 So our final speaker for this morning is Dr. Lorber from the EPA, excuse me, Dr. Lorber. Yes. Thank you for inviting me to speak here. It's certainly a pleasure since I've been working with panel members and members of the audience. In particular, I'm pleased that I didn't know actually that Dr. Stahlhau was going to be here and present. As you'll see, I take apart one of his pieces of work, but that's a bit in my talk. But I also want to acknowledge here that a lot of what I'll be speaking about results from work that I've done with Holger and with Lisa here in the audience. So again, folks, you two can speak up and correct me if I misspeak about some of the things we're doing together. I want to give an outline of my presentation. I'll be giving some background on what I'm going to be talking about. Before I get into phthalates, I did want to talk about bisphenol A, and this is where Richard's work has come in. Again, I look at one of his studies on BPA and point out as many flaws, but I have to be a little polite now that I know that he's here. But it does address, but I actually very much appreciate his talk. It leads in perfectly to mine. I'll spend the most of the time talking about trying to use NHANES data and particularly the amount of time fasting and see if we can learn anything about phthalate exposure and the role of diet versus non-dietary exposure. And that's what I'll spend most of my time. And also, I think it's informative that he pointed out the difference between the 2003, 2004 NHANES versus the later versions. And pointing out that in the later versions, they actually changed the rules on fasting, and that did result in a different distribution of information. So I'm glad I didn't try to meld the two. I stuck to one, and I think I stuck to at least the one that might have a more uniform distribution across all the participants. Finally, I'll talk about a database that I also tried to use to look at exposure from a forward perspective. And I think the panel has been given some information from Clark who tried to take the forward approach, which is take the concentration in the media and the food times the contact rate to get intakes in that manner. And there's actually a pretty good database put together by the American Chemistry Council that I tried to use to do the same kind of thing. And I'll give an example of how that could be used also to try to study this issue of the role of diet. So I just give you a little bit of a forewarning that my slides do contain. See, this is the forewarning. My slides do contain a fair amount of numbers and words. And I'll try to step you through what you should be looking at and what's the meaningful part of each slide that I present. Here I'm trying to start you off with a little bit of background on intake estimates that are in the literature. This was a good study put out by Wormuth in 2006 looking at exposure of Europeans to phthalates. And he again did a forward based analysis where they combined contact rates and concentrations. They looked at many different phthalates. And I just want to point out that this individual identified those which might be dominated by food and those which were dominated by other pathways. And a particular note here, everyone seems to believe that DEHP is dominated by food. This individual also had the second highest exposure to DNBP, which he also claimed was dominated by food. Bring that to your attention, because the next two talks I give up, people concluded just the opposite that DNBP was not dominated by food. And in fact, some of the analysis I did also kind of clouds the issue a little bit on whether or not DNBP is dominated by food or not. But anyway, other estimates in the literature, so this is a forward based approach. The other major approach is looking at urine concentration and back calculating based on a creatinine type of an understanding to get at estimates of intake. And for DEHP, we see he estimated three. And certainly in the literature, we're in that range when you start from urine to work your way backwards. So that's kind of a ball park for DEHP on a daily basis, three microgram per kilogram per day. Here's another very good piece of work by Fromm in 2007. And basically what they did is they took 50 individuals, took an estimate of their intake by analyzing the food they ate. And that was one measure of their dietary intake, then took urine measurements and back calculated the total intake. And if their intake was dominated by food, then maybe their intake as calculated by their urine measurement might equal or be close to the intake that's calculated by their food. That's the hypothesis. What these are, just there's a lot of data and results in the paper. I just called out this particular set of graphs. These are cumulative frequency distributions showing the intake on the x-axis and the frequency in the y. So if you look down, you can say this number intake, 60% was less than this intake and so on. There's two curves and these are actually points that are connected, dots of the 50 individuals. One of these curves represents biomonitoring the other dietary. It doesn't really matter which is which, but you can see that when you take these data points from these individuals and look at what their intake was based on assuming it was food only and then versus urine, you see that there's almost a one-to-one correspondence with DEHP. In other words, the exact amount of intake estimated by food was calculated as estimated by urine. I mean, it's just kind of awe-inspiring here to see how well it lined up. But it didn't line up with DNBP at all. I mean, there seemed to be a very large discrepancy. So again, we see this type of a study contradicting what Wyrmuth found when he said he thought DNBP was food-dominated. This would seem to appear that for a reasonable number of individuals, a fairly robust study, food did not dominate their intake. That's another reference you could get more detail from. This is some of Holger's work on fasting where he took these willing-enable young volunteers and said, you know, why don't you stop eating for a couple of days and let's take your urine and we'll measure it for, here are a bunch of DEHP metabolites and here are metabolites of other phthalates. And it seemed like he developed a rather compelling argument and these are averages over many individuals. Is that right? This is not just one. These are averages. And he developed a fairly compelling piece of evidence that after about 12 to 18 hours it was really low. So, you know, this really does look like their exposure over up to two days of fasting. Their exposure had been dominated by food because after two days it was actually nothing left in their urine, essentially. Whereas for these other ones, there was an initial drop, but then it peaked up again. So that looks like some of their exposure over the two days might have been occurring in other than food sources. So, you know, you can actually study this issue from this perspective, it's experimental. But what I try to do is look at NHANES, but before I get into NHANES a little more background, this is some other work that data that Holger had generated and that I and Holger had used in some modeling work. This looks at the metabolites of DEHP and urine over time from two individuals where they had a large exposure, the exposure was stopped. You measured their urine after the exposure stopped. In one case it was a self-dosing of a very high amount through oral. In the other case it was a very high dose through intravenous. So it was a blood donation procedure. But in both cases Holger was convinced and had right to that the exposures were high and then it stopped over time. So when it stops after a while and you're measuring urine, eventually the third metabolite, the tertiary one, shows up to be the highest because you're not exposed anymore. The secondary one, the first one is MEHP, that comes and goes. Then you get your secondary and tertiary. Over time the tertiary, this is the ratio of the tertiary to the secondary, the third to the second. Over time the third dominates, okay? So that would be, now if you try to use, you can kind of see in your mind where I'm going with this eventually. If I use fasting data and people fasted for a long time, what is the ratio in their urine? Does it match this type of information where the tertiary is exceeding the secondary by a factor of two or more or one or so? And both these individuals you can see a pretty clear and ambiguous trend that over time once you stop the exposure you're going to get a ratio that increases. So it's a little more background here. Rather annoying. Luckily, Richard was here earlier to discuss the PH, the fasting hour parameter in NHANES. So I'm not going to really spend that much time. It does become interesting if you start to try to understand what their instructions were and were they possibly exposed to things that could have had phthalates in them even though they claimed to have fasted. And they were, I don't know why they were just so worried about supplements or mints or cough drops, gum or snacks or whatever. And in fact there were questions that you're not supposed to do this, but then when you got there there were questions, did you in fact do that? So I didn't try to call out those that said, yeah, I was chewing gum, sorry. But we don't really know. And I think that Richard talked a lot about recall uncertainty, whether there might have been other ingestion, possibly related exposure. I don't know if phthalates are on toothpaste, but who knows. So that was a good introduction on that. I'm glad I don't have to spend a lot of time. Now I'm going to look at Richard's paper on Bisphenol A. It was a paper that was brought to my attention by Holger and he said, you know, what gives here? I thought we might find something in NHANES, but Richard didn't find anything. So what's going on? He applied the 2003, 2004 data using the fasting hour to find, here's N. And this is, you're going to see a lot of numbers displayed in this way, where I show the number of participants, the fasting times. And he looked at the median, and again in a creatinine corrected basis. And between those who claim to have last eaten within the past four and a half hours, there was a median of 2.6. And within this range, there's a median of 2.3. And then the rest of time, there was this median. So you're looking at medians, and he showed a lot of other data. I just pulled out this one piece. You would seem to say there is no relationship with time on fasting. So even though the premise and the thought was BPA was related to food exposure, this analysis within NHANES and the fasting seemed to disprove that. And that was perplexing. And Richard provided a bunch of possible explanations. And I should point out he excluded individuals under 18 years old. I'll get to that in a minute, and all older folks, and also greater than 24-hour fast, which I did as well. So I took it out of the database myself. And of course, this is all unweighted. Everything I talk about within NHANES is unweighted. I'm just looking at numbers. And I took it out on my own, and I did not first exclude the younger folks. And I looked at urine volume, and I looked at urine creatinine. And I looked at slightly different, similar but slightly different fasting periods. And I broke it out in regions of four. He went between eight and 24. One thing I saw, and I looked at means, and I looked at medians by urine volume, and urine creatinine. And one thing that struck me is that these numbers don't look very well behaved. They kind of bounce around. There's obviously no trend when you look at the mean of a population. And there's also a lot of bouncing around when you look at median. But when I did it on a creatinine basis, and I was a little discouraged where Richard was trying to say negative stuff about creatinine. Because I was thinking, well, this is well behaved. I mean, this kind of gently goes down. And this kind of gently goes down. I really felt like this issue of hydration. When you use urine concentration, you could get a low mass. But it's a tiny bit of urine, so it's a high concentration. Or similarly, God could be drinking like a lot. And then he has a big volume of urine and a pretty meaningful amount of chemical. But it shows up as a low concentration. So I was thinking what this demonstrates is that this issue of hydration makes urine volume a tough metric to work with. And so I decided not to work with it. Now, as Richard was pointing out, this one's not too much fun either. Because there's a lot of variation with men and women and black and white. And it's just, but I didn't do any of that breakouts. And after seeing this talk, maybe I should. But I was kind of feeling pretty good that let's just stick with urine creatinine based on this analysis. Another thing I noticed when I did this is that I did see actually a small difference here between this first grouping, 0 to 4 and 4 to 8. Richard didn't see that. I did see it, but what I did is I looked at everyone. Richard only looked at those individuals greater than 18 years old. So I'm wondering here, the medians were consistent with his except for 0 to 4 hours, does that mean there's something important about the individuals which he took out of his set to look at? So there are issues, and we can talk about them, but I decided to call out those younger individuals, less than 18, greater than 18. And I see something there, I didn't break out 6 to 12 and 12 to 18. I just said 6 to 18. And as I understand it, there are issues with the 6 to 12 group. They are not given any instruction on fasting at all. So having not called those out could have created issues for me. But if you just kind of eyeball the numbers, and that's all you're gonna do today is just eyeball, because I didn't do rigorous statistical testing. If you eyeball the numbers, I'm seeing a slight decreasing trend here with fasting for the, maybe more than slight. Maybe something that could be proven to be statistically significant for the younger folks. And then the older folks which Richard looked at, sure enough, these are almost exactly what he found. There's a slightly declining trend that may have been masked by Richard combining all the way from 8 to 24. And when I broke it out, maybe you see something declining. I'm not sure, that's again up to the statisticians to tease something like that out. The other comment I'll make is that when Richard was focusing on medians, if you look at means, do you get different information? And Richard touched on that when he was pointing out the very high numbers for the zero to four hours. You get some very high people. And I'll point that out later too. Then when you focus on medians, you're looking at the middle of the population. But if there are a critical mass that are way up there, that draws the means up. And that might be information you don't see if you focus on medians. So there's a lot of issues here. Means versus medians, age, race, sex, it never ends. But I tried to make my study a little more manageable. So for the remainder of my talk, I'm going to be focusing on these fasting times. And I'm breaking out by these age ranges. I'll be maybe looking at too many numbers, but I will be displaying. And you can look at them at your leisure, means and medians. And I will be displaying just to get a sense of the robustness, the number of individuals. So that's my analysis of BPA. Let's look at DEHP and what else for phthalates we can learn. Remember I talked about that ratio. The key ratio in the study were these two individuals. Well, if you actually look at NHANES data and look at that ratio from the same groupings of hours, I was hoping to see a trend that looked like the experimental individuals. And I didn't quite. I saw a gently increasing trend. And that's supportive. There may be a signal here. That's supportive, but it's not really strong. It supports the notion that this is proof, that dietary exposure is important for DEHP. But it's not as strong a signal as I was hoping to get. So that's one piece of evidence. Let's look at some other basic results for creatinine and this fasting as a function of age. Again, these are the kind of tables I've been generating. Did I see anything for MEHP? This is the monoester, the first metabolite formed. I think I probably did. I see some decline in mean. I see some decline in median for the younger folks. I also see something for the older folks, but it kind of lags a little bit. You don't really see the decline till here. And you see a decline over here. So I would probably look at this at least without the statistics and say there might be something here for MEHP, the first. I don't know if it's a really strong signal. It tends to show up more in the later hours in both sets of individuals. I'll also point out because it becomes important when I look at other phthalates that the concentrations between the younger folks and the older folks are fairly comparable. There's not a big difference. When you look at the others, there will be a difference. And you might ask yourselves why. I don't know why. I looked at this as a tertiary metabolite now. I'm going to another DEHP metabolite. I'm not as pleased about seeing a declining trend. I might see something when you look at the later hours of fasting. Now it's starting to decline. Maybe here, although even here are the medians. It looks like a population that don't have a lot of change in their central tendency. You don't really see a lot for adults. Maybe you see something for kids. Again, this is very difficult data to use. I'm trying to see if I can see something jump out at me. I didn't see much. I tried to make a statement in the end of these slides that say maybe a trend for less than 18 for medians when you look at the difference between the first, well, not even that. Medians might show difference. But look at the means. They kind of change. So you really have to dig into the distributions to see whether or not you see anything. I thought maybe a median, not means maybe a trend for declines by 16 to 20 hours, but really not much here upon visual inspection. But again, I think that's the tertiary metabolite versus a primary or secondary. Let's look at some other phthalates here. Again, the same kind of information. Remember I pointed out the difference between younger folks and older folks in concentration. Just by looking up and down the columns, you can see that younger folks have twice as high concentrations on a creatinine basis compared to adults. I'll interpret that a little bit in a minute. Wasn't true for the MEH to the DEHP metabolites, but I don't know why it's showing up for others. So one of my conclusions for this BBZP is that clearly differences are the function of age, but little evidence of a trend with fasting times. You don't see a lot here. Maybe, and for a lot of these, you see a difference between the first grouping and subsequent groupings. Like here, if you just eyeball it, maybe 46 is higher than all the rest. 27 is higher than all the rest. Don't really see it again for adults. 19, these means are looked pretty similar. Medians look similar. So you're not seeing any trend for older folks for this, and maybe only a slight trend for younger folks. So a lot of information. I don't know what's coming out of this at the end, but I tried to look at a couple of difference, not just DEHP. So then I looked at DNBP. I found that also difference is a function of age. It seems like the younger folks are exposed to a higher amount or have higher creatinine. Let's get to exposure in a minute, as compared to adults. And maybe a slight declining trend here, which would support a conclusion about food being important. Certainly, it looks like a declining trend when you look at 0 versus 4, 0 to 4 versus others. 54 is the highest, 31 the highest. So maybe something is, maybe there's a signal. Not quite sure. I know there's a lot of information you can stop me if I'm going too fast or if something comes up. Finally, the last one I looked at is DIBP. Again, some difference between younger and older in terms of the concentration. But if I look at these numbers, I'm not seeing anything for adults. And frankly, I'm not seeing anything for children. Nothing is jumping out at me visually. So this would suggest that DIBP is not a food-related exposure. Maybe it's easier to see in that than some others. OK, let's take a pause and see what I have said. DEHP metabolites were consistent in less than 18 and greater than 18. I talked about concentrations. The concentrations of the others, however, were higher in the younger as compared to the older group. So what does that mean? Well, one thing we can say is that younger folks excrete less creatinine as a grouping. So on a mass basis, they're excreting less creatinine, and yet their creatinine concentrations are higher. So if you want to try to understand what that means, the mass that children are excreting of chemical, the mass of chemical not creatinine, might be the same between children and adults. Let's run that by. That's the second bullet. The mass might be the same. In other words, the children have a higher concentration, but they're excreting less creatinine as compared to adults. So on a mass of chemical basis only, they're similar, but children also are lighter. So on a mass per body weight basis, children are going to be higher than adults when the trend is for creatinine concentrations to be higher for children and adults. I guess if you want to look at a bottom line, if that's a little, the math is too crazy, just look at this. Creatinine-based concentrations might be a good first indicator for trends in body weight-based exposures. Just as a first indicator, because you again use a difference in creatinine excretion but a difference in body weight, and maybe they cancel out. So when you're looking at DEHP metabolites, the suggestion is they're on a body weight basis, children are equally exposed than adults, because as I showed, the DEHP metabolites showed similar creatinine concentrations. For the other three that I showed, the children were higher creatinine concentrations as compared to the adults. So maybe it's also higher on a body weight basis for children as compared to adults. So trying to run some numbers by can get a little blurry after a while, but that's sort of the lesson there. Let's also step back and look at some review of some of the things I've tried to talk about. I think creatinine-corrected concentrations are expected to be more stable. Again, as we learned this morning, there's a difference between whether you're talking about black or white people or male or female, so maybe it's not as well-behaved as I had initially thought, but I think they're still probably better behaved as compared to urine volume concentrations. I'm not really sure what to use. I'm not really sure where to go at this point. Do I want means, medians? How about scatter plots? How about box plots? I'm gonna show a scatter plot. Maybe you really do want to focus on the 95th percentile and up when you're looking at potential health impacts. So what is the metric that helps us understand the questions? I mean, this goes back to the Tom's talk too. Pose your question correctly. What is it do you want? Do we want to look at means, medians, upper percentiles? I'm not sure. I tried to look at both. I did give an example in Richard's paper where when you looked at medians only, it looked stable, but maybe if you looked at means, you might be seeing something that you didn't see before. And I'll show you that graphically in a minute. BPAs show what appear to be evidence for food exposures when looking at younger individuals but not for older individuals. So I took that after looking at BPA first. I said, okay, now I'm gonna break out that grouping because it might be important for phthalates. I'm not sure I saw as much, except maybe for MEHP as I thought I saw when looking at BPA. And then this ratio of tertiary secondary showed upward trend with, but not as steep as implied by experimental data. So that was a little discouraging to me, but life goes on. And actually as you'll see, there's another part of the story. One thought I had is that if their relationship was gonna be seen between fasting hour and a DEHP metabolite, you'd see it with the first one. You'd see it with a monoester. You would probably not see it. That's just a hypothesis and I think it was born out by the data because the first metabolite gets out of you quickly. So if you've been fasting eight hours, you probably don't see it anymore, whereas you might still be seeing the secondary and tertiary metabolites. So if you're really looking to understand whether or not diet influences with use of the fast hour, then focus on the first metabolite, in this case MEHP. And I think I saw something between age groups and between different time frames. But I didn't see a relationship for the tertiary metabolites. I didn't see a whole hell, I didn't see a whole hell of a lot for the other three monoester metabolites, despite at least one researcher saying diet explained the exposure for two out of the other three that I looked at. So maybe something to learn from the differences between zero to four versus subsequent hours, I tried to show that, but maybe you just have to try to do more statistical testing with the data or maybe there's just nothing there to see. Let's discuss some of these subtleties and see if maybe we can learn something. If we look at, again, Holger's data, we saw a break around here where all of a sudden it dropped and there was pretty low throughout the rest of the time frame. So I asked myself if I look at the data on a scatter plot basis, am I also seeing something different about this area versus the earlier area, as was seen here? Maybe I do, actually. Maybe there is at least some prevalence of higher, an indication of higher exposure as compared to what Holger was finding in his fasting experiments. I'm not sure, it's just a discussion point. There's very, the best fit suggests a line that declines, it's linear with a very, meaning very poor correlation obviously with very low Rs, negative showing a declining trend. One other thing to point out by when you look at the scatter plot versus looking at means or medians, the scatter plot, I mean these are 890 points are on this figure, but you only see maybe how many, 50 or 80 that you can actually see. Everything is clustered near the bottom. So when you actually start to focus on the mid to central tendency, you lose information, and this by the way is a straight chart, it's not a log chart. So again, if I was to display this information on a log Y axis, then everything would look pretty evenly spread out. You don't get a feel, if you will, a sense that there are some meaningfully higher people in the population until you put it on a regular axis and not a log axis. So how you look at the data, again, frame the question, that was a point that Tom pointed out, what are you interested in? If you really are interested in higher exposures, then maybe you don't want to be so hyper-focused on the median, because you're going to be the same for all these population groups. All the different fasting, the median is probably very similar. So 6, 1700 people on this, and yeah, how many do you see in this range? So just kind of a thing to think about, food for thought if you will, unintended. Another, some other issues here, just think about this. It provides an indication, hopefully, an accurate one of the last possible exposure. But there's no information on the last time the guy peed, or the lady urinated before she came in and gave the sample. I mean, if it was just a half hour earlier, maybe you lost all your information. So we'll never know. So we're kind of maybe the experiments that like Holger did or Fromm did, where you actually collected over time, maybe that's the only way to get at this ultimately. Maybe NHANES may not be a good way to really get at this. The time during the day, and I'm glad Richard went into this in a fair amount of detail, because that also is an issue. Do you take a morning, an afternoon, or a night sample? And I actually went ahead and looked at this. And furthermore, I don't even know if the generation of creatinine changes. I would assume it does. When you're sleeping, you may not be generating as much. So the morning creatinine-based concentration is simply influenced by the rate that you've generated creatinine, and maybe nothing else. I don't know. These are all sorts of confounders. But when you start to break out the data between morning, afternoon, maybe you're actually seeing something, maybe something new is coming through in the data. So I took another look at MEHP, the monowester. Now I looked at morning, afternoon, and evening. Again, the fasting hours and the end. And I tried to bold what I'm seeing. I said, whoa, I think I see something here. The evening samples are showing a good relationship. They seem to be showing eyeballing without statistics. A good relationship between fasting time and concentration. So now I'm finding anything. Why am I all of a sudden finding something in the evening that I haven't found before? Well, one thought is when you've taken the evening sample and you're claiming to have fasted a long time or fasted a short time, what's being incorporated is maybe you did have breakfast, maybe you had lunch, maybe you had an afternoon snack or a mid snack. You may have had a lot of food exposure so that the true, this is a more true indication of really having fasted. And the zero to four people probably had a lot of exposure. So maybe this is more informative, simply, intuitively. Richard tried to talk about this a little bit. It's more informative to look at the evening sample with regard to past exposure, to look at it another way. If you come in the morning and you say, yeah, I ate two hours ago, I had a light breakfast. All you had two hours ago was a light breakfast followed by maybe nothing for eight or nine hours or 10 hours. Whereas if an evening person came in and said, yeah, I had something two hours ago, he may have had something, he may have had a dinner two hours ago, but also with a lunch or breakfast and snacks. So his hyper, he might have been very highly exposed, whereas this individual reporting only a low fasting time might have been hardly exposed at all. But if you group the data all together, you're not gonna ever see that. But you really have to start to maybe call out also as a function of time. So I found that to be interesting and I went back and tried to do the ratios with only the evening samples. Now all of a sudden, maybe I'm seeing something. I'm seeing these triangles. Now you are seeing a different ratio of metabolized tertiary to secondary when you only look at the evening. So maybe now I'm a little bit happier that something is informative coming out of this analysis. You can also use modeling to try to study the relationship between what comes out in your urine versus how you were exposed over the course of a day. This is work that was done by Lisa and she's here in the audience to explain it better, where you use the model which takes input. You have a dietary or have any input you want and you can specify when the input occurs and then you can simulate when the excretion would occur. And these are three different simulations described here. You only had exposure during one meal and you avoid it on the hour. So you can see that the peak comes shortly after you had that exposure. It was a morning exposure and it peaks and then it goes down to pretty expected behavior with only one exposure, one hour voids. But really if you look at the three meal scenarios, meaning you were exposed for breakfast, lunch, and dinner, the same total mass per day of exposure. But now you're breaking it up into three different amounts. You can see that generally you don't see the peak but you also see most of your high concentrations in the afternoon and early evening hours you see almost nothing in the morning for each. This is three similar days really. And here's your average. So again, how are you gonna frame the question? Are you interested in the peaks? Are you interested in the average? But for our purposes time of day becomes critical and this can be explored with a model and this application supports the notion that sampling later in the day might give you the highest concentration and the best information. I'm gonna switch gears and finish up with looking at the whole thing from a forward base. I've been looking at it from a backward base, looking at NHANES and the fasting. What if I now look at exposure from a forward based approach and I'm using the American Chemistry Council Database and I guess it's a proprietary database but then they let me use it and of course if I go forward any further with it I would do it in consultation with ACC but I evaluated it in my eye as a very comprehensive compilation of data, phthalates, all of them in all media including food including what's out there for consumer products which is next to nothing and I'll make that point later. The databases on Excel spreadsheets and includes generation of average concentrations for every chemical, all the metabolites in the food and the urine. I mean it's biological as well as exposure media and every chemical and every matrix and what I did is use it to generate a forward estimates for DEHP and actually another one too. I might point out that it's the same approach used by Clark who provided information to this panel and in fact he used some of the same data that I used in this and I'll show one example just one brief example, DEHP and Poultry and this is a little snippet of their database which showed the actual parent compound, DEHP and Poultry. Each line represents like one study and one thing to point out here is that the page in LaCroix from Canada 1995 was heavily relied upon by Clark in his work and what they did is when there was a non-detect they highlighted it and gave a detection limit. They figured out averages assuming that the occurrence was at half the detection limit and they did straight averages with each sample having equal weight. It was not like an average per study and in the average of studies it was each sample has equal weight. I want to also point out that for example this fried chicken from Japan had the really highest here. This was about 15 versus everything two and less. So this fried chicken from Osaka, Japan really kind of drove the average for Poultry up. I'm going to point it out because obviously it becomes important in what I found. So when I've used that simple approach right to the bottom line, dust ingestion, these are contact rates. We get them from any number of different sources and here they are, I could use them continually in the grams of exposure of food per kilogram per day. Here are the concentrations as gleaned directly from the ACC database. You'd multiply one times the other, get the intake and the bottom line is 15.3. Well I wasn't too excited about that because as I said initially literature had put DEHP in the range of one to five and now I've used the database and I found that it got as high as 15. Well but also I pointed out to you that Japan really, in the case of Poultry skewed the data. So the question is do I want to just use the database as is or do I want to carefully select what I use from the database? So the question is do I develop rules of exclusion based on what I want the numbers to show or do I do it independent of what I want the numbers to show? And you will have to decide. But I kind of knew I didn't want to say 15 and after looking at it I really thought that those studies from the Far East were pretty high. So you can agree that I should have excluded them but I did, I only wanted to look at, let's call them more recent data from 1980 and also didn't really was not too thrilled about this Chemical Manufacturers Association 1986 study. It's not publicly available, it was not rated as high quality by ACC in their database. So I got rid of that, I got rid of the Far East studies and only looked at latter studies and redid the calculation, you can see 4.5 was now dropped all the way down to 0.9 when I got rid of this fried chicken from Japan. So I redid the numbers with that exclusion and they came out to more pleasant looking 4.2 intake for DEHP and you can just eyeball it and see that it's clearly dominated by food compared to these. There was some dust, ingestion, exposure but obviously food dominated. I did this for BBZP again, just another use of the database, I got a small number and if you go back to my first slide with the warmest data I put out it also identified exposure in that range for this chemical. You could say food dominates here, it's not quite sure. Really it's only this matrix here of dairy which it came through, a lot of these very little was found, I put zeros in when there was essentially very little data and there was all non-detect rather than put a number in here, I thought it was more sensible to use zero and anyway these are concentrations in other matrices. You could claim it's food dominated, that's a problem with doing this. I'm not sure, again the quality of the data, you almost have to look at it in a case by case basis but a problem with that is that there's still a big need for data and consumer products and when you actually went into the database I only found one reference with data provided for both of these chemicals and on consumer products and that was for a very limited set of consumer products from Japan. So it's hard to really do a forward based analysis if you're gonna rely on just air, dust, water because I'm sure and you all know you get exposure from the shampoos, from the deodorants, from other matrices. So that's gonna always be a limit for use of the ACC database but if you're doing this, what a great place to start. I mean it's probably worth doing in any case. I'm at the conclusion here, what are my next steps? I'm not really sure, I'm working with my two collaborators, Holger and Lisa, hopefully I'm gonna be trying to refine these analyses maybe looking at instead of, you know, based on what I heard this morning, maybe I should be looking at further breakouts of the in-hanes, not group at all, maybe looking at creatinine but maybe then male versus female or white versus black. See if I can tease out some of those co-founders or confounders and maybe come up with something that's statistically more rigorous than just looking at numbers but we'd like to try to take this further and see and of course any input from the panel obviously would be very appreciated. And of course I have to put this on, you know I don't speak for EP, I speak for myself even though they cleared my talk and said I could go here yada yada yada, here's my disclaimer. Okay, I actually have a couple of other slides, I don't know if I wanna go through them, I wanna stop, I did the same thing for DNBP just to point this out, the morning, afternoon and evening I was discouraged to find that DNBP has the same trend, suggesting once again maybe DNBP is food after all even though Holger's data and the FROM data suggested DNBP was not food dominated, if I look at it this way, I begin to see the same thing I saw earlier. So I'm not gonna go through these other ones, there we go. Thank you. So let's start with some specific questions and then we'll go to a general discussion and please guest speakers, you can weigh in as well. Thanks Matt, very interesting. I have a technical question relating to the ACC database which probably you may not be able to answer because it's not your issue, but I found it very interesting to note that there were analyses of the unmetabolized pound salad shown in that food. Now we have learned that contamination is a serious issue, how did they get around that? Right, well that's an obvious one, I mean you'd have to go back to, these was just a compilation of literature references and what the literature reference said. There was a judgment for each reference in there and they judged it on a scale of one to four so they provide their qualitative judgment on how good the study, how good the data was. I don't know whether or not they considered what you just pointed out in their judgment of one to four. The CMA data which they judged as four least valid versus one. The Pageant LaCroix was evaluated as a one, that's 1995, Canada. So did they, I don't know if someone's, I think Steve Risotto was here, maybe do you have a comment on that or, but I wouldn't even know how to know, how would you know? If you're looking at a 90s study in the literature and they're reporting it, you can surmise, maybe you know the individuals and know they have a sloppy lab, I don't know. It's not a problem of individual sloppiness or otherwise the problem, even the best and a little chemist will have to grapple with this problem because contamination is so ubiquitous, but I mean this is nevertheless the analysis of the ACC database is very suggestive. You know the exclusion you've done, very interesting and gives ballpark figure-wise. Seem to be there, for BBZP, I can do it for all of them that are in there and see if it correlates well with how others have done it differently, potential use of the database. I think it's similar to what Clark did, I didn't see his paper, but he did what I did. He used Page and LaCroix, I don't know if he went to dust and air and water and so on. Yes. Okay, thanks. I'm still tired from too much jet lag last night. I don't see how the urinary data at this point helps us with our task, which is dealing with the risks to consumer products. In fact, when you ended your talk, you said, no. And does that, what does that suggest to you that it has to be done first before you can truly delve into the issue of consumer products? Well, I think your mission is even more focused on children if I'm not mistaken as well. And of course, NHANES doesn't really provide a lot on that. I was in talking with Michael pre-meeting about information in products, children's toys. And I don't know what's out there. He informed me that there's been some out there and there's some new stuff out there. I know that you know as well that in our field, we figure out as best we can, the rate at which it gets released from the product and the rate at which it gets mouthed by the child and so on. So you can continue the forward-based approach. You need more data and you need, obviously cooperation from manufacturers or someone to get the data. When I looked in there and saw that the only thing was one study in Japan for a very small number of cosmetics, it was discouraging and it's not a news. It's just not, the data's not out there. I don't know how to, that's a good question. How do you actually find that? Because we can do our fancy models. We, you do them all the time I do them. We can probably guess what gets into the scalp from use on the hair. If we knew what the concentration was in the shampoo, we don't. So how are we going to do it? I was looking to see if you were reinforcing my frustration. Because my frustration has been for the last six months that I don't see where we have the emissions data or the topical data to help us determine whether or not there's any absorption through the skin, whether or not we really have updates. I think we could get started just on general stuff even without specific, we just don't have the product. But our goal is to deal with specific products. And I don't, as an academic exercise, I think you're saying is what I would say, we can do something. But beyond an academic exercise, I think what you're suggesting is we need more data that's very product specific. I would think so. To help us get where we want to go. I think that what's also being, another part of that equation would be some way to kind of validate any estimates that you make. And that's where maybe Holger's kind of work. Metabolism comes through where you can get at body half lives and excretion rates and maybe even take the initial penetration models we could devise and see what shows up in urine if we can incorporate metabolism information. So there's another step in there too. I think that step is there. But I think from what you were showing this morning is that with the dominance of food, diet, how we're in the profile of data, is this material or information that we can use pharmacokinetics to tease out the influence of toys or other things? Because it's so overwhelmed by this other influence. Well, one thing that I wanted to use Van Haines, and as far as I felt like I could get if I could get anywhere, is just the issue of diet versus non-diet. If I can get that far with a few of them, that's important. And what's your conclusion? I think we did. I think MEHP was, I think, supportive of a dietary when you add some of the other things. But if you looked at the other ones, there's not a whole lot in Haines. Some of the others, there was clearly nothing showing up in my visual. And I think statistics would bear it out. The BBZP, I believe you can look back at the slides, or a couple, it's just not there. So that's a good start. If I can use this fasting hour correctly done, incorporating male versus female and everything, you're supposed to incorporate in a multiple regression, if I can say one really looks more like dietary than these don't look like dietary, that's a good first step, wouldn't you think? That at least gives me a chance to figure out what bull park I'm in. And where to start, you're looking. Where to start further investigation? Right, but the further investigation is the key question that has to be put on the table for this committee, is how do we get to the point where we request the kind of information need to do the investigation properly beyond doing basically the 0 and 1. This one is clearly dietary. This one is not clearly dietary. And maybe there's one in the middle where you don't know. I think, again, I don't want to harp on products, but if you have that, just seeing your mind, if you had all the uses of BBZP, all the uses of the ones that you want, and you actually kind of know that it's in this brand of shampoo and something, and you have that product list in front of you, wouldn't that be an exciting start? That would be, that would be, I think, a very major start. And also, getting information which we can get about how much time people, how frequently we need to do it. I think, Jack and Boya, we do that stuff where we are. You guys do it too. I mean, that's what we're in the business of. And Versard has done this for years. So there are people who can ask that question and come up with an answer and a good start. The key is the emissions and the amount of material. And we're finding this, anytime we get to consumer products now, whether it's volatile organics or something like phthalates, the missing link is the emissions. How it gets released from the product. How it gets released, the bioavailability, the emissions, the, but even if you started with a basic guess, it's 5% per year or 2% per year. Let's start there. Is that same reasonable? We could take it the rest of the way. You know, coming up with basic. Or 2%, 1% per use on the head, you know, or something. If we can come up with some uncertainty surrounding the number, it would be easier. Okay, thank you. I can, I find myself in total agreement with you too. I too would like to express a deep concern and surprise about this lack of data in consumer product. And I'm, I'm struggling to find a convincing explanation for this because surely it is not rocket science to analyze these products. So why isn't it done or has no one done this? I agree. We being recorded, I don't know. Yeah. It's, it's a difficult question to answer because there's so many products and so many phthalates. I mean, the people who make the products know what's in them more or less. The, you know, we have some data. It's not zero. There have been people, FDA and others, for example, a few other publications that have looked at personal care products. We have a lot on the toys, other things. I mean, all I can say is we can do testing up to a point or our lab can do testing in a, within a certain, you know, in a reasonable time if the list is not too long and we, you know, we can, that's something we can do. We could do migration rates and that sort of thing. Well, I guess a following on that question is have a task, a charge, but the data that we need to truly address the risk has not been provided. And, you know, I went to the phthalates cumulative risk document from the NRC and it really nails the question to the wall by the only page in which exposure is mentioned specifically, page 22. And the bottom line is the relative contributions of exposure to the total bottle burden at various ages are unknown. And you know what? When you go back in the rest of the document, it's true. There's no data. There is not. It's true. NRC. Yeah. Well, I know. That's true. Good point. But the point, I know, I have my doubts about the NRC sometimes, although he's a wonderful man over here and does a great job when he chairs an NRC committee. But the point is, is that even with that, there are no references. And that's, it makes it hard to do a real risk assessment that can help us meaningfully decouple what the issues are we have to do in terms of banning phthalates from products. Alger, am I wrong or you've been at this a longer time than I? You are perfectly right. I remember when I started doing phthalates and it was the first presentation I really have been to. And a lot of people said, oh, we know how they walk. We know how they talk. And now 10 years later, we're still sitting here and we don't know how they walk and how they talk. We don't even know where they're in. That's an issue. Mike, is it possible to commission some analysis? I mean, it would greatly enhance what we can do. I mean, I think, yes, your point is well taken out so many phthalates. Your point is also well taken out so many consumer products, but I'm sure we can, we can come up within five minutes flat with a priority list. If we could come up with a list, we can take that to our lab and see what they can do. That's that we can do. The other thing we can do is to ask as much, get as much information as we can from the people who make the chemicals. They don't know what's in every single product, but they can tell us more than we already know about what phthalates go into what products. I don't even know the specific name of the product. If you can give me a range of values, I'll give us a range of values of the amount that's in generally, you know, shampoos. We don't need... And give us a range, that's fair because it doesn't deal with proprietary information. Right, right. Dealing with this, trying to, I think, what Matt is driving toward is if we're gonna do a risk assessment, that's the kind of information that we can have a risk assessor use in the analysis of the data, correct? We need generic information, fairly basic information. Couldn't you go backwards though? I mean, you were talking about backwards forward. So you have a lot of this urinary data and you said, for instance, the butylbenzol is likely not diet. So you can make assumptions that it's coming from products. It could be coming from his main issue up there was dust. Oh no, dust, going from a product into the dust. Do you wanna make that leap of faith? I'll tell you, the abyss is pretty big. Well, there's some data on levels of these satellites and dust. But you don't know what the source is. You have no clue. You do know a little bit about it. You know where, for instance, just picking on butylbenzol, you know where it's used in terms of different adhesives, different products. If you don't know where the dust was taken and you don't know how, again, it's just, Well, the dust, you what? I mean, the studies they've done in, It's a dynamic, you're looking into, it's just like looking into dynamics of pharmacokinetics in the body. The dynamics of the distribution of any product in a room that ends up in dust, just like something that ends up in urine, can have multiple sources. The child could have been touching a table. He could have been playing or she playing with his mother's shampoo, taking a bath. You have no clue to know whether or not that material. I agree. It's not all known, but there is some information about the levels in dust, the use of it in certain products, the percent of that product cosmetic may contain 0.1%, 1%, et cetera, of certain phthalates. But that's what's available. But I think what we're asking for is more. That's what we're asking, because going backwards is not going to buy for you, because the NHANES data is spot samples. Have no clues to where these data came from. How these people, how the... Well, there's other data from data sets, where, you know, smaller data sets where you may have better information. No, the NHANES, it wasn't collected for this purpose, so, yeah. I'm not deriding NHANES. What I'm trying to do is ensure that we don't use the wrong approach to solve a problem that can come back to haunt us if we don't... I know I'm an outside guest, but I did talk about problem formulation. Is the committee charged with conducting pathway risk assessment and quantifying the risk by substance by pathway? I think that's the answer, but I think that's a crux of this issue, because that drives how much data you need to make the decision, but I don't know that that's what you're charged with. Yeah, I think we all agree that in an ideal world, we would have all the information that we need and we could do the kind of thing that I think Paul really wants to do, but we live in the world where that is not the case and we have to deal with what we have available to us, and so it might be useful, and I think the committee has to discuss, whether if we were to take 25 different cosmetics and look at six or seven or eight phthalates, that that would materially increase our ability to complete our task. I'm not so convinced that that would, but I'm open to being convinced, but I think that's a discussion we need to have. Absent that, then I think we really are forced to deal with the data we have and that comes primarily from NHANES, where we look at exposure. I mean, we know how much is there. We don't know where it comes from, but in terms of our charge, I think we can still use that information with appropriate defaults and uncertainties and variables that we have to program in, either mathematically program in or provide textual information that deals with them, why we did X and not Y. That's the uncertainty is just too large. Thanks, you too, for reminding us. We really think, I think we really need to look at the program formulation stage. As far as I can see, there are elements in our charge where perhaps exposure information is needed, but there are also elements where exposure information isn't needed at all. Right. So I think we need to go through that and critically examine. Yeah, I mean, the charge is, the bottom line is to, we have three permanently prohibited phthalates, prohibited in children's products and three interim ones are, what we have to decide is whether to continue the interim ban on those three and whether any other phthalates or phthalate substitutes should be prohibited in these products. But, and then it's a long list of eight buts. Examine all the health effects, all the sources of exposure, all the compounds in isolation, in combination, so on and so forth. And to consider things like personal care, all products in so on. So what we've been struggling from the day I saw this is what exactly do we need to do and we have two approaches. I think we need to do both approaches, the top down and bottom up. The bottom up is the harder, I think, and we don't have the data to do it for all the exposure sources but I think we can do a pretty good job at least on the children's products. And the question is, is that enough or how much do we have to do, try to do bottom up for everything or do we focus on children's products? Because I mean what we have from NHANES and other kinds of data is total exposure. And I think what we're really what we're asking is what's the exposure from these children's products in combination with total exposure? I think if you look at the total exposure, you're looking at your benchmark to say whether or not the other exposures have any or significant enough to warrant further consideration for the permanent ban. But without the other part, you can't make that judgment because you don't know what products you're looking at. And so it's got to be both ways. You don't, it's not either or, it's how we use these information properly to help us define the problem in a way I guess that Tom would define it because I know how he thinks and being able to use it rationally to make a conclusion. I think that maybe the idea of having some four or five phthalates and some reasonable products that are used by children haven't tested, that can go a long way to filling in a major gap. I think we have to say that we're gonna focus on a smaller range of phthalates in possibly a smaller range of products. Otherwise, we'll be like the grad student who never graduated. I mean, and just to make things a little more interesting, everything's changing. Everybody's reformulating their products to take away the phthalates. So what was in pharmaceuticals or personal care products during the 0304 NHANES data collection is probably not there today. Can I just ask a question? We've had a lot of discussion about the the phthalates with some food products. So and I wasn't thinking of it as a yes or no as more of a percentage in the sense of can we look at, say, DEHP and say 70% exposure comes from food? I mean, is that possible or is it really just yes or no on these chemicals? Can we get pretty good estimates of percentage? Because in my mind, we're thinking about products, yes, but if it's, and I think that may be what Russ was talking about, if it's food, it's not a product. So at least that narrows the window down. I think you can get a, you can begin to get a handle on that question. And I think the work that Clark did, which is similar to the work that Wormuth did, which is similar to what I was trying to do at the database, take all the literature out there on food and non food and combine with contact rates. I mean, you could just look at the figure, the chart I generated and add up the non foods and add up the foods. And you can see that the foods were, you know, three out of 4.2 or whatever it came out to be. So there's your start. It's a start. If you have, through other means, you know, good feeling that food dominates and I think we do for DEHP, that would be the way. I mean, I was discouraged that BBZP, the other one I looked at, at least when I went through the math that I did, food still came out high because it showed up high in dairy in the data sources that were in the ACC database. Now, does that mean BBZP generally is dominated by food? It's much lower. It's 0.38, 0.4 versus four, it's much lower. But my simple math would have suggested food, but I don't know. I don't know. I have to go back and look at the dairy studies that drove, would drive that conclusion. Are those things published? Yeah, well, most everything in the ACC database, which is why I like it, is publicly available. Not everything, the CMA study was not publicly available, but most of it, I don't know, 80% to 90%, or more than 90% of everything is out in the literature. It's not all sorts of proprietary data. It's public data. Stop me if I'm speaking too much as a guest. So this is a dilemma with background risk and a susceptible population. And the question to ask as well, so if food is 85%, does that change the hazard of these toys? That's the point I was going to make, is that let's take this hypothetical. We find that phthalates are at a level that we find to be, that we would say that these just remain banned. And we further find that 85% of the exposure comes in food. Does that mean that based on our charge, which relates primarily to toys and consumer products, that we would now say, well, we're not going to recommend that they be banned because most of it's coming from food and we're interested in consumer. I don't think so. Right. So your charge is in considering and susceptible that changes the risk from what your charge is to look at, to be informed by that. But if toys are only 20%. The first thing you have to find out is whether they really are. And the only way to do that is find out from some studies whether or not that they are first leachable. That's the most important part. If they're not leachable, you can bang away at them for 20 years and they'll never be a problem. But if they're leachable, well, then you have a chance from looking at the titration of that material from the product to either the hair, to the mouth, in a child on how they use it. And that will give you an idea that this product may actually should have none in it. It has to do with the contact. And that contact has to be meaningful. And that's why I think these studies on a very limited number of phthalates, which we're considering and with a range of products that are meaningful for kids to use, put that other part of the equation together so that we don't feel like we're making an assumption about the result. Because I totally agree with Tom that even if diet is 85%, it could be that in a child who plays with toys in a way where they have a pica-like activity, well, that could be 75% of the exposure. So you have to be able to use that data properly to be able to define the uncertainties so that when we make our judgments, we're making it a meaningful way. Paul, I get back to my hypothetical. Let's say we looked at the three band and the three interim band phthalates. And we looked at 100 different consumer products. This is hypothetical. And we found that there were no phthalates. Those six did not exist in any of those 100. But yet we knew that children, pregnant women, et cetera, being exposed to levels that still were of concern. So it's coming from somewhere else. Are we, as a committee, going to say that we no longer think that the three that are banned should be and the three that are on the interim band should be removed from that? We're talking about consumer products. We're talking about very specific children's products. If they're not in the products, then we don't have to worry about it because they've already gone. If they are in the products and there are meaningful proportion of the risk, then exposure requires contact. If there's no contact, there's no exposure, you move on to the next product where, in fact, it exists. Because what you don't want to spend your time is batting at gnats for something that's already been removed. But clearly in terms of our charge and the time frame in which we have to complete it, there is just no possibility that we're going to get the kind of information you're asking for. I'm not sure I agree with that. I mean, Mike, can you do some short-term studies on a fairly sharply focused number of products? We've got data on children's products. We know what's in them. We did migration rates and wipe tests. They migrate out of PVC, so there's exposure. We also have data on children's mouth and behavior. So we know what they put in their mouth, how often, how long, what we do. So we have a fair amount of data on children's products. We could get more if that's what we need. We have to see the data on children's products because maybe it makes this discussion academic. Because if we have the data, we can review the data. Well, the reality is that the things, there's an interim prohibition. So to a first approximation, they aren't there. Can I strongly support what Philip said a couple of minutes ago, this line of argumentation? I do have my doubts that if it comes to deciding about ban or not, whether really exposure data from these toys or any other articles are that decisive, because let me amplify this thought. If we decide all of a sudden to relax or to not support an intermediate ban hypothetically of these three thalites that are named, I would bet a lot that what we will do then, in effect, is trigger a reverse substitution movement. All of a sudden, these thalites will again appear in children's toys. Is this what we want? I mean, we have to be sure about the consequences. Well, yeah, I mean, I think that's why we have to maybe do the hypothetical case. We did this before with the last chap and said, well, of course, while we're deliberating, the manufacturers are reformulating. And we did the risk based on the way things are today. And then we said, well, what if we turn back the clock to where they all have DINP, and then this is what the risk would be? And I think that's probably something like what we need to do now. I just pressed this button. It's me. But just to conclude this train of thought, therefore, it is for this decision, in my opinion, pretty immaterial, whether the stuff is currently in toys or not. There may well be toys now where you don't find the name thalites anymore in there. And that's good. So for deciding on ban or no ban, we don't need really this information. The information is needed for other aspects in the charge, but probably not for deliberating on ban or no ban. To answer Rick's question, the permanent ban, permanent prohibition applies to up to 12 years of age. The interim ban is 0 to 3. Or no, it's toys that can be mouthed is what it is, which is essentially 0 to 3, because that's when the mouthing activity is the highest. Where the permanent ban is now? Well, they define a children's toy as something made for a child up to 12 years old. And then for the child care articles are defined as something for children up to three years. And then the interim ban applies to the child care articles, which are 0 to 3, or toys that can be mouthed, which ironically most of the toys that can fit in a child's mouth are for older kids, somewhat paradoxically. I think everybody would agree that more data would be helpful. I think the question is whether more data are necessary in the context of the charge to us. I've been involved in a number of product surveys, and they've never been quick, and they've never been clear when you get done. And for us to assume that we could quickly get some answers that would allow us to go forward, I think there's a fairly high probability that that just wouldn't happen. We would still deliberate. Now that we have data, what's the second generation of information that we would need to clarify the first one? And it's not uncommon. Data callings from manufacturers is not something that happens quickly. That takes a lot of time to get that, as the relevant data are sorted out. If we said that we wanted to test 10 products, it would end up being 20 or 30. If we want to do it in one lab, it would end up being analyzed in two labs. We would have conferences to discuss the data, and we would have a second generation of testing because some of the results were unclear. I have my doubts that we could collect this information in the context of the timeframe that we're charged with, unless that can be significantly delayed to allow more data to be collected. If we collect, if we agree on certain data that we want to collect, it's common afterwards to wish that you had collected more or a slightly different data or under a different protocol. A compromise, possibly, rather than do nothing, or rather than believe that we can do a analytical survey in the middle of this CHAP function, the possibility is that we would proceed as we need to, as with what we have in front of us, clearly articulating what isn't there. And if we want to lay out an example survey of the nature of the products that need to be surveyed in order to do this, and the phthalates that we would want, and phthalates substitutes that we would want to have data on to be able to do this right, that could be part of this report, that we simply articulate in greater detail than the usual statement that more research needs to be done. And we would follow up on what Tom is saying is to be more vocal on what actually the need is and what questions are we trying to answer by suggesting that another survey be done as opposed to simply say that we need to have data on a wide range of consumer products to which children might be exposed. There's a lot that has to be done beyond that in order to do this right. The only way I'd augment that is that whatever data is available from CPSC now, from studies they've done, it'll be worthwhile for us to review because at a minimum that allows us to have a baseline from which to say we don't want a backslide from what the current situation is. So anything they have available, noting that we're not looking to reintroduce it but use it as a baseline from which we can go forward to say that the reasons why we don't want a backslide is reasonable, I think, set of data to conclude at the present time. Totally, Paul. And I think maybe we've talked about this earlier in these meetings and outside of these meetings, but I think it's time to get some of your CPSC folks to come and talk to us about what data they have and see how it would fit into this picture to help us. I mean, I can talk about what we have and we can do it this afternoon if you like. Do you want to? Any other comments? If not, we'll, yeah, Bern? Just to thank our three speakers. Obviously you've stimulated a lot of thought and discussion that we really appreciate your coming in and helping us. Thanks for having me. The biggest fear of any NAS chair is that nobody will ever read or consider things. And I've chaired committees where I think that happened, but that's certainly in the case for this. And I think your approach is very consistent with the kind of things we were trying to urge the agencies to consider in answering the question. So thanks for having me. I just want to again say thank you to the speakers and one thing that amazes me is how willing so many people are to help us. And we appreciate it very much. Break for lunch.