 And I will begin making a brief comment. We've been talking about implementation in terms of pharmacogenomics. I just want to mention that for the last couple of years, the NIH Clinical Center has been offering HLA genotyping when abacavir, carbamazepine, anaphanitoin, and allopyrinol are prescribed. By practitioners in the NIH Clinical Center, my colleague Dr. Barry Goldspiel, who is acting rather deputy director of pharmacy, co-chairs of pharmacogenomics staff committee with me, and Dr. Bill Flego in the Department of Transfusion Medicine is in charge of the actual HLA genotyping implementation. So I just wanted to mention that the Clinical Center is in fact moving in that direction. It is required for abacavir. Otherwise, it is a recommendation and clinicians may decide on an individual patient basis whether or not to order a test at any given time. So with that, again, this is open for general discussion. If I could just make a comment, building a bit on what Steve described, it seems as though, Steve, you have a system that, you know, you've got data collection forms, you've got the software. It shouldn't be all that difficult to transport that to other systems. And so recognizing that CERNR is not the most penetrant nationwide, but still it's one that's fairly standard. So recognizing that that's not your mandate to do, how could we facilitate that happening if that would be a useful thing to happen in other systems? Well, I think we would be perfectly willing to share the forms. I mean, the institution paid for the development of the system through a contract with CERNR. So it's the institutions. I mean, the whole goal was to come up with a system that would allow, if there were to be a cutaneous adverse drug reaction network, I mean, I was only thinking in the context of pediatrics because that's really what our focus is. But if there were ever funding for a network, we wanted to be able to participate in the network by having everything in place. And we would be more than willing to share, you know, whatever we can that would facilitate this occurring at more institutions wherever. So what does it take? I guess it takes probably a medical IT person at a given hospital or system to tweak it or engineering. I don't know. But it was our medical informatics team working with the CERNR employees who walk around with Children's Mercy Hospital ID badges on to make it happen. And that's the extent of my knowledge. Yeah, Mark Williams, Geisinger, unfortunately living in this world. So, you know, these things are conceptually simple, but actually realizing them is incredibly difficult because what usually happens is that these do get built as one-offs. And so they will work in CERNR, but they won't transport necessarily even to other CERNRs. And so I think the strategy would be to take advantage of, I think, what is a little bit of momentum that's gathering as a result of the GM7 meeting and the IOM EHR action collaborative meeting where people are saying we really have to solve this generalizability issue. And we have to use some of the standards that are available that are going to be, all certified EHRs are going to have like a fast healthcare, interoperable resources and those sorts of things. So I think that as we develop a strategy assuming that we want to move this forward that this is, you know, creating a network, creating standardized data entry and those sorts of things, engaging with the informatics community to build the resource such that it would be, at least in theory, interoperable with any certified electronic health record, not integrated within it, but sitting on top of it and able to communicate with it would be a desirable approach. But it's difficult to do that without sort of a national perspective because at the individual level it just doesn't move in a generalized way. So it's very challenging. Although as we've talked about it's a big country and trying I think to do this on a national level, I mean, eventually it could happen and yet you'd like to take advantage of sort of willing partners early on and if there were, you know, at least a handful of hospitals that wanted to partner and then those could grow and those could grow broader and maybe Josh, you wanted to comment on that? Yeah, I was just thinking that, you know, I think it is great to have these interoperable standards and I think we need to continue to shoot for them and, you know, certain, there's been some great demonstration projects like SMART and things like that that could run across which are these apps you can put across different SMART-enabled EHRs. But, you know, optimally to be effective it's got to be integrated within the EHR in a way that sort of fits all the standard workflows people are used to and I do think as long as you know you have standard outputs and think about what those outputs look like then the work you've done in CERNR to design a form that works well and the data elements you want to collect could be properly mapped to something that looks similar and epic. I know we could probably do something similar for instance in our EHR and of course ours is homegrown so we have the ability to sort of do, you know, kind of whatever we want. But I think that those elements, if you standardize the outputs you could cross them across different systems and Mark is completely right that, you know, what works in one CERNR doesn't necessarily work in another CERNR, same with Epic, etc. But it gives you a great head start. It's a lot easier to adapt code that's in one CERNR to another CERNR application so I would think that model is a good step forward. I mean the thing that I would, I completely agree with what Josh has said but I think what you really want to do is you want to have an eye on ultimately where you want to be so that you don't create, you can create something that works with a small number of partners but if you create it in such a way that it only works with that small number of partners and doesn't have a vision for extensibility, that's not a good outcome either. There's a lot of resources that ultimately won't get you where you want to go. So I think starting with a smaller group across multiple EHR systems and demonstrating that it can work will then lower the energy barrier to extend it more broadly if that's the direction that things might go. We had talked yesterday at our break-off and we'll hear about this about whether or not, you know, even something within groups that are participating in the Sentinel program could potentially utilize something like this, which again would have a lot more links behind it if we could pull something like that off. Well, and maybe if I could just make one quick comment in response to that, you know, what we seem to be talking about is not, at least doesn't sound like research and so, you know, for NIH this would be a kind of a tall mountain to climb. On the other hand, you made the very good point that then it provides the infrastructure on which you can do research. So then we'd be interested. So is it possible to sort of leverage this with the variety of patient safety initiatives that are ongoing? And I'll turn to our FDA friends to help us to know who the people are to engage. I would think AHRQ to some degree, our Agency for Healthcare Research and Quality, and potentially the CDC. But I don't know. So I don't have Mark or others if you want to comment. So I don't have a very specific answer, but there are a lot of stakeholders in the government who are interested in this. And actually, the Deputy Secretary of Health had recently convened a report, which I sat on as a steering committee member, to look at various aspects around three areas of safety events and how to leverage informatic systems and so on to improve the both the research aspect but also the patient management aspect. And they brought in some IT specialists. AHRQ was one of the groups that was involved all right across the whole government, actually. So we can find out who are the people to go to. But I think that the vision here is to have a system in place, informatic system in place, which you can use both for patient management, so that is early recognition of events and interventions, the research aspect, and then the public health aspect. So those are the three-cornered hat. And if you really do it right over time, you build a network so that it becomes basically for these rare events especially, you know, sort of, you know, encompasses hopefully the whole health care system or as much as possible. The other aspect, and Dr. Hoopnagel is going to talk about this, is that besides the data elements which are important to accumulate, you really do need a clinical brain to look at individual cases. You need this kind of causality analysis, which also has its own methodology so that when patients are being seen at the bedside by the clinicians, they're actually calling what those cases actually are. And there needs to be a system in place where in addition to the data elements, there's a kind of smart narrative which has its own consistency across the system where their narrative synthesizes what the patient actually had and what the expert brain thought, what the diagnosis was. So, and that FDA has a lot of experience with this as an issue is not only to get information bits but to actually get the differential diagnosis, the causality analysis, the phenotype, and so on. So you still need that as part of the collection and documentation. And this will be especially important for genomics where when you start doing whatever your case control study designs are, you won't have confidence that the cases that you accrue to do genomic analyses on that are bona fide cases, that they're not something else because otherwise you won't have a good research plan for it. I just want to say other group to talk to be ONC. Other questions or comments? Well, a comment that one problem is we can come to diffuse and get all ADRs when this meeting is focused on a very serious one that does call for research type of investment. One thing about these forms that we're very involved with is they can be used to calculate causality scores, severity scores to put these things in context including you can include a what we call a completeness score which basically you have enough data to decide are the cases too confusing to consider because when you take all ADRs, a lot of them aren't as clear cut as the ones you get presented at these meetings. They're a mess. The patients, we have patients who are taking 25 drugs. So not much you can do with that. So these types of scores can be built into these systems and you can give an output to the person who puts it in. And it can even include a recommendation like do HLA typing or something like that. But we have been unsuccessful in getting people to actually submit things to us through these types of computerized systems. That does raise a question. What a challenge for us is polypharmacy and on average patients are leaving the hospital with eight different medications. So it really makes the association very complicated and I think there's some areas here for research and how to approach that problem. So maybe to change the topic a bit, Cynthia, you commented or when you presented that educating your clinicians was a key part of the effectiveness of your initiative and that the way you did that was through a letter to them, pre and post. And I just can't imagine that our physicians would pay attention to a single letter. So there was probably more than that. Yes, that's right. I mean it was really that the director of medical services said this is now standard of care. That's what made the difference. In 2009 we came out with a newsletter which summarized what was going on in the research, validating reporting the 1502 and there was no change at all in the carbamazepine SJS. So it was really convincing at the ministry of health level and telling people this is now your basically like your clinical practice guideline. And I think we've been feeling more and more that it's really getting these kind of recommendations into clinical practice guidelines that all is really the way to go. No, they don't listen to they don't read drug labels. And in terms of clinical practice guidelines, we're fortunate I think to have the clinical pharmacogenetics implementation consortium or CPIC that is developing guidelines, but they're not a professional society. And so what you'd like is a professional society to say, you know, this is this is our standard of care. And we heard yesterday how one of the challenges with carbamazepine is that there are multiple specialties that prescribe it. So you don't have a single specialty, but is there some way to to unify those or approach at least the major ones? I mean, I think you identified, you know, several that are likely to be prescribing this and then maybe kind of target them that way. Well, yes, one of the HSA staff, you know, presented at the Singapore Epilepsy Society. So like even here today, there's no neurologist right in this meeting. So unless I'm wrong. Yeah, well, you know, and not around the table, we do have have neurology colleagues from our neurology institute. Oh, okay. Yeah. But I think we need to reach out to the disciplines which are actually prescribing the drugs. So that's where they'll influence the decision whether or not to do the test. So Taiwan, one of the one of the initial interventions that helped was actually targeting off label use of carbamazepine is off label use a problem in Singapore, because I guess that also speaks to the issue of the cohesiveness of the providers you'd be speaking to in terms of how to prescribe carbamazepine in the constituency groups that would be involved. Yeah, I'm not familiar enough. I do know it's used by dentists, a fair amount, but you know, for tried, but I think that's in the label, right? Neuralgia. Yeah. So in as far as inappropriate uses came up yesterday with allopurinol, that a lot of the cases we see are just being used for asymptomatic hyperuricemia, and that is not in the indication. But uric acid is standardly measured in these health screens. And then when it's just a little bit bumped, I think people start getting, taking allopurinol. And I think if we eliminated that population, we could get, even without genotype, we could get some impact on the cases. So I've just one comment, Mr. Leader. I liked your presentation a lot. And I think one challenge is definitely getting as many patients into these documentation systems. But another is harmonizing the system so that we're using all the same and guaranteeing the quality of the data. And your comment on rash, which is for dermatologists, really a problem there. And getting that sorted out. And your comment on phenotyping, good phenotyping is really something that I think is important because those like Munir and those who are doing genotyping experiments afterwards, I mean, if they want to clean up the results, they have to be sure that the phenotypes are correct in this system. And that's a challenge. I should just mention that Dr. French has joined us for our second day. We very much appreciate your coming from, from Geneva. So welcome. Maybe two or three random thoughts. In terms of support. Can you step closer? Oh, sorry. In terms of support for these kinds of projects, national registries, and work thereof, remember PCORI, the Patient-Centered Outcome Research Institute. This could be a really interesting project for them. And then thinking about patient-centered kinds of programs. I think yesterday, you know, was mentioned the idea of social media and engaging patients through programs like Patients Like Me and other social media programs. You've already heard how important this is to actual patients and patients' families. And homegrown grassroots registries could be potentially much more powerful than top-down approaches as much as we think we know about this problem. The last comment, a little bit different from the first two, you know, we've heard a lot about over the course of the last day, you know, what's the offending drug? There's polypharmacy and chart reviews that take hours and hours to try to figure all of this out. And then I've also heard, I think Teri mentioned yesterday, that maybe there's this common pathway that we should be thinking much, much more about. And with, again, sequencing of HLA, which, again, is possible and will become better and better, maybe we don't need to worry so much about what the offending agent is. We should be thinking about HLA, essentially, as the main offender, genetically. And as many of you know, there are various kinds of statistical tests, burden tests, aggregate tests that might be very useful in essentially determining overall, what is the role of fine variation in HLA with regard to Steven Johnson's syndrome, really irrespective of the offending agent. Thank you. Mark, you have another? Yes, this is just in response to Teri's questions about guidelines. One clarification and then a comment. The CPIC guidelines presume that you have the genomic information in hand and provide guidelines about what to do with the information. They don't actually provide guidelines about who to test and who not to test. And so that is of some relevance in a situation where we're really talking about should we test or not test. The second thing is, again, in my role as wet blanket over the last two days, I would just once again point out that at least in the United States, there's been a fair amount of scholarship that shows that creating a clinical guideline does not, in fact, rapidly change practice, unfortunately. Some of it is because the guidelines are frequently constructed in such a way that they're not adequately explicit. They would say things like should consider or make, you know, what we refer to in the informatics world as weasel words. And so it really gives clinicians a lot of latitude. But again, the role the role the guidelines can play is that you can then use those to develop clinical decision support that can really work well. And the last point I would just make is one more lever to push, button to push, particularly related to severe adverse drug events is the liability point particularly in the United States, although it sounds like this is an issue in other countries as well. Hospital leaders don't like to hear about the fact that, hey, we're exposing ourselves to liability because we're not doing something to prevent these sorts of things. And so if you're like me, where you use any tool in the toolbox to get what you want, you know, this is just another weapon to add. It's nice here. I just want to add that the professional guideline is really important, I think because when you, I mean, the American Society of Professional, like American Society of Rheumatology, when they update the Gaul Treatment Guidelines to suggest actually be 50 or one testing in Asians or in Thai, it was published. And then you have a PDF downloadable all over the world and our clinicians read it. So when they talk with the head of the rheumatologist in Thailand, they concern about that recommendation because it was published in your guideline. Yes. So, and also the FDA recommendation here is also used in the other part of the world. So if you have a recommendation update in the FDA or the professional guideline, I think it has a worldwide effect. Yeah, but I would say that we struggle this because like for a back of year, it's almost absent. 5701 is almost absent. So it's not cost effective at all in Asia. So just follow, you know, just follow what is coming out of the US is not always the right thing. And we're struggling with allopurinol because of the limited alternative medications. If you are 5801, there really aren't a lot of other choices. So I think we're thinking more along the lines of increased safety monitoring rather than changing the drug. So I am a little bit curious though, in terms of the 5701 that is not at all prevalent in Singapore, do you have a back of your hypersensitivity? Very little, but we also have very little HIV because for anyone who's coming to the country to work, you have to take an HIV test and you don't get your entry pass if you're positive. But we have a back of your testing in Thailand and this is a standard recommendation and probably is because I mean, if there's no guideline recommend to test is before used. So we will conclude then this general discussion and as Terry mentioned earlier, there will be a group picture taken, the group photograph. Thank you very much. So before you go, thank you very much Juan. And before we go to break then we're supposed to gather in that back corner there and the quick more quickly we can do that and effectively the easier we can get on to break. So please, please do that. Thanks.