1 1 2 QUANTITATIVE RISK ASSESSMENT ON THE PUBLIC HEALTH 3 IMPACT OF VIBRIO PARAHAEMOLYTICUS IN RAW OYSTERS 4 5 * * * * * * * * * * * * 6 IN RE: PUBLIC MEETING 7 * * * * * * * * * * * * 8 9 10 11 The following proceedings were held at the 12 Grand Hotel Marriott Resort, One Grand 13 Boulevard, Point Clear, Alabama, 36564, 14 August 13, 2005, commencing at approximately 15 12:00 p.m. 16 17 18 19 20 21 22 23 MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 2 1 A P P E A R A N C E S 2 PANEL: 3 MR. DONALD KRAEMER (FDA) MR. JOHN BOWERS (FDA) 4 DR. ANDY DePAOLA (FDA) DR. MARIANNE MILIOTIS (FDA) 5 DR. JOHN PAINTER (CDC) DR. MARK WALDERHAUG (FDA) 6 7 DIRECTOR: 8 DR. ROBERT BRACKETT, DIRECTOR, CENTER FOR FOOD SAFETY 9 AND APPLIED NUTRITION, FOOD AND DRUG ADMINISTRATION 10 11 PUBLIC MEETING ATTENDEES 12 13 COURT REPORTER: 14 KAREN T. McDONALD, CSR 15 16 17 18 19 20 21 22 23 MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 3 1 I N D E X 2 3 OPENING REMARKS: 4 DR. DONALD KRAEMER - PAGE 5 DR. ROBERT BRACKETT - PAGE 6 6 OVERVIEW OF RISK ASSESSMENT: 7 DR. MARIANNE MILIOTIS (FDA) - PAGE 11 8 9 QUESTIONS TO PANEL: 10 PANEL INTRODUCTION BY DR. KRAEMER - PAGE 35 11 QUESTIONS TO PANEL - PAGE 36 12 PUBLIC COMMENTS - PAGE 67 13 14 15 16 17 18 19 20 21 22 23 MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 4 1 P R O C E E D I N G S 2 DR. KRAEMER: Thank you for coming to 3 the public risk assessment. This may not be where 4 you wanted to be, you probably wanted to be 5 somewhere else. I've got a couple of ground rules 6 to give you an idea of how the public meeting will 7 be structured. 8 We have done registration and we 9 encourage you to register if you have not already. 10 Some folks did pre-register. That wasn't 11 necessary, but helpful. 12 We're going to start with some opening 13 remarks from Dr. Brackett. We'll go then through 14 an overview of the risk assessment to give you an 15 idea of what we think the most relevant points 16 are. We'll then open the floor up for discussion, 17 and we strongly encourage you to ask questions. 18 We have a panel of experts here that's 19 about as good as it gets with respect to risk 20 assessment. So I think they should be able to 21 answer your questions. 22 And then at the end we will have some 23 time for public comments. We've had three MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 5 1 individuals who indicated in advance that they 2 would like to make a public comment, and we'll 3 start with those. Two of those will be joining us 4 by phone. So just in case you start hearing some 5 background noise, it's because they will be 6 joining us on the phone. Right now we have an 7 empty phone line but we expect them to be joining 8 us. So if that technology works, we'll start with 9 those and then move on to the rest of the 10 comments. 11 I should also mention that the 12 discussions here are being transcribed, just so 13 you know that there will be a record of this 14 meeting. It will reside in FDA's dockets, so 15 anybody can take a look at those comments at 16 anytime in the future. And because of that I will 17 ask that if you're going to speak, please use one 18 of the microphones and please identify your name 19 and your affiliation so we can have that all 20 recorded. 21 And with that, I'd like to introduce Dr. 22 Robert Brackett. He is the director of the Center 23 for Food Safety and Applied Nutrition of the FDA. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 6 1 DR. BRACKETT: Thank you, Don. And I 2 would like to welcome all of you to this public 3 meeting. For those of you who have been involved 4 in public meetings at FDA before, you may 5 recognize that this is something that we very much 6 value. This is one way that we can, quote, get 7 the information to you directly, but more 8 importantly hear from you. 9 And that's why I would like to reiterate 10 what Don just said, when the discussion portion 11 comes up, please do give us your opinions. If you 12 also have opinions that you could put in writing, 13 I think that's even more valuable to us as we 14 emphasize the docket. 15 FDA and ISSC have been working on the 16 Vibrio parahaemolyticus outbreak situation since 17 about the late 1990's -- '97, 1998 -- and 18 developed at that time an interim control plan in 19 1998 that was then later modified in 2001. And 20 this plan calls for states in which oyster-related 21 Vibrio parahaemolyticus illnesses have been traced 22 to monitor the pathogenic Vibrio parahaemolyticus 23 species and to close the affected waters where MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 7 1 certain levels have been exceeded. 2 The plan was specifically designed to 3 prevent very large outbreaks, especially those 4 related to the strain 03:K6. But since the late 5 1990's, we haven't seen that strain again. So 6 many of the control measures really don't apply to 7 that quite as much anymore. 8 In 2003, the FDA asked the conference to 9 begin to consider some preventative controls, and 10 that's something that we are -- the FDA -- is 11 interested in preventing, not necessarily dealing 12 with the illnesses. Preventative controls for the 13 so-called sporadic basis of Vibrio 14 parahaemolyticus recognizing that the original 15 plan, the interim control plan, was not really 16 designed to address the sporadic case. 17 And since that time, CDC has since 18 estimated that there are about 2,800 such 19 illnesses that occur each year in the United 20 States, and that they are associated with the 21 consumption of raw oysters. The FDA initiated a 22 number of years ago -- as many of you know and the 23 reason why we're here -- a risk assessment for MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 8 1 Vibrio parahaemolyticus in oysters. That was 2 initiated in 1999 and took a number of years to 3 complete. This public meeting is one of the first 4 reactions to the publication of the risk 5 assessment. 6 Risk assessments are very important to 7 the FDA, and they will continue to be even more so 8 in the future as we take a risk-based approach 9 toward preventing food-borne illness. At that 10 time when the risk assessment was being developed, 11 we did involve the general public in the 12 development of the assumptions involved in the 13 risk assessment as well as some of the directions 14 it should go. And many of the ISSC members were 15 also instrumental in that process in developing 16 and refining the risk assessment at that time. 17 We held other public meetings similar to 18 this and issued formal requests for your input via 19 the Federal Register, which is the official way to 20 do such things. We also issued the risk 21 assessment as a draft in 2001 to see if we had hit 22 the market right from that. And the risk 23 assessment that you're here about today really MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 9 1 takes into account all of the information we got 2 from the draft risk assessment and the submitted 3 data that we have received. 4 There are really two goals for this 5 assessment and for this public meeting. In terms 6 of the risk assessment, the first goal was to 7 determine factors that contributed to the risk -- 8 or the increased risk -- of becoming ill from 9 consumption of pathogenic Vibrio parahaemolyticus 10 in raw oysters. The second half is to evaluate 11 the public health impact of different control 12 measures that could be used. 13 So, again, what's causing the increased 14 risk and how do we minimize that risk. And we do 15 think that we've accomplished these goals and that 16 the risk assessment can be a very useful tool for 17 both the government as well as the private sector 18 and the ISSC for the appropriate risk mitigation 19 steps that could be used in the future. 20 In the ISSC meeting that will follow 21 this meeting today, both federal and state 22 regulators, the shellfish industry, and other 23 state boards will be deliberating that very topic MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 10 1 among many other topics that are important to the 2 shellfish industry. 3 The purpose of this particular meeting 4 here is, first of all, to provide you with a 5 summary including the methods and the results of 6 the risk assessment, to provide you the 7 opportunity as key participants in the risk 8 assessment to get your questions answered about 9 the details of the assessment and especially to 10 those that this is of great interest to you 11 particularly. And thirdly, to provide you with an 12 opportunity to make a statement for the record 13 that is part of the public record of your views on 14 the risk assessment or how it should be used. 15 As Don mentioned, the meeting is being 16 transcribed word for word, and the contents 17 including your comments, the public comments, will 18 become part of FDA's docket on the subject. So I 19 do encourage your participation in the question 20 and answer session. As Don mentioned, you have 21 some of the world's experts on Vibrio 22 parahaemolyticus risk assessments at the table 23 here. So I would like to offer you the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 11 1 opportunity to really get your questions answered 2 by those individuals. 3 So with that, I will close. And, again, 4 thank you for coming and sharing your interest and 5 we look forward to your comments. 6 DR. KRAEMER: Thanks, Bob. The next 7 portion of the program is a formal presentation. 8 We have some slides here that we'll go through 9 that will take you through the points that the FDA 10 thinks are the most important for your 11 understanding of the risk assessment. And for 12 that presentation we have Dr. Marianne Miliotis. 13 Dr. Miliotis is from FDA's Office of Science, and 14 is the project leader for this risk assessment. 15 DR. MILIOTIS: Thank you, Don. Good 16 afternoon, everybody. Basically what I'm going to 17 do for the next 45 minutes or so is I will go 18 through an overview of the risk assessment. We 19 will start with why we have a risk assessment of 20 the Vibrio parahaemolyticus of raw oysters. We'll 21 go through the approach which includes the 22 objectives of the time line of the risk assessment 23 process and the components of the risk assessment. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 12 1 I will also go over some of the relative results. 2 And in the interest of time, I would 3 like to go through all of the results and show you 4 what we have in your copy of our technical 5 document. I will also discuss the what-if 6 scenarios for the prevention strategies which will 7 be evaluated and the impact of the Vp risk 8 analysis. And I will end my presentation with 9 some of the major conclusions. 10 Why do we conduct risk assessment in Vp 11 oysters. It's the leading cause of seafood- 12 associated illnesses, bacterial gastroenteritis in 13 the U.S. The CDC estimates that there are 14 approximately 2,800 cases of Vp food-borne 15 illnesses in the U.S., 62 percent from the 16 oysters. 62 percent of all VP cases are 17 oyster-related. Outbreaks in the U.S., in the 18 Pacific Northwest, Atlantic, and the Gulf Coast in 19 1997 and 1998, which involved over 700 cases of 20 illness brought many concerns of patients to the 21 forefront. The majority of the patients had a 22 consumption of raw oysters. Implicated oysters 23 came from specific growing areas. There was a MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 13 1 direct relationship between the consumption of raw 2 oysters and illness. That was the first 3 introduction of the pandemic strain of VP 03:K6. 4 Okay. How did we approach this. 5 Basically as Dr. Brackett mentioned the objectives 6 of the risk assessment is twofold. First we 7 determined the factors of what contributed to the 8 risk, and second, we evaluated the different 9 control measures and came up with the strategies. 10 This is just a time line of the risk assessment 11 process that operated in 1997 and 1998. The risk 12 assessment was initiated in 1999. 13 We held two public meetings within 1999. 14 The national association conducted the criteria 15 for this. This was basically to involve the 16 general public and our stakeholders, and just like 17 Dr. Brackett said, to gain some input information 18 from you. We had a public meeting for your 19 comments. Between 2001 and this year we took your 20 comments into consideration. Any new data that is 21 published, we use new model techniques. We 22 revised the risk assessment. It was reviewed and 23 the report was published the 30th of July 2001. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 14 1 Okay. What did this model comprise of. 2 In my further risk assessment, both the 3 variability and uncertainty existing for 4 parameters used for the model. The variability is 5 an apparent property such as water and air 6 temperatures, whereas uncertainty may be decreased 7 with the acquisition of research and more 8 information such as the ratio of pathogenic Vp to 9 total Vp, the number of oysters consumed and the 10 frequency of consumption. To show the variability 11 and the uncertainty that exists for each parameter 12 our model input expresses distributions. This 13 allowed us to show a range of values instead of a 14 single-point estimate. 15 Data sources included published and 16 unpublished literature and the reports produced by 17 various organizations such as the state shellfish 18 control authorities, the CDC, the shellfish 19 industry and the Interstate Shellfish Sanitation 20 Conference and state health departments. 21 Assumptions were used when data were incomplete. 22 We used new data and information received during 23 the comment period, and those were incorporated MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 15 1 into the model. And we validated our risk 2 assessment using data computed in the risk 3 assessment and so on. 4 Okay. I'm just going to list some of 5 the key assumptions. We considered the pathogenic 6 Vp as well as the tdh positive. We've assumed 7 that the growth and survival of pathogenic Vp in 8 harvested oysters is the same as in total Vp. 9 Based on studies -- the growth studies -- in 10 oysters at 26 degrees and lab studies at 26 11 degrees and other temperatures, they found that 12 the growth of Vp oysters was a quarter of that of 13 the growth rate in which are the lab conditions. 14 So we assumed that the growth rate is a quarter 15 growth at all temperatures. We assumed the lag 16 time to grow the Vp in oysters after harvest is 17 negligible. And consumption patterns by immune 18 compromised and healthy populations are the same. 19 This risk assessment was conducted in 20 accordance with the FAO and WHO framework for 21 conducting risk assessments comprised of four 22 components: hazard identification, hazard 23 characterization, exposure assessment, and risk MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 16 1 characterization. For this risk assessment, 2 hazard identification describes Vp in raw oysters. 3 Hazard characterization/dose-response 4 characterizes the relationship between the levels 5 of Vp ingested and the frequency and severity of 6 illness. Exposure assessment describes the 7 likelihood of ingesting Vp at the levels of Vp 8 ingested by eating raw oysters containing 9 pathogenic Vp. And the risk characterization is 10 an integration of hazard characterization and 11 exposure assessment to determine the risk of 12 illness. The important part of this step is 13 determining the uncertainties associated with 14 these predicted illnesses. 15 We also conducted a sensitivity analysis 16 and a validation of the model. And as I said 17 earlier, we used the base-line model to develop 18 what-if scenarios and to look at the impact of 19 different intervention control measures. 20 The hazard identification, as you'll note, Vp was 21 discovered in Japan in the 1950's. It's a natural 22 inhabitant of the temperate of tropical coastal 23 waters. It's predominant factors are tdh and trh. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 17 1 It causes gastroenteritis and on rare occasions it 2 can cause septicemia. And it's associated mainly 3 with oyster consumption in the United States. 4 This is a schematic representation of 5 the hazard characterization of the dose-response 6 model. It basically fit a curve to the human 7 clear data. This gave us our dose-response 8 relationship. We then adjusted this curve for the 9 uncertainty in the dose-response because of the 10 limited data from the clinical studies. This 11 shift is commonly referred as anchoring the risk 12 assessment. So we anchored the dose-response 13 model to CDC's surveillance data of approximately 14 2,800 Vp illnesses per year. This adjustment 15 represents the effect of the apparent difference 16 between the dose-response in the human body and 17 human study under controlled conditions versus 18 back in the general population when the oyster is 19 associated with the food matrix to get our 20 dose-response model. 21 And this is a graph of our dose-response 22 model. The solid line is a curve that is fit to 23 the clinical data. The dotted line with the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 18 1 adjusted curve that was anchored to the risk 2 assessment of the CDC's surveillance data. As you 3 can see the dose is much higher. A much higher 4 dose is needed to cause illness. And once we 5 adjusted it in the original clinical trial, this 6 is the effect of the Vp. For example, to get 50 7 percent of people who are ill, you need 100,000,00 8 Vp per serving. Whereas if you went by the 9 clinical trial, you need about 3,000,000 or so. 10 So it's a big difference. 11 Okay. Exposure assessment, this risk 12 assessment is a product pathway analysis. We 13 modeled the steps sequentially from harvest 14 through post-harvest and to consumption and 15 illness. We therefore divided the exposure 16 assessment into three modules: the harvest 17 module, the post-harvest, and consumption. 18 The harvest module describes the 19 presence and levels of pathogenic Vp in the 20 oysters at harvest. The post-harvest module 21 relates to the post-harvest handling practices and 22 processing at these levels. The consumption 23 module estimates the dose of the level of Vp MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 19 1 consumed per serving. 2 Because of the differences existing in 3 the oyster harvesting practice and climates in the 4 different areas in the United States, each module 5 was modeled for six separate harvests into 6 geographic regions. You have the Pacific 7 Northwest non intertidal. The Pacific Northwest 8 intertidal and the Mid Atlantic region, the 9 Northeast Atlantic. Then we divided the Gulf 10 Coast into Louisiana and non-Louisiana because a 11 survey in 1997 showed the duration of harvest in 12 Louisiana and non-Louisiana was much longer than 13 the other Gulf Coast-based areas. This duration 14 has impacted post-harvest growth. 15 Okay. This is a schematic 16 representation of the exposure assessment showing 17 the harvest and post-harvest and the consumption 18 modules and how they fit together. The top 19 colored bubbles and squares are the harvest 20 modules. These all show the parameters included 21 in the different modules. The middle white one is 22 the post-harvest module and the darker is the 23 consumption module. This representation also MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 20 1 shows the results from the white module and the 2 harvest module can pathogenically be oysters at 3 harvest. It helps the parameter for the 4 post-harvest module and the ending result of the 5 post-harvest module helps the parameter for the 6 consumption module. 7 These are the results for the exposure 8 assessment of Vp per gram in oysters at harvest, 9 post-harvest and in levels of Vp per serving at 10 consumption. This table provides the predicted 11 mean levels of Vibrio parahaemolyticus both total 12 and pathogenic at harvest from each of the 24 13 regions and seasons of the population. Across all 14 regions and seasons as you can see the predicted 15 levels are much higher in the summer and spring 16 than in the cooler months. The predicted levels 17 in the Gulf Coast region are considerably higher 18 than the other regions due to the warmer water 19 temperatures. 20 The levels of Vp in the Mid Atlantic and 21 the Northeast Atlantic in the summer are higher 22 than the Pacific Northeast because of the cooler 23 temperatures in the Pacific Northwest. However, MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 21 1 due to intertidal harvest, the exposure of the 2 oysters to temperature allows additional growth. 3 And therefore it results in an increase in those 4 levels to levels higher than in the Northeast 5 Atlantic. 6 This table shows the predicted mean 7 levels for total and pathogenic Vibrio 8 parahaemolyticus per serving of oysters at 9 consumption. The consumption levels are derived 10 from the post-harvest levels and modified by the 11 serving size. We assume the serving size to be 12 the same for all of the regions and seasons, 200 13 grams. As to be expected the mean levels of 14 pathogenic Vp are higher in the Gulf Coast than in 15 other regions. 16 As I said, the risk characterization is 17 a combination of exposure assessment and hazard 18 characterization. You combine the final output of 19 the two to provide us with the risk of illness per 20 serving. We multiplied the risk of illness per 21 serving by the frequency of servings to give us a 22 risk of illness to you. And then we multiplied 23 that by the probability of gastroenteritis MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 22 1 progressive to septicemia based on some CDC data 2 we had on the population to give us the risk of 3 septicemia to you. Okay. So once again, our 4 results of the risk characterization, our risk per 5 serving, the risk per annual number of illnesses 6 and severity. 7 Before I continue, I would just like to 8 point out to you the risk assessment model 9 predicts risk of illnesses contributed to the 10 oyster source, to the regions where the oysters 11 were harvested. It could be different from where 12 the illnesses occurred or where it was reported, 13 so keep that in mind. 14 Okay. Here we see the predicted risk of 15 the serving associated with the consumption of Vp 16 in raw oysters. The risk of serving, again, the 17 predicted risk of individuals becoming ill either 18 by gastroenteritis followed by septicemia if he or 19 she consumes a single serving of raw oysters. 20 Because of the high levels we have in 21 the Gulf Coast, once again, there's a higher risk 22 in the Gulf Coast than the Pacific Northwest 23 intertidal and the Pacific Northwest stresses the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 23 1 lowest risk. In this table, it shows the 2 predicted breakdown for the particular number of 3 illnesses for every harvest regional and seasonal. 4 Basically we took the mean risk, as I said, and 5 multiplied it by the number of servings per year. 6 And this is different for every regions and 7 seasons. And once again, the Gulf Coast is 8 higher. And once again we see a difference 9 between the warmer regions and the colder regions. 10 Okay. Here we go, here's the number of 11 septicemia cases per year and the total number is 12 seven which is a fraction of the number of the 13 total of gastroenteritis total illnesses caused by 14 Vp. 15 I mentioned earlier that our model 16 allowed us to form a sensitivity analysis, which 17 is a systematic evaluation of model parameters, 18 model input and assumptions. In other words, our 19 sensitivity analysis was conducted to determine 20 which model input factors had the strongest 21 interest for a predicted level of illness. This 22 represents this type of evaluation showing this 23 figure. This graph is referred to as a tornado MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 24 1 plot, and it shows the rank and magnitude of the 2 factors from highest to lowest of the number of Vp 3 illnesses. Here we show the Gulf Coast as an 4 example. If you look at the technical document, 5 it has all the other regions and combinations. 6 As you can see from this graph, the 7 model prediction of risk interest is most at the 8 level of Vp in an environment and second is the 9 percent of pathogenic Vp at the time of harvest. 10 The length of time oysters are unrefrigerated 11 after harvest is also an important factor. This 12 ranking is similar to all the regions and seasons 13 except for the Pacific Northwest intertidal 14 harvest. Here for the Pacific Northwest, the 15 second and one of the most influential factors are 16 the air and moisture temperatures. 17 Okay. One of the most difficult 18 problems facing risk assessments is trying to 19 determine whether our model is an accurate 20 representation of what is actually going on in the 21 real world and to convey this to our public and 22 especially to our stakeholders. In other words, 23 we need to validate our model. We compared our MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 25 1 data to real world data not just to the risk 2 assessment. Exposure predictions compared to data 3 of the levels of total Vp in oysters at retail 4 based on the ISSC/FDA retail study performed in 5 1997 and 1998, and also on data collected. And 6 then some of our harvest data was compared to data 7 collected by the Washington State Department of 8 Health for harvest levels in the Pacific Northwest 9 intertidal. 10 For the risk characterization, we 11 attempted to validate the model in the same 12 manner. And for this we used the data reported to 13 the CDC on the number of Vp illnessess. 14 Okay. This slide shows the comparison 15 between predicted levels at retail in the open 16 blocks and then a survey result is enclosed in 17 both black circles. Here is the different regions 18 and seasons in the Gulf Coast. Again, this is 19 just for the Gulf Coast. The technical document 20 shows the results for all of the other regions. 21 On the X axis we have the different 22 seasons, and the Y axis is the density levels of 23 Vp. Here you see the model predictions of mean MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 26 1 levels at retail are very similar to those 2 obtained by the study with one minor exception 3 which was during the fall for the retail on the 4 Gulf Coast. When we went back, we found out that 5 from 1998 when the retail survey was being 6 conducted, the water temperature was higher than 7 the average annual water temperature. So we went 8 back and re-ran the model using the warmer water 9 temperatures. And in the red block is the result. 10 And you can see, it's much closer to the actual 11 data obtained by the FDA retail study. 12 As I said before, the surveillance data 13 report to the CDC are the only data available to 14 compare our model predictions of illness. We did 15 a seasonal comparison, and what we predicted was a 16 surveillance based on illnesses. And here we see 17 that although the others are exactly the same, 18 there is a temporary season that is a very similar 19 pattern between our predicted number of illnesses 20 and the CDC surveillance. However, when we 21 compared the illnessess on a regional basis, the 22 agreement between the surveillance data and the 23 regional predictions was less clear-cut. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 27 1 As I mentioned earlier, the risk 2 assessment predicts illnesses based on the source 3 of the oyster, the region where the oyster was 4 harvested. Whereas the surveillance data 5 basically is on illnesses where illnesses were 6 reported or occurred. Of 715 oyster-associated Vp 7 illnesses reported to CDC between 1998 and 2003, 8 only 18.4 percent could be traced to a specific 9 harvest site. Based on this data, CDC estimated 10 the percentage of illnesses attributable to each 11 region, each harvest region. So using these 12 numbers we were able to more accurately compare a 13 predicted illness with the surveillance data. 14 This is shown in the last two rows. The total 15 attribute of illnesses is where the CDC estimated 16 based on their studies and the model-predicted 17 illnesses are on the bottom row. As you can see, 18 the models may be different, but the patterns are 19 very similar. Most of the illnesses occurred in 20 the Gulf Coast, attributed to the Gulf Coast, and 21 the second highest in the Pacific Northwest. 22 Okay. As I said, for the risk 23 assessment model, again we used it to evaluate the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 28 1 likely impact of our prevention strategies and 2 control measures of the predicted number of 3 illnesses. 4 The post-harvest mitigation control 5 center include evaluation of the impact on 6 different reduction levels on the risk of illness. 7 These reduction levels represent a range of 8 potential mitigation controls. For example, 9 immediate refrigeration, which is cooling 10 immediately after harvest. There's approximately 11 a 1-log reduction of Vp. Again, it depends on the 12 season and the region. Sometimes it's more than a 13 1-log and sometimes it may be less. A 2-log 14 reduction, which could be freezing followed by 15 cold storage. And a 4.5-log reduction or greater, 16 examples of this could be mild treatment, ultra 17 high pressure or irradiation. 18 We also evaluated reducing the time to 19 refrigeration after harvest, overnight submersion 20 of intertidally harvested oysters, and sample- 21 based control plans. Okay. Here, I'm showing you 22 the slide which measures that control reduce the 23 levels of Vp in oysters, It also reduces the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 29 1 predicted risk of illness associated with the 2 pathogen. Treatment such is needed is 3 refrigeration to decrease the number of predicted 4 illnesses by approximately 10-fold. Treatment 5 causing a 2-log decrease in the levels of Vp 6 illnesses reduce the possibly of illness at 7 approximately 100-fold. And treatment causing at 8 least a 4.5 log decrease in the number of Vp 9 reduce the predicted illness and state it's 10 unlikely that illnesses will be observed, less 11 than one. 12 Here we see the predicted reduction of 13 illnesses associated with the reduction in time of 14 harvest to be initiated for Summer harvest of the 15 Gulf Coast oysters from one to four hours after 16 harvest. The top solid line represents cooling 17 with ice and the bottom dotted diagram represents 18 cooling with conventional refrigeration. So 19 depending upon the specifics of the scenario, the 20 predicted reduction of Vp illness are Summer 21 harvest of Gulf Coast oysters is a .6 percent 22 reduction if cooling in four hours after the 23 harvest -- conventional cooling -- to 96, 97 MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 30 1 percent if cooled rapidly within an hour on ice. 2 Again as you can see there's a much 3 greater reduction in illness with rapid cooling if 4 oysters are rapidly cooled using ice. Here, we 5 see this table shows the impact of different 6 control levels at harvest and at retail. Assuming 7 it was possible to identify and exclude oysters 8 from the raw market which contained very specified 9 levels of Vp either at harvest or at retail, the 10 risk assessment evaluated the impact of illness as 11 well as the diversion of post-harvest from the raw 12 oyster market. If we could control or if we could 13 exclude all of those oysters that had definite 14 different values, different levels of Vp, either 15 at harvest or at retail, we would reduce the 16 number of predicted illnesses. As you can see in 17 this table it would also increase the number of 18 oysters diverted from the raw market. Or it would 19 require modifications on handling practices after 20 harvest. 21 Here is the Gulf Coast of Louisiana. 22 This is an example excluding all oysters with at 23 least 10,000 Vp per gram harvested in the absence MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 31 1 of subsequent post-harvest treatment reduce 2 illness by 16 percent and three percent of the 3 oysters which have been diverted from the raw 4 market. Post-harvest of all oysters containing 5 10,000 Vp per gram were excluded from the raw 6 market with a 99 percent reduction in illness. 7 But 43 percent of the oysters with Vp were 8 diverted from the raw market. And as the control 9 level decreases, more illnesses are reduced. The 10 more illnesses that are diverted, then the harvest 11 diverted has increased. 12 Studies by Chaney, et al and Andy 13 DePaola here show that levels in the Pacific 14 Northwest intertidal harvesting, levels of Vp, 15 decreased after overnight submersion. So the risk 16 assessment aside from the evaluation look at the 17 impact on this overnight submersion on illness. 18 The results revealed that there was a 90 percent 19 of 10-fold reduction in risk of illness, if 20 intertidal harvested oysters were less submerged 21 in the water overnight. Again, further research 22 is needed to determine if this reduction could 23 actually be achieved when the oysters are MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 32 1 submerged. 2 So in conclusion, I'm almost there. 3 Anyone exposed can become infected and develop 4 gastroenteritis. There's a higher probability of 5 gastroenteritis developing into septicemia in 6 subpopulation with chronic medical conditions. 7 The probability of illness is more likely when Vp 8 levels in oysters are higher, as you saw in that 9 slide I showed you. For example, there's about .1 10 percent probability if there's 50 Vp per gram and 11 a 50 percent probability of illness if there are 12 500,000 Vp per gram. There are seasonal 13 differences in illnesses if more illnesses are 14 compared in the summer months than in the cooler 15 months. 16 And there are also regional differences 17 in illnesses. There are more illnesses occurring 18 in the warmer regions like the Gulf Coast than in 19 the cooler regions as you can see from here. As 20 for the CDC data, there are more illnesses in the 21 Gulf Coast than the Pacific Northwest, Northeast 22 Atlantic than the Mid Atlantic and then Pacific 23 Northwest dredged. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 33 1 The single most important factor related 2 to the risk of illness caused by Vp is the level 3 of oysters at the time of harvest. This was true 4 for all regions and seasons. Water and air 5 temperature at the time of harvest are the major 6 factors influencing the initial levels of this 7 pathogen in oysters. 8 Preventing growth of Vp after harvest 9 reduces levels and consequently illness. So 10 reducing time to refrigeration will reduce 11 illness. Post-harvest measures aimed at reducing 12 levels in oysters reduces illness too. A 4.5 log 13 reduction reduces illness to less than one annum. 14 A 2-log reduction reduces predicted illness by 15 100-fold. And immediate refrigeration reduces 16 illness by approximately 10-fold. Overnight 17 submersion of oysters harvested intertidally can 18 also reduce risk of illness by approximately 10- 19 fold. 20 As a result of this risk assessment, 21 assumptions are made and we anticipate other 22 updates as we obtain more information to reduce 23 the uncertainty. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 34 1 There is a risk assessment team of all 2 the FDA people involved in the risk assessment. 3 On behalf of the team I would also like to thank 4 everyone who provided comments to us in 2001. I 5 would also like to acknowledge all of those who 6 provided information and guidance throughout the 7 conduction of our risk assessment, and especially 8 thank John Painter and the CDC staff for their 9 assistance in providing the epidemiological data 10 for the dose-response model and the data analysis 11 used to compare the model for the risk assessment. 12 Thank you. 13 If you need further information, there's 14 a web site provided. You can download the 15 document or at the bottom of the agenda there's 16 also directions of how to contact us directly to 17 get a hard copy of the risk assessment. 18 DR. KRAEMER: Thank you, Marianne. 19 Okay. The next section of our session here is an 20 opportunity for you to ask questions. I would 21 like to say that we've intentionally structured 22 the afternoon so that we have an allocated period 23 of time for information sharing. We know that MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 35 1 there are people that would like to provide public 2 comments, and we provided a specific segment which 3 will be at the end of the session at which time 4 you can provide public comments. We are very much 5 interested in both, but we would like to keep the 6 two separated just to keep it clean. So at this 7 time what I'd like to do is introduce members of 8 the panel here who we hope will be able to answer 9 your questions, and then we'll go to the question 10 and answer session. 11 First we have Dr. Mark Walderhaug, who 12 is a microbiologist in FDA's office of plant and 13 dairy foods, what we call land foods. That's 14 another office within the center. And then we 15 have Mr. John Bowers, who is a mathematical 16 statistician in the division of mathematics. Then 17 we have Andy DePaola, who is a microbiologist and 18 Vibrio expert at Dauphin Island, Alabama and works 19 for the Office of seafood. And you've already met 20 Marianne. And Dr. John Painter, who is in the 21 food-borne and diarrheal disease branch at CDC. 22 And that's the panel. 23 You can ask your questions individually MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 36 1 or just openly. I'll open that up to the panel in 2 general and we'll try to give it to the right 3 person. And, again, if you have questions, please 4 step up to the mic and introduce yourself and your 5 affiliation. Does anyone want to take the first 6 step? 7 Al Sunseri: Al Sunseri, I'm with P&J's 8 Oyster Company in New Orleans. I wanted to ask a 9 question about the retail study used in this risk 10 assessment done in '97 and '98. 11 Have y'all considered doing a study 12 after that time since those were the years in 13 which there were spikes in Vp illnesses? 14 DR. DEPAOLA: Yes. Thank you for that. 15 Actually that study was begun in June of 1998, 16 and it continued through May of 1999. It was just 17 a season off. There was the 03:K6 outbreak in 18 Texas. In fact, they were originally thought that 19 they were going to be part of the study, but they 20 were closed down until November that year. So 21 none of those oysters actually made it into the 22 study. 23 There was a representive study on the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 37 1 coast. We had nine states throughout the country 2 where we collected the oysters twice a month, two 3 samples each time, mostly from restaurants. 4 We're planning to initiate another study 5 with a very similar geographical design to keep in 6 the same collection states. And we're working on 7 that as we go. We will use some of the same 8 methods that we used in 1998 and 1999, plus we 9 will implement some of the newer methods on 10 realtime PCR, based on NPM and PCR so that we'll 11 have a greater sensitivity in detecting the 12 pathogenic levels of Vibrio parahaemolyticus, 13 which at the time the methods that we had 14 available were not sensitive enough to detect 15 those levels. 16 (Brief phone interruption.) 17 WILLIAM ATHAWES: William Athawes, New 18 York State Department of Environmental 19 Conservation. 20 MR. KRAEMER: Let me ask you to hang on 21 just a second. I think we had a couple of people 22 join us on the phone, and you're going to be next. 23 But I just wanted to let them know that we're MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 38 1 right now in the question and answer session, and 2 we will be doing the public comments when we wrap 3 this up. 4 AMY MCDONALD: Okay, thank you. 5 WILLIAM ATHAWES: This is a real minor 6 point. It's got to do with where New York falls, 7 whether it's the northern-most Atlantic state or 8 the southern-most northeast state. It's 9 information I need and I can ask the question 10 later and I propose some comments. 11 DR. MILIOTIS: It's the northeast 12 Atlantic. 13 WILLIAM ATHAWES: So it's the southern- 14 most northeast state. 15 DR. MILIOTIS: Right. 16 ROBERT WITTMAN: Robert Wittman, State 17 of Virginia. I have a couple of questions just in 18 elaboration. When you talked about the scenarios 19 that are evaluated outside of the immediate 20 refrigeration, you described a reducing time of 21 the refrigeration and overnight submersion of 22 intertidally harvested oysters. 23 Can you describe what specifically the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 39 1 time and temperatures and the parameters were in 2 that scenario and specifically what the overnight 3 submersion entails? 4 MR. BOWERS: Let me try to repeat the 5 question, and you can correct me if I'm wrong. 6 You're asking about what the specific 7 model assumptions were for the intertidally 8 harvested oysters? 9 ROBERT WITTMAN: Yes. 10 MR. BOWERS: The assumptions were that 11 as the tide recedes you have a four-to-eight-hour 12 period of time where the oysters are exposed to 13 the air temperature. And it's not just the effect 14 of the air, but there is a radiative heating 15 effect that may occur depending on whether it's, 16 you know, a bright sunny day or not. 17 And we looked both a study that was done 18 by Andy DePaola and another study done by Chaney, 19 et al., from a university in the Pacific 20 Northwest. And we were looking at how much the 21 oyster temperature was elevated, and it looked 22 like we had a good deal of samples that were 23 covering both cloudy days and sunny days. And it MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 40 1 looked like the outer temperature was elevated 2 zero to up to ten degrees celsius so we assumed 3 the uniform distribution of up to zero to ten 4 degrees celsius td upper temperature range for 5 those oysters exposed for 48 hours. 6 BILL DEWEY: Bill Dewey, Taylor 7 Shellfish Company in Washington state. I'm not 8 sure if I can articulate my question because I'm 9 kind of confused and I'm not sure I can express it 10 that well. It's related to how the model takes 11 into account -- how the risk assessment takes into 12 account -- the pathogenic versus nonpathogenic 13 Vibrio. 14 On one of your key assumptions, you 15 acknowledged that the tdh were positive for 16 pathogenic. And then as we go through your 17 presentation -- the slides aren't numbered -- 18 there was one where you had predicted mean levels 19 of Vibrio parahaemolyticus per serving of oysters 20 at consumption. And as you presented that slide, 21 you said you predicted mean levels of pathogenic. 22 And then when you get to your validation slide in 23 the back of that exposure assessment, you MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 41 1 indicated that that was total levels of Vp. So I 2 guess I'm just confused. 3 And also in this industry as we've 4 learned in the last few days at the prior meeting, 5 it's a very small percentage of the Vibrio that 6 are pathogenic. And how is that being taken into 7 account? Is that a consistent percentage? Is it 8 assumed that the pathogenic is always three 9 percent of the total? Or how are we dealing with 10 that? Help me out here. I'm afraid I'm not 11 articulating this well. 12 DR. WALDERHAUG: I think I understand 13 what you're asking. I want you to know that the 14 way we did this was to model the percent 15 pathogenic in the Pacific Northwest differently 16 than the way we modeled in the gulf and the middle 17 and Northeast Atlantic. And other than taking a 18 simple percentage, what we did we took a random 19 value from a distribution with a mean that had the 20 same mean as the percentage that we have for most 21 of the data with respect to percent pathogenic. 22 The way we ran this model was we took 23 that percent pathogenic, and each time we did an MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 42 1 iteration or a sample of the simulations, we 2 picked a new value for that particular value, for 3 that particular percent pathogenic, and then ran 4 that through and simulated how much pathogenic was 5 present. At the same time we kept the total Vp as 6 well. And the reason we kept the total Vp is 7 because the retail study to validate only looked 8 at totals was to pathogenic. So that's why we 9 held onto the total even though only the 10 pathogenic was the one that was used in 11 calculating the risk. 12 Now, it gets a little complicated 13 because after we did 10,000 samples with the 14 random selection within a particular distribution 15 for the percent pathogenic, we then used new 16 values that represented our uncertainty about the 17 percent pathogenic and then used those as primers 18 for another distribution. This has gone into 19 great detail in the technical document. 20 But we feel that it gives a sense of 21 both the randomness of the percent pathogenic for 22 each meal consumed, but it also reflects the lower 23 uncertainty with respect to how well we understand MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 43 1 percent pathogenic and total. 2 DR. DePAOLA: And when he said we did it 3 different from the Pacific Northwest, we relied on 4 several studies -- environmental studies -- of 5 oysters where the Vibrio parahaemolyticus was 6 isolated and was later tested to see the frequency 7 of tdh. In the Pacific Northwest, the studies 8 were in fairly good agreement that about three 9 percent of the isolates are pathogenic, whereas in 10 the gulf and the Atlantic coast the data suggest 11 that .2 percent of the isolates there are 12 pathogenic. 13 VICTOR GARRIDO: Victor Garrido, 14 University of Florida. Looking at one of your 15 slides when it talks about the factors that 16 influenced the impact of controls, the first two 17 are actually the pathogenicity and the quantity 18 for amounts of Vp in the waters. But all the 19 control that we're looking at are actually putting 20 a burden on the industry to time of refrigeration. 21 Are we happy with the numbers that you guys used 22 for the risk assessment, or do you think that 23 we're going to need some more to fine-tune the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 44 1 risk assessment and not directly put the burden on 2 the industry? 3 MR. BOWERS: I think I understand your 4 question. And sensitivity analyses can get quite 5 complicated and perhaps misinterpreted. But the 6 way this particular slide that you're referring to 7 was run, the top one is the level in the 8 environment for a particular oyster-causing 9 illness on a particular occasion if we knew what 10 that level was. It's not referring to perhaps 11 like the mean level in the environment from what 12 your collection of oysters with a distribution of 13 values went out to the consumers. If we re-ran a 14 sensitivity analysis looking at the output of the 15 model, predicted risk versus a predicted mean 16 level at a harvest site from which oysters might 17 vary about that level, it would be a different 18 sensitivity analysis than the one that's presented 19 here. 20 VICTOR GARRIDO: Maybe suggesting a 21 control that would fall on the shellfish control 22 authorities to go out and monitor the oysters for 23 the levels of sensitivity, then they take the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 45 1 measures to open, close, or control harvest. 2 MR. KRAEMER: I'm supposed to be the 3 moderator, but really you're moving into the risk 4 management side, so I'll venture into this, but 5 not too far. 6 I think that suggestion is something 7 that should be considered by the ISSC which will 8 be convening right after this meeting. I will say 9 that the tornado plot in the first two bars 10 suggest that the two most important factors are 11 factors that man can't control other than by 12 determining when you harvest. You can go to 13 another region or another season and you would 14 find the levels lower. 15 But you're correct, those are not things 16 that are sort of hand in hand, if you will. It's 17 when you move down to the next issue that's 18 associated with how long they're exposed and this 19 is where the industry can have the greatest impact 20 other than avoiding certain seasons and regions. 21 RON KLEIN: Ron Klein for the Alaska 22 Department of Environmental Conservation. I have 23 a question on the dose-response model. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 46 1 Can you help me understand the 2 assumptions that were used in moving the curve 3 from -- as I understand -- the clinical trial data 4 to the expected? 5 DR. WALDERHAUG: The only assumptions 6 that are really based on moving that or shifting 7 that dose-response of the curve is the fact that 8 that's where we had to move it to get -- based on 9 our exposure -- the number of illnesses that we 10 anticipate that we want to match based on CDC's 11 estimates. So there were no assumptions with 12 respect to sensitive populations on those slides. 13 It was strictly done to line the results with what 14 we would expect from epidemiology. 15 Now, there may be reasons why it's a 16 good move because of the fact that you may have 17 differences between the amount you consume and the 18 amount you really wind up getting exposed to. But 19 we didn't make any assumptions along those lines. 20 Those are just possible reasons for why there 21 would be a dose-response shift. 22 RON KLEIN: Okay, thank you. 23 CASANDRA SHAW: Casandra Shaw, Louisiana MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 47 1 State Health Department. I have a question 2 regarding the predicted septicemia cases. Is that 3 based on the underlying population, the normal 4 U.S. population underlying conditions? Or does it 5 somehow take into account the difference between 6 consumer level data from people with underlying 7 conditions and without underlying conditions? I 8 was not sure about that. 9 MR. BOWERS: The assumption is that 10 there's no difference in the consumptions patterns 11 between at risk versus -- 12 CASANDRA SHAW: So you assume that the 13 consumption between people without and with 14 underlying conditions is the same on that model? 15 MR. BOWERS: People who are at greater 16 risk for septicemia consume moistures at the same 17 frequency as people who are otherwise more 18 healthy. That was the assumption. 19 CASANDRA SHAW: Okay, thank you. 20 BRETT BISHOP: My name is Brent Bishop. 21 I'm with Pacific Coast Shellfish Growers in Puget 22 Sound. Mr. Bowers, when you were talking about 23 the assumptions of intertidal harvest, you MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 48 1 mentioned that the tide could be out anywhere from 2 a certain number to eight hours. Granted we have 3 a high tide every twelve hours, and even highest 4 oyster beds -- let's say a plus six-foot level -- 5 will never be exposed to that. We never get more 6 than six hours. I am not a mathematician, so I 7 don't know if this is significant to the formula, 8 but I just thought I should mention it. 9 MR. BOWERS: That's a good point, a 10 relevant point. If it was possible, we would like 11 to know -- I've seen that there's some areas that 12 perhaps may be in the range of four to eight 13 hours, as much as four hours exposed to the air 14 from the time that the oysters are exposed until 15 they're transported. If there's a distribution of 16 different beds that have different ranges of 17 exposure, and that information could be provided 18 to us, that would be an improvement. 19 BRENT BISHOP: Thank you. The highest 20 elevation of which oysters will grow in my area is 21 plus six. And from maybe five or six thousand 22 days of standing out there with a watch watching 23 the tide go out, you never get more than three MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 49 1 hours before to three hours after before the tide 2 comes back in. 3 MR. BOWERS: In that area, what 4 percentage of the Pacific Northwest harvest does 5 that represent? 6 BRENT BISHOP: Most of Puget Sound. 7 DR. DePAOLA: I'd like to say one thing, 8 that this will be a work in progress. John will 9 be here some this afternoon and later on. And his 10 e-mail address will be provided, and we look 11 forward to the opportunity to do more specific 12 risk assessment modeling. 13 RHONDA TALLEY: Rhonda Talley, Pacific 14 Coast Shellfish Growers Association. When you 15 were showing that I believe was a 10-log reduction 16 for oysters that are resubmerged after the initial 17 harvest, and I might have that log reduction 18 incorrect. 19 DR. MILIOTIS: 1-log. 20 RHONDA TALLEY: But the question that I 21 have is when you were looking at that 22 resubmersion, at what level are those oysters 23 being submerged? Is it below the thermaline? MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 50 1 Would you address that? 2 DR. DePAOLA: I can speak to the study 3 that we did in 2001. We were there late July or 4 early August, and as it turned out it was a period 5 that was much cooler than what typically occurs. 6 There were a lot of cloudy days. And we did this 7 over a period of a week or so. We would go out to 8 the highest point, and when the tide went out in 9 the morning, we would collect oysters there and we 10 measured the density. And sometimes it was more 11 than six hours later. 12 In the afternoon, we would go back to 13 that same location and collect oysters again. 14 Generally the water temperature is running about 15 18 degrees centigrade and that seemed to be very 16 low constant throughout the period. The 17 temperatures in the air would get up to about 25 18 and sometimes the oyster temperature would get up 19 to 30, but it wasn't below a thermaline, and we 20 pretty much conducted it in surface waters. 21 MARK SOBSCY: Mark Sobscy from the 22 University of North Carolina, Chapel Hill. If I 23 understood what I thought I heard, you actually MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 51 1 adjusted the dose-response relationship in some 2 way to be consistent with disease burdens based on 3 CDC estimates; is that correct? 4 MR. BOWERS: That's correct. 5 MARK SOBSCY: Knowing that CDC's food- 6 borne disease surveillance system has seriously 7 underestimated the total burden of food-borne 8 disease in general and that there's a great deal 9 of uncertainty associated with those estimates -- 10 and CDC is the first to admit this -- first of 11 all, why would you do that? And secondly, how 12 would you do that taking into consideration that 13 this is an effort to do quantitative analysis? 14 MR. BOWERS: Well, your question has two 15 parts. And one is, I think all the people in this 16 room would agree that there's some degree of 17 under-reporting. I mean, when you're looking at 18 reported cases, this is not all of the cases that 19 are occurring. So there's the issue of what is an 20 under-reporting factor. I would leave that for 21 John Painter to discuss if you want to go there. 22 As to why we would want to model not 23 just reported illnesses, but all illnesses is MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 52 1 we're concerned with all illnesses occurring 2 whether or not they're captured by surveillance or 3 not. And CDC's best estimate of the under- 4 reporting was a factor of 20 arriving at a total 5 number of 2,800 per year, so we anchored our 6 dose-response to that. And we felt that was 7 reasonable because the clinical data controlled an 8 exposure to healthy eating and the fact that there 9 would be a different dose-response and that 10 ingestment was reasonable. 11 MARK SOBSCY: All right. Well, with 12 your indulgence, I'll ask a third question. 13 Are you planning to actually do any 14 epidemiologic studies respective to actually 15 determine the national burden of Vp disease? Is 16 that something in your game plan? Because frankly 17 I think that's the only way you can get the 18 estimates of what that might be. 19 DR. PAINTER: As you know it's very 20 complex to try and estimate all cases that are out 21 there when there are cases that don't seek a 22 physician's guidance or the physician doesn't ask 23 for a stool culture, and the laboratory of the MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 53 1 stool culture test tube doesn't use an appropriate 2 monitor. So the accounting process is a 3 complicated business. 4 And we have several ways of trying to 5 estimate that. A very exhaustive study was done 6 in 1997. And in 1998, it was published by Paul 7 Meade where he established a factor for most 8 moderately severe illnesses. And that was a 9 factor of under-reporting of 21. One could have 10 chosen a higher degree, and given the dose- 11 response data it probably would have been more 12 consistent with that. I think that we chose a 13 fairly conservative number for an estimate. 14 In addition to that study, our branch 15 has an ongoing project called Foodnet which seeks 16 to determine the incidents of food-borne illness 17 in the population. It's a very labor-intensive 18 and expensive program. So it targets a fraction 19 of the U.S. population. Currently there are nine 20 states involved that represent 13 percent of the 21 U.S. population. We compared some of our data 22 with Foodnet's data. And Foodnet's data actually 23 suggested a slightly higher number of total MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 54 1 estimated cases. 2 However, the Foodnet states do not 3 include the Gulf Coast. There is the potential 4 that it is not as representative, even then they 5 have a slightly higher number. So we decided to 6 use their more conservative number of 2,800 just 7 because we feel that it has for several years 8 consistently represented the national 9 surveillance. 10 As far as getting a more precise 11 estimate, we expect that Foodnet will perhaps be 12 getting more sensitive and perhaps expanding to 13 other sites. And we believe that one of the 14 elements that they are doing that can help us is 15 in the laboratory surveys that indicate what 16 clinical laboratories we're using appropriate 17 stool cultures to be able to detect the Vibrio 18 parahaemolyticus. And that's certainly something 19 that changes over time. And we hope that we can 20 get an accurate estimation of it. 21 I would guess that each year the total 22 -- if one were to redo an estimation, it would 23 vary based on certain assumptions. But I think MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 55 1 the number that we came up with represents a 2 fairly consistent and perhaps somewhat 3 conservative number. An effort to get more 4 precision would I'm sure interest all of us. And 5 the ability to do that and determine whether the 6 number is 2,800 or 2,900 is certainly a question 7 of managing resources as to how much that increase 8 in accuracy would drive the decision-makers for 9 the risk assessment. 10 WILLIAM ATHAWES: WILLIAM ATHAWES, New 11 York State Department of Environmental 12 Conservation. This is also related to 13 epidemiology and illness reporting and the risk 14 assessment. 15 As the individual who reports the number 16 of illnesses to the FDA and ISSC, I have a 17 question regarding what appears in the 18 interpretive summary. And forgive me if I don't 19 know the entire quantitative risk assessment yet. 20 But on page 24 is an interpretive 21 summary. You have number seven Vp illnesses in 22 our state that was reported for 2002. When we 23 reported to the FDA in 2003 for the 2002 calendar MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 56 1 year, we reported a total of ten cases of which 2 only three even had oysters as one of the foods 3 consumed. I was interested in knowing where you 4 got the number seven from. I double checked with 5 out epidemiological people last week and they 6 confirmed a number of three. 7 DR. DePAOLA: Page 24? 8 WILLIAM ATHAWES: Page 24 is on my 9 computer. But I believe it was an interpretative 10 summary in 2002 for New York state the number 11 cases were seven. And as far as we can count 12 there's only three. Now, this doesn't seem like a 13 whole lot in the grand scheme of things, but it 14 does again build a faith confidence on some of the 15 inputs and therefore some of the outputs. 16 DR. DePAOLA: What's on this table 17 doesn't necessarily reflect what gets into CDC. 18 For CDC there's only culture-confirmed cases and I 19 think here we have clinical cases where a culture 20 may not have been submitted to CDC. 21 WILLIAM ATHAWES: Well, again these are 22 alleged cases that were associated with raw oyster 23 consumption. The New York State Health Department MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 57 1 only reported three cases. How did you get seven? 2 MR. KRAEMER: I suggest that we need 3 some time to better understand your question and 4 conduct some research, but we'll get back to you 5 as soon as we can. 6 AL SUNSERI: Al Sunseri, P&J Oyster 7 Company in New Orleans. My question is on the use 8 of total Vibrio parahaemolyticus numbers in the 9 retail study. 15,000 I think was the number. 10 What happens when shellfish are 11 refrigerated at the proper temperature over a 12 seven-day period to the pathogenic Vibrio 13 parahaemolyticus? 14 DR. DePAOLA: Our assumption is that the 15 same thing that happens to the total population. 16 And under refrigeration, there is a reduction. 17 And I think after about a two-week period, it's in 18 the neighborhood of about a 1-log reduction. It's 19 more precise in the risk assessment, but that's in 20 the ballpark. 21 MIKE VOISIN: Mike Voisin with Motivatit 22 Seafoods in Louisiana. On page nine under the 23 regional comparisons on the bottom graph in your MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 58 1 handout, you show in the whole line total 2 attributed illnesses and then model predicted 3 illnesses. It appears that in the Gulf Coast, 4 your model prediction is over a hundred percent 5 greater than your total attributed illnesses. 6 I'm confused. Is the model off by over 7 a hundred percent? Am I reading this wrong? Is 8 there something I'm missing? 9 MR. BOWERS: No, you're not reading it 10 wrong. If you double 44 percent, that's getting 11 closer to the 90 percent. So if you want to 12 phrase the discrepancy between the model versus 13 CDC data based on that 100 percent would to be 14 fair. 15 I would say that the total attributed -- 16 you know, the model is generating a predicted 17 illness, and the analysis of the CDC is also -- 18 it's based on an analysis itself. So it's not a 19 perfectly accurate attribution of the illnesses 20 either. So it's not comparing one thing to gold, 21 it's comparing two things which have some measure 22 of discrepancy versus what is real. 23 MIKE VOISIN: So in your discussion, MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 59 1 then the Atlantic coast is off by 1,400 percent, 2 the Pacific Northwest would be off by 400 percent, 3 and the other Pacific wasn't modeled, does this 4 create a lack of confidence? 5 Or does it in your mind create some 6 level of confidence in what you're trying to 7 accomplish in determining risk? Or is it just that 8 both sides have such different assumptions that we 9 may not be able to assume anything? 10 MR. BOWERS: That's a very good 11 question. I can only speak for my opinion on 12 that. There is a difficulty in identifying the 13 cases epidemiologically and one can do the best 14 analysis that they can to try to hammer that down. 15 And the same applies on the modeling side as well. 16 And the fact that this disagreement does 17 exist doesn't necessarily mean that we don't have 18 a good model or that we don't have good 19 epidemiology, but that either one is perfect. 20 The difference is troubling but we do 21 the best we can and we hope to improve in the 22 future. We hope that CDC can do some good 23 epidemiology in the future and that we can do MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 60 1 better modeling in the future. 2 DR. PAINTER: We found those differences 3 striking as well. There's no doubt there's a 4 difference there. And our conclusion was that 5 there's probably some undetermined factor that 6 accounts for the difference of level of illness 7 that's reported from the Pacific Northwest 8 compared to what was modeled in the risk 9 assessment. That may be an unknown factor, a 10 strained specificity, that wasn't included in the 11 model. It may be a factor related to the 12 pathogenicity, the percent pathogenic or some 13 other pathogenicity factor. 14 We found in reviewing the model, we 15 thought that overall its assumptions were sound 16 and that that disagreement was something that one 17 would hope would disappear with improved data, 18 both epidemiology and as part of the risk 19 assessment. But that within certainly the Gulf 20 Coast, the assumptions certainly seemed to us to 21 be valid, the calculation of the total numbers. 22 There is clearly a disagreement. But 23 whether or not that invalidates the model is MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 61 1 really a very different question. We don't feel 2 that discrepancy of number of volume refers to it 3 as having a high percentage disagreement. The 4 importance of that percentage difference is not so 5 great. Meaning, for instance, that the rank order 6 is perhaps more important in determining where one 7 would focus attention. 8 That may be more a question of the 9 people using the model rather than the models 10 themselves. But there is going to be some 11 difference where cases are reported and what's 12 predicted by the model. 13 MR. BOWERS: I'm glad, John, you brought 14 up the question about the model. I didn't bring 15 that up, but I'm glad you brought it up. Thank 16 you. 17 MR. KRAEMER: Mike, I think there was a 18 part of your question that was a risk management 19 question when you asked what level of confidence 20 we have in the model, or can we assign confidence 21 to it given that magnitude of difference. And 22 that truly is a risk manager's decision. And at 23 least in the meeting that's about to happen, we MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 62 1 are all risk managers. 2 And I think the beauty -- if you will -- 3 of a risk assessment is it's very transparent. 4 You're able to make judgements for yourself 5 whether you think that that piece of data is so 6 compromised as the model that it's not useful. 7 But I think you heard Dr. Brackett in 8 his opening comments suggest we believe the model 9 is useful. We think that while it is something -- 10 while it is a disagreement between two data sets 11 that we would like to resolve, in fact we have 12 some research underway to try to resolve -- it 13 doesn't so fundamentally compromise the risk 14 assessment in FDA's opinion as to make it not 15 useful for modeling. 16 TOM DRUM: Tom Drum, New York State 17 Department of Environmental Conservation. During 18 the presentation a comment was made on the number 19 of illnesses that could not be traced back to the 20 source. I missed that figure. That was my first 21 part. 22 DR. MILIOTIS: I'll answer your first 23 part. 715 oyster-associated cases were reported MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 63 1 to the CDC in a period of time. Of the reported 2 cases where the illnesses occurred or where they 3 were reported, of those only less than 20 percent 4 were actually able to be traced back to the 5 source, the regional harvest. 6 TOM DRUM: Okay. So the second part of 7 my question is, the information is based on the 8 CDC reports from where the incidents actually 9 occurred, for example, the restaurant or the 10 store; am I correct, where the product was 11 purchased? 12 DR. MILIOTIS: No, from the region where 13 it was purchased, where the oysters were 14 harvested. For example, if people reported 15 illness in the Mid Atlantic, but the oysters -- 16 the culprit oysters -- were traced back to the 17 Gulf Coast and the Pacific Northwest, the harvest 18 region, not necessarily the region where the 19 illness was reported. 20 TOM DRUM: I'm not clear on what you're 21 stating. 22 DR. MILIOTIS: Okay. When illnesses are 23 normally reported, they're reported where they MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 64 1 occur. For example, if I eat oysters here today, 2 they may be the oysters from the Gulf Coast or 3 they may have come from the Pacific Northwest. I 4 go back to Maryland tomorrow and I'm ill. And 5 when I go and file a report, it's going to be a 6 report that comes from Maryland, which would be 7 the Mid Atlantic region. But that's not where the 8 oysters were harvested. 9 So when we report -- when the CDC 10 obtains data, surveillance data -- the data is 11 provided where the illness was reported. Like my 12 illness would be a Maryland illness. And what the 13 risk assessment does, it predicts illness from the 14 source from where the oyster was harvested. 15 TOM DRUM: Okay, thank you. 16 ROB WITTMAN: Rob Wittman with the 17 Virginia Department of Health. I have a question 18 concerning your slide that's on page eleven of the 19 handout, the impact and control levels at harvest. 20 Dr. Miliotis, you spoke of essentially a 21 1-log reduction in the time of harvest to the time 22 of retail. If you look at the guidance level and 23 the implementation of a 10,000 level for Vp, and MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 65 1 you see illnesses averted to 16 at harvest and 99 2 percent at retail. 16 percent at harvest and 99 3 at retail. 4 Are there other considerations that go 5 in there? It doesn't seem logical that only a 6 1-log reduction would result in such a wide-spread 7 in illnesses averted in a 10,000-level implemented 8 at harvest versus at retail. And that goes for 9 the rest of the table too. There seems to be a 10 big difference in harvest and at retail for the 11 implementation to be up at that level. 12 DR. DePAOLA: The 1-log would be an 13 average reduction. And what the 10,000 may 14 represent are examples where there's much more 15 than a 1-log increase. And the reason would be 16 that there was a regular time and higher 17 temperature from harvest to refrigeration. 18 So what we have in harvest is that only 19 a very small percent of oysters -- only three 20 percent -- are over 10,000. This is an unusual 21 level to find at harvest. It occurs occasionally. 22 So when you take those three percent off the 23 market, you've averted 16 percent of the cases. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 66 1 But during the distribution, harvest practices, 2 and other things, by the time they reach the 3 consumer, 43 percent are now exceeding 10,000. 4 And if you take those off the market, then you 5 reduce 99 percent of the cases. 6 BOB COLLETTE: Bob Collette, NFI 7 Shellfish Institute. Could you comment on the 8 strength of the information used to model the 9 predicted growth of the pathogenic Vp relative to 10 total? 11 DR. DePAOLA: Okay. We have two 12 different: the growth of pathogenic versus the 13 growth of the total. That was largely an 14 assumption, The strength of that is not that 15 great, because it was difficult in the past to 16 enumerate pathogenic. We do have some recent data 17 from Alaska -- actually just a few weeks ago -- 18 that helps confirm our assumptions. There are a 19 high percentage of the isolate pathogenic and we 20 did find that they grew proportionately to the 21 total population. 22 So for the purpose of this risk 23 assessment, it was quite an assumption. When we MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 67 1 inoculated, artificially contaminated, the 2 oysters, we did find that 03.K6 strains would grow 3 at a comparable rate. But this did not appear to 4 be the result. 5 MR. KRAEMER: Any other questions? I 6 want to suggest that many of us on the panel will 7 still be here throughout the ISSC. And, of 8 course, we're available to answer questions. Or 9 if you'd refer to contact us after the conference, 10 we're also available there. 11 What we'd like to go to now is the 12 public comment period. And at this point you're 13 free to obviously say whatever it is that you feel 14 compelled to say about this risk assessment. 15 And of particular interest to us I think 16 are some of the questions that Mike Voisin raised 17 earlier about the utility of it and where we go 18 from here. That's a fair game I think. 19 We did have three individuals that based 20 on the instructions in the Federal Register 21 announcement announcing this meeting pre- 22 registered to make public comments. And so I'd 23 like to go to them first just to make sure that we MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 68 1 have time for them. We have about an hour and a 2 quarter, so I think we probably will have enough 3 time. So don't worry about being able to make 4 your comment. 5 The first comment we have is from Dr. 6 Raoult Ratard who's a state epidemiologist from 7 the Louisiana Office of Public Health. Are you 8 here? 9 DR. RATARD via telephone: My question 10 was answered. 11 MR. KRAEMER: Okay, thank you very much. 12 We had two participants who could not be here but 13 asked to make a public comment via the telephone. 14 So we're going to try out this technology. 15 We have someone from Public Citizen. 16 Are you on the line? 17 ZACHARY CORRIGAN via telephone: Yes. 18 This is Zach Corrigan from Public Citizen. Can 19 you hear me? 20 MR. KRAEMER: Yes. Can the group hear? 21 Someone in the back, can you tell me if there's a 22 problem? Okay. Good. All right, go ahead. 23 Thank you. MacReporting Service PO Box 1734 Fairhope, Alabama 36533 (251) 929-0941 69 1 ZACHARY CORRIGAN via telephone: I'm 2 sorry. I'm havin