 Okay. First of all, I'm going to say that I expect us to be having the exact same conversation in 15 minutes when I'm done as we've been having for the last half an hour. So whether or not I contribute anything to this is yet to be determined, but I think that we've all been hitting all afternoon on the issues that we have at hand and the difficult choices that are going to have to be made in a collection of interim and longer term solutions to get to our scientific ideals and our policy ideals as well. So I was just tasked with policy and ELSI broadly. I'm not sure exactly what that means other than a large bin to go through. So I am going to begin by my very simplified view of the scientific aims because I think whenever we're talking about policy or ethics issues that that's the scientific goals are where we need to start and the design that we put into the science has huge implications for what our policy choices then need to be. And I think that that again has already come out this afternoon as we've talked about what are the priorities that we're going to have. So again in simplified terms we want to build data resources with sufficient power and flexibility to ask really big questions and find really small answers. And what I mean by that are the smaller genetic contributors or the rare variants where we're not going to see really big effects. We also want to have high powered statistical analysis capability as well as biologically relevant science so that all different scientists from different sectors can use the data and ask the questions of the data that they would like to. And we want to work across disease disciplines using multiple data types and common structures in terms of platforms or phenotype measures so that we can really again maximize the amount of data that we're able to use in our analysis. So we set out a pretty tall charge in terms of what we're trying to get to and there have to be compromises made along the way I think to be able to just just for practicality purposes and that can be because of policy issues but it can also just be from technical issues that I think we've been working through and so it's going to be an iterative process to get to where we want to go. So I'm also going to as you'll see as we go through this I'm going to put out some basic context to think about this and I'm going to ask a lot more questions and I'm going to answer and I'm not going to walk through the individual models in particular with with one exception because I knew that UN was going to do that and the papers are there and a lot of the work went into those papers and laid out some of the specific issues. So to talk first about systems of oversight and just to acknowledge this is a proxy for ethics. Our oversight systems are not ethical concerns and there are different questions and different ways of thinking about this but the oversight systems are sort of how we embody the ethical principles that we have developed through research and other societal conversations and this again is it is a systems and a relational process. So there are oversight and policies that are made at the level of the project as we've heard some projects and programs are releasing different p values others are not and so there are these these make big differences into what happens with the science even down at the individual program level then there are local oversight issues in terms of the institutions that you belong to the states that you live in in the case of the US where our state systems are all over the place in many cases and then finally there can be high level policies and regulations which come to bear then and can trickle down to some of these other issues but they're going to be at the more principled or 30,000 foot level. There are also then this an interaction with systems of trust that we have to to think about and that have an effect on how we answer the questions that we've all been asking this afternoon and here we have again a relational system where we have the oversight and the policy systems we have researchers who are trying to accomplish particular goals and we have the public who were asking to participate and to contribute their samples and their information to this and we require their trust in order to do this and part of that then speaks directly to their willingness to give us this information and to be a part of this and what are they willing to sacrifice in terms of their privacy or in terms of of their autonomy as we'll talk about going forward through this and all of this again is what we're trying to where we're trying to accomplish the research that we want to move through this space and so we have to think about all of these things together and and how do we fit them together how do we make different choices in different ways to accomplish different aims things to think about again always the science what are the priority aims what what is it that we most want to accomplish and we could have a different answer for for some of those different goals from a policy perspective what are the questions that we need to think about for any model whether it's the four models we're talking about today or some various iteration of them they are I think and this is not a comprehensive list but the ones that we talked about largely in the papers or that I was thinking about have to deal with participant autonomy and choice and how their data are used which comes down to questions around informed consent and what we've all you know talked about broad consent narrow consent best practices for consent consistent consent all kinds of issues in different ways to try and speak to that issue of the conversation between the researcher and the participant and the understanding that both have and how the research will be conducted at the time and into the future participant privacy interests what are they what does privacy mean and I'll explore that a little bit further the potential for recontact in terms of going back to participants and again this could be for additional analysis as we've talked about earlier or for just return of results which has to do again with the participant interest in coming into the studies at all we also agreed pretty early on that this is really complex and we you know a longer term situation and so something that we want to keep in mind because this is important for science but that we're not necessarily going to solve in the next day and a half and intellectual property issues this is a little bit more mundane but it is fundamental to what we're trying to do in terms of how the data are used how different sectors of the research community use the data and how they remain available for others to use in the future other questions what's possible within existing frameworks and what will require new development we've been again wrestling with that all afternoon and something to that I think is important to think about from the context of the way I look at things every day is is how can we scale this and how do we track it anything that we propose it's easy to come up with a different option to do something but there are technical limitations and there are policy limitations and all of that has to do with the fact that we are based in this systems model where there are relationships and shared responsibilities and we have to think about how we're going to follow those through in a manner that we think is credible that's beneficial to the research process and that can maintain public trust so I'm going to now throw in just a few sort of sidelines that are not directly relevant to the things that we're talking about today but they're intrinsic to what we're doing and and represent the variables that we have to contend with as we're trying to craft policies around these different issues and one that has come up and we've been talking about recently already is the identifiability issue there have been papers beginning with the Homer paper two recent papers from Eric Schott and Nancy Cox's lab again showing with the power of statistics what we can do to resolve unique patterns within these data and that is going to continue to happen and it's something that we have to contend with because of the policy and regulatory implications around identifiability and around what individuals think about this and understand about this in terms of what does it mean to be identified by your genomic sequence and that is also in and of itself a changing question probably with time or has different answers we also have in the U.S. at least our regulatory system with an advanced notice for a new proposed rule which would completely shift how we manage and oversee human subjects research and the protections provided to human subjects in our country and there are some particular parameters that are on point for this discussion naming and again taking a position on the concept of identifiability naming genetic samples as considering them as identifiable which is a big change for our legal system but trying to then balance that acknowledgement with saying well the risk is just informational and then proposing that we can deal with that informational risk through data security and so again it's trying to reach a way to mediate the risk from the identifiability through particular situations but there are questions about scalability and applicability. The other trade-off in that is that under the proposed new rule then consent would be required for everything to go forward which again represents a very different model from how samples and data sets have been put together to date and so could have enormous consequences for existing data sets and what we could analyze going forward versus having to get new consent or re-consent for everything else. So this is something that's out there it's in a common period with an unknown immersion date to come forward so we're just waiting to see but anything that we decide here will be affected by how the the rule comes forward. We also have a presidential bioethics commission which has decided to take on this interesting topic of large-scale human sequence information and what it might mean for privacy and data access. They also had a request for information and I'm not going to go through this but just want to point out the amount of text that they put out in questions that they're asking the country and they're going to have a report in the fall making statements from this high level commission about the ethics involved in these decisions and again you can just see from the number of red lines they're looking at privacy, balancing risks, what stakeholders think whether they're participants or researchers, health IT and other data security measures it's completely unclear what they will say or how their how deep their recommendations will go and then again another question what affect those recommendations will have regardless of what they may say. So we just have to wait and see but they could change the nature of our conversation and finally then of course we need to keep in mind the participants again and what is it that they're seeing in the midst of all of the science that we're talking about and the headlines that are coming out about whole genome sequencing, the number of individual genomes that are being put out there, the capacity for genomic medicine, etc. and researcher conduct and right now I'm really going to just focus on some of this. We have two issues from the last few years which are well known around the Havasupai tribe and the Henrietta Lacks story that really speak to this role of what the participant autonomy is and what choices do they have about how data are used in the future which again is highly relevant for us as we're thinking about building sort of immortal resources where we have data and the data are there and we can use them for any questions into the future theoretically. We also have within the last two years our president and secretary of state having to apologize for some horrid behavior of researchers and how they treated participants in the respect that they showed or did not show for the participants in a particular study and again that made headlines and newborn screening and the use of newborn blood spots and again how research the research system has chosen to use the newborn blood spots and what participants but what the general public knew or did not know and understood or did not understand about how their their blood spots of their children were going to be used and so this is all in the news and this all goes into the context of how we maintain trust and how we're going to need to make policy decisions because it's it's not always logical there's a lot of public relations issues in it and a lot of emotions that can come to bear. So again these are the four models we're going to come back and and look at for today the open data access, streamline control, certification of researchers and a research comments and then the central analysis and I'm not going to go through them individually in great detail with with one exception as I said but just talk about them across the board. I'd rather focus on these issues for consideration that we want to think about and so from the ethics perspective and so autonomy and consent I think are fundamental to what we're trying to do and so again questions that we ask what are the participants willing to let their data be used for how much of a choice do they have in that and that's a changing paradigm from what we used to think the choice should be that they have and what we're thinking today the choices should be that they have and related to that is what are the limits of informed consent both from the the standpoint of broad consent and is it feasible to ask for broad consent what does broad consent mean what do participants understand about that and also from the perspective of can we use informed consent as the vehicle to solve all problems if we put it in the informed consent is that enough and we can go on I think there are a lot of questions right now over whether or not we are really pushing that that one mechanism too hard and expecting too much from it and then the realities of the research paradigm and I mean this from the perspective of we don't know what we're going to want to do in 10 years and then we're dealing now with the fact that we didn't know five years ago what we were going to want to do even though we thought you know we saw lots of exciting possibilities but we're now living with some of those consequences in just a short time and also another reality we have to contend with again is the relatively distributed nature of investigators in the way research is done and they have relationships with their own institutions they're bound by different state laws different country laws and all of this has to come together if we look at the at the consent issues across the different models there again questions that I've asked they won't necessarily go through all of these they're asked and in some cases raised with a lot more discussion in the documents but again with the open access I just heard just before coming up here Brad expressing doubt as to whether or not open access is ever okay because we just don't really understand what we're asking participants to assume again even if they've consented fully to what's going on with streamlined controlled that the controlled access model that's really built around consent completely everything at least in the NIH system is around consent groups and it's structured to abide by data use limitations which stem from consent so it is looking for efficiencies and incremental improvements that we can make to just make that process better but I think consent here is really attempted to be the the the ultimate principle that that everything else flows from and so there are questions about again brought an open consent what governance is needed to assure conformance or compliance with the consent in the data use limitations from the research comments perspective you know who would certify these researchers to abide by the data use limitations that's a big question there was a comment before that it would be sort of the ultimate compliance mechanism because you could just remove their ability to access any of the data but I think that's an open question from a policy perspective of can you do that and and what kind of enforcement do you have based on who would do the certification and so that's something to think about and of course what you know we'd want to see some reciprocity if we're going to go in that route to make it easier to work between databases in the U.S. databases in Europe and other places around the globe from the central perspective Pearl in the document and I think she'll talk later we'll raise some questions about just the technical issues with that and and if you still have data use limitations how are you actually going to to manipulate the data in a way that can respect those when you're pooling them all together so privacy you know the question we've come back to for many years and it's not getting any easier though the possibility to resolve unique patterns is getting easier is what does it mean if you do identify a unique pattern what exactly are the risks the open access model just assumes that there is no privacy and it accepts it through the informed consent process and moves on but there's still questions then again what can you really articulate about this risk and can you quantify it we know that there can be certain things and certain harms assumed just from being identified as a participant in a research study that may be a harm to some people it may not be a harm to others and so it is hard to articulate again on a across the board or global way all three of the other models I think attempt to manage the privacy risk in different ways but again it comes down to a question of can you really manage the risk and what is the risk what is the context of privacy these days with all of the other information that's out there about us and so I think this is a place where we do need to really start having the longer term discussion about the trade-offs between privacy and advancing research and what that means for an individual versus what it means for society and where are different groups in assuming the risk of this privacy risk again or potential harm our participants willing to move on and acknowledge that to go forward and if some are our others recontact of participants and intellectual property I don't I think that they are important but again I don't think that they are things that we will deal with deeply at this meeting so we can acknowledge that being able to recontact either for return of results or to be able to do additional studies is an ideal but there will be policy work required and so what we need to do in the interim is do research scholarly research quantitative qualitative normative on how to do this and we have some programs at NHGRI underway on this and also we need to collect the experience as Eric mentioned this is happening now some studies are doing this and we need to learn from it so that we can integrate it and move forward in an iterative way intellectual property principles they vary across research sectors but I do think that there is a cultural baseline coming forward that some of these basic data are not appropriate and some willingness not to try to patent everything that comes out of these first line studies but again we have questions about enforcement and what could the implications be for those that don't accept that and what are we as a research community willing to risk by putting the data out there and taking our chances governance again coming back to public trust and accountability for researchers for the research system for funders for everyone involved in the system intrinsic to a lot of the models that we have is a promise of conduct we will abide by the data use limitations we will you know only use information in certain ways we will put in place certain data security standards but to maintain that there has to be a reasonable expectation from the public's perspective of transparency and compliance and consequence if something is not followed this as well as other technical issues bring to bear questions about scale and practicability which comes to bear and in our thinking and in our decisions about the different choices and decision points that we will have to make over the next few days to reach our scientific ideals and so that's going to be I guess where I want to stop with the theoretical and I thought it might be helpful if I talked about the controlled access model particularly the one that I work most closely with at NIH and talk about some of the things that we are trying to do to improve the system that speak to some of the issues that were raised in the the white papers on this point so first of all one of the things that that we're talking about is trying to build a standard lexicon for data use limitations so that investigators and data access committees understand what given consent groups mean and decisions can be made more efficiently and more consistently and investigators can understand what they're asking for so hopefully make better choices when they're selecting data sets to request we're also looking at a way to build a centralized data access committee so that those that request multiple data sets can have one data access committee review process which should shorten the time to make this go forward similarly for aggregate data and for those who aren't the deep data users but want to look at the higher level information we're trying to design and think about the again trade-offs and different balancing acts that we need to do to simplify or create another more straightforward way to to access all of the aggregate data aggregate data perhaps through one simple request rather than multiple requests which is how it needs to be done now and then also as has been mentioned here is as being an ideal is trying to build in filters within db gaps so that you can sort the data that you're looking for by particular variant or by some of the other factors study type data use limitations some of the other basic things that would make it faster and more straight and straightforward for investigators to identify the right types of data for the questions that they want to address and for just at the end just coming back to all of this is about balance it's about choices it's about costs and benefits and costs and benefits from various perspectives so sometimes one variable or one concern will weigh heavier than others and in another case or another choice it's going to be the other way around and so where we want to try to get through all of this and through the discussions in the next day and a half is just to try and come to as much of an even balance as we can that's reasonable and then along the way we need to assess the policies we need to assess how we're doing the work we need to learn from our experience and we need to adjust what we're doing so that we don't get into a fixed situation and we can continually have a dynamic system that can keep up to the extent possible with the science as it's moving forward and that with that I will stop and open the discussion back up. Yeah I have a comment regarding the models don't necessarily assume that people change their minds which happens maybe not so frequently but it does happen so I would think a model that takes into account that should be an active informed consent model and the other aspect that I was thinking is if you can monitor what analysis are being done then you can prevent the types some types of theoretical re-identifications that are likely to be very rare so in that case the model of accessing the data but not necessarily downloading could be very important. Right and that was one thing that I I forgot to mention in terms of one of the options right now is going to the cloud and that came up earlier and again NIH does have a working group that's looking at how to do that and trying to build some case scenarios and Vivian Bonanzi and Dawn Poicer are co-chairing that and maybe I think have plans to call some of you in this room to help them think through the issues because from a policy perspective I really like that idea of having people not download all of the individual level data they can go in they can ask their questions and it stays in an environment that we have more control over though we need to make sure we really have more control over it and so I do agree with that. Ewan. Laura when it's kind of interesting when in these discussions do you feel that it's about the risk of identifiability or de-anonymization or is it the risk of harm I mean in other words to what extent does the policy get driven by the harm of that would occur to the participant when de-anonymized and therefore should or is there a concept of the gradation for different styles of cohorts depending on the data that they have? I think that it depends on the conversation. I think that we are starting to see more often than it came through in the white papers that perhaps some studies based on the nature of the data collected would have a different standard for what was okay to be released or what security measures you would put in place or need to have in place for access to different things. I do think that the discussion has too much turned on the question of identifiability or not and made it more of the binary decision and that it that's why I really think now in particular we need to go forward and have the discussion about the societal values and how do we manage that across different participant groups so we don't lose certain groups or are there cases just like different types of data that for certain groups maybe we have different security procedures in place or access mechanisms in place. Just to crystallize that it's from us moving thinking about the risk of identifiability or the probability of identifiability to the risk of harm. I mean that's the shift in thinking this through, right? It is but the risk of harm it's so hard to talk about because we can name things but the likelihood of them happening is a question and the interpretation of that harm is I think will be very individual there will also be some implications by groups but it will mean different things to me and to you. Sorry final question has anybody actually done has does anybody know of a de-anonymization moment? Did anybody know the moment? Has it ever happened in the last five or ten years? I to date I don't know. Has anybody published the fact that it's not obviously happened? Yeah we published a paper on it last year but it wasn't with respect to genomic data it was just with respect to mainly health data there's no known cases of genomic identification for malicious intent there are demonstrations like the work that David has done and others but there there are no published accountings of explicit identification for malicious gain but that doesn't mean it doesn't happen remember that's that's proof by absence, right? But it's an assessment of probability. Yeah it's a big point for probability. So I'm curious there have been a number of comments that have been somewhat dismissive of the idea that we could do a better job with software and security systems of at least avoiding inadvertent misuse of data. So I take and we'll accept Lincoln's point that there's no incredibly there's no perfectly secure system that the NSA wanted to hack into any system or you know other cyber crimes could happen but at least my view of how these things currently work is that we fill out these papers and there's this very careful review but once the data is actually put on someone's computer they're actually on their honor to do the right thing in what often is an incredibly complex set of approvals and in my mind this is not about a central analysis server or whatever especially given that it was changes to the common rule which actually make me quite nervous about about what's going to happen if we had better information systems just to sort of at least avoid inadvertent misuse of data where it wasn't just you guys read something send us a letter and then we're on our honor never to do it again that would seem like a good thing but I've heard a number of comments this meeting like I can't see how we would do that and I just note my iPhone and Google and the world we live in like information technology is remarkable and also like you can put your ATM into any ATM machine in the world and it can query another bank securely and find out how much money's in your account and do the transaction but we seem nervous about the idea that we could simply annotate a file with only diabetes non-commercial and somehow a computer system could at least attempt to honor that unless it was intentionally hacked why are we so skeptical that things that are already done routinely in all others the whole other cyber economy somehow can't be done for this very important medical application so I don't know if I'm so much skeptical it's ignorant but ignorant not just you but others being stated as can't be done David so sorry this is not so one of one of the big differences between those is that those tend to be closed systems so if we were to come up with a a contain set of software that we all agreed on and that would be the only thing accessing accessing the data I agree that it could be possible to do many of these things or even to place some sort of extra layer in between right that that everyone was required but then let's be honest as a group and say we might be philosophically so ill opposed to the idea of anyone having a system like that that we're saying it's impossible but we should say is we'd think philosophically we have personal belief systems that tell us what can be done and we might say for some settings our personal belief systems will allow us to you know I'm saying so under some settings we could do such things we might not choose to let's not say they can't be done I totally agree money is another issue a lot a lot more investment went into those people here it can't be done they say it's not worth discussing because it can't be done so I'd like to bring the conversation to a slightly different level and that in terms of what you were talking about is that the harms are always contextual and in the context of what the what the benefits would be okay at least by my view of looking at these things and that there was right now okay in our society we're in a really difficult position with respect to health care overall because that we aren't improving people's health outcomes but we're using more and more of society's resources to do it that's a serious problem right so instead of talking about how we're going to protect people's privacy that I think I wanted to bring the discussion to the point that if we don't figure out a way to take the basic science of people in this room and combine it with medical practice we're all screwed because right it's not going to work so then okay we have to and the public realizes this too so you have extremes on all sides but we have to have a way of having that be the value proposition and then talking about okay the the the harms or the risks of these different things in the context of that value proposition so I just would like this group to please make sure that that that we don't forget why we're here and and I completely agree with that and do you think that that is a choice that we need to make um through some high-level national dialogue to to have some of those questions and that will need to be informed again by looking at the choices and the way different groups different individuals weigh those trade-offs to be made because I think that's information we'll need to have and coming to any kind of even interim conclusion about how to make that value and that's what we really need leadership because that the people in this room can only contribute to that if we have the leaders setting a framework for us in that regard and I think those leaders are the national institutes of health the Sanger center the the funding bodies that take the responsibility to give us a framework that and then Debbie so I'll ask a question first and then I'll make a quick comment um the the question that I have has a lot to do with the discussions that I'm hearing today and it's not exactly clear to me what exactly this resource is being set up for um to the best of my knowledge it is not in design in its design to figure out how to give better medical care or health care per se directly it is if there is a design to support research but we are not tied into electronic medical records in this system now that's a whole different research program that's a whole different research issue now maybe I'm being heretical on this but but I'll I'll stake that right now okay but that's fine um so so on the flip side what I I I agree with what you're saying in terms of you know why can't we just put the day on the cloud or something like that I mean data is data is data I think that there's a more pressing challenge with respect to the policies that you set around that data um but I will say that we've done a lot of work in fraud detection and intrusion detection and extremely complex systems and and I don't know if you've ever been subject to having your credit card shut off you know this is one of the concerns that that people have when when you're working in these types of complex systems that you you detect things and then you start investigations and a lot of these investigations may lead nowhere and the things that you really do want to detect you just have no idea what you're trying to detect and so it's nice to audit it's nice to have all the audit logs on hand but unless you have the models of the workflows of how the information is going to be used it's very difficult to audit them and so we'd have to think very hard about what these workflows are so just land it on the table okay so Debbie's been waiting so I just wanted to I I agree with David I think that making the data accessible like we can globally access everything else is actually a much better model and I'm not saying it doesn't have access issues to it but I will say that every researcher every student every could carry much of the data that we have on their computers and in some form and they're not supposed to but that doesn't mean they don't and I think it's unrealistic if people in this room say it doesn't happen because it does and you know we can find it we can eliminate it but doesn't mean that it didn't happen but that's a potential brooch I think centralizing letting people slice and dice things the way they want if there is such a thing that could be made available people will go there and do it you know I mean that was the thing with the variation server it's just a way for people to slice and dice a data set which anyone can download because the information it's in there and if you think it's hard and if you think it's hard it's not even in db gap it's in db snip and freely accessible but if you think it's hard to audit this centralized system where there might be some work controls how hard is it to audit every graduate student postdoc walking around with data on their laptop and possibly leaving it somewhere like that's a lot harder to audit and that's where we are today actually whether or not people want to acknowledge it that's what can happen that's why I'm bringing it up because I think it's a disaster waiting to happen like well I want to just chime in with the I mean it seems to me that you know we don't want to make to be in the situation where we recognize that there are some samples that need a lot more control but we do have a lot of samples that are more open and it would seem to me that what we want is a technical solution that enforces the right data access so that the samples that we are broadly consented that it's easy to access them consistent with their consent and the samples that are least consented and the most restricted are protected and it seems like that's a technical issue that can be partially addressed by some computer system I mean there's there exists computer systems that can enforce that and that that would be consistent with our goals for broad access we want the samples that are accessible to be available to us easily and those that are most dangerous to share should be should have the highest barriers and and at the same level they're sort of orthogonal to that is there's different levels of the data I mean there's individually identifying reads all the way up to you have a common variant which probably can't tell you anything about the individual at the single site level and and we need to find another technical solution that that handles that scale as well that I can access different types of data at different scales depending on the sample depending on the type of data I mean it shouldn't the discussion be around what how do we implement such a system and what how do we articulate those requirements such that the things we've promised to patients and samples are are enforced but I again I I don't think in the in this debate I don't think it's it's it's useful to set these up as either or there's a whole variety of things we can do to improve access and and systems and I think we get ourselves into trouble when we set these things up in a in a kind of we either do this or we do this or we do this and I'm I'm quite attracted by the idea of this shift to risk of harm rather than risk of identity because I think it helps place a lot of these things to take David's analogy that the other thing key thing in the ATM scenario is that system has accepted a level of risk of fraud. Yeah I want to say that I want to say that my earlier remarks have been misinterpreted to to represent an absolutist position that I don't hold all I was saying was that the the central server cannot replace the the certification and or DAX or whatever societal regulation policy we have because there's always a way of hacking into it there it's exactly the same with credit card fraud we have lots and lots of technical ways of making it hard for people to commit credit card fraud however the old however the ultimate way of of controlling it is to have laws that put people in jail for doing for for for executing credit card fraud same same thing with ATM machines you can take a sledgehammer and break into the ATM machine if you want to there are laws that prevent you from doing that and that so we need to have both and that's all it was saying Eric following up on that did your committee in preparing for this discuss local IRB approval because they seem to be that is the one place where investigators local investigators tend to listen to in terms of their standards of behavior um so I don't think we talked about them explicitly in that way I mean I think that local IRB education and some providing some more explicit guidance I know would be greatly appreciated for the local IRBs and Pearl may be able to respond more to that so that there is some consistency in what they think they're supposed to do I think right now because we're operating in a space that's outside of the regulations there's a lot left up to individual interpretation for that because it's it isn't clear and that's one thing that the ANPRM would do it would it would set a very clear standard and everything would flow from that so it would make some of the choices easier it may just have trade-offs and costs to what could be done with the research that we might not like so I do think the local IRBs is is a linchpin in how this is done because of the shared responsibility and it's one that's and it's hard to get out because it is distributed and because the IRB in common rule as it's set up is intended to be locally heavy I guess in its emphasis so that issues of individual populations can trump other larger concerns and that's something when we're looking to build larger repositories that we have to contend with Debbie I think it it comes back to I think everybody in this room would support something more global whatever that model will become but I think in the end there has to be what Lincoln brought up what is the punishment for doing wrong in a system like this and you know I often ask the people who are in the data access well what if somebody doesn't you know do right by this and they say well there's not much we can do I mean I think if we're going to do something like this we have to have serious measures and say what it is will happen to somebody if they do something wrong in the system I don't think you can just I mean sure am I going to let everybody have my financial information and not know that if they stole it that they they get arrested that's that's nice to know okay and and I think the same we have to think about people's data in the same way there has to be rules set up as what happens to people if they do something off I mean we have it in scientific fraud I mean people are not allowed to to get access to NIH funding I don't know if I think it's that strong but I don't know so that is a question that we've asked the bioethics commission to ask themselves because they are at a position to be able to make that kind of recommendation about this and I think again as a community there are some papers that I've seen out there that propose a criminal penalty system for this kind of information as solving the problem because there's then a set of expectations and if you don't follow it then you know it's beyond fines and and other things so I think that's definitely something that should be in the consideration in the back just following up on the issue of of governance and accountability and so on and I think of course it's important that there's compliance but I'm also curious whether there's any thinking going on about governance with respects to factors external to the research community so not not so much just ensuring that researchers use the the data and access it appropriately but what about protecting for example against federal government saying you will now make the start available to law enforcement agencies or things like that because those things I think also fall under governance and how might they be incorporated right so I think right now we have put in place with our existing controlled access model that that federal government wouldn't be in a position to go back and compel someone to release the individually identifiable information that they have that's part of why we don't hold any information so that we don't have it to provide and then we just encourage I mean again we're working in our current system with our existing tools and so we encourage the certificates of confidentiality those aren't perfect there's a lot of distrust of how far they would go we've looked at ways to strengthen that kind of grant we haven't made any progress yet but it's certainly something that I think people are worried about but I'm not sure that it's as likely of a harm as people may think that it could be I think that's one of those cases where fear perhaps has taken over fear of a possibility has taken over in terms of the size of the likelihood just just to follow up I mean law enforcement I think is just an example but I think there are I mean many other things right I mean if if you know laws laws change all the time right and if you know another I mean another secondary use might be for pharmaceutical companies using it for marketing or things like that right and again that's why we were very again in the existing model it's very careful to say it's for research purposes and why we're so tied to consent and data use limitations because those requests that would come in and why there's an individual request process they wouldn't be acceptable under our systems that we currently have I think it also speaks to what Debbie was talking about about wouldn't it be nice if there were a broader legislative system that talked about what appropriate uses were for genomic data and had some penalties attached to that we're not in a place right now where we have that but that would certainly is something that I've thought about a lot and again goes back to the model of the financial systems you know there are there are certain people that can have access to data to do certain things and maybe that is where we are to instill the public trust and confidence in the system with this kind of information and not just genomics but health information more globally probably too. I was wondering if you wanted to comment on to what extent if the phenotype information is obtained from electronic medical records from health records the HIPAA would actually cover the inappropriate release of the information. So with regard to the phenotype information again I think it would depend on what phenotype information it was and what else it was linked to because it could be that individual phenotype data points just like genomic data in and of itself doesn't fall under HIPAA unless it's attached to some other identified private health information and so things could be released and I do worry much more about phenotype information being released generally and what people can learn from that and how that could be pinpointed back down to an individual than the genomic information. So yes Jeff. I wanted to emphasize the distinction between the public perception of risk and actual risk and I don't think there's any question that the public is moderately concerned about genetic risks of a variety of different sorts I think stigma and discrimination are far and away the biggest concerns but also as we conduct our focus groups we've heard probably a dozen times that folks are concerned about things like cloning. So I think the public has a limited understanding of these issues and I think in many circumstances they're just plain wrong and in fact I think the empirical fact is that this type of research has been extraordinarily safe in low risk and I don't think anybody wants to say the risk is zero now or forever and certainly the sorts of efforts that folks are making with this very discussion and sort of safeguards being discussed are all justified to think about but I think it's also entirely justified to say that the track record for this type of research has been extraordinary and so I think that the genetics community has been in a defensive crouch for quite a few years about these risks and I think it may be time to do a couple things one would be to make sure we collect all the data on the adverse event so we can give accurate depictions of in fact what the risks are but at some point I think it's quite justified and I would say probably now to really push back with the public and with potential research participants to say the track record for this type of research is extraordinarily good the risks aren't zero but experience and justified by quite a few years of this type of research now the risk is really very low with the kind of prudent safeguards that are in place so I guess I would welcome both that data collection piece and some additional efforts to begin to push back so that the public really recognizes that this is potentially valuable stuff with extraordinary limited risk so if I could just respond to that Laura yes the I think there's a overemphasis on risk I actually agree with the comment and it's actually very reassuring and I think would be a very positive part of public engagement to to provide reassuring data about how low the risk of re-identification is when we talk to participants and if you look at empiric data on participant views it seems pretty clear that there is a whole other aspect of the conversation that I think is equally important and it starts really with engaging in the with the public so I think the public has an interest in how their data are used they're interested in receiving reassurances that data are used to generate societal benefit they view those resources as important resources that should be used to achieve purposes that they value and so I think it's I think it's it's great and important for us to reassure people about the risk of re-identification but I think it may be far more important to talk about why the kind of data sharing we're talking about today has the potential to enhance benefits of research I think this is really the point that David was making I think we really need to shift the conversation toward here's what we're doing and here's why we're doing it and here's where we see a pathway to benefit that matters to all of us and I would think that that strain of the conversation and the data strain of that and some of this go into this conversation about the value prospect that we could have which also gets to UN's you know desire to move more to risks of harm and you can put all of it together. Can I follow up quickly on that? I very much agree with Wally but I also I would make a distinction here again too between harm which I just said I think is extraordinary low and wronging people and I don't think we've done a very good job with the latter I think it's been quite common for folks to be asked for their samples or data for one purpose and we turn around and use it for something entirely different that happens all the time and it's a huge problem and I think the public may well lack trust in us by virtue of not being transparent in that regard so I think that's a huge ethical issue doesn't translate into harm but I think it does translate into wrongs. It's really a reframing of the conversation around respect rather than risk alone. And then Carlos and then Pearl and then I'm sorry but I have to go pick up my children. I think the other thing to think about is how this is going to evolve over the next five years or ten years or as this generation that's used to put information on Facebook and trying to organize and so on is evolving right so folks now you know the notion that you put everything out on Facebook you know people don't want everybody looking at it or trust Facebook with all their information or trust Google with all their information so you know why would they trust NIH back researchers right and so people are already sort of very you know I think that that pendulum is swinging in terms of how comfortable people feel sharing all kinds of information and I think EMRs and associated genetic data are just going to be part of that whole big discussion and so I think it's really critical to engage it and thinking about that long term. I completely agree and there are patients like me and other organizations in the director consumer movement which are showing how much people are willing to put this into play for the aims of improved health. I just really want to agree with both Jeff and Wiley but I think we need a good PR campaign in that while we're fighting to say trust us you know we really can't show that people have been hurt in any way. The HIPAA privacy and the disclosure issues state laws on disclosure and it's a dime a dozen of lost laptops and it's now on the front page of most of our papers so I think if there were some way I agree of getting the message out we got to get people and the press thinking I think a little bit differently because you know again the high tech is just killing us. Okay I'm not sure if there's a more discussion or is it dinner now? All right thank you.