that to actually detect outbreaks or identify sources of infection.

DR. WHITE: I think that is probably why we do it. In terms of outbreak assistance I doubt that will be of much help at this point because we are not real time. Our isolates are going in probably two years from when they were collected. So really, what we are trying to do is expand the data at PulseNet so that if there is an outbreak and they submit a pattern and it does have a hit with one of ours, maybe they can start focusing their efforts on a particular food.

So, if they say I have a Salmonella Heidelberg and it is a match with a Heidelberg from our retail foods from ground beef, maybe that will help the health personnel to start focusing on where it is going to go.

However, we do have numerous pulse field patterns that are identical among the different meats, like Newport or Typhimurium. So, it depends on the particular organism as well.

DR. ALTEKRUSE: Just to take it one step further, the reason I am interested in that is because one of the triggers that FSIS that FSIS is likely to use for going into an establish and doing a really intensive assessment of their food safety system is the finding of some epidemiologic evidence, like a pulsed field match between human isolates and isolates coming from that plant.

DR. WHITE: A lot of it too -- when people do those comparisons without, I think, a good epidemiological study to back that up, it is -- it just tells you there is a hit, but we need to do a lot more in between to see what is going on. So that is why we do the pulsed field.

I think it is very important to expand that database. They don’t have much in their database from actual foods. Now we are expanding with Paula’s isolates and CDC’s isolates as well. Hopefully, we will get all of those in the system, and when we do have outbreaks they may be able to better target their intervention strategies.

DR. ALTEKRUSE: I would like to ask you how -- what were the criteria to select the 10 sites that you include?

DR. WHITE: Okay. That is a great question. The 10 sites were already existing as part of the FoodNet system. So, they were already in existence.

DR. VOGEL: In one of your slides you mentioned that it was a samples from a diagnostic laboratory. Are you obtaining samples from diagnostic laboratories yourself? Or is that a sample that was obtained from USDA and then shared? If you are doing it yourself, it seems like it is duplicative use of sources of resources here.

DR. WHITE: Yes, I agree. That was a diagnostic isolate from another study. It is not from a NARMS study. Remember, a lot of us are also Office of Research principle investigators. We have our own studies underway as well besides NARMS. That was a study that Dr. Zhao and Dr. Walker are PIs and were working with diagnostic labs. But we do the same thing. We do susceptibility testing and pulsed field.

So then, when we run it through the system and we get hits, it would just happen to come out as one of those.

DR. VOGEL: Can I ask another question?

DR. WHITE: Sure.

DR. VOGEL: On Campylobacter you were testing for Doxycycline why did you choose that instead of Tetracycline like the other arms?

DR. WHITE: Yes. Doxycycline I believe would be the actual drug that would be used to treat human illness. I think more so than tetracycline. But we have gone back to tetracycline for the microdilution.

DR. KOTARSKI: Yes. I come back to a comment that I made earlier in understanding the purposes of the sampling strategies. Currently it looks like you are consisting sampling the same number from different -- getting the same number of meats from each store.

So, on the surface it looks at your sampling strategies are designed to look at prevalence of Salmonella or Campylobacter of meats in stores. It is not set up exactly to examine exposure to humans.

DR. WHITE: I agree.

DR. KOTARSKI: Okay. So that is understood. Now you have generated data to show that the prevalence of Salmonella actually is very low in certain meats and that point has been made earlier. I haven’t looked in detail at years 2003 and 2004, but I do recall, for example, your ground beef data for 2002.

You had very few isolates. On the order of nine Salmonella isolates, as compared to hundreds maybe of isolates in poultry. That being the case, to look at resistance trends of nine isolates in ground beef from one year to the next, probably your sampling error associated with only obtaining nine isolates of Salmonella, for example, in ground beef does not permit you to look at temporal trends that you would like to do without a lot of variability, just because of the number of isolates that you get.

So, I come back to the original point. What is your objective in your sampling strategy for the retail meats? Is it to understand the prevalence in meats, per se? Understand resistance prevalence in retail stores? Or is it to understand the prevalence in food consumption?

I think that those objectives should be clearly stated and then select your sampling strategy accordingly.

DR. WHITE: Okay.

DR. ALTEKRUSE: Many of the serotypes that you are isolating, for example, from poultry aren’t serotypes that are associated with human illness, and I wonder if those get the same priority or perhaps they are tested later. Have you thought about it?

DR. WHITE: In our system?

DR. ALTEKRUSE: Right.

DR. WHITE: Well, the first thing is serotyping takes the longest amount of time, besides pulsed field. So actually, as soon as we get something in it is susceptibility tested. So, they are done right away before we do serotyping. You know what I mean?

We have different groups working on the bugs at the same time. That is the best way to get things done the fastest, not wait until they are done. Otherwise it would take years for us to get each year done. But the majority of our Salmonella, as we see it, are the top five that CDC sees in human illness.

We do see some other ones that we don’t traditionally see in human illness, but across the board they are typical household names that most people know. Heidelberg, Newport, Typhimurium and so forth.

DR. VOGEL: You mentioned that you are sampling E. coli and Enterococci from only four of the FoodNet sites, and I understand why, because of the number of isolates. Have you looked to see if there is any geographic variation in E. coli or Enterococci? If so, might you be better doing a systematic sampling from all 10 sites or all 11 sites instead of restricting yourself to four sites?

DR. WHITE: Yes. In our first annual report we tried to break it down by state, and I think at least with Enterococcus we do see a species distribution with -- Maryland sees more faecalis than faecium compared to the other three sites.

We haven’t look at the 2003 data yet to see if that holds up in terms of a trend. So, it is definitely something to throw out and look at the data.

E. coli seems to be different, we don’t see resistance associated with one state more so than another. They are almost mirror images for the two years. We try to pull out those things to see if there are differences between the states as well.

We did that with our Campy. If you look at Campylobacter isolation, you see large variations between the states in terms of what they isolate over the year. Okay. Thank you.

DR. YOUNGMAN: Thanks, Dave.

In the interest of time, if we can press ahead, our next presenter is Dr. Paula Cray, who is going to be talking about the animal arm of NARMS.