 So one of the disadvantages of coming this late in the program is that much of what I had wanted to talk about has been pretty well covered already. But one of the advantages is that I get to reflect on what I've heard the last couple of days. So I'd like to begin with two reflections before I dive into my material. The first has to do with the interface between clinicians and laboratorians. And I'm going to argue that we need a new paradigm for collaborative dialogue. In 1950, a physician would order an x-ray on a piece of paper, put it in the mail, send it to a radiologist sitting in the dark somewhere in a different building, and would get a report back that said something like shadow of unknown significance, additional clinical correlation recommended. And that really is still our paradigm. Our major concession to the 21st century is that we now send it by fax instead of mail. But I'm still filling out a form with some cursory clinical information. And the laboratory is doing an interpretation and putting that interpretation on a piece of paper and sending it back to me. And now it says variant of unknown significance, family studies recommended. We need a new paradigm. We live in a world of crowd sourcing and open access and networked intelligence. But simultaneously we live in a world of gene patents and proprietary interpretive algorithms and competitive research. And I really think that there needs to be more dialogue. And I'm not pointing fingers. Clinicians are just as much to blame. I have a telephone, you know, and I have email and I could engage in a dialogue. And I know I have examples, lots of examples where that dialogue has occurred to the benefit of the patient and to the benefit of an accurate interpretation. But we need to really build that more fundamentally into our algorithm. The second observation really, and I'm going to chastise us a little bit because what I've heard the last couple of days is that we know the questions, we just need the answers. I'm going to argue that we don't know the right questions yet. It would be foolish to believe we know the right questions. If we knew the right questions, there wouldn't be so much heritability that's missing in action. And we're talking about collecting information now that's going to be relevant to patients 60 and 80 years from now. I want to remind you that 60 years ago I had 48 chromosomes. You know, it was a couple of years later that it was discovered that the right number was 46. 80 years ago DNA hadn't been invented yet. And to imagine that we know enough now to think that the data we're going to be collecting now will still be relevant or interpretable 60 or 80 years from now is a little bit foolish, especially in the era of epigenomics. And what if we collect all this genomic data and realize that it's the changes in the genome over time that are most relevant or the environmental influences that we didn't put into the medical record now that are really going to influence our interpretation 60 or 80 years from now. So enough said about that. Let me actually cover the topic that Mark assigned to me. And I think we all understand what the word integrated means. But there are no two integrated systems that are built the same. The first decision that needs to be made is what's in your integrated system and what's outside your integrated system. Does it include dental care? Does it include medical transport? Does it include speech therapy for autism? Or is that part of the integrated educational system? I mean of course around that issue there's consensus. Both are convinced that it belongs to the other system. But in my system one of the fundamental elements and this will be familiar to our European colleagues or Canadian colleagues is that the payer and provider is integrated into a single system. So in my system an order by me or by any of my physician colleagues constitutes a coverage decision. And what that means is that we are charged with making good evidence-based decisions based upon our assessment of clinical utility. Is it rationing? I would argue that it's not rationing. There are no rules other than clinical judgment that guide our assessment of clinical utility. But we do have a fixed dollar amount available to us in a capitated prepaid system and we are responsible for being good stewards of those resources and providing the maximum benefit to our patients within that dollar amount. Within my system we have preventative care, primary care, specialty care, subspecialty care and that really allows the subspecialists to have a very great impact on the way that primary care and preventive care is practiced. I mean my colleagues love it when we take things over for them. My system would not only tolerate, it would love a rule that said that genomic tests can only be ordered by a geneticist or that neurogenetic tests can only be ordered by a geneticist or a neurologist. We really have the opportunity to influence the way that care is provided. And, you know, Howard's data was relevant yesterday but the ready availability of clinical geneticists in my system probably would make clinical geneticists the first source of information rather than up to date for primary care physicians. We have easy electronic systems for expert consults where we're carrying live phones so that a geneticist is always available with a hot phone to answer a telephone call and we're even mandated to have same day access so that a patient seen in primary care who requires a genetics appointment and is in the medical center that day can be seen that day in genetics. I know that's unimaginable and in less integrated systems. Prenatal pediatric adult and end-of-life care, there is a limit here. I think the average tenure of our members is about 20 years which is very, very good for an American healthcare system. But remember we were just talking about 60 and 80 years of actions and people do move and one of the side benefits of the great recession is that people are moving less frequently and changing jobs and therefore employer paid healthcare less frequently but we do lose patience over time. I'll skip down to test ordering performance and interpretation and this will lead me to a comment on my next slide. These are some of the foundational elements of an integrated system. There must be some element of scale. There is virtual integration and the electronic medical records greatest contribution may be to allow that kind of virtual integration. I don't think I'll say any more about the electronic medical record because it's been covered pretty well today. It has its limits. It is a wonderful tool. The model A4 was a wonderful tool in its day. It didn't have onboard navigation and it's silly to think of it as a wonderful tool now. The current versions of the electronic medical records simply don't have the ability to deal with genomic information. Group or virtual group practice I've already touched upon and internalization, the more that can be brought into your system the more integrated it will be. But there are some tests and some procedures that are done so rarely that even in an organization of our scale they really can't be done internally. So one of the things I've been fond of saying in the past is that Kaiser Permanente is like a small Scandinavian country and there are two things wrong with that analogy. First of all it's not small for a Scandinavian country. It's a large Scandinavian country. Maybe Northern California is a relatively small Scandinavian country but the other issue gets back to my comment yesterday is that the other thing is that we're not all Scandinavians. It's a very diverse population with a huge Asian population on the West Coast and a very large African American population in Atlanta and Cleveland and the Washington Baltimore area and a very large Latino population especially in Southern California. And obviously the captive population with the rich database really allows a lot of research ideas. I heard somebody float that idea yesterday and many of you have heard presentations on the UCSF, Kaiser Permanente Northern California collaborative effort to collect genomes and link them to the electronic medical record. So I'm going to leave you with these questions. I don't have answers to these questions but perhaps they can be the basis for some conversation. Which of these principles apply in less integrated systems because of course most of you practice in less integrated systems? Mark's already mentioned accountable care organizations a couple of times. I don't think any of us can predict what's going to happen to healthcare reform and whether accountable care organizations will really emerge or not but to some extent they are designed to function as virtual integrated systems. And finally how will the evolution from genetic testing to genomic testing foster or hinder integration? And I can give you an example on either side of that coin. I mean right now when I'm doing a genetic test I have to go to gene tests and figure out which one of 20 or 30 laboratories around the country I want to send my test to. You know in the era of whole genome or whole exome sequencing all laboratories are going to be doing it. Perhaps there will still be proprietary interpretation algorithms but once my own lab is doing genomic sequencing which I'm predicting will be one to two years from now exome sequencing at least. Next gen sequencing is probably the more accurate term to use than I may be less dependent upon a disintegrated model of relying on laboratories all over the country. And you know will the amount of data available simply exceed the ability of an integrated system to store and track that data? Is it feasible to imagine that 80 years from now somebody will refer back to a newborn screening genomic panel to call some data on how someone might respond to Coominin? So I'll pause there and take some questions. Thank you. Thanks Bruce. Questions? Yeah. You said that you know if you start now with the genome as it is right now then maybe there's the epigenome maybe there's you know other mutations throughout our lifespan etc. But all of those things are true but it's the same framework in terms of handling it. So I'd argue that you know we should start now anyway and just waiting for the perfect solution. There's things we can do now. There are low hanging fruits that we should go after and the systems will be able to handle these other genomic based things that come along later on. I absolutely agree with you. And by the way I think it's really good news that we haven't figured out the right questions. After all who would want to spend their life's work in a field where all the good questions had already been imagined and asked. I mean that's what science is all about. It's not just about collecting data. It's about figuring out the right questions. Bruce, so have you added data points? I mean when thinking about going forward and collecting data is there a discussion about different fields that you would want to add into Epic or that you've systematically gone after? Certainly not systematically and I have to admit that as a clinician I'm one of the worst defenders that well it's one of the differences between clinical medicine and research is that as a clinician what is most valuable to me is a narrative report. It's looking at a consultants report and being able to see how they thought about a case and what they considered and why they came to the conclusion that they did. And free text fields is not the way that you want to go if you want to aggregate data and have it available and searchable in the future. It's all checklists and using specific terminology and that's hard for a clinician but without it research is much more difficult. Yeah, Howard? I've got a philosophical question but since you've got the podium it gives you the right to pontificate. There's been the discussion throughout the last two days about the problems of people moving from one system to another be it an insurer or healthcare provider or whatever. So I would pose the philosophical question of thinking of the integrated system as the entire health system make it simple and say the U.S. health system but ultimately it would be the international health system. What do you see as a way to operationalize making all this data interoperable and interchangeable? Wow, that's a big challenge. I can tell you that there has been some pilot work between Kaiser Permanente and the VA who are probably the two systems that have the largest integrated medical records and they're entirely different electronic medical systems and yet the pilot work shows that they probably are interoperable to some extent, that they can crosstalk. Now, that's the extent of my knowledge. There are people in both organizations who would know more about that than I. Somebody used the analogy of the Sony and Betamax we cannot get into that situation. We absolutely need an interoperable data system and that's as ambitious as I'm willing to be. Imagining that there will be integration beyond the sharing of data in this country is a little bit beyond my imagination right now. Mark. So one comment that relates to that and then a question. This is an expansion of a point that I made yesterday which is the idea of certain data being treated as proprietary and we talked about that in the context of variant interpretation and being held by individual laboratories and not being made available which I think is a problem that needs to be addressed but I think there's also issues with data that is within health systems where while there are some barriers to sharing the information that relate to privacy and HIPAA and other things of that nature there are also examples where systems are holding onto the data because they say that gives us a competitive advantage which in the long run is detrimental to improving health from a societal perspective so I just would put that on the table something that I think needs to be part of the broader consideration about how we can make sure that we never use data as our market differentiator. Let me tell a little story that you may be familiar with there is a gold mining company that had geologic data that it could not interpret and geologic data really is the goal of gold mining companies they lock it in safes, it is their most proprietary information they couldn't figure out where to dig for gold so what they did was they put their geologic data on the internet and offered a half million dollar prize to anyone who could interpret the data they had 77 entrants and they awarded a half million dollars to someone who had a creative algorithm for interpreting the data and they mined three billion dollars worth of gold based upon that interpretation so we have to rethink what is proprietary and what is not Thanks Bruce