[This Transcript is Unedited]
Hubert Humphrey Building
200 Independence Avenue, S.W.
Room 705A
Washington, D.C. 20001
Workgroup on Quality Members:
Liaison Representative:
Staff:
Agenda Item: Call to Order and Introductions -- Review Agenda/Intent of Hearing
MR. HUNGATE: We are still missing John Lumpkin from the first panel, but his plane wasn't arriving until 9 o'clock and we all know the vagaries of arrivals at DCA. So, he will get here when he can.
I want to welcome everybody to this hearing. It has been preceded by a lot of discussion and comment and time and there has been a lot of background that has gone into this and before we go into that, let's go around and introduce ourselves.
We are on the Internet, I believe. Is that right, Donald. So, all that we say and do is available to the general public if they wish to avail themselves of that. At some time in the future, it will be summarized into comments and observations.
I am Bob Hungate. I am the chairman of the Quality Workgroup, a member of the National Committee on Vital and Health Statistics, which is charged with the advice to the Secretary on health information policy and our specific charge relates in the quality area. I am also the principal of Physician Patient Partnerships for Health, which is really an advocacy directed at patient participation in the health care system and broadening the dialogue between physician and patient.
With that, let me pass around and ask each to -- I don't think there are likely to be conflicts of interest on this, but if you have such, please identify it. No need to recuse yourself if you don't.
MS. MC CALL: Good morning. My name is Carol McCall. I am vice chair of the Quality Workgroup and a member of the full committee for NCVHS. I am vice president with Humana. I run our Center for Health Metrics and I have no known conflicts.
DR. SCANLON: I am Bill Scanlon. I am a member of the Quality Workgroup, as well as the full committee and have no known conflicts. I am also with Health Policy R&D.
DR. FRIEDMAN: I am Dan Friedman with Population and Public Health Information Services.
DR. FITZMAURICE: I am Michael Fitzmaurice, senior science advisor for information technology to the Agency for Healthcare Research and Quality and liaison to the full committee.
DR. KIBBE: Good morning. I am David Kibbe. I am the director of the Center for Health Information Technology at the American Academy of Family Physicians.
DR. VILLAGRA: Good morning. I am Victor Villagra. I am president of Health and Technology Vector, a consulting company that concentrates on disease management and technology assessment. I don't think I have any conflicts.
DR. LANSKY: Good morning. I am David Lansky with the Markle Foundation.
DR. ORTIZ: Good morning. I am Eduardo Ortiz. I am staff to the NCVHS Quality Workgroup and I am at the Washington, D.C. VA Medical Center, where I am the associate chief of staff. I am also director of clinical informatics and I am also a staff physician on the inpatient and outpatient medical services.
MS. KANAAN: I am Susan Kanaan, a writer for the committee.
MS. GREENBERG: I am Marjorie Greenberg from the National Center for Health Statistics, CDC and executive secretary to the committee.
MS. HOLMES: I am Julia Holmes and I am a staff member of the Quality Workgroup and I work at the National Center for Health Statistics.
DR. CARR: I am Justine Carr, member of the committee and the Quality Workgroup. I am a physician at Beth Israel Deaconess Medical Center and a director of health care quality there.
MS. JACKSON: Debbie Jackson, National Center for Health Statistics, CDC, committee staff.
MS. JONES: Katherine Jones, CDC, National Center for Health Statistics and staff to the committee.
MS. MATHEWS: Erin Mathews, American Society of Clinical Oncology.
MS. BOYD: Lynn Boyd, College of American Pathologists.
PARTICIPANT: -- program associate, Robert Wood Johnson Foundation.
MS. CHRISTIANSON: Susan Christianson, health information technology group at the Agency for Healthcare Research and Quality.
MR. ROHDE: Dan Rohde, American Health Information Management Association.
DR. HUFF: Stan Huff with Intermountain Health Care and University of Utah in Salt Lake City. I am a member of the full committee, just a visitor today in this subcommittee meeting.
MR. HUNGATE: All right. That is great.
Is there anyone on the phone? Not yet. There will be later. Thank you for reminding me.
John Lumpkin had a plane that arrived at 9:00. I have already covered that, I guess, but he will be here soon.
Today's hearing really has two purposes. One is a very long range kind of a visioning purpose and the other is kind of an immediate do we need to get something taken care of now that we will be glad we did later if we do it now. Let me try to characterize those two this way.
This workgroup started out looking at improving the claims information by adding more information into the claims transactions in order to improve the measurement of quality at the administrative end of the system, the pay for performance side, if you may use that in the broader context.
We conducted hearings on that and found a real conflict between expectations on the provider side in the provision of information and the payers in the use of that information. From that set of observations, we concluded that it was probably not going to be very productive to continue to work on trying to improve the claims information, that it was better to move toward the electronic health record and put our efforts forward toward where that is going to come out.
So, that is the background that has put us to this point. That is a pretty big -- when you start to try to look forward, it gets to be harder and harder to make everything work. So, we had a one day retreat basically. We drew in a lot of people, whom we have had before us or participating with us before and talked about this broad topic and the content. Brent James was there. Steve Jenks was there. Don Detmer, a past chair of this committee, was there. So, we drew on their experiences in debating this subject as well.
That created, I think, a fair amount of excitement within the group, in terms of the potential to really try to position the kind of things that need to take place over the transition to electronic health records in order for us to really do the job of quality improvement that we think is possible in the system.
I think it is pretty universally accepted that electronic health records are necessary in order to deal with the information content that is so rapidly growing, that there really isn't any alternative but to do that among those that work closely with the system. That said, getting your expectations right and understanding what you have to do in between to make it come down to where you get the answer that you want is another game. That is where we felt the visioning might fit.
Central to the expectations and the firmest conclusion really coming out of the retreat was that the secondary use of clinical data was the clean element and the critical element in terms of improving the coherence across the measurement system for quality for health. So, that is a central kind of content that should be kept in mind as we go through the context of today's hearing. That was quite adequately covered by our own Stan Huff in the just past NCVH full committee meeting. So that we have just had a recent dose, if you will, of where does this fit.
The secondary uses raises a lot of content issues. Did you collect enough? That kind of really led us into the short term question of whether the electronic health record is being looked at sufficiently carefully to understand the breadth of what you have got to have in there on downstream as you go through the learning curve of understanding what you really need to have versus what you thought you need to have.
There are institutions that have been on this learning curve for a long time and there are a lot of folks that are just starting. So, we are trying to understand that kind of a content. The longer term content is directed at let's call it the goals for the measurement system. We have had discussions within the workgroup about is it quality per se we are talking about? Is it health per se we are talking about?
We have been advised by some of our testimony to speak in terms of performance measurements and these are semantically different terms. They are all interrelated. We have tried in the positioning of this first panel to set it up so that we got as good an understanding as we could of the individual health, population health, the various perspectives within which quality might be judged at the end user point. So, that is an understanding. I have invited Dan Friedman to be a reactor to this comment from his vantage in the earlier development of the health statistics for the 21st Century document that came out of the NCVHS.
So, I am trying to make sure that we get a very broad understanding but yet don't lose sight of the fact that as Willie Sutton put it, you go where the money is and there is an awful lot of health care expense and relationship to health involved in inpatient health care, which is where much of the pay for performance work tends to end up. So, that is the other piece, the short term link into the content of this first panel.
Now, sometimes I confuse people by my expression of what I perceive to be, but I hope that we have set this up so that the discussion, which follows presentations is intended to be committee and panel so that there is room to -- and we have allowed more time for discussion than for presentations because I think that the talking back and forth is where an awful lot of our gain will come.
Justine, did you want to --
DR. CARR: I think I would just like to thank Bob for framing that and add also that we -- I think the real question that we are looking at is what quality will come out of the electronic health record. I think in terms of primary use, we know there will be efficiency, availability, timely -- you know, all of that, but in terms of the secondary uses, what is the -- what do we get in the current configurations of electronic health records and I think we feel particular urgency as the growing member of required core measures or pay for performance measures are being developed does not -- clear crosswalk to being able to respond to those growing -- that growing list of metrics with the electronic health record. I think what we are trying to say is that can we anticipate the building blocks necessary within the electronic health records so that we can achieve not only the primary quality goals, but the secondary use goals that we will have configured it in a way that is dynamic and can respond as new evidence comes out, new questions, new metrics, will we have an electronic health record that will keep up with that, that will deliver on the efficiency and also give credible information about quality.
MR. HUNGATE: The next is such that this morning you are kind of put here as users of information representing that viewpoint. This afternoon we will go to those who are providing the information to electronic health records and we will try to work those two back and forth.
Are there any other background questions that should be covered now before we begin? Okay. Do you have any preference, Panel, as to who goes first? Well, David has got his stuff up. So, you get the honors.
Agenda Item: Users of Electronic Health Records -- Panel 1
DR. KIBBE: Why don't I sit here and then I can advance the slides from here sitting down, if that is okay with you all.
First of all, let me say on behalf of the American Academy of Family Physicians, it is a real pleasure for me to be here and speak with you again. Just by way of orienting everyone, the American Academy of Family Physicians has about 60,000 active practicing members. It is an organization that represents disproportionately small and medium sized medical practices. Somewhere in the neighborhood of 70 percent of our members practice in groups of five or fewer, although we certainly have members who are practicing in very large practices as well.
The division that I run, the Center for Health Information Technology, was established in 2003 expressly for the purposes of helping our members to acquire electronic health records that are standards based and affordable. We see a growth in the use of commercial electronic health record systems from about 10 percent in 2002 to 30 percent now in 2005. I thought what I would do in the context of your questioning and getting to this major issue was to first of all acknowledge that from our point of view, we see the question that you are getting at in terms of the infrastructure in small and medium size medical practices, which is more than the electronic health record, but it is becoming the combination of the suite of proper applications that are used in a practice and what you can do with that and the opportunities there for a quality measure collection for reporting and opportunities for feeding back to the practices and getting the improvement cycle going.
I would agree with you that particularly in light of the rapid growth of electronic health records, not only by large groups, but by very small groups, if we don't think this through in the manner that you are trying to, we will end up without the capacity we want to be able to get the quality measures of the routine use of these products.
I thought I would give you an update and comment on some of the rapid growth in use of electronic health records among family physicians because I think that this goes counter to some of the impressions and ideas currently in conventional wisdom. This is a very rapid moving field and I hope this news will be useful to you in your discussion. I also want to offer some commentary on affordability and interoperability of the standards that are now making their way into electronic health records and with the partners of informational sources and that I think because that is so critically important in getting the data out of these systems on thinking through how the data gets in and is managed inside the system.
Physicians in small, medium sized medical practices using electronic health records are not only users of information, but they are producers of information. So, it is really a supply change model that I would suggest. Clearly, there is no question the connection between electronic health records and even small practices, the quality of care is becoming obvious. At the top of the list really is this chief benefit of instantaneous access to medical records by the doctor or by the nurse basically anywhere they are and that includes home or at the hospital.
At the bottom of this list, you will see the collection of quality performance data as a routine byproduct of the use of the system. The reason that that is the bottom of the system is not because it is not important or it is the least important issue, but in terms of a hierarchy of needs and medical practices, it is probably the bottom one.
So, if we are going to raise it without any federal pay for performance program that really changes the reimbursement system, we are going to have to do some work with the vendors and with the physicians themselves. I want to make the point very quickly that it is important, I think or we think, to understand that HIT has the option and use issues in ambulatory care versus those in the inpatient environment or large enterprises are very, very different. On the left hand side of the scale of this chart you see the ambulatory care picture where -- which is characterized by a small revenue per encounter, large, large volume. Whereas, in the hospitals, you have very large revenues per encounter, but relatively small number of admissions.
So, the leverage here issue is enormous in the outpatient environment. I want to point out to you that the vendors in the outpatient market are not the same vendors in the inpatient market. The standards in some cases that are used in the outpatient environment and the ambulatory care vendors are not the same and the buying cycle and the infrastructure upon which they are building is not the same.
Much of the business in the outpatient and ambulatory care market with respect to information technology is new business. So, it tends to be more innovative and that is an important thing to keep in mind.
Well, I am really sorry but we didn't transfer that one slide. That is too bad because that is a very important slide. This is a graph, supposed to be a graph here, that shows that we did two surveys, formal web-based surveys, one in 2003 and one in 2005, showing in the first instance that 24 percent of our members were using electronic health records in 2003 and 46 percent in 2005. Because those were web-based surveys, we discounted them significantly and said, well, maybe it is half that But we also in 2005 did two surveys that were paper-based, very large numbers of physicians that were paper-based, so you would not expect a bias.
We asked a very simple question in both of those. One was physician profile and another had to do with immunization. Are you using then a commercially available electronic health record in your practice? Both of those surveys came out right at 30 percent. In combination with a number of surveys that have been done at the state level, we feel very confident that 30 percent of America's family physicians are now using electronic health records.
This information is very important to grasp, I think. And the slide did come through. We asked the physicians in these surveys who aren't using electronic health records yet why not? What are your barriers? And notice that the blue in 2005 and the red in 2003, in both of those years can't afford and electronic health record is over 50 percent of the respondents who don't have an electronic health record. But then note what happens versus in 2005 versus 2003 with these other barriers.
In each case, decreased productivity, risk of vendor going out of business, security and privacy issues, worried about partners' acceptance, the respondents in 2005 are less worried about these barriers, even though they haven't purchased an electronic health record. This bears out a lot of information that we have from the field that our members and perhaps just physicians in general in the ambulatory care space are more ready to purchase electronic health records.
As a matter of fact, if we were to grow much faster than we are growing now in terms of adoption, the capacity would not be there because the major vendors who are selling in this market are already at 4 and 5 and sometimes 6 months waiting list. So, it is an important issue to consider in terms of the capacity of the markets. This didn't come through. I am sorry. This slide also shows that overall satisfaction with EHRs is high, with a few exceptions. I would reference the October issue of family practice management for these graphics and I will certainly try to do something to help us get this -- I should have brought my own slides today.
This is another piece of good news and that is it is starting to address the issue of ownership. We did a survey of 26 vendors in 2005. We asked them and then we verified with some practices, what is the total cost of your system for a practice of three doctors over three years. In other words, include not just the software fees but the hardware costs, assuming you are starting from scratch, the training fees, the implementation fees, the third party software fees, because you always have to pay our friend, Bill Gates, something if you are using a Microsoft environment.
To cut right to the chase, there is a figure here that is really quite remarkable is it came out to about $7,200 per doctor per year over those three years. I mention this because it is often quoted in Washington and other meetings like this, that it is $30,000 a year per doctor or $40,000 a year per doctor. Now, it can be. There is no question because the standard deviation of this information is huge. If you want to pay a hundred thousand dollars a year per doctor for an electronic health record system, you can do that. There are vendors who will sell it to you but this is well under $10,000 now for most of the available and reputable commercial EHRs. I think that while that is still expensive, you can look at it -- one data point to judge this by is if in a smaller medical practice, we are doing transcription, we are paying eight to ten thousand dollars per doctor per year just for the transcription.
If by implementing an electronic health record in the data entry now is starting to be done by the office staff or the nurses and the doctors, you can pay for it right off the top, not just for the software, but for everything. So, the issue on return in investment is becoming much, much easier to justify and I think that is why we have seen this really rapid growth, a tripling of our members' use.
Now I want to talk very briefly about exchange standards because interoperability and connectivity is really the key driver behind the value proposition for electronic health records in these practices. Physicians are beginning to understand that this is no longer simply about documentation and putting the beautiful paper into the computer. It is about transferring information from A to B, getting it into your electronic health record, being able to view it when you need it and making work flow changes in the office that not only increase productivity and achieve cost savings, but provide better value to the patient, more convenience, less waiting time, answering the phone, getting refills done much more quickly.
There are several areas where this interoperability and connectivity progress is being made that are worthwhile noting, I think, in the context of this issue of quality because to the extent that that theses efforts allow data to be gotten into the electronic health record, they also allow that information to be stored and gotten out at some point.
The continuity of care records, one of those standards is now a fully validated standard. It is known as ASTMCCRE2369 and it is a major advance because it is an XML standard for interoperable exchange of core summary health data between electronic health records and for portability of patient and consumers and as patients and consumers start to take more responsibility as we think they will for their own health information, they may also start to become a source of this information electronically.
E-Links has made a good progress in promoting national industrywide laboratory results reporting. We are not where we would like to be nationwide. But this -- and this has been an effort that has been more than simply a standards issue. It has also had to deal with industry practices and problems associated with the laboratory industry. But it is getting more dependable now, that if you buy an electronic health record in Toledo, Ohio and you want a contract with laboratory company A or B, that you will have and still interface for your electronic health records.
I think we made extraordinary progress in e-prescribing standardization and conformance of students over the last year or so. We have still got a long ways to go but the same thing, if you buy an electronic health record in Toledo, you can now get sure scripts, a computer to computer exchange transactions for e-prescribing and for renewals of prescriptions, at least some of the major drug stores in Toledo and other places.
As you are probably aware, Docket(?) has created a schema for physician office, EHR and the national database connectivity with respect for downloading quality and performance measures or uploading from the perspective of electronic health records.
Although I think that there are some real problems technically with that whole schema in the way that it has been set up and I say that as the project director for Docket. So, part of the responsibility is mine in not getting it right, I think it is still the right idea in that if we can work with the electronic health record vendors in such a way that specific data sets that need to go to a data aggregator that carry the particular reference, performance, quality data can be done, but we need to make it a national priority.
At the American Academy of Family Physicians, we are very, very disappointed with this G code scheme that has been put forward as a voluntary means in part because it backs off of using real data and collecting real data from the practice and substituting made up codes for it.
So, in summary, I think the challenge with respect to this longer term process of assuring that physicians and practices that use electronic health records can reliably dependently and accurately export quality and performance and cost information to data aggregators, is to really try to leverage the early successes we are having now in EHR adoption, to assure that this quality measurement capability is before us.
I think this is -- several things I want to highlight. One is I think we need to continue to work with the Federal Government and state governments and private health plans to help finance affordable standards of ACHRs. If there isn't going to be financial help for small or medium sized medical practices, I think the adoption and transformation is already going to occur and will occur and will continue but it is going to be slower.
If we could find a way for even small amounts of financial help or tax breaks to be made available to practices, we would speed it up. I think we need to work with vendors and physician groups to pilot the automated export of quality and performance data from practice electronic health records to data aggregators.
I am a co-chair with George Isham(?) of the Ambulatory Care Quality Alliances Data Sharing and Aggregation Subcommittee and we are working with CMS and within ACQA to make recommendations regarding the pilot project that will do just this. I think that there are a number of different ideas and technologies that need to be tried out in order to get this additional robust clinical information into a data aggregation, hopefully, all-payer database.
We need to continue progress within the AQA that has already been made to standardize measures, standardize the measures. We have got a good starter set. We need to get it out there and start using it and to develop policies and standardizing procedures for data collection sharing and reporting. I think this is a very, very important group. I think that we have made good progress in a very collaborative fashion. There has been a lot of compromise and a lot of progress in a short period of time.
I would encourage us to continue to see that group as a place where some of this work can be done. I think I will stop there. I think that is plenty at this point.
Thank you very much.
MR. HUNGATE: Thank you. That is very helpful.
I think we will try to get through all the presentations and then get into discussion and questions. So, write down your questions as we go and keep them, so we don't lose them. But let's go on in that way.
Who would like to go next? Would you like to go, John?
DR. LUMPKIN: Sure.
MR. HUNGATE: We would be delighted if you would do that.
I have to give John a few kudos before he gets to start because when I joined the NCVHS, we had some work to be done that required a report to get generated. Without John and Marjorie, whose institutional memory and participation was critical to getting that done, we wouldn't have made it. So, welcome back to the Quality Workgroup and NCVHS.
DR. LUMPKIN: Great. It is a thrill to be back and see so many familiar faces, get to sit on the table from this side. Although I am getting paid the same amount as I was before.
Let me talk about a couple of things. I thought a lot about what I should say when I was coming here and thought maybe it would be best to start off with the context, what we are talking about improving quality within the electronic age as we are seeing the transformation occur. It is a transformation, which really is quite surprising. I think I saw a recent article that identified that 10 percent of hospitals now have some form of electronic health records and they were lamenting the fact it was only 10 percent. I can remember the days when we were sitting here saying, oh, my God, it is only 5 percent and we are not sure about that 5 percent.
So, progress is being made and this becomes a very important time to consider the issue of quality. I heard the end of David's presentation. There are some very important issues that we will need to address. But let me put it within the context.
At the Robert Wood Johnson Foundation, we just completed a survey of business leaders in 2005. We did this in relationship to the issue of covering the uninsured. But the No. 1 issue -- and when we asked this of leaders, what is the most important issues to you related to health care, 52 percent said the issue was affordability of health care costs. Twenty-five percent said covering everybody because of the impact and only 12 percent for quality. Now, this is a shift. It is a shift that is reflected by the rising cost of health care, the fact that we have returned to double digit inflation in health care over the last three or four years, although that seems to be moderating a little bit.
But basically a big shift in the link in 1990 is when business was really pushing on the issue of quality, the issue of affordability has come there. The impact, they believe is that 79 percent are concerned about their ability of their employees to pay for their health care. Many expect their employees to drop the coverage that they have because they can't afford their portion of that health care. If you paid any attention to some of the recent strikes that have gone on in GM, the transportation staff in Philadelphia recently, as a result of that settlement, employees are shouldering more and more of the cost of their health care.
Now, we also completed a survey and this one has not been released, but will be released soon. This is part of our approach to going out and querying health care leaders in some target communities. In fact, we queried over a thousand health care leaders. What are the issues they are seeing?
These health care leaders in these target tracking communities are seeing health care expansions, primarily in the areas that are most lucrative, cardiology, procedure-based facilities. They are also expecting to see substantial cost increase in the future. What is most interesting, really in contradistinction to the business leaders is that almost none of them are looking at cost control strategies and cost control strategies can be done well or can be done poorly and we have seen in the 1990s sometimes when the issue of cost outweighs the issue of quality.
What is different is is that we have better ability to measure quality, but we also have different tools. We know that quality is a challenge, the work by Elizabeth McGuinn(?). Half of the care that is given doesn't meet the standards when you do a significant amount of chart review. The problem with quality, this particular study, which involves close to 7,000 patients who had their charts, they were called up, people said, you know, the researchers said can we look at your charts and compare them to the quality of care that was given.
This was an extremely costly study. It cost in the neighborhood of about 12 to 14 million dollars to do this study. The data is overwhelming. Obviously, in a different world, in a different environment, this kind of study could occur almost with no cost if we, in fact, start shifting that data into data aggregators. So, we have a quality problem where people are only receiving the right care about half of the time and much of our health care spending has no value, which ties into the issue of cost.
Now this is related to a project that is done by the Dartmouth Health Atlas. If you are not aware of that project, this is where they take basically Medicare data and they aggregate the Medicare data and they start looking at the cost of care, not charges, but the actual cost of care and they look by regions and there is great variation in these regions, from 15 percent above average to 15 percent below average.
Now, the interesting thing is you can say, okay, well, I can understand. I can go to a hospital or a health care provider and they are going to cost most because I am getting better quality care. Well, we are not so sure about that. First of all, there is a direct correlation in many of these areas when you look at the records where the number of visits to cardiologists, for instance, seem to be better correlated to the number of cardiologists per hundred thousand in the population rather than the amount of cardiac disease. If you look at a disease such as fractured hip, where there is really very little discretion on whether or not you are going to put a pin in somebody's hip. That is going to be a line basically flat across the bottom. It is independent on the number of orthopedic surgeons.
But other procedures, these sort of preference procedures are where you are seeing the high cost of care. Now, is the quality the same? No. When you look at the lowest region, in other words those who are charging the least amount in the country and you compare mortality, it is slightly hire in those regions that charge more. And if you compare it to those who are providing aspirin at discharge in those regions, that cost more, not charges, actually costs of care, adherence to providing aspirin at discharge after a heart attack, myocardial infarction, is actually less in the higher cost regions than it is in the lower cost region.
So, we have this great variation in cost of care with no associated increase in quality. This is a huge problem to get our arms around. Then there is another component and that is that the variation that occurs is not just from region to region. There is also variations by who you are. This is an interesting study that was done by David Narrin(?) in 2002 and what he looked at, a whole number of quality measures, particularly looking at African Americans and whites. As you see in this particular chart, on the left, the African American -- the white rate is twice that of the African American rate. The principle is fairly simple. If a kid who is under 17, 5 to 17 years old, has asthma, goes into the ER, the health plan ought to follow up with them within a week. I mean, that is not an unreasonable thing.
Yet the rate for the African Americans is half that for the whites. The interesting thing is a colleague of mine at the foundation when we were seeing this presentation leaned over to me and said, you know, John, the real problem is they both stink. But if we are going to be able to address quality and to do the kind of measurement and to identify the issues related to disparities, then we can't do that unless we can also identify who is getting the care and what their demographics are, not just enough to know the population as a whole.
It is interesting. Some of the critiques and the follow-up and the analysis that is going on and the violence that occurred in France was because the whole system in France, which they considered everybody is treated the same, had the inability to measure disparities in their society and that our ability to address these issues is not by sweeping them under the floor and assuming that they don't exist.
If you look by state, what is even equally interesting is if you look by state and you take the states on the bottom axis, on the X axis, is the percentage of African Americans in the state and on the Y axis are the adherence to guidelines for use of beta blockers after myocardial infarction and you plot that out, you get a straight line plot. The more minorities there are in the state, the less adherence there is to those guidelines. I want to say that very clearly, for everybody.
When you look at the data in communities, those communities that tend to have higher rates of African Americans and other minorities in the communities tend to have lower quality measurements for everybody who lives in that community, not just for the African Americans in those communities. That is related to the fact that over 80 percent of the care provided to African Americans and other minorities, Hispanics and Asians, provided by 20 percent of the providers and those providers tend to have less access to subspecialist care, less access to other kinds of resources, which enable them to give the kinds of care and, unfortunately, have the potential to have less access to electronic health data.
When you focus on quality, the good news is there is increasing evidence that not only -- and you look at this particular chart and you say, okay, the disparities over this ten year period of time -- and this was not an effort to reduce disparities. It was to improve the quality of care of people getting hemodialysis. They went back and looked at the literature to see what was different between whites and blacks. Basically, the gap narrowed, but the most significant thing is is that for African Americans whose adherence to the guidelines increased from 36 percent to 84 percent, almost a threefold increase.
So, inherence and use of quality data to improve quality can have an impact upon everybody in the community, as well as reduce disparities. Now, in order to think about the model and the kinds of ways we want to aggregate data, we also have to think about how we are going to address and improve quality. This model was developed by Ed Wagner and his -- ICIC, Institute for Chronic Illness Care, for Improvement of Chronic Illness Care. This is what is called the chronic care model.
This model looks at and says that it is more than just what is happening in the delivery system and how it is designed, but you have to include decisional support, clinical information systems and also involve what is going on with the patient so that if you think about the best quality interaction, it is between the informed, activated patient, through self-management and if you can envision that with an electronic personal health decision support system and the prepared proactive practice team, who is using electronic health record, also with decision support, you can get the most high quality interaction and providing that data is going to be an important component of it.
Well, why haven't we solved the problem of quality? Well, there are a number of factors. Individualistic culture of medicine. I can remember still my senior resident's voice in my ears when I was a junior student, the first time on my floor saying you assumed, you relied on somebody else's data. That is the way physicians are taught.
The second is is that quality is invisible to consumer and providers. Providers by and large, on the vast majority want to do a good job. They just don't know how well they are doing. There is no way to compare it and neither can the consumers. There is no business case for providers to adopt quality improvement because they just don't know where they stand.
The fact that we have this whole system where all the different parts don't connect. I want to kind of end up with a couple of quick thoughts. First of all, many of you may have seen this. David is probably a little bit upset because this is like the old version of what Connecting For Health has said for their motto, which is one of the four models adopted by the Office of the National Coordinator. But I want to use this particular motto because it is so much better at making my point than the new model.
This is that we have two things to talk about. We are going to talk about this area that is shaded, which is aggregating data to measure quality, but we shouldn't forget that where we really want to be is here, assuring quality where patients are getting the care.
Now, the challenge to us as we begin to develop it and to use the committee and think about data standards in relationship to quality, it is kind of -- you know, I am not a baseball fan, but there is a book that was published called Money Ball. If you haven't read it, you don't have to bother because I am just going to tell you the important part of it, unless you are a baseball fan, which I am not, but it is an interesting story.
Billy Dean(?) was the manager of the Oakland A's, had one of the smallest budgets in professional baseball. Yet, he was able to be extremely successful because he began to collect the data and look at it in a different way. What other teams looked at, they looked at, okay, who had a big name. Who was athletic? How fast could they run? How strong were they? And, you know, how fast could the pitcher throw the ball. Fast, young pitchers, expensive.
What he did is did a really intensive data analysis and rather than hiring big name pitchers, he hired pitchers who had a lot of ground outs. So, it wasn't strike outs that he was -- you know, if you could get a pitcher who strikes out 60 percent of the players, that is great. But the other 40 percent are hitting and getting on base, as opposed to a pitcher who 60 percent of the time, they are grounding out and get strike outs 20 percent of the time, you are ahead of the game.
He began to look at the data differently and the assumption was and what they found out was is they were looking at the wrong thing in making their decisions about putting together a quality team. Our challenge in health care is to make sure that we are, in fact, looking at the right thing.
First of all, we have to design the systems, electronic health record systems in the right way. In 1993, we put a system in Illinois called Cornerstone, which is a paperless system to manage maternal and child health. When we put the system in place, what we learned is if we designed a system to meet the needs of the patient and the provider, the caregiver, the client, that we had all the data we needed in order to measure quality. So, that has to be one of the principles first and foremost that we do the systems, that we design them in such a way that they don't impede care.
Now, the data that is currently being collected, the National Quality Forum, AQA, other kinds of organizations are developing quality measures and we want to have a system that can do that in a reasonable fashion, in other words, that bottom arc of the diagram. We have to recognize that data is going to be collected for pay for performance and how that expands and how that works out as other specialty societies are getting more granular, that we want to collect data that is going to address the issue of disparities reduction. So, race and ethnicity data, as the committee has said before, is important to collect.
But increasingly, the granularity and quality measurements is going to be increasing and the kind of measures are going to move for unit of measure from plan to provider, from plan to hospital, to plan to individual physician and other caregivers. So, we have to think about this.
So, the final thoughts are quality measures and what we currently use take that lower arc. In other words, their way of moving knowledge, patients should be given beta blockers. The thought is is that if we measure it, then docs will actually order beta blockers.
So, if I have a measurement that is done in 2005 and I will only order beta blockers for the patients who are discharged from the hospital 40 percent of the time and I realize I haven't adhered to the guidelines, then maybe my patients in 2006 will get better care. Very fudgy, the knowledge is that is the right thing to do.
But it is much better than nothing and that has to be the first step. But we also have to be aware of the Wayne Gretsky principle. Everybody asks him how come he was so good and he says most people skate to where the puck is. I skate to where the puck will be. So, we have to begin to think about how we can measure and test our measurement systems because I know how most of these measures got put together. There has been some analysis, but the data collection costs are very expensive, which is you get a committee together and say what seems like to be a reasonable set of measures.
Then we put them through a process, which is called a consensus process, which is not that you collect data on thousands and thousands of patients, but you have people sit around the room and say, yes, yes, that sounds like a good measure. But we need to go beyond that. With the volume of data, it is going to be possible through electronic system, we need to think about how we can begin to measure the measures.
Then, finally, to move the unit of measure from provider to patient centered measure. Instead of thinking of measuring quality in terms of what the provider is doing, we need to think of how we can move that paradigm into measure how the patients are receiving the kind of care that they need.
Then, finally, this always has to be kept in mind, Arthur C. Clarke, any sufficiently advanced technology is indistinguishable from magic. Until it looks like magic and until the providers can use it in order to get their -- to provide care to their patients and patients can use it in order to get the kind of care and to manage their own health, it becomes an obstacle and not a utility.
Thanks.
MR. HUNGATE: Very good. Thank you.
We need to have a copy of that for our archives and use.
DR. LUMPKIN: I will.
MR. HUNGATE: Very good. Thank you.
Who would like to go next? Do you two have any preference?
DR. LANSKY: Thank you, Bob. Thank, everyone.
I will try to take a somewhat different view, I guess. I want to try to approach this from the point of view of the patient or the consumer and I have been talking to consumer organizations and representatives for some time about these issues. As a result, as John concluded, I would probably try to focus my comments less on the settings of care and the technology that are in those settings and more around the information needs of the patient and the society as a whole.
I am not a big fan of the current measurement models and as John just alluded to the consensus process and so on that has been around for the last few years, having been involved in that for a long time, I want to make a couple of comments about where we are in quality measurement and the degree to which that is taking us in the right direction and whether or not the information technology we are now bringing to bear will be if primarily applied to the current quality measurement strategies, I don't know that will be terribly fruitful.
So, I think there is an opportunity for your deliberations to think about the quality measurement framework in a broader sense and then the application of technology to that, but first thinking about questioning some of the assumptions we have in the quality measurement environment.
Finally, I want to talk a little bit about national policy as a whole and try to keep at least my comments focused on whether we are getting as a society value for our health care investment as a challenge to the measurement community.
I think one of the broader themes I want to talk about is that we have created a level of granularity and quality measurement that frustrates our ability to make judgments about health care and health and I think the opportunity again for this group is to add to the existing work, which is very useful on an operational level, a policy, analytic level that I think is missing right now.
First of all, I think that the current inadequacy of the technology infrastructure, of the IT infrastructure, the shortage of electronic health records and so on is not the principal problem facing quality measurement. We have been at this for quite awhile now and it is important to keep in mind the primary problem facing quality measurement is the lack of will to measure important aspects of quality.
Similarly, I think the lack of introduction of IT in the health care system has been the lack of will to commit to introducing IT into the health care system. Other societies all over the world have been far more successful and aggressive and major health systems have been successful and aggressive in solving these problems. So, I don't know that we should approach this as a technology or architectural problem primarily.
So, will and strategy, I think, are the two things that need additional attention. I want to just make a couple of comments from my years in the quality measurement business on how it may pertain to the issue in front of us today. First of all, the issue of human health should be a paramount one in the discussion of measurement and of IT and it tends not to be. We had very little discussion of how the IT infrastructure will measure improvements in health or decrements in health or disparities in health.
Almost all of it has been again at a very granular level of data items, coding systems, terminologies, vocabularies, which are difficult to aggregate up to assessments of health and health improvement. I think that is an important challenge that needs to be addressed. I do think this is a big moment we are at right now; whereas, we are about to extend a large scale financial investment in technology and in infrastructure and in burden and everything else. If we do that in the pursuit of fine grained measures, which don't answer fundamental policy questions and resource allocation questions and social commitment questions, it will be unfortunate. We will essentially aggravate the silo fragmentation problem we all complain about and we will, in fact, institutionalize it in technology, which I think would be a mistake.
On that same note, I think we have a conceptual fragmentation, which I think this body has been successful in the past of addressing, with some very important reports. I hope you can do that in this arena as well. Our measurement strategies right now tend to be insensitive to the life continuum, the movement of each of us through phases of life, health, illness and so on, tend to be insensitive to the disease continuum stages of illness within a particular disease paradigm or across multiple disease paradigms, tends to be insensitive to hand off between settings. For the most part we are building measurement paradigms that are setting specific or practitioner specific and not around the continuum of experience in the health care system.
It tends to be insensitive to the interface between the patient's family and system and how well that interface operates, which is where most people actually experience health care. And it tends to be insensitive to the relationship between financing and care delivery, both in terms of availability of care and types of care, but we tend to treat cost and clinical services as if they are independent from a measurement point of view. I think it is important in this design phase with the IT work to assess whether we can do better at integrating those various things that we -- I guess from our own mental simplicity and for business reasons tend to fragment.
As I mentioned before, I think we should be very cautious in instituting process measures as the fundamental mechanism of quality assessment. It is very valuable for enterprises, who are in the business of delivering services to understand their processes in minute detail and to do the best possible work they can to optimize those processes and evaluate them and measure them and approve them.
I don't in any way want to diminish the importance of that. I am not sure that is an appropriate role for the national policy structure to address, at least not in the absence of having a set of national goals and measures that address those goals and evaluate those goals. So, if we aren't able in society to measure health improvement and health gains from the investments we are making in various sectors of health care. I think it is a distraction to then suboptimize around at a national level identifying processes of importance even though they are very important.
Through the IT work one other concern I have from my past experience is that we are risking digitizing the silos that we have instead of creating a digital climate, which could allow us to transcend those silos. Silos have arisen partly for service delivery reasons, but now increasingly for business reasons because we have institutions and categorical funding structures and so on, licensing structures, which perpetuate a set of delivery models that may or may not be well suited to the burden of illness in society. If we now institutionalize an electronic environment, which reinforces that rigidity of our current business structures, it will make it even more difficult to really reengineer how care is delivered.
So, I think we should be cautious about that and that gets back to John's concluding point. It is a worthwhile thought exercise as you do the long term part of your discussions here, to center your analysis on the person and say the person is the ultimate source of every piece of information that we are transacting in this health care environment. We should at least entertain the thought of architecting the IT environment and the quality measurement environment around that person, rather than around the various settings, which institutionally have developed over the last 75 years or so and we are stuck with.
Then to evaluate how to get from our current institutional environment to a truly person centered information and measurement environment. If, as many people are, if we are at a point where we believe this century needs to see significant redesign of the health care system, then this is a key moment to develop the metaphor or the model, as you said in your summer retreat, which is based upon the person as being the central source of information and decision-maker in the resource use in the system.
Let me change tacts and make a couple of other comments on a different direction. That was kind of an overview of the past. I want to talk about what is going on forward a little bit. There is an awful lot of digital data available today that we are not yet taking advantage of and I want to at least raise the question of whether the electronic health record as we tend to discuss it in rooms like this is where the puck is going to be three years or ten years from now.
My guess is that it is a worthwhile exercise to imagine a digital information environment, which is not particularly reliant on electronic health record as we now think of it in a physician's office or in a practice setting, that we are already seeing a very large proportion of the relevant health information that is now available in digital forms, in distributed networks. Roughly half of all the radiology images are now digital and available in PAC(?) systems and similar systems. Laboratory, national laboratory systems account for 50, 60 percent of all the laboratory data.
As we saw with Hurricane Katrina and the Katrina health strategy, close to 70 percent of all the dispensed medications are now available on national digital networks. Those all exist now without having infrastructure per se in the physician's office, similar to what we think of as an electronic health record.
If we extend that analysis and, of course, the claims data, which whatever liability you want to assign to it has some information about diagnosis and so on, procedure, treatment. It is worthwhile to think about a network metaphor instead of an institutional metaphor for the availability of health information that will constitute the platform on which we do quality measurement and other things and it is certainly worthwhile to look at other sectors. I think my fear at this point is we are taking with essentially a 10 to 15 year old technological approach and electronic health information and now trying to fund it and propagate it and have it widely disseminated while at the same time we look at the world of instant messenger and the platforms that Google is proliferating rapidly and see a much different approach to information acquisition, dissemination, dissemination and reuse than we envision in this setting specific electronic health record.
So, I think a more comprehensive analysis of network models of information handling is timely if we are going to be where the puck is likely to be in five years or ten years. Certainly, if we look around the world, the models that are emerging for health record systems and health storage, health information storage, are moving very rapidly in very interesting ways and, again, in our society we are not doing nearly as much of that.
Let me close with one example specifically of where I think we could be approaching the problem with a different mindset. When Congress was debating the Medicare Modernization Act and the creation of the Part D Drug Benefit, we did some work to evaluate what would be the appropriate quality measurement infrastructure to put in place at the same time as the drug benefit was being established and the kind of questions we were thinking should be asked.
But nowadays we estimate we are going to spend $720 billion on this Medicare Drug Plan. It seems like a worthwhile question, whether there will be a health benefit to the American public, to the Medicare population by virtue of a very significant public investment. I certainly am not aware of everything that is going on, but I am not aware of any public discussion of establishing an evaluation framework for the health benefit that is going to be gained by the Medicare population by virtue of this benefit.
In that case, in the case of the PBMs, prescription drug plans that are out there, they have very sophisticated electronic information systems in support of their existing business. They already have the ability and they already practice with their commercial customers the drug interaction, dose checking, age related dose checking. They do alerts. They do reminders. They communicate with the patient. They communicate with the physician. They do generic substitution. They do all kinds of administrative look-ups. They are also quite good at doing administrative -- doing operational measurement. So, for example, they have good estimates of dispensing errors, more so than the retail pharmacies typically are able to evaluate.
It seemed very reasonable as we looked at it to say from the existing electronic infrastructure in the pharmacy dispensing business, it is possible to evaluate both safety in the sense of dispensing errors and other errors associated with the operational procedures of the organization, service quality issues in terms of customer service and clinical outcomes in terms of whether it is alerts, reminders adherence, whether people are successful staying on protocols, complying with regimens, whether they are achieving desired health outcomes.
Are people taking Lipitor, having their cholesterol successfully lowered at six months or twelve months? Now, from the point of view of the government spending a vast amount of money on these services, it seems like a basic expectation that we would know whether this investment is achieving reductions in cardiac risk factors or cardiac outcomes. In fact, I am appalled that we don't have in place an infrastructure to tell the public whether those effects are being realized by virtue of this expense.
But there has been essentially no discussion about it. So, I think that -- to put in place a set of expected outcomes from major public investments and to expect those who are contracting with the government to provide that information as part of the service and to put in place the necessary infrastructure to report that information would have been a discussion worth having. That would not have required EHRs to be deployed across every physician's office in the country. It is something that could have been done by alternative and essentially available information infrastructure.
The challenge I think is to articulate an appropriate role for public oversight or evaluation of public expenses in this particular example. I think differentiating what is the role of the public sector in defining criteria for quality measurement and I would argue that that should be first on at a high level of fundamental public value and then setting in place the measurement infrastructure to support that and then evaluating how does the IT environment need to be addressed to accommodate that public goal is the kind of pathway that would be worth following.
Thanks very much.
MR. HUNGATE: Thank you. Very provocative thoughts as were the ones before.
Now it is your turn.
DR. VILLAGRA: Thank you.
First of all, I would like to thank the workgroup for distributing the background paper that -- I understand it was put together by Susan and I found it very, very helpful. While I have to confess that I was very confused about what the exact purpose of today's iterations might be, having read that paper, I remain confused, but now at a higher plane and that is --
MR. HUNGATE: Join the crowd.
DR. VILLAGRA: What I gather from reading the background is that the workgroup has been engaged in a very impressive quest in terms of breadth and depth to define the framework that will allow imbedding quality into the very fabric of the National Health Information Infrastructure, more specifically into the electronic health record. So, I will address very briefly four dimensions of this electronic health record quality linkage. The first is in agreement with you, David, is the need for an overarching driving force for population-based quality improvement.
The second is a series of system attributes that immediately surround the deployment of electronic health records that renders it a useful tool for the advance of the quality agenda.
The third is the role of a payment system and its link to financing of adoption of electronic health records and quality simultaneously and fourth is the issue of access to aggregate data and the rules that govern their use to advance the health of the public.
My experience with electronic health records has been through the lens of disease management programs. There are no data that I know of regarding how many nurses, health educators, pharmacists, nutritionists and other allied health professionals are using electronic health records but from my empirical experience, I can tell you that the number is growing very rapidly.
Examining the disease management phenomenon is useful, I believe, because it emerged outside of the traditional delivery system and in doing so, it escaped many of the financial constraints, technological barriers, cultural legacy and the inertia inherent in affecting change from within. You can think of it as an experiment designed by interdisciplinary teams of physicians, IT professionals, financial experts or just actuaries, underwriters, lawyers and marketers, all brought together to develop this model.
And the model, while far from perfect has been adopted widely by payers, both in the United States and increasingly abroad in many versions. It is also being adopted increasingly by hospitals and large group practices as a vehicle to improve quality, to decrease disparities in health care, to improve bottom line performance and to improve consumer satisfaction with the care experience.
DM, disease management, particularly in its early stages of when significant portion of the fees were at risk based on the attainment of certain goals and certain outcomes provides an especially valuable paradigm to understand the link between electronic health records and population-based quality improvement.
Electronic health records play a critical role in disease management. These records are continually populated with patient derived information, information supplemented with administrative data and more recently lab results. The work group and the Subcommittee on Standards and Safety has already pointed out the gaps in electronic health records gathered in this fashion, but in spite of its shortcomings, disease management programs and with its core of this type of electronic health records are driving real advances in quality on a large scale and with the speed commensurate with a sense of urgency, we all feel is needed to narrow the quality gaps.
Without electronic health records, disease management programs simply could not operate on any scale or with the speed they do today. However, electronic health records are decidedly not the driving engine of the system. The real driving engine is driving quality, is shared values between stakeholders of, in this case, disease management, service providers, patients and payers. These shared values encode the utilities derived by each stakeholder by participating. The shared value creates a non-zero sum gain of all involved or the proverbial win-win arrangement. This system rewards demonstrable improvements of clinical quality, patient satisfaction and cost containment. When fees are at risk, the arrangement penalizes performance failures.
These shared values are intuitive. They are easy to describe and easy to understand. They permeate all the actions in complex systems delivering these programs, including improvements in the functionalities of electronic health records and its supporting database analytic. A surveillance system while very clumsy and slow at this point, but that aspires to look very much like the equivalent of the Blumberg of health care or the tick tapes using financial world to monitor performance is the aspiration of this particular model.
Reports generated from the surveillance system supports of accountability of petition, collaboration and learning. As utilities are realized or missed, then myriad of elements comprising the system respond usually in concordance and in unison with corrective action for additional refinement as the case may be to maximize those utilities all formed around the shared value that I spoke about before.
These values and their corollary utilities become then the driving force that direct the technological specifications of a electronic health record, its functionality, supporting database and it pulls the system to higher levels of organizational efficiency and complexity, expansion and self preservation.
This has to do a great deal with its ability to persistently survive over the long term. So, for me, the first lesson learned is that the role of electronic health records should be subservient to the attainment of broad, shared value and objectives rather than the goal or a goal in its own right.
The second thing mentioned of electronic health record quality equation as I have experienced it through disease management is that attaining superior quality requires the deployment of a different organization of care and a different delivery system than was readily available. The traditional delivery system as organized today is conformed poorly to absorb the increased information output and the increased efficiency derived from the adoption of electronic health records.
Furthermore, tradition al settings of care, such as the physician office could not possibly accommodate the activities of ongoing patient education, motivation, support for lifestyle changes and so many of the quality imperative that will be underpinnings of real change in the future. Without a concomitant effort then to transform the delivery system, much of the promise of the electronic health record to improve quality would probably be unrealized.
For example, our system more and better information will demand staffing levels, rolled admissions, communication infrastructure and a physical plant that must be capable of managing large scale coordinated actions. In the case of disease management programs, these requirements demanded the deployment of highly sophisticated call centers. These structures did not exist five years ago. These are staffed by especially trained nurse, equipped with the most advanced mass communication technology available, Internet-based, telephony, electronic system monitoring and others. The sheer volume of information available to clinicians through this admittedly imperfect electronic health record could not find its way to patients without an organization of the type I just described.
So, if the committee welcomes metaphors, as Susan pointed out in her background paper, I would say that electronic health records that as we conceive them today would be the equivalent of putting a Ferrari engine into a Model T Ford and expect outstanding performance.
The third element related to shared values that I spoke before is really the need for a dramatic change in reimbursement strategy in favor of outcomes-based payment or reimbursement instead of production-based payment. The theoretical underpinnings of such a payment system are in its infancy and requires a great deal of attention. Payment reform that explicitly rewards quality is one of the greatest allies for linking electronic health records and in quality simultaneously and concomitantly.
A good example of how misaligned payment system can afford the best of intentions and excellent technology is in the findings from a recent study by Robert Miller called "The Value of Electronic Health Records in Solo or Small Group Practices in Health Affairs." I don't know how many of you are familiar with this.
The study looked at 14 practices that adopted electronic health records, small practices, and it showed that, first of all, the primary stated reason for implementing the health record was to improve quality. Only a few of the 14 practices, however, engagement in substantial quality improvement efforts. On the other hand, several practices experienced significant and apparently unexpected benefits in revenue generation and practice efficiency, allowing them to recoup their initial investment and increase their net income.
This is then a very good example of how the quality agenda was at the forefront of these groups implementing electronic health records. While they realized some of these practice substantial financial improvement and bottom line performance. The quality agenda seems to have been relegated to the background.
The fourth dimension that I simply want to put on the table for discussion perhaps is the ability to aggregate data as it collects itself in a network environment such as the one you described, David. The disease management model had the ability to aggregate data on discrete populations and had the legitimacy to use it. Payers and their partners seem to have at least to my knowledge had this legitimacy to use this information has not been challenged. But with the accrual of the sparse albeit interconnected information, I think one of the real challenges will be how are these data aggregated, how are the leverage to improve the health of the public or uses such as epidemiologic surveillance and I have not read very much about what would be both the technological requirements, the requirements for the point of view of the protection of privacy and security of these records, but this is another dimension of the ability of electronic health records and electronic data in general in quality and their linkage in terms of how they can be put to good work to improve the health of the public.
Thank you.
MR. HUNGATE: Very good. Thank you very much. Very helpful content and I look forward to first Dan's reactions and then discussion between the panel, among the workgroup. You know, I think the discussion is where we make our greatest progress. So, looking forward to that.
DR. FRIEDMAN: Thank you, Bob.
In his introductory remarks, Bob more or less implied that I was here in my role as institutional memory and I am a former member of the committee and I can be either described as an emeritus member or an escapee member or an exiled member, depending on ones perspective.
I am not comfortable in the role of institutional memory. Last week actually I had a hard disk crash and it reminded me of the fragility of memories both electronic and human.
I would like to reflect on five different themes that I have heard today and these aren't any original points, but they are more just quickly raising a few things that were mentioned. The first is sort of inherent in the discussion and inherent in Bob's initial remarks were really three different issues, what I think of as three different issues, and I just want to name those because I do think they are different and I think they need to be differentiated.
One issue is the role of what I think of as the desk top electronic medical record in improving the quality of health care of individuals. A second is the role of desktop electronic medical records in improving the health of individuals and the quality of health of individuals and a third issue that I think is distinct is the potential role of electronic medical records in improving the quality of health at a population level.
A second point that I would like to make is that I think that we need to acknowledge, discuss, debate, whatever, the possibility that the desktop electronic medical record can serve to improve the health care of individuals. It can certainly serve to improve the measurement of health and of health care of individuals, but that does not necessarily mean that it will improve the measurement of health of populations.
A third point I would like to make relates to the role of what Dr. Kibbe and what John Lumpkin referred to as data aggregators and in passing, Dr. Kibbe mentioned the importance of all payer data aggregators. Certainly, if we are going to talk about the possibility of a desktop electronic medical record in improving the measurement of the health of populations, it is all about, quote, data aggregators. You know, whether that is a national data repository, whether it is a federated model, et cetera, et cetera, it is all about the role of data aggregators and it is also, obviously, all about scaling up.
But the issues involved in scaling up are partially architectural. They are partially technical, but they are also, I think, largely political issues and I don't mean political necessarily in the sense of partisan, but that is part of it. I think we need to recognize that the discussions in other countries as Dr. Lansky referred to, there is a lot to be learned from those. They are sobering. They are thought provoking and the discussions change as the politics change and as administrations change.
That is especially true when it comes to data aggregation issues. John mentioned -- and I don't want to misquote you here. So, I am going to try to state this very carefully. My interpretation was that it is difficult to make a business case for quality. You want to restate that in a way that you are more comfortable with, John?
DR. LUMPKIN: No, I wouldn't say it is difficult. There is a difference between the -- I have to start someplace else to get back to explain what I meant. The way a market works is that you have a free exchange of information and that drives people to achieve a product that is an ideal product, depending upon the state of the market. So, if you have a market that doesn't value quality, then you are not going to have that as one of the criteria. Volume may be a criteria. We saw that at the end of World War II, where the focus was let's get out a lot of consumer products because we haven't had anything for the last four or five years. But the quality of those individual products pretty much staying.
When the Japanese entered the market, then the basis of competition shifted from how many products can you put out to how high the quality of those products were. Right now, the market in health care is not -- the competition is not based upon quality.
DR. FRIEDMAN: I guess I would add to that that the competition certainly isn't based upon the secondary uses of desktop electronic medical records for population health measurement purposes. I wish I could find a briefer way of saying that, but I think that is a point that, you know, to the extent part of this discussion is the potential use of desktop electronic medical records for population health. I think that is a point that needs to be acknowledged and discussed.
Fifth and finally, I want to get back to John's about Michael Lewis's book, Money Ball. John said he isn't a baseball fan and if we had had this discussion a month ago, I would have said that is because John is from Chicago and I am because I am from Boston and Chicago hasn't won a World Series since 1917 and we won one in 2004.
But it is a month later. But I think one of the things that is really disturbing and instructive is the extent to which even in baseball, there is a more vibrant, active and public discussion of measurement, of lead quality measurements of team quality measurement, of player quality, than there is around population health. I mean, it is really quite remarkable and has quite remarkable contrast.
I am sure one could -- our current Supreme Court nominee could speak to this, but, you know, this is a -- let me just read you the names of a couple of statistics. This is a book called Mind Game: How the Boston Red Sox Got Smart and Won the World Series and Created a New Blueprint for Winning. This is from a group that is one of the many groups that markets baseball statistics very successfully. EQA, equivalent average, a completely new measure that they have invented. Best single season equivalent average, career best equivalent run, fielding runs above average; career best fielding runs above average, pitching runs above replacement, et cetera, et cetera, et cetera. I mean, there is -- we just don't have this going on in population health. We don't have this going on in population health. We don't have this going on certainly in the U.S. in terms of measuring the quality of population health and we certainly don't have the discussion going on around how electronic medical records might be used to improve the measurement of population health.
Agenda Item: Panel and Workgroup Discussion
MR. HUNGATE: This should be an interesting discussion. The range of content is impressive and appreciated.
David.
DR. KIBBE: Can I start to make a couple of comments? Unfortunately, I am going to have to leave. I have to catch a plane and I won't be here for the entire discussion. I apologize for that, but I couldn't avoid it.
I think this has been extraordinary. I have been listening very intensively. I think everybody here has something uniquely different to say. I would like to make a couple of comments. One is I agree with David Lansky's criticisms of the desktop computer and what is happening with respect to some of the inefficiencies that are occurring when every small institution, medical practice, all the way up to a hospital has its own infrastructure and I would love to see a world in which we had three or four very, very good companies that were providing very low cost ASP model network systems.
The problem is is that I don't see that happening very quickly. What I do see happening is, one, even very small medical practices are very intensely interested now in inquiring their own infrastructures. Some of them are buying ASP model systems, but it is still their own infrastructure. I think that is even more intense at the level of a hospital. So, I don't think that we can fight too hard up hill against that trend because that trend is really motivated by a set of professionalism, of values. You know, the electronic health record for physicians has been three years away for 20 years and it is finally here. You know, it really is three years away.
The other thing is is that what worries me about some of the network systems is that I see we are building around the country extensive, very expensive private networks. The larger the organization, of course, the larger their network is able to be and a lot more private it has to be. I think we are going to expend amounts of money and effort and probably go through a lot of pain over the next five to ten years as these private networks establish themselves more and more vigorously and powerfully in certain parts of the country and use that private network in some way the way Prodigy did when early -- before the Internet came along in a predatory kind of market-driven way.
The reason that I mention that is because I think both of those trends in some ways are going to make it more and more difficult to get the data that we want to do population studies on patients. So, we have to understand, I think, that this situation in terms of what we want, that is, how can we get the best quality, most accurate information in the very quick turnaround time into a trusted and competent data aggregator so that we can now analyze it and get it back to people so we understand what the quality is, efficiencies are and compare.
I think actually it is going to get worse before it gets better. I think we are at risk for things really falling apart.
MR. HUNGATE: What time do you have to leave?
DR. KIBBE: I have to leave in ten minutes.
MR. HUNGATE: So, let's make sure that any questions that we are particularly -- for David, we get asked in the next ten minutes. Can we do that?
Right now, I would like to pose an issue a little bit in advance. I am wondering -- and I think speaks to part of what David Lansky was talking about. Is it going to be essential to have some kind of population-based measurement system, which becomes a benchmark against which others can judge their own performance? Is there an essential missing ingredient at the population level that could be a better enforcing function for the change that we need?
DR. KIBBE: David's point wasn't -- one of your points, we don't have the political will to do that and I think that is one of the issues. How do we get that?
MR. HUNGATE: That is a question. Others?
DR. SCANLON: I wanted to ask you a question and it is related to what Justine said about the issue of need for dynamic capacity within electronic health records. Because I think of it as a threat. When you encounter the need to retrofit because the demands of the external world are changing, it lowers your enthusiasm for doing this. Given that you said that we are -- I think you are one of the more optimistic people I have heard in terms of the growing enthusiasm for physicians adopting particularly small practices, the question is how much of this has been a problem in the past, how much of it do you see as a problem in the future as we -- you know, pay for performance is now getting a little bit of momentum and we are starting to think of sets of measures and regardless of how ill-conceived they may be, it has gotten some momentum and we may see reality there.
How much of a problem do you see it becoming and then I guess what can we do about this? I mean, where is the fix in this?
DR. KIBBE: I am not sure I know the answer to the fix in this, but a couple of comments about it. One is, I think David is right is that even if you have the best electronic health record in your practice, let's say you use one from G.E. Centricity. I was just at the G.E. Users Group giving a keynote yesterday in Dallas, or -- you pay a lot of money for a system and it really has a lot of information in it. There is still this enormous gap between what you have got in your electronic health records database and what the health plan has. The health plans aren't sending you all of that data.
You understand what I am saying. So, there is this issue of where does the data repose in any way that would allow for it to be aggregated in a meaningful, all payer patients centered way. The approach that we have taken the continuity of care record is to say look, focus on getting a set of data about a patient in an Internet standard, XML and allow that to be extensible so that we can create a vehicle, if you will, a vesicle or container for that information, which would be completely independent of the electronic health record or the database from where it came. That may or may not work, but I think right now the problem is we don't have any centralized, competent, trusted entity to aggregate data from all these different sources.
If we took that upon ourselves, who would pay for that? That is a huge issue.
DR. SCANLON: I think also in your slides when you talked about the uses of a record in an office, the last line, which essentially was about how do we retrieve this information for external purposes. It is one thing for a physician to be able to review an electronic record. It is another thing to be able to extract something for all patients. It is that capacity that I think we are worried about some in terms of --
DR. KIBBE: I think the continuity of care record has enormous potential for that in the future because it really creates a file format, a structure for structuring data that is completely vendor neutral, completely information system neutral. It really only cares about the data and the structure and it is using an Internet standard that is well used and well understood and the tools and skill sets are already developed in other industries. But I think that, you know, what we will have if we aren't careful is we will build these systems and people will buy them and use them and we won't be able to get the data out that we need from that particular source when we decide that we have got a data aggregator here to get it.
MS. MC CALL: Thank you. I have got some questions for you, David. I wanted to thank you before I asked -- for the remarks that you shared with us today and the enthusiasm I think that is obvious in the comments that you shared.
The questions that I have -- it is actually kind of two parts and the first one is related to the enthusiasm that you see on the part of practices and specifically small practices. Do you see in that enthusiasm an increasing adoption, a desire to actually have measures or some sort of feedback on -- I don't know if we want to call it quality or whatever, that somehow that that will come as a part of EHRs and that somehow that they do want it?
DR. KIBBE: Well, I think that one of the things that is a bit of a disconnect here and it comes up in Bob Miller's article that was -- Rob Miller's article that was published. It was a very good article. I think that when a family physician group comes to me, to my center for help, they are very enthusiastic and they are focused on quality, but they are not defining quality as outputting measures from my system and getting it back.
They are defining quality in a more sort of global but practice centered way. How can I and my practice and my colleagues function more efficiently? How can our work flow be better?
They often get also confused with and comingled with how can my practice do better financially because -- those issues are tied up, but the thing that they are not thinking about -- I think this is the question you want -- they are not thinking about or a small section of them are, a small subset, how can I use this system to export all of my hemoglobin A1Cs on all my patients to someplace that will help me get the information.
MS. MC CALL: Right. Then the second part of the question because that becomes important and I think it gets to maybe some things everybody has been talking about, which is lenses that we view things through and what do people see. Before I ask the second question, I want to introduce another metaphor, which is that of a nervous system. And I think that then what you had talked about in terms on the secondary uses of data and if you think of this kind of synaptic response and then going off to a brain, but some things don't go to a brain. But some things don't go to a brain. Some things just kind of loop right back.
I put my hand on the stove. It is immediately going to come off. It didn't go up. I didn't think about it. It just came off and it sounds like what the immediate feedback that docs want, practices want, is not about going off to a big brain and secondary uses of data and all of that, that what they want is something more immediate. What I can't tell is whether or not they actually envision something more profound coming after some, you know, cognitive chewing.
So, with that as a context, I guess I have a question now. You talk about the continuity of care record and can you and then some other people talk about the continuity of care record and how we think about that versus a use of EHR and versus clinical applications and I think that the vocabulary we used, the concepts we build are still very fuzzy. Can you help differentiate what you think a good vocabulary should be? Is an EHR the same as a continuity of care? Is it the same as a clinical lab or are there distinct concepts that are important for us to tease out and make distinct as we move forward?
DR. KIBBE: Let me take a crack at that because I have to leave and you all will have an opportunity because this is a very important issue. The semantics here are really confused. When a physician calls me and says we want help with an electronic health record, they are talking about a set of software applications that includes most of the time billing, scheduling and what used to be called the electronic medical record, clinical information. They are looking for -- and this is a big breakthrough in the last year or so -- they are looking for an integrated system of software in their practice that will do a whole range of functionality.
Now, sometimes people use the term "electronic record" or "electronic health record" to mean actually a file, think of a Word document or a PDF document or a spreadsheet and that is an electronic health record. What is confusing about that is that that could be something small and highly structured and very specific, which the continuity of care record is, right, or it could be all of the records that a patient has ever had, including images and tons of documents and so forth.
The term "electronic health record" and now we have patient health record are clearly undefined and we have no idea what we are talking about most of the time. Let me finish by telling you what the continuity of care record is and isn't in that context so you will understand. The continuity of care record was designed to do basically two things. I have an electronic health record from Company A and I have an electronic health record from Company B. This is the integrated software program. It could be Hospital A and Hospital B, but let's talk just about ambulatory care.
These are data islands. There is absolutely no way to get any information from this system into this system because these are proprietary databases. You can't take a data set out of NextGen(?) and import into e-clinical works, electronic health records. It doesn't work. So, a group of physicians primarily said, look, one of the first things we really have to have is a limited operability that would allow a defined data set of health information to go from this computer system to this computer system and be read and understood in the same way that when you take a Word document and import it now into a Word Perfect application, they can understand each other.
It ought to be clinically relevant. So, the CCR was designed to include sections like patient demographics, problem lists, diagnosis. Any physician who looks at a CCR immediately gets it because we work with this information all the time, whether we are in an emergency room or a doctor's office or whatever.
So, the CCR is a highly defined content standard that structures these data in XML and now it allows this company to say, okay, we will just export that data set to the XML. We only have to do that once because then we can export it the same way every single time. This company says all I have to do is parse that data from the XML into my database and anytime I am presented with this file, I can read and interpret it. That is what the CCR is.
So, it is in some ways an electronic health record. It is i some ways a personal health record, but it was designed primarily to get a data set from Company A computer to Company B computer in doctor's offices.
Does that help?
MS. MC CALL: It does. It helps a great deal. I guess when you had talked about it before because I am not familiar with its technical specifications, I heard something different. What I heard was the data cloud in the sky, that somehow it was the longitudinal cradle to grave that had everything about me.
DR. KIBBE: It is more like a snapshot. However, it could be a very powerful snapshot because if you have got this data set about yourself, if it includes your immunizations and your medications and your problems, there is a lot of information in there which could be utilized for quality measurement and if all of these information systems can read and write to that, then it becomes very efficient to use that particular data standard for that purpose. I think we will see that kind of thing emerge.
MS. MC CALL: Okay. Thank you.
MR. HUNGATE: Michael, did you have a question for David also? He is running, but he --
DR. FITZMAURICE: A couple of maybe more technical questions. One of them has to do with the continuity of care records. I see it listed as a standard E2369. Is it an American National Standard? It has been balloted and it is finally approved as an American National Standard?
DR. KIBBE: It is a fully balloted standard under ASTM International. This is the largest standard development organization in the United States and it is fully accredited with ANSI.
DR. FITZMAURICE: So, it becomes a American National Standard when ASTM gets done with it. Is ASTM done with it?
DR. KIBBE: Yes.
DR. FITZMAURICE: So, is it now an American National Standard, do you know? I don't know.
DR. KIBBE: I don't know if it has gone to that, but it is fully balloted within ASTM. It will be published by ASTM within the next few weeks.
DR. FITZMAURICE: Okay. Then that would make it an American National Standard, I believe.
Next question, in one of your slides you say e-links has made progress in promoting national and industry-wide laboratory results reporting between clinical labs and practices with electronic health records. There is a coding system that has been promoted for widespread use, LOINC, logic observations, identifiers, names and codes is what LOINC stands for.
Do they use LOINC or do they have their own proprietary --
DR. KIBBE: E-Links uses LOINC and it also uses HL7 2.4. So, it is an agreement around those two standards in the main.
DR. FITZMAURICE: So, this is an example of harmonization not of more diversity.
DR. KIBBE: Oh, yes. This is very much using the existing standards. And everything on that page I think with the exception of the CCR is about using and harmonizing these.
MR. HUNGATE: I think we ought to take a quick break here for about 15 minutes. We have got a lot more discussion to do. I am hoping the rest of you can stay for that. Is that okay?
Let's be back, oh, a little bit after 10 after. Let's shoot for that.
[Brief recess.]
MR. HUNGATE: We have got about 45 minutes here for discussion, continued discussion. And I would like to try to separate it into the two kind of discrete content frames of the broad, long term where are we going with this, what are the barriers, what are the limitations, what is it we are trying to get done, what are the possibilities. And the more immediate is the EHR as it is currently conceived going to be adequate in its content to do the things we want to do?
So, if we could maybe split the time 20/20 in that sense and first start with the broad one. I want to kick that off with a question that I think ties together all three of your presentations and see if you think I am on the right track. It seems to me that the payers are trying to
-- they have recognized the variation in cost between places and are trying to find ways to justify making decisions based on the variations of cost that are not anti-quality. That is what they are trying to do. It is the motivation.
It seems to me that some of that has shifted from an emphasis on managed care in the classical way to the disease management tools that are appearing.
Maybe it is not completely that way but there is at least some of the movement in that direction. My question kind of is directed at does the emphasis on disease management offer potential to yield a common health measurement system in a disease by disease way that would serve the broad aggregation of measurement of public health and evaluation of the health effect that David Lansky raised in his content. That is a broad question. I hope it is clear.
DR. VILLAGRA: Let me say a couple of words about the concept of disease management because I think there is a lingering misconception that this is a disease by disease kind of retail proposition and although it started that way, in fact, in fact, freestanding organizations that offer these services sort of bore the names of the type of services they provided. Mayfield(?) was maternal and child and core solutions was heart and that kind of thing.
When managed care decided to implement disease management programs for lots of reasons and put these organizations or these services at risk for the totality of the patient claim experience. In other words your point of entry was diabetes, but once you are in it, you are you with all your problems, health problems.
And you are accountable both on quality and financially and satisfaction for the totality of that experience. These organizations all expanded their competencies, horizontally to comprise many diseases clustering around these type of patients simultaneously. What that produced was a development of electronic health records that was able to house the information across diseases. Word processes that accumulated information especially from claims by that disseminated and moved information across specialties across specialties by necessity and developed information systems that was able to manipulate all this information with a big financial risk on their back and so some pressure to perform or essentially be outcompeted by somebody else.
That integration, there was one more dimension of this integration, which I just not having a name for it, I called it meta-guidelines in a paper I wrote a couple of years ago, which is the interaction between clinical practice guidelines when they coexist in the same patients. Some recommendations being synergistic, some being antagonistic but on a real person, the ability for these disease management programs to have the business support to do all that.
When you put all of these things together, I believe disease management as a concept can hold many of the elements of Ed Wagner's chronic care model. It is already invested in technology, mass communication infrastructure, aggregation of clinical content and it has developed a business model that so far has proven viable.
I will stop there because I think that that is as far as I understand its potential role. If managed care, let's say, would retreat from decisions about medical care at a very granular level, and -- decisions back to the delivery system, we would not have very good quality care and I think disease management is a potential repository for an entity, maybe even a future institution that could fulfill that function.
MR. HUNGATE: John and David?
DR. LUMPKIN: Well, I will jump into it. First, a couple of terms. When we did the report in 2001 on the National Health Information Infrastructure, we were very careful to stay away from the terms "health records," either personal or provider-based. As I struggle for terms, I am kind of like resting on the term "personal health information system," "provider-based health information system, because not all electronic health records are the same. The key issue is it is not just do you have a way to record what is going on electronically, but can you provide the decisional support that enables an individual to meet the challenge of Dan Mathis, which is the kind of decision making, which exceeds the ability of human cognition.
Having said that, the promise of disease management has to be taken in context with the transformations that are going in the health care system and there are other trends that would tend to move away from disease management. Do you need disease management if you have a really good information system that provides decision of support so that every doc can manage diabetes in the way that it ought to be managed.
On one sense you could say, no, you don't. On the other sense, what we have learned in a number of areas, and maternal and child health is perhaps the best, is that if you provide services in a comprehensive way to patients or to individuals, you have better outcomes, particularly because that means that there is continuity of care and there isn't always continuity of care, even if you have this whole identified source of health care.
But the other disruptive factor is the trend toward the combination of health savings accounts, which are tied in with high option plans, consumer driven health care, increasing the activities by payers of care to shift costs to the individual, as well as, you know, all these other trends, in which case the information needs become a lot different because what you have with the disease management systems is you are sort of fixing, you are cobbling together the system, where fundamental changes is that you can't really get quality if people don't know what quality is and they won't demand it.
There is increasing evidence and I was just at a presentation yesterday at the Clinical Scholars Meeting, where there are some surveys that are being done that the awareness of the public of quality, that there are discrepancies in levels of care that their variation of care is increasing and that if we can have consumers and purchasers of care engaged in making decisions based upon some understanding with transparency of quality and price, it is our belief at the foundation that we can begin to do different things in the health care system.
In fact, that is a major area that we are going to be doing some of our grant making in the future and trying to do that in communities to drive that. Within that context, I still believe that there will be an important role for disease management, particularly with those who have chronic disease just because of the fact that that kind of comprehensive care can be seen as an adjunct to the care that is given in other venues.
DR. LANSKY: A couple other thoughts. I think this is -- Victor described it -- we are talking about a way of virtual integration of the delivery of care and coordination of care, as John just said. I think from an IT point of view what is happening is one can imagine it happening at either of two levels but not in the middle. The upper level, in effect, is the payer or disease management entity as an aggregator of information across a number of sources. As you heard Paul Shields, I think, testify to the NHII committee a few months ago about the health plan, national health plan's idea of especially integrating all the digital data under plan auspices, but it would have the same potential theory, principle, to provide the integration and apply division support or intelligence for that database and do a lot of good things with it.
The other locus of integration is the patient themselves or the family and the personal health record, personal health knowledge system, whatever we want to call it is an alternative focus or locus for integrating that data. Within the middle are all the individual care settings and data collection points and they are each so fragmentary in their contact with the person that they are not able to be the integrators. They are really suppliers of data to some other function, whether it is at a higher order or a singular order of functionality.
But I think the challenge is -- for me, one of the reasons we keep putting emphasis on the patient as the aggregator is they are the only one who can supply all the patient source of data about outcome symptoms and quality of care experience, changing needs and so on. None of the individual professional data sources can do that. I think there is a reason for work like you are doing to continue to analyze how can we enable the person to be the virtual integrator of the information flow around them and around their lives.
The last thing I want to say about disease management, in a sense to me it is one instance, one of the more successful instances of system innovation and redesign and it is not the last. What we want to do is allow there to be an information environment and a quality measurement environment in which this model and a thousand other models yet undefined will proliferate but also be tested as to whether they actually make a difference in improving health.
MR. HUNGATE: Very good. Thank you.
Eduardo, you had a question next?
DR. ORTIZ: My question is directed basically at David, but, obviously, anybody feel free to answer. Indulge me for a minute on my preamble so you understand the context of my question.
A lot of the potential benefit of electronic health records is to improve quality. As I was listening to David, it seemed to me that he had some concerns about what we currently use to assess quality of care. So, you know, he has talked about process measures. So, we obviously use process measures a lot. As an example we will say, well, what percentage of patients in the practice who have coronary artery disease are receiving a beta blocker or aspirin therapy or statin? That is a performance measure to indicate quality.
We do a lot of that. It sounds like there are some concerns about that. We also do intermediate outcome measures, which is a step a little further along than process measures. So, for example, we might say, well, what percentage of patients that have hypertension have their blood pressure controlled, less than 140 over 90, which we do know that we have good data to show that that is an intermediate outcome measure, but we know that that is actually related and linked to important patient-oriented health outcomes.
So, you know, that might be a reasonable measure to use but there is a lot of intermediate measures that don't have that association, that we don't know that they necessarily improve important patient oriented outcomes. So, a lot of these measures now, it seems like -- and that is what a lot of places are using. That is what the VA uses and that is what Kaiser is using. That is what the pay for performance thing -- these are the types of measures we want to use. I think we use a lot of these measures, you know, one, because they are easy to collect and measure and analyze. So, that is why we do it because it is there and available.
Some of them we do because there is some evidence for it, but not all of them and probably not the majority of them. There is not a lot of good evidence. So, it is obvious and I agree with you that these are probably sub-optimal in terms of measuring overall quality of care. So, in terms of the content of quality measures and substance, this is a really important issue moving forward with electronic health records.
My question is, you know, what are your thoughts or recommendations as to what we should be measuring and what is needed to move this forward besides what we are currently doing. That is directed at David, but anybody else feel free to chime in on that.
DR. LANSKY: I did say in my comments that I think it is very worthwhile for a physician practice or an enterprise like the VA to measure the kinds of things you just listed because certainly having done work in the outcomes field the criticism correctly was that it wasn't actionable for some who is operating a specific program or shaping a practice design.
So, I think you should be measuring the things you are measuring. I wouldn't want to take us away from that. My comments today were more directed at national strategy and what is the role of broad public policy as established in a building like this in setting a both measurement and IT agenda. So, one of the concerns I have from a sort of microscopic point of view is that the consumer as they experience their experience their own health care is not terribly sensitive to, for example, the hospital measures that are now part of the CMS measurement set for pneumonia and cardiac and so on are not very consumer relevant measures for someone who is suffering with heart disease or some family member of someone who is suffering with heart disease. They don't talk about angina symptoms, shortness of breath, life experience, lost work, productivity, lost quality of life with the family, things that are immediately experienced by the person and are outcomes that that person would like to improve and that they think they are making a financial investment in the ability to improve those outcomes that they experience.
So, at a national strategy level I think it is unfortunate that we don't have a way to help people understand whether they are making to a provider or that we are making financially on their behalf through public funded programs are producing a benefit that they experience in their health.
For those intermediates, those primary intermediate outcomes that you described are important, but as you say only to the extent they ultimately contribute to improvements in health. Some of them are going to be sort of hard endpoints like stroke reduction and some of them will be quality of life improvements, like angina symptom reduction or improved mobility for someone with orthopedic concerns. That is where I think some of the focus should be and one of the reasons the measurement issues I have said we should give some attention to the personal health record as a platform. If it is possible there to assess cancer patient's pain and angina patient's pain or disruption, an asthma patient's ability to sleep through the night or attend school, those are the kinds of endpoints that we are just completely neglecting.
One of my concerns about the EHR launch is that that platform makes literally no accommodation for patient supplied endpoint or outcome.
DR. ORTIZ: Let me ask for some of these things then, David, is it your opinion that we know enough to know what we should be measuring, for example, someone like the cancer pain thing and the angina scores and we know what it is, we are just not doing it or do we still need a lot of work even in determining what we should be measuring to determine whether we are providing good quality of care?
DR. LANSKY: I don't think that -- there is, obviously, much more we could learn about how to do it well. I don't think the state of measurement science is any less in those outcome categories than it has been in the process measures. About 10, 15 years ago, there was an acceleration of attention on the process measures because the data systems were available through primarily claims to permit us to do that. So, there may be relatively more progress in the last ten years or so on that. But there is a great body of work. One of the ironies is because of the clinical trials environment, we have had a very rigorous set of patient outcome measures in the world of the FDA and clinical trials for a long time, but they haven't been brought over to the performance assessment and the evaluative arena very much.
MS. MC CALL: Just a clarifying question on that. I am sorry. To understand that if you think about the measures that we have, that can be derived from claims and think of that as having end measures, that there is a lot more that would be available that could be captured, assuming standards and harmonization and all that if they won't attempted to be derived from a claim but attempted to be derived from systems if designed appropriately that there is a body of knowledge that could be captured and codified that is much larger?
DR. LANSKY: One example I brought I mentioned earlier, I was on a panel, an IOM panel that did this report on measuring quality of cancer care in Georgia and I thought the process was very interesting for me and I thought the measurement approach they came up with in this project was a very balanced one in terms of the issues Eduardo is raising, where there are a number of process measures that the people on the ground in cancer care felt were very important indicators for themselves of quality of care delivery day by day and it was also a set of patient centered measures and public health measures in effect and the long term outcome measures that created a balanced view of cancer care.
So, I think to your question, Carol, I think we know enough, an example like this particular project, to compile a set of rigorous, well-documented measures that cover that whole suite.
DR. LUMPKIN: But I think that the key issue, what we don't really know, goes back to the story of the guy who is walking out one night and he sees this man clearly intoxicated, who is looking for his keys. He proceeds to help him for half an hour, look for his keys under this lamppost and it finally gets a little bit frustrating. He said, you know, let's reconstruct this. Now, where is the last place you remember seeing your keys. He said, well, over there by my car. Why are you looking here? Because I have a light here.
Much of what we have done in measurement has been driven by the data that is available and there is a lot of science on, you know, how do you measure outcomes. My memory is blocking because -- but we had a lot of hearings on that, of the committee, on measuring, you know, scales of function. The problem is how do you get that data.
Part of the challenge I think is to sort of create the case for collecting that data in a way that is meaningful. Again, at the meeting I was at just the last two days, there was a set of 86 measures that were being vetted for looking at surgical risks, reducing surgical risks for people over 65, who are undergoing surgery.
One of those measures that the expert panel came up with was the position of the -- appropriately positioning the patient on the table by the anesthesiologist. What is the value to the anesthesiologist to document that? When it becomes a value for the anesthesiologist to document it, then the measurement of that quality and that encounter becomes really -- it doesn't cost anything to do. So, we have to think about the interplay between systems development and systems analysis of what is going on and how to improve care and getting the providers bought into it at the same time that we are doing with measurement.
So, we have to be careful that our conceptual models of measurement enables us to get to the granularity that ultimately we should be able to get with an electronic health record and we think about it sort of as peeling the skin of an onion.
DR. VILLAGRA: Just a quick comment on the granularity versus integrated outcomes measurement. I am beginning to see a significant uptake of activity, quality related activity, aimed at improving specific metrics to cash in on pay for performance schemes, whenever the money is available. This is done in such a focused way that I am observing that this may happen at the expense of other important processes that make the care experience a good one from the patient perspective and medically that good in general.
When you think about the contributions of measurements to be able to assess quality with the level of granularity that enabled you to change systems and to improve constantly, you can superimpose that on a patient centric view where they do not -- they are not thinking of particular measurements or particular metrics of what just happened. I am related to somebody who had a fractured arm and experienced infinite transactions in the health care system. The best integrator is a highly complex cognitive process that takes place in the head of the patient.
If we can devise a mechanism that creates a business case for that impression to be captured and measured, the one I came up with is that I would ask this person to assess the quality of care she received under the care of this orthopedic surgeon and the team that took care of her three months later and that that particular assessment would be an authorization for payment for, let's say 80 percent of the fee that the orthopedic surgeon charges and that six months later, that particular patient has to -- this person is integrating in a lot more information that the particular metrics that we can devise, objective, as well as subjective, and that that person could authorize payment for not the remaining 20 percent, but they say 30 percent of the remaining payment, based on satisfaction with care, incorporating technical outcomes and objective outcomes and build a business case that truly empowers the data collection system in a way that we are, I think, generally speaking very shy to allow to happen because it would be in somebody else's control.
MR. HUNGATE: Very good. Thank you.
We have two more questions on this and then we shift to the other folks.
DR. HOLMES: One of the purposes of these hearings is for the committee to develop a kind of work plan for the next year or two years. One of the things that we struggle with is trying -- is going back and forth between being very narrow in focus about, you know, what should the electronic medical record look like or something versus being very general. I was interested in David's comments to the effect that if we just look at the electronic medical record, that is perhaps too narrow a view, that we should look at digital networks of information sharing.
Then on the other end, we shouldn't just look at quality measurement, but the outcomes of quality measurement, which is after all the purpose of the whole endeavor. Given that perspective and given, you know, our attempts to lay out, you know, what we might do over the next couple of years.
Would the three of you maybe comment on what you think would be most useful for us to pursue in terms of supporting the advancement of quality and health care?
MR. HUNGATE: Please proceed.
MS. MC CALL: Okay. Top three wishes.
DR. LUMPKIN: I will start off because it is a challenging question and one which I much preferred when I was sitting over there.
I think that we have to be careful that we are not looking so far ahead to where we want to go that we trip over the rock, that we don't see in front of us. To that extent, we need to do a combination of things. One is is that with the small penetration that we have in electronic health records, expecting that it is going to get larger over time is to look at those kinds of quality applications that can be inserted into the health record. Let me give an example of a standard that I think would be important to be developed now and that is the common interface between a decisional support object and an electronic health record so that the data can be fed into these objects, which can be built by the specialty societies and other organizations to abstract data in a common way.
So, rather than building one for each one of the vendors, building their own decisional support stuff, there is a common way to feed that in to engines that could be then built by specialty societies, by other organizations. That would be one thing. Second is to continue to resist the paradigm of medical records and think about what can be done when the data that you are trying to churn is digitized. That is to go back to the work of the committee in 2001 and think about these in dimensions rather than -- in systems rather than individual documents, that the goal of all of this is to push knowledge down to the point of service. So, I think defining the terms is going to be very important.
Then, third, is think about ways that the current system can be implemented in ways that will enable a patient-centric concept of quality.
DR. LANSKY: I will throw a couple other ideas on the fire. I support John's comments, too. I think at this stage of development where we are all very uncertain about what is going to happen, I would essentially work through scenarios for the next couple of years and I would develop a set of architectural assumptions or architectural alternatives, one of which is essentially broad EHR adoption, another of which is more of the network digital database model, where labs, pharmacy, images, encounters, claims are out there and can be accessed.
So, lay out two or three technical architectural models. I would lay out a set of assumptions about availability of data for public disclosure in which essentially the issue of rights or intellectual property are in play because I think that needs much careful analysis. The Katrina health work that we did was instructive because we discovered that in the space of a week in a crisis that everyone agreed upon was a humanitarian crisis, business issues and legal issues and technical issues could all be relaxed and things could be done in a week.
As soon as the crisis had passed, those walls came back up and we did not in any way develop a durable solution to what should be. The solvable problem is that in a week, virtually 70 percent of Americans could have their medication list available online in a week from now, but the legal and business issues are significant and they are meaningful. They are not trivial.
So, I think that analysis needs to be done. As we look at this digital network, whether it is in EHRs or in PBMs, wherever it may be, what are the rules or expectations about availability of that data for uses of the quality assessment. When conversely is it -- we have to operate on the premise that only voluntary disclosure by individual data holders is the only way that data would be made available. That is a very important issue.
A third thing I would do is what the AHIC is doing now, identify a set of use cases, short term. I would say a three year time frame and basically identify for the sake of discussion three use cases that you would like to have answered and then use those to drive analysis. Essentially, if you could analyze in the context of architecture and policy, if you could then drive scenarios for three use cases, such as the one I proposed was what is the value of the Medicare prescription drug benefit and to whom and to provide value and to what degree and be able to analyze the quality if you will of that program.
I am sure you can think of a dozen other use cases. Pick two or three use cases and then develop essentially what would the work plan look like to implement both the IT infrastructure and the information requirements to analyze that.
DR. VILLAGRA: A couple of thoughts. I would try at all costs to avoid the potential error of allowing the electronic medical records and all the technology surrounding the electronic records to get so far ahead of our executive capabilities against the information and the knowledge that we generate that we will create a tremendous amount of frustration much like what we had -- we did when we developed guidelines and they sat there and they sat there and nobody did anything about it in spite of enormous knowledge that it embodied.
So, that would be to understand what is the delivery infrastructure that is capable of executing in parallel the acquisition of new data and new knowledge related to implementation of a health record. That will be the first one.
The second would be an examination of who or what are the potential aggregators of information. I would suggest that a platform that you can use to examine that function of aggregator could be built on the function of care coordination because care coordination has an information knowledge function, but it also has an operational correlate.
I believe that a concept like disease management but not it as it is out there today necessarily, could be a place to start and go from there. But I suspect that the institution or the entity that will function both as aggregator of knowledge, aggregator of information that would allow analysis of population health and be able to intersect with a delivery system under the umbrella of care coordination does not exist today. Simply, we don't have it. What are the requisites for this entity or institution?
The third one is just very general and take, perhaps, the Institute of Medicine report and more clearly articulate what we want as a country in terms of health care. This is a much longer term or position but remember a few years back, Ron Widen(?) and Orenn Hatch put a bill in front of the Senate that was completely ignored that essentially called for an examination, systematic examination of what we want out of health care, given limited resources and what are the tradeoffs that we are willing to accept and somehow create what I call share values that are simple, easy to communicate and easy to understand for everybody.
We know what it is that we want to measure at the end of the day.
DR. SCANLON: It actually didn't get ignored. It is part of the NMA and the work is underway right now.
MR. HUNGATE: Okay. Carol? And then we will segue over to the other -- you need to run, John?
MS. MC CALL: If you can listen while you are packing up.
Actually, I want to extend this theme. Okay? So, you have some wish lists and I love the items that are on here. Okay? So, the question that I now have are what do you believe are either the compelling events or the burning platforms that can be used to kind of ignite some of these issues because not all of these can necessarily be done or
-- order matters, okay, as we try to grow toward the solution. So, think about what those could be and I have two in mind that I want to ask specifically about and get your opinion, but there may be others.
Those two, the first one are the chronic care improvement program pilots. Again, another investment that the government is making and it is a smaller investment than PDP, where if you make almost a trillion dollar investment, one would hope to see a business plan for how we are going to see what is coming out. But could that, in fact, be one of those? And I ask specifically around that because we actually at Humana, we won one of those awards in the area surrounding Tampa. As an intervention and as a model that is person centric, it is wonderful. It is delightful. It takes into account the person and the families and the caregivers and everything from just what is the experience of your life and perhaps your death, as well as some of the process metrics and some of the things that are more classically defined as measures.
So, they are both opportunities for research and which we will do, but are they a fulcrum because the country will expand that. So, that is the first one.
The second is P for P. We have talked about this as a workgroup before, kind of waxing and waning on whether or not that is, in fact, a compelling not event but wall of activity or a burning platform and I keep thinking about whether or not -- you know, it is based on claims. What if it were based on something that could come out of an EHR? We have talked as a group about whether or not you could have a partial P for P, kind of a lower case if it is based on claims and upper case if it is based on something that is based on something that is based on output, exhaust, if you will, from an EHR or some sort of clinical application. I don't know if that is something that you see could be used.
So, that and any other compelling events that you see out there.
DR. LANSKY: I think the British Excellence Model is the closest example of using P for P in a way that provides additional rewards for an IT adoption within the more outcomes oriented approach that they have. They do capture data, which is beyond what is available from claims. So, I think that is a good place to go.
I do think the two models you gave as examples are at opposite ends of the spectrum, where the chronic care improvement projects, like disease management and the other virtual integrations I talked about are methods of integrating care. I think all three of us, at least this morning, have talked about the challenge of integrating care across fragmented systems. So, by coordinating payment or integrating payment or oversight through a central contractor, that facilitates the ability to do outcomes oriented measurement and patient centered measurement.
Conversely, most P for P is fairly fragmentary and it is paying individual providers for their small episodic contact with a patient and they are not really trying to -- they are not capable of addressing the continuum.
MS. MC CALL: Yes, I did not mean to imply that those were things that would get at or compel the same items on the wish list. They would touch on very different attributes.
DR. LUMPKIN: I am just going to toss one thing out here. I think if you are familiar with the concept of disruptive innovation, I think the disruptive innovation, I think the disruptive innovation in health care is going to be personal health records or actually personal health information systems.
Two million people currently have health savings accounts with high option plans and they are in the position -- and that number is growing -- of having to make health care decisions that heretofore have been put in other people's laps. Their desire to have the kind of information that they need in order to make those decisions, I think, is going to be a big push and it is going to be a push that is going to be felt in many of the places, which make policy because when you can't get the data and you won't need to make the decisions, you are going to get angry and where do you go but to your elected officials.
MR. HUNGATE: Interesting observation.
MS. MC CALL: Do you believe that the laws are currently -- if you as a consumer, as a person said, look, I want my data, do you think that the laws are currently in place to enable you to get it electronically?
DR. LUMPKIN: Well, the point is -- and that is the reason why I think it is a disruptive innovation is is that if people want the data and more and more people are asking for it, providers who are going to look at the bottom line and saying unless I go electronically, I can't meet this demand. So, it becomes another factor that is going to be pushing towards adoption of -- in addition to the ones that David was mentioning.
MR. HUNGATE: Thank you, John.
DR. VILLAGRA: If I can just add one additional thought to your question about compelling ignition points. The small office or the large physician practice, adoption, quality improvement processes that are helped by electronic health records and similar type of aids, I think, boosting the pay for performance initiative is definitely worthwhile and it has already ignited the response and the enthusiasm by physicians in practice, but the other area where it has not been exploited sufficiently is the potential role of quality in managing risk, meaning medical legal risk.
This is something physicians are extremely sensitive about. Any call to put together a, let's say, continuing medical education program or management systems that lowers medical legal risk and essentially decreases the likelihood of being sued, that is met with a tremendous response on the part of physicians. I think if we can continue to explore some of these areas where I call the utilities that are derived from values that we explore and flesh out, we can get a lot more mileage out of the existing processes.
An example would be if a payer has information about all patients who are 50 years of age and who have no claims for, say, colonoscopies and that information can be fed into a quality improvement loop, it also has important correlation in risk management because the vast majority of legal action against physicians, particularly primary care physicians is failure to diagnose and colon cancer and breast cancer are at the top of the list. This is just a minute example but I think a broader exploration of these potential utilities that align data and quality with other benefits would be very helpful.
MR. HUNGATE: Very good. Thank you.
We have not gotten to the second part, the short term issues relating to the adequacy of the EHR. The cultural issues, the other kind of barriers has taken that time. Let's see if there are some immediate questions that should be addressed to that point now before we take another -- at least spend 15 minutes there and make sure that we have got any questions that are sitting right now covered.
But I think we are going to come back to that more this afternoon when we get to the developers. So, questions in that context?
You asked one earlier. Do you have any follow-up questions, Bill, about --
DR. SCANLON: It is more, I guess -- you mentioned that there is enthusiasm for pay for performance and I guess I am not sure -- I would say there is avid interest. I am not sure which direction it is ultimately going to go because I think John's presentation and yours both set up the fact that we are potentially on the verge of a transformation of health system and the marketplace and in that transformation there are going to be losers and -- in John's presentation to the issue of -- that if you look at it from the buyer's side, we are very concerned about costs and like to see something done about that and actually think of information and the flows of information as a tool to try and influence to try and influence the future path of costs.
I guess I am somewhat pessimistic about pay for performance and in the introduction, I didn't say I am from Washington and having been here in Washington and watched the resistance to incredibly rational small changes and the successful sort of blockage of small changes. I guess I worry about the ability to have something that is going to require sustained major change. I don't know how -- I mean, my cynical side says what we need to do is we need to sneak up on everybody and have this transformation occur subtly over time and suddenly everybody wakes up and says, wow, things are different.
But beyond that, I don't know if you have any suggestions in terms of how we influence this shift in a way that is going to be more palatable. Pay for performance here, you know, we are talking about if you go look at the Medicare Payment Advisory Commission's recommendations, we are talking about budget neutrality, take away sort of money from the bottom and give it to the top, relatively modest rewards. If you look at the British model, where in the U.K. for the primary care physician, they put a big chunk of additional money on the table.
Now, they are starting off from such a different base, that was in some respects a relatively easy thing to do. Our situation is so different that I kind of have these worries that we are not going to be able to manage to keep this moving forward. Once it starts to pinch, you know, at this point nobody's had any money taken away, at least from the public program.
DR. VILLAGRA: What I would ask is I understand the challenge both operationally and politically of the need for budget neutrality -- I would like to know more about what is the potential for operating within even a budget neutral environment that rechannels health care away from inefficient, ineffective and excessive care in a way that will not impact quality and that will allow for redistribution of income across the spectrum of that variation that we know exists.
I understand very well the political barriers to implementing something like this are enormous, but the only alternative in the near future is to continue to pay for production and that is a dead end. I mean, I do not see with the advent of new technologies that we all know are going to be not only expensive but are going to be extraordinarily good. If we don't tackle the pay for performance and the methodology that allows us to shift from pay for production to pay for outcomes in some measure, be that in a granular way or be that in a more global patient centric way, I think we are just postponing perhaps a political solution to all this that will just make it easier on all of us.
DR. SCANLON: The one slide of John's where he showed the relationship between supply and the use of services, I mean, it almost comes down to if we really start to recognize the quality of services and resultant effects on health, that the solution, we are only going to do the redistribution is we are going to have to redistribute the location of providers because as long as we have that concentration , we are going to end up in this production model where we reward production.
MR. HUNGATE: Okay. Are we through with the topic for this morning? It seems to me that we are.
Thank you for excellent contribution at all levels. You are welcome to stay for the afternoon and participate further if you have the time and interest. But you don't need to.
Let's reconvene at 1 o'clock.
[Whereupon, at 12:05 p.m., the meeting was recessed, to reconvene at 1:03 p.m., the same afternoon, Friday, November 18, 2005.]
A F T E R N O O N S E S S I O N [1:03 p.m.]
MR. HUNGATE: I think we had better begin to proceed, although we are missing a few and go through the introductions again.
I am Bob Hungate, chair of the Quality Workgroup, member of the NCVHS, which is positioned in this discussion as the adviser to HHS on health information policy. The workgroup specifically working in the quality area. I chair the workgroup. I will move to my left.
MS. MC CALL: I am Carol McCall. I am vice chair of the Quality Workgroup, also a member of the full committee of NCVHS. I am vice president of the Center for Health Metrics, with Humana, as well as on the board of directors of Green Ribbon Health, which is one of the recent awardees of the Chronic Care Improvement Pilot. This company has been started in Tampa.
MR. HUNGATE: Dr. Fletcher, it is to you now.
DR. FLETCHER: I am Dr. Fletcher and I am the VA representative. I happen to be the chief of staff at the VA here in town and we have pretty much a complete paperless and filmless hospital, which I will be talking about.
Thanks.
DR. HUFF: I am Stan Huff. I am with Intermountain Health Care and the University of Utah in Salt Lake City. I am a member of the committee, of the full NCVHS committee and a guest here with the Quality Committee and will be speaking on behalf of IHC and our electronic health record today.
DR. RUCKER: Don Rucker. I am with Siemens Medical Solutions, USA and we are a large vendor of health care IT and imaging. I am also on the clinical faculty at the University of Pennsylvania.
DR. JANES: Gail Janes, staff to the committee, CDC.
DR. ORTIZ: Good afternoon. Eduardo Ortiz. I am staff to the Quality Workgroup and I am also at the Washington, D.C. VA Medical Center where I am the associate chief of staff for informatics and a staff physician on the inpatient and outpatient medical services.
MS. KANAAN: Susan Kanaan. I write for the committee.
DR. HOLMES: Julia Holmes. I am a staff member to the committee and I work at the National Center for Health Statistics.
DR. CARR: I am Justine Carr, member of the committee and member of the Quality Workgroup. I am a physician at Beth Israel Deaconess Medical Center and director of health care quality.
MS. JACKSON: Debbie Jackson, National Center for Health Statistics, CDC, committee staff.
DR. SCANLON: Bill Scanlon from Health Policy R&D, member of the workgroup and the committee.
MS. GOVAN-JENKINS: Wanda Govan-Jenkins, National Center for Health Statistics, CDC, staff to Standards and Security Workgroup.
MS. ALTON: Migna Alton(?). I am program associate for the Quality of Health Care Team at the Robert Wood Johnson Foundation.
DR. FRIEDMAN: I am Dan Friedman. I am with somewhere between small and minuscule consulting company called Population and Public Health Information Services.
MR. ROHDE: I am Dan Rohde. I am with the American Health Information Management Association.
MR. HUNGATE: Okay. We are all set.
This morning we focused very much on expectations from the electronic health record in terms of the users of information that might evolve from that. It focused a lot on the barriers and complications that exist in making this a reality in improving quality. The intent of this afternoon is to start from the other side of it and work from the EHR and say how is the EHR positioned to serve these various demands.
I should express that there is concern on the part of the workgroup about whether we have adequately understood the needs of the quality agenda in the rollout of EHRs as they occur. So, we are looking to understand better what we need to worry about in terms of that meshing downstream.
Agenda Item: Developers/Suppliers of Electronic Health Records -- Panel 2
That said, I would like to turn to Stan Huff, who helped us very much to understand the secondary uses of data, which has become a core focus of the attention of the workgroup. That will follow by Peter Geerlofs, then Dr. Rucker and Dr. Fletcher.
We try to hold comments to maybe 15 minutes to allow more time for discussion back and forth, which is where we seem to make a lot of progress.
So, Stan, look forward to hearing.
DR. HUFF: Just to start off, I would just like to acknowledge that the things I am going to present in fact are the work of a lot of people and I would just like to acknowledge them as mentors that as well as co-creators of the systems that we are going to talk about.
Just a brief introduction to Intermountain Health Care. We are not-for-profit corporation with 22 hospitals; 1.4 million patients that we have electronic medical records on, 24 clinics. We have around 450 employed physicians and roughly another thousand or so physicians that are closely aligned and use our inpatient facilities.
My presentation is coming from the perspective of involvement in electronic health record systems that span the health system, which was developed by Homer Warner and others starting in the 1970s, a system I worked on when I was AT&T, Bell Laboratories in the eighties. Then electronic health record system that IHC developed with 3M during the nineties and then a new system that we are developing today. So, there is some perspective on different systems, different environments that I am bringing to this discussion.
Just a quick summary of system design considerations, some obvious things. Speed is very important. Response time that you have for your users is incredibly important. Business events trump good system design. That might be obvious, but in at least two circumstances they were very good systems and because the systems weren't selling as well as the producer had anticipated, you know, those ventures were cancelled basically.
So, a lot of times good system design can be trumped, in fact, by the fact that it is not succeeding, you know, for whatever reasons as a business. I want to emphasize in this context, as well as later slides, that good people are the thing that actually make changes happen in the system. So, that is true about the system design. It is also true about actually getting quality outputs from the system. I think we have been successful at IHC as much because people have had longevity in their position as by the fact that we have made good systems because the people are there to be able to make a promise to the clinicians about capabilities and then they stay long enough to actually fulfill those things and it often takes five to ten years to have any impact on the true infrastructure of large organizations.
So, that is a consideration. Patient centered longitudinal records, data sharing from a common repository and I could -- I won't go into the details of all these things, but that basically reduces the cost of interfacing and makes it possible to modularly change out a subsystem and make it possible to migrate to newer applications within a given area without perturbing the whole system.
Common terminology services as a modular part of the architecture and that is really important in terms of maintainability and using the system for decision support. Formal information model for the data that you are storing and, again that comes down to really being able to use the information that you are capturing, have it consistent over the lifetime of the patient and being able to share that outside your institution with public health and other people.
Modular architecture, we, again, believe in trying to purchase the best software for a given purpose and then integrate using standards and common terminology. Again, build decision support in from the ground up. There are important aspects of maintaining decision logic that you want to consider in the upfront design of the system because decision logic is going to change rapidly over time. You want to be able to able to develop it quickly and so you want that to be a modular part of the system and to be a centralized repository of rules and protocols rather than implementing them as part of the applications themselves so that you can change the decision logic without having to redeploy applications or redeploy software.
So, it is a matter of configuring terminology and writing new rules rather than deploying new software when you want to change and do a new kind of report or do a new kind of alert or do a new kind of protocol. Then standards are the future in terms of interoperability. I won't talk too much about that, but it is very important to adhere to standards.
So, a driving assumption in the creation of the systems that we work with at IHC has been that we want to provide vast, least expensive, highest quality patient care that we can possibly provide and our belief is that we can only meet that goal by using computers in appropriate ways in the system. So, that relates to another principle that has come up at least in a peripheral of some of the conversations. Some systems are designed with the assumption that what they are trying to do is bring data to a physician or a nurse, who makes a decision and goes forward. That is a very good thing.
That is certainly better than a written paper chart, but in fact, what we are trying to do is design a system where the system is an active part of taking care of the person so that we can real time do alert. We can have the system an active part of implementing protocols and we find that there are some things we can only do because the system is an active part. So, that leads to, in fact, a somewhat more complex architecture than you would have if your goal was just to bring data to clinicians.
So, the specific kinds of things that we are trying to support are real time decision support. We are trying to do data sharing, which is not only data sharing between different hospitals within our system, the data sharing outside of our system with public health, with clinical trials, multi-centered clinical trials, all that sort of thing.
We are also trying to share decision support and there is no market for that right now because there is no common infrastructure based on which people can share that decision logic, but we feel very sort of passionate about having that ability because it takes so long to develop the computerized protocols that one institution can't bear the cost of doing all of the things that need to be done. So, we are really committed to that, but we haven't been able to find a marketplace or a forum where that can happen yet because of the lack of standards in this area.
Of course, we want to do bio-surveillance. We want to do data analysis. We want to do clinical research and clinical trials and those kind of things. Just so people know when we talk about, you know, real time patient specific decision support, we are talking about alerting reminders, critiquing, protocols, patient management, all these things where the computer is very active in taking care of the patient.
This is just kind of an overblocked diagram of the way our system is set up today. We have a number of ancillary systems that represent a lab system, radiology system, pharmacy system, blood gases, x-ray pack systems, those kind of things. We have inpatient and outpatient systems that are all contributing to that clinical data repository down in the lower center of the system. All of the things that are going through an interface engine and in the interface engine are where things get converted from whatever the representation is in these external modules, external systems, into a standard structured coded representation against which we can do the decision logic and against which all patient care and all patient data access happens. The standard structured coding that is happening by reference to the health data dictionary that is shown in pink there, which in that data dictionary, we have our own concepts that we store in the database, but we have known correspondents, which is mostly one to one to LOINC codes, SNOMED codes, in our case drug codes that are coming from the first data bank, drug knowledge base.
So, kind of a big overview. Just a few statistics. Our health care data dictionary has over 875,000 concepts, 4 1/2 million relationships, 5 million different representations, different names for the same thing and that is supported by ten people, who work again in the corporation to support mapping of the concepts to the ancillary systems, as well as creation of new concepts and new data structures as we need them to store data in the database. We have a team of 26 people that are providing interfaces and we have 60 plus different kinds of interfaces, both HL7 and X12 interfaces. It is roughly a 50/50 split between clinical data interfaces and what I would call administrative interfaces. The X12 benefits claims, all that part, versus XL7 for lab, pharmacy, pathology, microbiology, all of those kind of things.
Because we have multiple hospitals and other things, that 60 different kinds of interfaces, they get replicated because they are talking to multiple institutions. So, we have over 700 actual interface instances and that results in about 3 1/2 million transactions a day. All of those are not against our clinical data repositories. You know, roughly, again, half of those are billing sorts of transactions. So, to just talk about the quality issues, we then instituted a lot of computerized protocols or simple alerts or other kinds of quality measures.
One of the things that we have implemented is, in fact, care related to diabetics. The things that we actually instituted were threefold. There is a report that physicians can pull and what the report shows is for a given physician, the hemoglobin A1C level for his patients, his or her patients, versus the hemoglobin A1C level for all of the patients, all of the diabetic patients in IHC. This spans both inpatient and outpatient environments.
The second thing that they can do or second part of that diabetic is the creation by the system of a personalized diabetic report and so when a diabetic is seen in an outpatient clinic, the report looks at the medications, looks at the hemoglobin A1C. It understands how often certain kinds of tests, such as hemoglobin A1C, ophthalmologic exams and other things are supposed to be done. The computer makes recommendations about what should be done. So, that is easy for the clinicians to follow.
Then the third thing that they can do is pull a report and it shows their patients who are out of protocol, if you will, whose hemoglobin A1C is high or who haven't had a test that they should have had or haven't had an ophthalmologic exam and so it is easy for them to get a list, sort of a to-do list of patients that they need to see or that they need to take action on.
A combination of those things basically you can see over time that the percent of patients that are getting a hemoglobin A1C test has increased over time. So, this is a process measure as were talking about before. Correspondingly, if you look at the number of patients where either their hemoglobin A1C was not measured or it was greater than 9 1/2, the percentage of those patients have come down from 34 percent down to 19 percent.
So, you know, we have seen rather dramatic improvement because of the computerized protocol in this area. This represents a plot of adverse drug events and initially adverse drug events were reported in our institution manually and there was a suspicion that there were tremendously more of those than we were finding. So, we put in computer protocols that watch for -- you know, just watch for high levels, toxic levels of drugs from the laboratory that looked for treatments from pharmacy to treat adverse drug events, that sort of thing.
So, what you see, for instance, is that in the first measurement we have was before we implemented the computerized protocol. So, you see this huge jump. That represents not actually a change in clinical care, but the fact that we were now detecting many, many more adverse drug events than we ever detected before and then the subsequent decline in the number of adverse drug events represents a real change because we implemented protocols, we were able to analyze how these adverse drug events were happening and create systems and changed to the way we provided care, that decreased that number of reactions. So, the rates today, 2004 to 2005, it has remained pretty stable at about 270 per year and, you know, the difference between the previous rates and this rate basically represents about a $1 million a year net cost reduction at LDS hospital along within the institution.
Another thing that we have done is looked at elective inductions for pregnancies less than -- where induction was less than 39 weeks and, again, the computer's role in this one isn't actually -- the computer is not active in implementing this protocol but the data from the computer allowed us to understand what was happening so that we could institute change. So, it was easy to find the data to show that, in fact, we were inducing labor in many more cases than what would be warranted by the clinical condition. By tracking that, you can see this dramatic reduction in the number of patients that induced before 39 weeks. You can break that out and show -- this is broken out by a bishop's score. The bishop's score is a score that gives you an idea of the probability that induction is going to be successful. You want to have a bishop's score of 9 or greater and if you have a bishop's score that is lower than that, then it is unlikely that -- you just have a higher risk that induction is not going to be successful.
So, again, you can see sort of for those people with a bishop's score of 10 and a bishop's score of less than 8 and then the overall decline in the number of induced elective -- induced births. This just represents basically the cost savings that accrue from that.
I won't go into a lot of the detail but especially if you can prevent -- so, one of the complications from improper induction is the fact that you then end up with cesarean section because of the stress of the mother or the fetus because of those early inductions. Of course, if it is the first birth of the mother, then it is very likely that subsequent births are going to also be cesarean sections so you have this -- you know, if you don't do it right the first time, then you have this building of costs that you know are going to happen in the future.
So, again, this just represents the fact that, you know, the cumulative savings since 2001, since instituting this, or roughly $10 million in cost savings to Intermountain Health Care by introducing this particular protocol. Another interesting study and this one, again, is a hundred percent computerized. This represents something as simple as bilirubin testing on babies and, again, you can see, you know, basically a step function change in behavior when we instituted the protocol and it basically just said we want to do a hundred percent of testing of babies to make sure who has high bilirubins and who doesn't.
You can again -- a combination of first measuring and understanding what is happening and then instituting protocols to manage that show basically, you know, a very important decrease in the number of babies that have bilirubin above a certain level. One of the most interesting ones is these counts down here on the bottom are the ones that you really care about because these are babies, whose bilirubin was greater than 25 milligrams per deciliter and those are the babies then that are potential for brain damage or hearing loss or other serious kind of complications.
You will notice that we are actually down to zero here and at census time basically, the last 18 months, we have had no -- we have had zero babies with a high level. So, also this chart is showing basically the rate of readmission of babies with -- and you can see that the readmission rate is dramatically decreased. Now, one of the things that is interesting in terms of the Quality Workgroup, this and another protocol, this quality improvement actually cost Intermountain Health Care money because the cost of testing in sum doesn't compensate for this decreased rate of readmission. Now, IHC continues to do it because, obviously, if we didn't do it, you would have all of this, you would have that continuing set of babies that have either brain damage or hearing loss or other kinds of problems, but it points to, in fact, the need to align in this whole scheme to align incentives with the other programs that are going on.
This is a situation where, in fact, we lose money doing this but it provides better care. There are a number of these others and I have probably gone over my time here, but prophylactic use of antibiotics in surgery, nosocomial infection monitoring, rule-based billing, reportable diseases, clinical research are all other kinds of initiatives that are going on at IHC.
So, recommendations, this is really kind of a synthesis from things I have been thinking about and things that I heard this morning, you know. Just in time processing has some advantages here. You know, I think one of the perspectives that has come to me is so if we had interoperability and everything was perfect in terms of systems that could interoperate in understanding standards and we could export and import data, it still wouldn't equate the data. There is still something else that has to happen for quality and at Intermountain Health Care we have been successful because we have been enabled by a good EHR, but in and of itself it wouldn't have had a quality impact. We have an impact because we have clinicians like Brent James and Dave Burton and there are many others in IHC, who worry about the people process of creating quality improvements. They are the ones who create reports and provide feedback to the clinicians and that work with the clinicians to say now that we have good data about what has happening, what change should we make in the system in order to have an actual increase in the improvement and quality.
I mention again that there is a need to align incentives because the good data that allows you to do these studies takes time for physician and nurses to enter and to get good data and they are not often the -- don't benefit as greatly from the data as some other people do, either the patient or other administrators. So, we really want to try and figure out a way that people who collect the data can get benefit from the data, as well as all of the other people. I mentioned sort of the perverse incentive about newborn bilirubin. The same sort of thing happens with community acquired pneumonia. I didn't show the statistics.
We implemented protocol also around community acquired pneumonia and, again, we lose money on that protocol because what it means is that we send people home who have sort of appropriate scores that will allow them to be treated safely at home and we actually lose money because we send those people home, rather than admitting them. We get more reimbursement -- we would have gotten more reimbursement had we admitted those patients. So, again, there is this sort of perverse incentive in some cases to do the wrong thing because of the way the reimbursement schemes are set up.
So, I think the sum of all of that says how do we initiate change in the practice of medicine and the EHR enables that, that it comes down to people processes. As Larry Weed(?) has argued and I have to agree with him, we can't make perfect physicians. Physicians don't -- can't know and can't remember everything that they need to do and so the systems can compensate but it takes an ever-increasingly smarter group of people to tell the computers what is good and to be able to implement that in protocols that actually change. So, we need research in order to understand what data to collect and what measures are good measures of quality and how to implement those and we need a way of sharing the computerized protocols across institutions when once they are developed and once they have been tried and tested.
So, I will stop there. I apologize for taking more time that I should have.
MR. HUNGATE: Very good. Thank you. Make sure that we get a copy of that to put into hard copy for everyone to have.
Let's shift now. Peter, are you ready?
DR. GEERLOFS: I am.
MR. HUNGATE: Janine is just getting the computer up. Let me take a minute for housekeeping detail while she gets that up.
Who has a train at what time? You have one at 2:30. At 3:00. And you have one at 2:30? Just one. We will watch to make sure that we make that work.
If you have any introductory things to say before your slides are up, Peter, go right ahead.
DR. GEERLOFS: Sure. First of all, I just wanted to say that Stan's comments really are a good segue into one of the things I want to stress in my few minutes here. That is just the notion that an EHR is nothing but a tool, just like a violin is an instrument and the tool is nothing without the processes, the people and the thought that goes into how that tool is used.
So, this industry is such an interesting dance because, frankly, we are sort of alternating between vendors who think they know how the tool could be used, users who use it in ways that are unexpected, which teaches us and then we iterate the tool to make it more and more capable as people kind of expand in their transformed thinking, if you will. So, that is going to be one of the things I wanted to talk about.
Just briefly, I am the chief medical officer at Allscripts and I will talk about Allscripts just very, very briefly. I am a family physician by training, did that for about 20 years and have also been in medical informatics since about 1981.
Do we have the slides up yet?
MS. MC CALL: We do have paper copies in front of us, though. So, feel free to actually speak to them.
DR. GEERLOFS: Sure. Why don't I just start.
Slide No. 1 is just a little bit about who is Allscripts. I am putting it up less from a sales perspective than from the notion that you may or may not be aware that especially in the last year to 15 months, this whole world of electronic health record and physician adoption of it has really kind of turned on its head. There has been a major sea change, tremendous increased interest, tremendous increase in effective utilization of these tools, which means that there has been tremendous learning on the part of organizations, you know, such as Intermountain, as well as the vendor community.
In our own experience, we have 20,000 plus physicians in most of the country, utilizing our tools. Currently, our physicians are writing in excess of a million prescriptions a month electronically, using our tools. I expect this number is going to double quite easily over this next year. We are primarily ambulatory focused and primarily large practice focused. So, our experience to date has been in clinics, typically, the smallest clinic, 20 to 30 physicians, on up to a thousand plus physicians. So, our learning as a particular bent to it because the larger clinics in general have been much earlier to adopt these types of products and they have the wherewithal, if you will, to put processes in place to really use them effectively.
The real challenge as I am sure you are well aware is in the smaller physician market where those resources aren't there. One of the things -- we have really been paying attention to this because it isn't just about the tool. It is how you institute best practices around abuse. We actually just recently published a book. It is available on Amazon, called Electronic Physician, which is a compilation of what we have learned in terms of tricks of the trade around implementing electronic health records in the kind of environment where we are. One other comment about our environment.
We don't have the luxury for our customers to take years to really adapt to these new tools because as a public company, you know, we are expected to be successful and make a certain amount of money each quarter. That is good and it is bad, but the good side is that it has really forced us to say, look, how do we take a group of, say, a hundred physicians, who may not have had a lot of say about the purchase of a system such as this. It may have been done for them. Had a lot of say about the purchase of a system such as this. It may have been done for them, although, you know, more and more certainly clinicians have important input, but often the vast majority of clinicians, who are faced with using a tool such as ours really haven't had much say.
They may or may not be ready but there are various stages of readiness. How do you, No. 1, build the tool? And No. 2, how do you build the processes and the teaching of processes such that in eight months, nine months, no more than a year, you can get that organization up and running. So, we really don't have those kind of luxury of time, which has been challenging but also it has been good in terms of teaching us a lot. So, what I want to do in these few minutes is I want to do two things. One is I want to share with you some of the insights, just a few of the insights that we have learned and then secondly, I want to get very much kind of down to where the rubber meets the road and I want to share with you just one example of an approach that we are taking towards helping our customers improve quality.
I am choosing this particular one because it is not a terribly technically difficult approach. It is not even one that requires many standards or intercommunication. All that is important and we are doing it, but it is so important to recognize that there is a lot of low hanging fruit out there and to the extent that incentives can be created to help physicians adopt these tools, independent of all of the bigger kind of broader things that we want to do nationwide. This is going to have tremendous impact if they are the right tools and if they are implemented correctly.
So, slide 2 talks a little bit about our philosophy.
MR. HUNGATE: Your slides are up now, Peter.
DR. GEERLOFS: Okay. Great. The slide 2 says "Adweenum"(?) at the top and Adweenum is an acronym and it has almost become politically incorrect. Yet it is still the acronym that we sort of use within our company. That is, if doctors don't use it, nothing else matters. That could be substituted for if clinicians don't use it and, frankly, if patients don't use it as we are implementing our personal health record. But, you know, it all really boils down to the relationship between clinicians and patients and in the ambulatory -- in the hospital information system world, you can put in a system and if the docs don't use it effectively, you haven't really totally failed.
You have got a laboratory information system and a pharmacy system and a radiology information system and typically those people use those systems quite effectively. In the ambulatory world, it is all about the docs. If the doc is not effectively EHR, then in effect that organization is not realizing any return on investment, either from a quality perspective or a cost savings perspective. So, this is very much kind of the center of our universe and it has caused us to do a lot of thinking about why it is that docs don't use these systems because, you know, we have had the technology to implement EHRs nationwide for probably 15 years. So, a lot of it is cultural. A lot of it is will.
This whole notion of the phases of technology adoption has actually become a very important sort of learning tool for us. I don't know if you are familiar with it. I will go through it just very quickly. The notion is that there really are three phases whenever any individual human being is introduced to a new technology. So the substitute of phases, basically I can only really understand the new technology from the context of the old technology. So, the classic example is when cars were first introduced what were they called? They were called horseless carriages because people really couldn't understand this new thing, except in the context of the old thing. This is one of the reasons why it has been so important for us to get rid of the term EMR, electronic medical record, because that is a highly substituted term. It really is taking a computer and taking the paper record and putting it in the computer. That is really kind of what an EMR means.
Frankly, the paper record stinks and has major problems with health care for a hundred years and we all know that automating a bad process typically leads to a bad automated process. So, the big challenge, frankly, for vendors and for physicians in general is to as quickly as possible move past the substitutive phase into the next phase, innovation. Innovation is where you kind of have learned to play the instrument a little bit and you start to say to yourself, gee, you know, I bet you I could do something differently, more effectively, more efficiently with this tool.
What we have discovered is that once you get a critical mass of clinicians saying that, they are on their way because they are owning the tool. They are starting to be creative with it and what we find is that inevitably an organization that has a critical mass of clinicians who are innovative within a year, you start to see true transformation happening. Transformation I like to define is you wake up one morning and you realize that you are using this technology in ways that you really could not have dreamed of using it before you had the technology.
Now I say all this because I think it is terribly important. I mean, as policy makers or advisers to policy makers, one always -- I think you are always tempted to kind of work from the top down. How do you create the incentives, which I think is terribly important? How do you create the policy that can drive us in the direction that you want? But I think it is also important to recognize thinking from the bottom up, which is to the extent that we are beginning to achieve critical mass of clinicians effectively using these tools and some of them get to this transformative stage.
We are getting to a level of understanding about how these tools need to work that we could not have imagined, frankly, a couple of years ago. So, really the challenge is getting physicians who first start off past this initial substitute of change. What we have discovered is that there are certain drivers helping them do this. First of all, if you talk to an individual physician and I talk to probably 15 to 30 physicians a week, unfortunately, if you ask them what is most important to them in terms of adopting EHR, almost all of them are going to say it is speed. I need something that is going to speed me up. Very few of them will say I am doing this because I want to improve my quality. That is just a fact of life.
However, if you talk to that same physician a year later and they have figured out how to use this thing and are now being creative, either innovative or even transformative, almost all of their questions are how can I do this in a way that helps me with P for P or how can I do this in a way that helps me track, you know, my hemoglobin A1Cs, et cetera. But the whole interest in that clinician shifts overnight.
So, really my plea, I think, to this group is -- it is like the old Gertha(?) quote that if you really want to move something, you simply have to begin it. In the beginning it is getting this in the hands of as many clinicians as possible and I think you are going to see a lot of magic happen and, of course, now this exact same argument, I think, works for getting it in the hands of as many patients or consumers as possible. So, this whole PHR concept, I think, is going to take off very much the same way that the EHR is taking off.
I want to speak just very briefly to deterrents to adoption because I think there are -- well, I don't know that I want to call them myths or sort of ideas that are floated out there that really in the real world can be challenging and that is that if you take someone sort of outside of the industry directly and say what do you really want in electronic health record and often what they are going to say is they want a highly prescriptive device that in effect leads the clinicians through all the things they ought to do. So, you know, very much prototype driven, i.e., do this step first and then the system is going to go out and check what is going on and using rules and alerts, et cetera, and, you know, I think we very well may get there and certainly there are organizations that have successfully deployed systems that are highly rules based, highly alert and interruptive oriented.
But they can be very, very challenging, especially in our environment where we are really dealing with customers that need to get up and running very, very quickly. So, our view is that the decision support model that we think works the best -- and we certainly use ruled engines and we use alerts where it is absolutely necessary, things like drug allergies and that kind of stuff, but what seems to work the best in our environment has been what we call a referential decision support model. What this means is if you make it really easy for the doc to do the right thing and somewhat harder to do the wrong thing, most docs will very happily do the right thing because it both speeds them up and they get to go home at night, realizing they are doing the right thing.
So, I will give you some examples. That is really what I want to focus on, kind of for the rest of the couple minutes here. So, just an example of this is something that we are newly introducing called Guideline Templates. We used to call them Care Plan Templates. I will try to give you an overview of this quickly to give you kind of a sense of it.
Fundamentally, we have developed in house and have had reviewed across the country and actually by AAFP about 1,200 templates and they are going to grow to about 2,000 by the end of next year. What these are are sort of guideline documents that help lead the clinician through a variety of clinical situations. These are all inclusive. What I mean by that is so many of the sort of guidelines and rules based approaches have just because of the volume have been forced to limit themselves to kind of the big ticket items if you will. Our approach is that we need to create a tool that operates -- I don't care whether you are seeing a child with a cold, an adult with diabetes or, you know, a well child exam or prenatal exam, it really needs to cover the entire gamut of what clinicians do.
So, these templates really help maintenance, acute, chronic disease and they are highly specific so that, you know, the difference between asthma, moderately severe in adults versus asthma, mild in a child, same disease states but obviously the guidelines are very, very different. Second, they are patient centric. Fundamentally, what these are are a tool to enable the clinician to very quickly decide on a care plan and then effectively communicate that care plan with the patient, either in a written document that is a completely customized patient education document, including precautions, call 911 if this and home monitoring, et cetera, as well as a document that could be sent directly through their personal health record, so they have sort of their instructions, personal instructions from their physician on their personal health record.
Then finally they are designed to provide decision support and they provide decision support simply by the way they are organized. I will show you a picture -- I think, is it the next slide? Yes. If you look at the next slide, there is a lot on here. This is slide 4. This is a screen shot and if you just look at the center of the screen where it says "ianotropic(?) agents" and attention to and instructions, diet, lifestyle modifications, basically this clinician has brought up a template on congestive heart failure and when he brings it up or she brings it up, what it does is it has a list of meds, orders, patient instructions, patient precautions, follow-up, what we call health management plan. In other words, what should I be doing on a periodic basis for congestive heart failure? How often should I be getting a chest x-ray? What disease management programs should the patient be in?
In effect, all of this is together in a single document that they can scroll through. Once they do it once and have selected sort of the standard things they want to do, the system remembers this as default or he can specifically set it as default. So, the next time you see a patient, you have sort of your standard template and all you have to do is really say how is this patient different from your standard patients. In doing so, we have discovered that our clinicians can often complete a template in less than a minute and I will share with you in just a second really what that accomplishes.
But there are a variety of other things here that this picture doesn't show effectively. Fundamentally you can reorganize the template in any way. So, with the medications you can have a group that says here are my first line choices. Here are my second line choices of medication. All the clinician has to do is put a check in front of the med. If he wants more information, he can double click the med, basically bring up a monograph on that medication or he can do a very quick search on the scholar. As well, the organization can send to that specific medication item an organizational guideline, which could be just the text, a paragraph or it could be a whole web page that could at the point of care provide information about the right way or approach or protocol, what have you, to order that particular item.
So, probably an awful lot for you to digest, but I wanted to give you a little bit of a sense this is something that isn't technologically difficult to do, really speeds clinicians up and can really drive decisions. Just while we are here, the little green happy faces are formularies. These are drugs on formulary for this particular patient.
Obviously, it is possible to insert the guidelines and the sort of component of this from various stakeholders, not just that enterprise. So, the very last slide and then I will stop, so, as I have mentioned, this is sort of a non-threatening approach that really has had the opportunity dramatically to speed up what the clinician does. As a byproduct of spending that 30 to 40 seconds, the patient gets his problem list updated, his medication list updated, allergies updated, orders and medication list updated.
The assessment plan is automatically dumped into the note and the patient gets this completely customized patient education piece. It drives the health management plan, which is a grid-like control that shows the clinician all of the things that are scheduled for this patient and it automatically captures the most important discrete data. This is an important one because one of the great barriers, frankly, to the electronic health record is folks who say, hey, you have got to use pick lists to document the note.
You have to choose from lists and do what we call a structured note. This can be very, very time-consuming for folks and the truth is that most of the data necessary for research and for quality really comes from the assessment plan. What are the diagnoses and what was done for it? So, using care planning in effect, all of that is captured automatically and the last thought that I will leave you with, we are doing a lot with patient questionnaires where patients do either web-based or kiosk-based questionnaires prior to coming into the dock.
But we have set up the template so that automatically based on what I have selected, the patient can get a questionnaire in a specified time interval that is specific to what I have done. So, as a small example, a lot of patients who were prescribed ace inhibitors get a cough and they stop the meds and you only find that out the next time they come back in. The system can be set up so that every time I write a prescription for an ace inhibitor, the patient automatically transparent to me, the doc, gets a little questionnaire ten days later asking about the cost and with the ability for that information to come back and alert me.
So, I will stop here and hopefully in questions we can go further.
MR. HUNGATE: Very good. Thank you.
Okay. Dr. Rucker, are you ready to plug in there?
DR. RUCKER: My name is Don Rucker. I am with Siemens and maybe give a little bit of this sort of another industry perspective on how do you get statistics around quality, which I think is really the focus of the workgroup and how might you go about that and how might you think about that and some of the things that we have done and some of the tools that we think are particularly -- have some very high potential for really changing the dynamic of this.
I think sort of first what is implicit in a lot of what we have already heard is if you want quality, the best way to find it is not to look for it. That is, you know, minimally counterintuitive, but, you know, the reality is if you look at folks who have changed the world and improved the quality of, you know, their product or service, they have done it in ways that are remarkably different from sort of looking for quality per se.
Here are a couple of case studies of people who have fundamentally changed the way we live in America and really improved the quality of their service, but they have done it entirely through being very clever about process automation. I think that is sort of the focal point of process automation and how do we drive that in health care is the way to go. So, the four little case studies, the first is a very young John D. Rockefeller and Standard Oil and we sort of think Rockefeller as a robber baron, who somehow negotiated out lower rates on, you know, shipping his oil by train, but his real business insight was fundamentally having a uniform system of refineries and guaranteeing the quality of every barrel of kerosene. Right? This was, you know, Titusville, Drakes Oil Well and kerosene in the 1860s and 1870s was an ad mixture of dirt and kerosene.
John D. Rockefeller made his living and his fortune by saying he was going to provide 90 percent pure kerosene. So, I think those are the same numbers that Stan was looking at for the hemoglobin A1Cs, as I believe -- I think we are sort of right there.
Thinking, of course, in the 1870s, that was so rare and so novel, having standards, that he decided to call the company, the little company that he started, Standard Oil. Right? You know, now standard means we didn't get the alloy wheels and the power windows. We are not wildly dissimilar. And, you know, you could say, well, that was manufacturing but in a lot of ways it was actually a service industry.
Henry Ford, I think we all know the standardization there, but it is worth thinking about two more modern pioneers in service industry standardization, which is after all what we are about in health care. This is Mr. Crock, who founded McDonald's as a large business or he bought it as a very, very small business. He set about figuring out how to totally reengineer the making and delivery of hamburgers.
Now, you say, well, what does that have to do with health care? Obviously, present little in some ways, but that fundamental rethinking, that fundamental breaking down, what are the steps? How can we rethink the hamburger? That is I think what we need.
Another more recent pioneer in service industries, I just flew out of the Memphis airport this morning. This is Fred Smith, the founder of FedEx. So, the ability to have certainty, you know, this sort of gets to the IOM quality chasm things, you know, the timeliness. We are willing to pay roughly 50 times as much to get a piece of first class mail, absolutely, positively guaranteed. We still do this. The U.S. Post Office is pretty good these days. You know, first class mail actually -- I mean, when I was a kid -- I won't say when that was, but it was a long time ago -- you know, that was a sort of random event. This week or it might get their next week. Right now, there is no place for mail to hide, with the volumes of mail. So, it actually gets there pretty quickly.
But for that extra degree of certainty, we are willing to pay a small fortune. So, can we identify process automation in health care to drive this underlying statistic because we know that manual quality and manual quality data is just plain too expensive. Now, the problem with process reengineering is, you know, the committee is called the last thing, you know, vital health statistics. Maybe you should rethink the name and call it vital health processes rather than statistics because if you ain't good on statistics, you are sort of a little bit looking in the wrong corner of the world.
I mean, how do we do processes and what do we identify there. I think our target should not be statistics but what are the things we want to change and then work backwards to the variables. In a conversation I had last night with one of our customers down in central Mississippi, this fellow is a vascular surgeon and he was complaining that Medicare essentially pays him on net much less to put in a primary natafistula(?) than to throw some gortex(?) in. Primary fistulas take longer to put in. You know, an AB Fistula for dialysis access, what we are talking about, and gortex, but they almost never clot in the lifetime of the patient, unlike gortex fistulas, which can clot. His declotting business has gone down by 70 percent, you know, apropos that -- acquired pneumonia. The guy is losing two thirds of his business to do the right thing. If we targeted not a statistic about this, that or the other, but statistics, let's say, on primary fistulas to reengineer that process.
We would get the variance reduction, the quality we want. It is very interesting looking at all the things that Health and Human Services has done and Secretary Leavitt and, you know, the regional health initiatives. What is interesting to me is that that vocabulary and those issues are actually almost identical to a whole community of issues in computer science called Enterprise Software Architecture. If you pick up a J2EE, which is a sort of an enterprise software design paradigm, the issues of networking, of semantic interoperability, of security, of huge complicated processes that are not definable up front to reengineer, doing this over multiple sites, over unknown work flows are essentially the same. So, I think one of the things that the committee may want to look at in further work is getting in some people who do enterprise software. I know we have the regional health initiative grants out there that are joined in an experiment on that, but there is a very similar vocabulary and a very similar thought process and it may be a specific benefit just to get somebody, one or two people who are speakers on that area in.
Let me get concrete with a couple specific tools that I think can generate both interesting health care data and process reengineering. The previous speaker talked about that. We still as an industry totally get confused on structured versus free text data.
We impute the benefit of one to the cost of the other and the cost of one to the benefit of the other. You know, as the previous speaker mentioned, you know, the things a really important problem with -- you know, plans are much more important than signs and symptoms. So, I think there needs to be a conscious statement by the committee if it hasn't been and I think you may have already done that on that, there are some clever tools now with XML and other technologies that you can do here. I would disagree with the previous speaker and I don't think this is a complicated sociologic thing to get doctors to change. I am not saying it is easy. It is not easy, but when you look at doctor behavior, I have -- I was one of the principals in building the first Windows-based EMR starting on Windows 2.1 in 1988, not as long as the folks in Intermountain Health Care, but, you know, I have been doing this for a long time. I see very little physician behavior that can't be explained by the amount of time they need to read about -- read the screen, navigate from screen to screen and think about the work flows on that screen.
So, you know, every sort of microsecond of time on screens I think is really what we have to focus on and when you do that doctors are very good at picking up new technology and I think are very good micromanagers of their time. So, I would just sort of shy away a little bit from the grand cultural thing because I think if you make it easy enough people will come.
Other tools, natural language processing. This is a technology that is not quite there yet because natural language processing means we understand what the human brain does. That will be next year. You can today go in and do automated processing of large bodies of narrative text. This is -- some of our researchers have done some work here. There are multiple projects by multiple groups of researchers throughout the United States taking large bodies of narrative data and trying to do some classification of that. And, again, to give you some interesting quality data, for example, our researchers went and for the implantable defibrillators at South Carolina Heart Center, which is a monstrously large cardiology practice, they were looking for people who might be eligible for this device and with a 94 percent accuracy, were able to look at 61,000 patient charts and come up with 300 odd people who benefited from this.
So, I think there are some interesting tools where you don't actually need total agreement on vocabulary and statistics in order to do it. For angina, they were able to identify in an automated fashion who is on aspirin and beta blockers and ace inhibitors, et cetera. As you look on the right here at basically the same level of accuracy as nurse audits, again, very powerful industrial quality metrics that don't strictly require that.
Three more tools to think about. Vocabulary services. I think as a big fan of SNOMED over the years, I think we have sort of done ourselves a little bit of a disservice with those things in the sense that -- and Dr. Huff mentioned this or maybe implied this, maybe I am overinterpreting, but having a large list of many words is really not wildly helpful. It seems wildly helpful, but it is not as wildly helpful as you think because I may have 200,000 terms. I cannot put a menu on a doctor's EMR with 200,000 terms. I am going to get carpal tunnel syndrome by the time I am 1 percent down that menu. So, as soon as I have decided these are the 50 terms, I have semantically changed the nature of the communication because now I am shooting at a pool of 1 of 50 versus 1 of 200,000.
There are some subtle things. So, I would say that what we ought to look at is not just these raw vocabularies with a little bit of characterization of, you know, it is fish, fowl, it is a disease. It is a med. It is a symptom, but really look at richer vocabulary services. We do some stuff where we have very rich mix of a term dictionary, an entire physician ordering knowledge representation that is entirely integrated so you can put in all of these rich behaviors. You can do synonym searches against any part of it and really abstract that as a layer in toto and it can make a large difference in ease of adoption because it allows sites to have a nice separable tool for representing their site specific protocols and behaviors. You can obviously put in national protocols.
It allows us in a very, very subtle way to get around the absolute bugaboo of probabilities and rules and alerts -- the absolute bugaboo of rules and alerts, which is probablistic reasoning. I mean, that is why the AI community in 1980 was not anywhere near successful as those of us who spent time in it had hoped is because human reasoning is pattern recognition and probablistic reasoning. So, when you are doing EMRs, you have to figure out how am I going to accommodate that. That would be a whole talk and then some.
Another set of tools, work flow engines. Microsoft now, for example, in the Visual Studio 2005 has embedded an entire suite of workflow application tools. So, workflow engines you can think of as automated flow charts. Take a visual flow chart and automate each of those steps and tools for that because you can automate the entire process not just sort of the thinking or the alerting around it have been very successful in a number of industrial applications. For example, Intel has gotten rid of all the people in the bunny suits. They used to have Intel inside and people in the silver lame bunny suits.
Now on the Penny and Poor assembly lines, their fab, they hook up 450 different machine tools in a single human hands free automation. That work flow engine exchanges 170 million messages a day to get this and, believe me, they have 27 separate decision support aps(?) to make sure that power point works fine on our laptop by the time we get to it. So, there are some very rich automation processes here and you can say, well, health care is more complicated, but I am not sure that what we do in a hospital is intrinsically more complicated than a $3 billion fab that has to get literally hundreds of millions of items to within, you know, tiny fractions of a micron.
HL7, the 3-0 rim has done great work with process automation. It is maybe too rich for a lot of tastes in terms of the details. It is something that I think NCVHS may be able to lend a very powerful helping hand.
The last of the five tools I wanted to mention, just as sort of a thought process in terms of automatic capture of data, generation of quality, is voice over IP. Our telephone systems, if you go to any hospital or physician office and just, you know, close your eyes for a second, open them up, you will see a lot of people on telephones. You know, whom are they calling? What are the calling for? I remember I was trying to get a hold of, you know, Dr. Carr and folks and consultants and, you know, we mutually spent a lot of time on the phone, you know. You know, paging each other and then we would be in a room and they would be somewhere and we do a lot of that in health care.
That whole Lily Tomlin model, one ringy dingy, you know, point to point, you know, the tin cans kind of -- Alexander Graham Bell, that is actually history now. There is something called voice over IP, rich protocols, which you can decide, do I want two human beings synchronously talking to each other or do I want an asynchronous conversation. There are all kinds of tools to build stuff in. So, again, automation around telephony -- Vocera(?), you are familiar with a company that is out there in Silicon Valley, has done some interesting things there.
So, all of these new technologies I think get us wholesale quality data, takes a lot of analysis. Let me conclude with the simple moron version of quality. This is sort of something I sort of thought about over the last couple of years. I have been involved in redesigning our computerized physician order entry and have been lucky enough to sort of install, you know, some of this handiwork in about 20, 25 of our customers, enterprise customers. So, based on, you know, roughly about a hundred thousand physician orders placed today. The real issue on quality you can often net out, the cycle time.
You can say -- and, you know, the core measures do this. I think you have some mention of that in the committee document. If you want a single statistic in complicated environments on what is working, cycle time is a wonderful statistic. Here is some data published by Ohio State a couple of years ago. Now, you could argue with what the baseline was, but they took their time -- first dose of medications from 5 1/2 hours pre-CPOE, post-CPOE of 1 1/2 hours. Now, you might say that was not the most efficient pre. We have a life span, a four hospital chain in Rhode Island. They took their medication time from when the physician ordered it to when it was in a pixus(?) on the nursing unit from 90 minutes to 10.75 minutes. So, just under 11 minutes. Right?
If you are septic, you need an antibiotic, that is a 10 percent mortality reduction right off -- again, connectivity cycle time. The other thing, Bob and I were talking about cycle time over lunch. The other beauty about cycle time is physicians and nurses aren't trained about process reengineering. You know, some of us do it. We are obviously here because we have a taste for it. It you just say clinicians, this takes two hours. Why does it take two hours? That conversation empowers clinicians to reengineer their thing in a far easier way than saying here is, you know, Gerand's(?) book or here is Demming or here is, you know, some big sort -- Peter Drucker, you know. Here is some big engineering thing.
What are the components that empowers discussion. So, I think NCVHS could come up with a metric of cycle time. Last example, Cincinnati Children's measured by hour how many orders were placed pre and post-CPOE, you can't actually interpret all this data because it is too messy, but the first line, the pre-CPOE was 11:00 a.m. for orders placed. Post-CPOE, so they have all their residents and attendings rounding on wireless with CPOE on wireless carts. So, they are placing orders as they round. Patients are getting x-rays as they are walking down the hall, rather than going back to the nursing station and, you know, finding the chart and writing an order and then someone in the afternoon radiology comes.
They have taken -- this was the first couple weeks. They are now 1 1/2 hours earlier on average for every single order placed in that 400 bed institution. So, tremendous improvements on cycle time.
Let me conclude, the best quality data actually comes not specifically from collecting quality data, but as a byproduct of process automation. I think we owe it to ourselves to be very clever about vocabulary tools and services rather than just saying go to the large national vocabularies. If you want one stop shopping, you know, cycle time is wonderful and, you know, it is an exciting time. HHS has some great initiatives coming up with, you know, all the standardization, you know, the regional health initiatives, which are great, you know, experiments in nature.
Plan B, if none of this works, just pay for it. Lots of performance. Don't pay for five core measures or ten. Those will be gamed. If you pay for hundreds of core measures, people will have to buy IT.
Let me conclude -- sometimes you just need a good picture.
Thank you very much.
MR. HUNGATE: Very good. Thank you.
Dr. Fletcher, another switch on the plug.
DR. FLETCHER: Earlier I mentioned I was chief of staff at the VA here in town and it is kind of interesting to hear people's history in IT. I have been there since 1972. I think I have been in IT since about 1978 or 1979. So, we have been at it for a long time and have enjoyed it. I think the fact that IT has been available to us at that hospital is why a lot of us have refused to leave and I think I am one of those.
The questions that were asked were how does health improvement with widespread adoption -- how does improvement occur with widespread adoption interconnected with EMR? What I hoped to show is that we have a system that is widespread. We have about six or seven million patients that are all connected, all the hospitals, 180 hospitals and about 800 clinics are all connected to the same system and when we pull up a patient, we fundamentally -- and most of the patients, we can see all the data in anyplace on any patient and that is soon to be improved by the common health data repository, which I will be coming into.
Then there are several other questions which I won't go through in detail, but I think you will see that we will be answering as I move along.
The organization of our adoption was from the top down the director and staff became committed to it with the RM chief and then several of us clinicians who had a lot of experience helping to develop this throughout the system were on board, but most important to our committees and I think this is true of most hospitals is the presence of clinicians with high use, but little computer expertise at least initially. Some of those people have gained a lot of expertise and are now innovators in and of themselves as we moved along.
We have used this combination of expert and non-expert advisers, created local ownership whenever possible. I think a system that allows you to customize templates, customize reminders and set overall rules within each service even within a hospital makes the best ability to adopt it because the doctors in those specific areas take ownership of the product. We often package popular situations with unpopular, discharge summaries everybody loves, labs and the x-ray images which we have on all our systems at every place in the hospital. We say that if you don't do order entry, you won't get discharge summaries. If you don't write the notes, we will have to abandon the images.
In truth in our own system that was actually the fact. If we showed ourselves very good at order entry, we often were the first persons in the VA system to adopt images and we were the alpha site, for instance, for that. So, we would actually link these together and gain some compliance. There is a point when you don't have to work anymore at having people adopt it and I think everybody has been talking about that, that when about 60, 70 percent of use is achieved, all the doctors start using it and as a matter of fact will not go back, absolutely will not go back to a paper system because they see so many advantages.
Sometimes it is not speed, although the speed is extremely important, as was mentioned earlier. But they see a lot of other advantages in that they can see the record every time they see the patient, where in the past it was like 50 percent of the time you would actually get the chart when you saw a patient in our outpatient area. We never have that problem now.
Keeping that software intuitive and user friendly is extremely important because although we took time for those of us who were on the ground floor to allow adoption of this whole system, there are residents and interns that come in and in one day, they have to be using the system. So, it must remain intuitive and user friendly and I think all of us find that if it is providing real improvement in patient care, we are for it and favor it fairly strongly. The physicians are in back of it on those circumstances alone.
This is an example of the cover sheet, which I will not dally on any, but it has all the problem lists, medications and a bunch of tabs at the bottom that let you go through and order things, see the notes, see the discharge summaries. This is a note screen and over on the left shows that if we click on reminders, it shows the reminders that are important for this patient on this particular visit. Those reminders will look something like this and our system links the reminders with some action and we document in a plain text what we have done -- that text message that you see at the bottom of the screen automatically goes in the note when we have hit the reminders. So, we can actually solve the reminder problem very, very quickly with documentation.
The record frequently also order right out of the reminder and we can enter data such as the patient's most recent blood pressure right through the reminder. Anything that is important for that we are able to put down. As a matter of fact, the message at the bottom is often patient education, which actually the patient can see if they belong my Healthy Vet Program, which is a personalized health record, see exactly what we intend to have them do.
This is the kind of data that can be rolled back to the group and as you can see, it is a rank order, which I find extremely valuable and useful. The names are at the bottom and you can see immediately who is not doing well over on the right by virtue of the fact that they are over 40 percent of their patients have a hemoglobin A1C greater than 9 or not done. That particular person you will notice on the next slide is over on the left. He is now at 3 percent and that is a three month period.
I happen to have, as Dr. Ortiz will know, personally embarrassed that person by announcing what his numbers were and he claims he did nothing different from one to the other, but this is now 77 patients at a 3 percent less than 9 percent. So, I think seeing the data in rank order with all the other doctors makes the doctor move towards the better, even if it doesn't get him pay for performance.
Pay for performance is even better and you can kind of see that where it is green and blue, they are meeting or exceeding and if you see the previous slide, you can actually see the blue and green have markedly changed in only a three month period. So, we are getting better and better very fast on this particular measure. We need to because it was not good for us in our hospital.
This is a performance measure on the diabetic foot exam and you can see very quickly that the green and blue, while they change in relationship to each other, they are a little green and blue, they really didn't change much on this one. So, we are not making an inroad on that, although we are doing pretty well in general.
We can report that back by month and you can see the Dr. DD -- in this instance we have coded the doctors' names, but you can see that doctor has markedly improved month by month, but was very, very low to begin with, to see that -- for this doctor to see himself low in relationship to his colleagues helped produce that change I am sure.
Additionally, there is a second physician showing a marked change, actually even in the colorectal cancer screen. Colorectal cancer screen, this is patients who have returned the values and have three values registered, not just those that have been -- had the order given or been given the cards, but those that have actually returned it and he is obviously talking his patients into doing it more and more over that period of time.
We are able to pull the data. If we know the database, we can create automatic reports. We can improve our computer entry that way, but most importantly, we can improve patient care as you will see.
Here is the hypertensives in our hospital. Hypertension is defined by the database. Those patients who have had three values, three different days apart, above 140 over 90, are patients who are hypertensive. Those patients who have returned to below that number are listed in green. So that in 1998, we had 33 percent return and over time, gradually improving each and every year we are actually in 2005 up into the 70 percent bracket. We are obviously creating more and more patients, who are in the normal tensive zone, who were previously hypertensive and those that were 160 over 100, you can see on the red bar have gotten markedly improved as time as gone on and we are actually looking at every single patient in our hospital. This is 13,000 patients. We are not just taking a review, an EPRP review. As a matter of fact, we look at these numbers before the EPRP walks through our hospital and say oh, oh, we have got to pick this one up because we are not doing so well. We can anticipate what their review will be. As a matter of fact as time goes on and the electronic record gets adopted nationwide, I think you will be looking at the actual blood pressure values, not some secondary review of the chart because most of the pressures will be in the automatic record.
We can compare Baltimore, Martinsburg and Washington with our system and you can see that in each of the hospitals, the green got better and the red got less on one year to the next and now I am talking about 31,000 patients. All these vital signs are now rolled into the health data repository, which we are now looking at in the VA. That is one of the first things that got put in it and we are looking at a very interesting phenomena, which I will demonstrate in just a second. It looks like in September to January all of us did badly for quality.
You can see the green in Baltimore going to less green in January and you can see the red going to more red. What did we do? What are we doing wrong? Well, one of the questions you have got to answer is what did we do last year from September to January and the nice neat thing about having an electronic health record is you can examine the data over time and in large numbers of patients and you can see in this instance that in March of 1999 we were not so good and we are much better in September. Going from the left hand corner to the right hand corner, overall this curve is improving but every September we are better and every winter we are worse, every summer we are better, every winter we are worse, every summer we are better as we go through the whole chart. That is typically not just with this six month review, but if we take all of the blood pressures done in July and done in January, you can see we have a similar sinusoidal curve. So, we have actually developed the idea that there is a seasonal change in blood pressures which is reflective of these thousands and thousands of patients. These are now about 10,000 patients that go into these numbers.
If we have a different cut point at 160 over a hundred, we have the same affair, that it is different. Dr. Perlin(?), who is our chief of the PHA in the VA, asked a very interesting question as he say this data just a few weeks ago, what happens at 120? I imagine it goes over and above that number as well at that place as well.
Now, for hypertension we can feed back the data to the doctors on what percentage of their patients are severe, moderate or normal and the percentage of drugs they are on. We can do that by patient and you can see that in the diuretic column in the middle, there are many blanks, are encouraging the doctors use more and more diuretics. We can actually see the results of that encouragement.
I notice from August to February we actually got a big jump in diuretic use in the green zone and we actually are using more central acting -- more ace inhibitors, but that fell off and we have to now reenergize our antihypertensive program. Same kind of data can be brought up with LDLs. They can be markedly changed. This is a very short period of time, about seven or eight months. The LDLs below 120 went from 68 to 83. Our new measure is now below a hundred and that number we can very majorly address.
We actually take all the patients who are in orange and in red and bring them back to a special clinic and we very abruptly changed this number. It isn't a large number of patients, who are post-MI in our hospital, but we were able to handle that very fast and change the number for the better.
Some of that is done by, again, another report that goes out to the doctor and you can see at the bottom of the page, that the lowest result in 1995 is encouraging the physician to give a call very quickly and get the patient back on the medications because he is already 187 and 156. He may just stop this drugs to get up to that other higher number.
This is the result. Across the VA, we are benchmarking these performance measures for tobacco, beta blocker, breast cancer. You can see all the way down. We look at Medicare's numbers and we look at the best possible number elsewhere and the VA is on all 18 measures better than any others. It is partly this reminder system inside the electronic health record that allows us to do this.
The interoperability and connectability is very good. We have remote views for our data. This is the web that we use to get out data anywhere the patient has been seen in the VA. ECGs, we can see anywhere that they have been taken. We just map the news, which is a private system but we can map each of the muses to our hospital and immediately see in order all of the electrocardiograms and compare them. A remote image view just started in our hospital. I think we might have alpha tested it once again, but now it is going through the rest of the system where if the patient has an x-ray anywhere else in the system, it will appear in our record and we will see it in order.
We will move in January, I believe, to the health data repository, which contains all the notes from all of the A visits, which we can see now, but it is not as it will be when it will actually be part of that patient's chart as we pull it up.
We saw 560 patients from Gulfport Armed Forces Retirement Home. They were evacuated with the Katrina hurricane up to Washington. We put 16 laptops and three buildings over there, got on line through VPN access and immediately registered the patients and could see the drugs they had been getting either from the site that they had come from because we could still see into the Biloxi VA and the New Orleans VA or into a web site from their pharmacy. So, we could actually look at the pharmacy that dispensed it, see what medications there were, get them on their drugs immediately that evening and very, very quickly.
The patients who had been treated in the VA in New Orleans did not lose their records, none of them, and we could see them immediately while they were even in the process of being evacuated, very, very strong. This is a picture of the electronic health record that the patient sees. He actually sees all of his progress notes on all the discharge summaries and he can thus, if he walks into any other doctor's area and they have the ability to get on the web, he can actually give them access to this information and they can see it.
In addition, he can see his own reminder system so now it is not just the doctor that sees reminders, but the patients that seems them as well. And he can enter in blood pressures, blood sugars, cholesterols, heart rates, weights and you can see on this particular patient that the patient, once he started entering his weight, he gradually brought it down from a very high point.
Also, this same patient was hypertensive and came down to a much more normal level. This patient was perturbating with bursts of hypertension and he would tell me, he would see that the pressure was up and think what was wrong. Did I take too much salt in the last day and often that was the case? Sometimes it was simply he went off his medication. But recording it himself and watching it, he could get into a much more sustained period where the blood pressures were in reasonable areas.
This is the way blood pressures need and must be monitored. You can't just monitor them every three months or every six months, like doctors do. You have to be on a day by day basis to get any effect. This is one of the other ways we have been monitoring weights and blood pressures. In this instance, this patient had an effusion on the left chest. We took it off. It came right back up and then it went back more slowly and all of these weights come right into our own personalized health record but it is because the man steps on a scale in his home. It goes to a server in the -- at that time in VISN 1, now it is in Austin and then into our health record data so that we can follow them along.
We are able to take data that is in the health record and change the way we do care. These are pacemaker patients. We have actually been able to follow when the pacemaker fails and we have changed the calling system to the cut point. We changed it from calling every three months to every one month, once we see the data go down. So, you can take the data out of your database and actually change the care for patients.
You can also in the database analyze things like mortality and see that dual chamber pacemakers do better than single chamber pacemakers. As a matter of fact, we can break it down into the type of device and can see it. So, we will be using data in very unusual ways that we haven't even thought of when we are talking about a piece of paper once we know it is all in one system.
So, we can see that the electronic record can improve the health in many ways. It is obviously doing it we think in our system. The reminders are a major way we are doing that. There is major value in addition to being able to see the data from a remote site if the patient moves or as they often do in our system, move from north to south in the winter, for example. If they carry with them their own personalized health record, that is a window to what we are doing in the VA for anybody else to see. We would hope in the near future everyone else's data will be equally available on such a system. The ability to have the electronic medical record show the physician where he stands in his performance is very easy. I fully adopt the idea that they should be rewarded for good performance.
In the VA, we have got a performance component to our salary coming up in January and we are sort of saying if you have very good performance, we will give you even more money. We are not saying we are going to take anything away, although it looks that way to the doctor. We are changing medical knowledge. As we look at the medical knowledge, we are able to pull off these systems over time. We are obviously, able to get in decision support in a much better way once the electronic record is fully implanted.
Thank you very much.
MR. HUNGATE: Thank you. We need a copy of those slides, too. I hope that we have made provision for that.
DR. FLETCHER: They are here.
MR. HUNGATE: Have you got that already, Janine?
Okay. Good. She says she has a copy of it. So, I think we are all set.
You have got a train to catch. How quickly do you have to leave? Like now or have you got two minutes? Well, let's make sure any questions get addressed in your direction first.
Agenda Item: Panel and Workgroup Discussion
Questions?
DR. RUCKER: Obviously, your docs are very sensitive to sort of the breadth and scope of this data. Is there -- and how do you decide what things to focus on because I can imagine that you have so much data and so many opportunities to focus and you haven't increased the length of the office visit, I am guessing as part of this. How do you decide what to focus on and do you find that if you are not focusing on something, other things fall off? Or is there something that just isn't done because now they are so metric oriented that they are working on these 18 metrics and the rest of it just sort of goes a little bit by the wayside?
DR. FLETCHER: It is a very interesting question and one of the things we tried to make sure happens is that the performance measure is a logical improvement in health. When it isn't, people like Dr. Ortiz and myself holler even at Central Office and say what are you doing with this one. It doesn't make any sense and they not infrequently modify what -- their measure, not infrequently.
MS. MC CALL: Can you describe a little bit more of the structure of the day as well as you go through? Is this something that is done by individuals? Is it done by --
DR. FLETCHER: Well, it is done by -- I am not talking about two individuals who reflect what the physicians are saying back to the central organization, but the performance measures themselves are set centrally but by a committee of physicians. It is not just performance, quality improvement people, although Barbara Fleming is very good at this. John Perlin, who is now our leader, was the chairman of quality improvement in our hospitals. That is why we are, I am sure, so good at it. But very, very early on, John, when he first came into the quality improvement group saw the way the electronic health record was going to take us way up beyond where we had been before and has made that happen.
So, the measure gets set and the VISN directors and the directors are responsible and that responsibility is very specific such that if we are doing very poorly in comparison -- now, this is an interesting point because I think this is the key. If VISN 5 -- there are 22 in the system -- is not near as good as all the other VISNs. We get concerned and we get worried.
Now, VISN 5 may well be better at hemoglobin A1C than any other private organization around. I mean, our numbers, we are trying to get below 15 percent and the mean is about 10 percent and that is very good. The control of blood pressure may be better, but in comparison with the other hospitals, we are not good and we get called on it. As a matter of fact, the measures keep changing in their targets. This year the target will be met when it is 80 percent of the hospitals. So, it continually rises. Every now and then the hospitals don't do well as a group and it levels off. Most frequently, it is just continually, gradually going up. I think that is a large part of why we are doing so well. We are comparing each other with each other. It is important for the VISN director to be as good as the next VISN.
It is important for the director in our hospital to be better than Baltimore and I always look at Baltimore to make sure that I am not going to be hollered at that day. We actually then move it right on down to the practicing physician, but it has to make sense that controlling hypertension is good for the patient, controlling LDL and acute myocardial infarction, less than a hundred reduces acute infarctions.
MR. HUNGATE: This sounds to me like an excellent model of clinician control of the performance measures that self-assessment is enabled against.
DR. FLETCHER: I think the other point that you just made is that the record itself was largely pushed by physicians. In other words, there were physician user groups that define what the record ought to be. We had as our major goal improvement of care of patients and that -- and taking care of the patients rather than where some of them have all the financial incentives of the private sector, did not have that causing problems. Although I think the record speaks best even for those. We do an encounter at the end of our visits, which we then can send right out to the billing process and it helps that as well. So, if you have the records solid and good, all of those other things will follow.
MR. HUNGATE: Very good.
Carol. I am sure Eduardo can take over the question and answering questions.
DR. FLETCHER: That is okay.
MS. MC CALL: To continue on in the decision where to focus, once you have made those decisions about how many different measures are you actually -- is in that set that you are focusing on and actively trying to manage and to understand and to study?
DR. FLETCHER: That is a good question. It is a lot.
MS. MC CALL: Is it ten? Is it a hundred or a thousand?
DR. FLETCHER: I would say overall it may be more like 50, like it is breast cancer, it is cervical cancer, it is colorectal cancer in the cancer area. Ace inhibitors, when the -- less than 40 to be on board, when the patient comes in for congestive heart failure, it is discharged, instructions and so forth. It is hypertension, 140 over 90 and 160 over a hundred. It is in diabetes foot care, an eye exam, hemoglobin A1C and the blood pressures. So, it is -- if you add them all up, it is quite a few, but they fit the patient. In other words if the -- and they only come up if the patient needs it. In other words, if the patient is not a diabetic, you don't see them. That is the beauty of the reminder system is that you walk into a patient's chart, you may have to look at all the things he has and all the problem lists and then try to say, well, I should have gotten hemoglobin A1C because I do it every year.
No, no. All you have to do is hit the reminder and it will tell you if the patient is diabetic and it will tell you whether one has been done in the last year. If it doesn't come up at all, one has been done and it is less than 9. Now, if it is greater than 9, it will come up for you to repeat it every three months because we want it down as quickly as possible. But in actual fact, it comes up in relationship to the patient's basic disease. You will not see hypertensive reminder if there is no hypertension.
You will not see the heart failure reminders if there is no heart failure and so forth and so on. So, it may actually be as the patient walks in the door, your numbers, like 10, for that individual patient as you see than -- and it will be influence of vaccine and things like that. So, when I say 50 or 60, it is not that much of a -- it is not as big a burden as it seems.
DR. CARR: In charting, in the physician charting, how many fields are modular and how many are narrative? In other words, we were hearing about the word search, if you have a narrative, but I mean, your blood pressures obviously, go into a field. I am just trying to get a sense of what the encounter is like in terms of is it physician typing or filling in or how does this go?
DR. FLETCHER: The note that we bring up is, again, a customized template. Some physicians want a blank sheet of paper and other physicians want all the medications the patients are on, the problem list, recent lab values,
x-ray results. In my instance as a cardiologist, I want the stress test and the echo.
Now, I can select them and right click, cut them out of the note. I don't have to keep them in my note, but they are there in the note when I start and then I begin the narrative, which is the patient's current complaints, which I then type in and then what the patient does. If I am using the reminder for some of these things, if I have used that ahead of the note, they go in the note automatically as text, dependent on the box I click. So that things like blood pressure and lab and pharmacy are specific, concrete areas. Things like dysmion(?) exertion and so forth are not -- signs and symptoms are not at the moment. They are in the text.
MR. HUNGATE: Further questions here?
Eduardo, go ahead.
DR. ORTIZ: An important thing, too, though, is so it brings in structured stuff, like problem lists, medications, lab reports, vital signs, the subjective, you know, like the history and physical part, the physical exam, the assessment plan are pretext narrative. They are not structured templates, but you can build your own individual templates. So, for example, if I want to, and I have done this, I have built my own like comprehensive initial H&P template, where I have actually put all that stuff in there. I have got one for low back pain that I have built for myself. You can share it with other people so people can choose to have a structured templated note or not. I think most of the -- I think it is a mix. Some clinicians have built their own templates for their notes because they see certain types of patients regularly. Others just type it in on their own. But it does bring in, as Dr. Fletcher was saying, typical things like problemless medications, labs, et cetera.
DR. RUCKER: But is it correct to interpret that within the note there, other than automatically brought in, you are not filling out structured database fields that you can do, for example, a relational database query on?
DR. ORTIZ: Right. You are not. Exactly. When I see a patient, I am just typing in basically free text. Exactly.
DR. CARR: Just to reflect a little bit, so if we think about the various sort of modules, I mean, one part is the fields that get filled in. A second part is the -- well, the patient putting in it. Then third is the decision support and then fourth, I am wondering about what are the venues where the physicians come together to look at these data and respond to them. You generate those reports and then do they just go out electronically or are there discussions or is there M&Ms or --
DR. FLETCHER: Primary care has a meeting that they give these reports out. We give the reports out to the doctors, but they have a meeting where they go over it and they discuss it with each other. It is quite frequent that someone who is doing poorly just picks up a clue on how to do it better and they immediately jump the next time around. So, they do help each other.
In those meetings, we emphasize who is doing well and point those people out and then have those not doing so well go see them and see if they can't make a better run at it. We present them almost every week, but we have a medical executive conference, which I run. I am chief of staff and this data is given out at that time and that is exactly what we are doing as a hospital.
But as you can see, we can drill down and see what we are doing as individual clinics, as well as physicians.
DR. CARR: Phenomenal.
DR. FLETCHER: Thank you.
MR. HUNGATE: Thank you. Have a good train ride.
DR. CARR: I would just like to emphasize again that it is not -- you know, we have heard a lot today about it is not just one thing and it is not just one moment in time. It is serial data. But here I think we are seeing that it is not just the field, it is not just the data. It is really the aggregating component, the decision support or the decision -- maybe an analytic and then the venue with the physicians.
MR. HUNGATE: Carol.
MS. MC CALL: I have a question and actually this one is directed to Peter. Are you still there?
DR. RUCKER: I am.
MS. MC CALL: Wonderful.
We have heard about a little bit of context for this question, Peter. We have heard a lot about two systems now, both Intermountain, as well as the VA, both of which are a different model. They are a different organizational design and so what I would like to hear about is what you encounter -- obviously, your relationship and your experience is very different because you are a provider of these solutions to physicians that are not part of your company. So, I would like to hear you talk about what you see out there with respect to the processes that they have, whether it is some sort of kind of top down approach for how do we decide where to focus, how do we actually give feedback.
Then related to that, I do have a question about what your system is actually able to tell them. Both of these systems have a built in process and does yours have a process for actually bringing back metrics and reporting or is that something that they do on their own?
DR. RUCKER: So, we are in, I don't know, 160 some odd large organizations now and trying to think about generalizing their internal process for deciding sort of how to use the tool, if I understand your question correctly, sort of how do you use the tools to obtain certain objectives. So, for example -- and one thing that I have seen happen is the tool sort of enables you to instantiate at the point of care the behaviors that you would like to see occur.
So, what it makes possible is, for example, if you have got a monthly meeting, where you get clinicians together and say, okay, what would we like to focus on from the clinical quality perspective over this next month. Let's say they decide that it is diabetes, Type 2 diabetes. What they can do is they can think through what it is that they want to accomplish and large portions of that can be instantiated in the tool so that as they are writing notes or as they bring a patient up and see what we call the health management plan, which is somewhat of a different paradigm for showing the clinician the context of the patient. This is, in effect, a grid that is problem oriented. So, for the problem that the patient currently has in their problem list, this grid shows in effect, everything that has been done around that problem of a discrete nature. So, the bed, what was the last decision around a med, any labs that had been ordered ad hoc, any orders that are scheduled orders, you know, the hemoglobin A1C every three months, for example.
What was the last value of that? One click and they get a graph of all the previous values. When is it overdue. If it is overdue, there is an alert and the ability to order it right and cut it in that spot. So, the point is that what we are trying to do is create a tool where as clinicians become more and more creative around what they want to do with focused disease states as an example, how do they improve their efficiency and quality around a particular disease state.
What we want to do is create a tool that enables them to put I guess one would call it sort of rules, but also the referential information, which is the approach that we prefer, it guides the clinician at the point of care. So, that is the -- I am not sure if that answers your first question, Carol.
MS. MC CALL: Not quite. Let me try it again because what I heard you say is that this is a very kind of point of care decision support driven thing where I can make sure that I am focusing on the right things.
DR. GEERLOFS: As well as, obviously, any electronic health record accumulates information and can report on a population basis. But a lot of our focus is -- the bottom line is garbage is, garbage out. Unless you are collecting the right data, then the reports that essentially any good EHR can create are worthless. So, our focus is on how do you get compliance from the clinicians and how do you collect -- how do you get collect -- how do you get data that is useful without slowing down the clinician?
MS. MC CALL: Understood. So, let's assume for a moment that all of that is delightfully so and happening. How does the practice as a practice decide its focus, No. 1, is that a data driven process? Are you seeing that emerge or as they have been with it over time. Then the second part is how do they manage and see whether or not they are making the changes.
So, it is more a meta analysis. So, how do they go about managing the improvement, the process of improving, which is, you know, what we heard talked about?
DR. GEERLOFS: Sure. Okay. As a commercial system, the system comes with certain standard reports that are kind of pre-canned, that enable you to look at various aspects of patterns of care and as well, it comes with something we call query panel, which is the ability for non-informaticists, if you will, to create pretty sophisticated reports that pull out patterns. So, for example, if the decision is let's take a look at a particular disease state, such as diabetes and then looks at, you know, by physician, what are the hemoglobin A1Cs, what have they been.
So, all of those reports are possible and because we are providing in a way a totally potential tool, each of our customers chooses to approach how to use that tool somewhat differently.
MS. MC CALL: Yes, are you seeing them use that?
DR. GEERLOFS: They often start with the canned reports, which can give them trends around individual clinical performance, clinician performance, as well as trends around specified D(?) states or they can create their own. So, it is a difficult question to answer in a way. Our job is to imagine with our customers all of the possible ways in which they want to envision patterns of care, population care and be sure that the system is structured so the critical data elements can be captured and that, obviously, the reporting can pull it out.
So, we have a little bit less direct influence and even sometimes understanding of how some of our customers are using it. They often surprise us and they go beyond what we had even envisioned. That is how we can do our next level of product.
MS. MC CALL: Because you have a different role, a very different role than what we have heard about from some other speakers today, I was just wondering what you saw out there. You know, how well are they doing?
DR. GEERLOFS: It largely depends frankly upon the culture and sophistication of the organization. They have got a lot of academic medical centers that are very sophisticated and IDNs that are very sophisticated and we have a number of smaller practices, you know, in the 30 to 40 range that are much more focused on collections than on collections than on quality so far, although they are all starting to pay attention, obviously, to the whole pay for performance issue and they are starting to ask the right questions. So, fundamentally we are -- just like I said, substitutive, innovative and transformative around EHR technology. What we are seeing is that we have a wide range of customers who are at different phases, frankly, of understanding how to do this.
As a vendor, one of the things we are trying to do is collect experiences that have worked from among our best clients and share those and kind of create this best practices situation. I am not sure if this is where you are going with this, Carol, but I think that one of the great benefits, frankly, on a national level would be to sort of increase the discussion of what are the -- how to put it -- what are the best practices ways of using EHRs, whose ever EHR to get data out in ways that help practices improve in quality. I think that is an important national discussion to have.
DR. RUCKER: Don Rucker. Peter, I am guessing that the folks who are using your templates, the 2000 templates you outlined, in essence, you know, their use of those sort of guides what happens. I know on the inpatient side, the thing sort of analogous to your templates that we had to do were really work with our customers on what type of order sets they want. Now, we provide a started order set family that, you know, makes it easier for them to edit, but it really is in the large sites it is the governance around the base component, which in an outpatient EMR might or an EMR might be more in the templates on order entry is more on the order sets.
That gets sort of tricky because you have to do a couple things like you have to get buy-in from everybody in the departments. So, you usually use the same administrative hierarchy you are using for other things. You have to monitor that that is kept up. You know, some of our customers are fairly clever. They just put a finite life on every order set and the people have to renew it. So, you know, two years out, the order set just is taken out of the system and then they know that, you know, if people complain, it has been used and if people don't complain, it hasn't been used.
So, I think a lot of -- you know, in these high end systems, a lot of it is just, you know, the substrate that people are working with. If you are just working with sort of a fairly raw H&P template, you are not going to have the behavioral collection and if --
DR. GEERLOFS: I think that is exactly right. People can't expect of the typical customers that they had time to think all of this up from scratch. So, we have a little mantra within the company saying the content is key
-- is king, actually, is what we say. So, by delivering templates and we also do note templates in a variety of other contents -- what this does is it gives people that starter set that you were talking about.
It is amazing how when they see something they can very quickly understand how to make modifications that would make sense to them. But it is a lot tougher for them to start from scratch.
DR. RUCKER: I would think that content delivery is really, really a key part to the appropriate utilization of the tool.
MR. HUNGATE: I wonder if I could stimulate a little conversation between the three of you around the direction of what -- I come from an experiential background at Hewlett Packard, where I learned that we had to do a product three times before we finally got it right. I am very worried about getting the implementation in all of these offices done before we have got it right.
So, the question really is what do we have to worry about in what is called an EHR right now that we will be very happy to be worried about later but very sad if we didn't? Can you give us any -- the three of you give us some help in that content? What are the pitfalls that we ought to be sensitive to when we care about quality improvement being a major piece of what happens from these systems?
DR. RUCKER: Well, that is a hard question. You know, I mean, you can look at it a couple of ways. You can certainly look at it from just what is the quality data we want to measure. You know, it is the hemoglobin A1C. It is the blood pressure. That somehow you can -- that, you can fix after the fact, I think, in probably multiple vendor systems and multiple implementations.
You can fix sort of the data fields you capture. I think that sort of the bigger national question, for example, around office vista is, you know, how much do office practices have to invest in these things and do we want to sort of -- if we push too hard, it is sort of, you know, from the Washington perspective, you are going to force people to buy a lot of stuff that fundamentally they are not very happy with and have a backlash.
My sense is that is sort of part of HHS's worry about Office Vista. You know, I would be curious of Dr. Ortiz's impression on that. I mean, the VA environment is different than an office environment in some, you know, very fundamental senses. So, I think that is the biggest risk because, you know, these are expensive systems for, you know, undercapitalized physicians.
MR. HUNGATE: Yes, they are.
DR. GEERLOFS: I was just going to comment that, you know, one of the promising approaches is the EHR certification approach, which I am sure you are all aware of. Although the first level of certification is going to be fairly elementary, certainly the goal is to ratchet it up. I think that we are going to see certainly -- I mean, I have been doing this for too many years really, but certainly in the last five years, I have really seen a growing sort of coalescence of functionality. Now, how people actually do it, there is still a lot of creativity around that, but in terms of what is expected to be there in terms of functionality, especially the ambulatory EHR that I am so familiar with, has really been coalescing and, frankly, internally within our company, from a strategic planning perspective we are assuming that two or three years from now, that -- maybe five years -- that the fundamental basic EHR is going to be a commodity, that it is going to be standardized and that the real competitive areas are going to be around innovative ways of using content.
There will always be sort of cutting edge, sort of new things that you can do with this, but the basic bread and butter product, we really believe is going to move very rapidly towards commodity. Again, my main message today, if you hear nothing else, is that we really are seeing -- we believe we are approaching a tipping point around this. You have all heard that, but fundamentally what that means is that a very, very large number of clinicians in this nation are going to get to a new level of understanding about what this is all about.
I happen to be an optimist. I really believe that when they get to that new level of understanding through use of these tools, that in and of itself is going to have a huge transformative impact on health care and that is going to be along with all of the other things, you know, that we are doing on a national level.
But I guess I don't worry as much maybe about imperfect systems because I think there are market forces but I think the systems are getting better and better. The ones that are good enough for physician adoption, I have great trust that they are all going to be good enough to accomplish what you want to accomplish.
MR. HUNGATE: Eduardo, do you want to pick up in this conversation with observations as well, then Stan?
DR. ORTIZ: A couple of things. First, if you have questions directed to the VA that you would have asked Dr. Fletcher, I will try and answer them if I can. Second of all, a clarification just because I heard there was some confusion. Dr. Fletcher uses a lot of acronyms because he has been in the government for a long time. They are a second nature to him. So, because he kept throwing around the term "VISN," people may not know what that means. That stands for Veterans Integrated Service Networks. What it is is the VA is divided into regions. So, you have got approximately -- I am not sure what our final count is because it changes depending on different issues. But we have got about 160 to 170 actual hospitals in the country. There are about 850 outpatient clinics that are freestanding outpatient clinics.
Then we have nursing home services, home health care, et cetera. But then we have got 22 VISNs and each VISN is a region. So, our VISN here is VISN 5, which is Washington, D.C., Baltimore, Martinsburg, West Virginia. So, when he refers to a VISN, it is a regional network and they each have a medical director and a VISN director and they kind of work together.
So, in case you wanted to know that. He also said EPRP. EPRP is basically a system that the VA uses for a monthly chart review where people come in and do hand chart reviews. He was kind of comparing the data that we can get from our EHR to people coming in and doing chart reviews on a random number of charts.
So, that being said, the EHR -- I don't know if that is what you wanted me to talk about -- all I can tell you about that -- so, first of all, I know some about that, but I am not an expert in Vista Office EHR, but a couple of observations. One as you pointed out, the Vista system has worked really, really well for the VA, but we are unique. You know, we are not a typical small practice with three or four physicians. So, one of the areas that Vista has not been that strong in is things like practice management. So, that has been one of the issues in terms of when you roll this out in public domain and give it to physicians, is this something that is going to work? We don't know.
It might work in some situations. It may not work because it doesn't meet their needs. I think this is something in evolution right now. I mean, part of the process from what I was at HHS and I am at the VA is that let's -- you know, we know that Vista works well. It is a tried and true system and let's put it out there as something that will help at least alleviate the initial upfront cost of purchasing a system. It is a public domain system. We can put it out for free. That doesn't mean it doesn't cost anything because you still have to maintain it and update it and do other things and there are companies out there that are positioning themselves to be able to do that.
We have got a few pilot projects out there right now where they are putting it into place. They have put it in some Washington, D.C. public health clinics. They are putting it in some places in Louisiana. I am not sure what happened with the hurricane, whether that derailed that temporarily. A couple of places up in -- I think near Stanford, a couple of places. So, there are some pilot projects to figure out how is this going to work, what are the barriers, what works, what doesn't work. So, I think it is something that we are not sure. We just have to wait and see.
But I think it is a nice thing to throw it out there as part of the competition out there for giving away a nice proven system. So, that is kind of what I know, but I am not an expert in that field and there are other people in the VA that know more about that than I do.
MR. HUNGATE: Let me try another cut at another piece of the same question that I just tried to ask. You in your comments, Stan, talked about being able to transfer packages of protocols, decision support, process management tools and make them available in more than institution.
Now, for instance, could a tool developed at Intermountain Health Care get moved to the VA and incorporated in its system. The answer is no. Now, it seems to me that that is a productivity limiter on the system then, that we are not going to be able to afford to develop any decision support tools if we can't do that.
DR. HUFF: Yes. Several comments that I have been piling up here with your questions, just to -- one is I think in this conversation, we need to be careful to talk about the inpatient environment versus the outpatient environment because decision support in those two environments are very different things. In the one case you are doing fairly simple things typically that, you know, in the outpatient environment you are worried about alerts and some diabetic reports and other things that are fairly simple. In the inpatient environment, you are worried about much more complicated things, like weaning protocols from ventilators and in our case extracorporeal oxygenation protocols and cancer protocols that span large periods of time, take a lot longer to develop.
I mean it is fairly simple to do lab alerts or to do drug-drug interactions. That is well-known. It is a simple look up in a list. These other things have time oriented aspects and, in fact, there are lots of art in it rather than science and you have to sort of turn it into science as you develop the protocol because people, you know -- anyway. So, there are enough differences that I think you ought to be careful about to sort of separate the inpatient from the outpatient circumstances as we do this discussion.
So, to come back, one of the things that happens today is that even though you are recording coding data elements, there are sort of two patterns of things going on. A lot of vendors, when they install a system basically have a starter set of stuff and then people go ahead and create their terminology and their order sets and their entirely local terminologies. So, even within the same vendor, they don't have the ability to transfer knowledge because the terminology and the data, they -- you know, and that is a barrier to being able to make fast improvements. It means that even though I develop good order sets, I can't give my order sets to somebody else and have them be executable.
It is not that you don't get some value from it because I can look at what you did and say do I want to do the same thing and then I can reimplement and it saves me time in the design, but I really am doing the reimplementation in the sense that I have to map that to my own terminology, my own software to do it. So, it is one of those things that it comes way late in the game, that you see the pain of not standardizing from the start because it is easier to install it, to just let people make whatever they want. But then when you get to the stage where you want to share, then you pay a huge penalty to try and understand and retrofit the data and make it consistent across all of the sites.
So, that is one of the things that I think, you know, you were asking specifically what would you do? If you could, you would like people to install and actually sort of be consistent from the start and overall it would be a lower cost when you got to the endpoint where you are sharing decision logic then if you just let everybody do everything and then try and ask after the question how to do it.
MR. HUNGATE: So, limiting to the inpatient circumstance. Let's pursue that a little bit. What does it take to -- is it a standards issue in order to define that kind of a content well enough to make it happen? What is at issue here and what has to be described, I guess, is what I am trying to -- because it does sound to me like this is a place where there is a great potential productivity in quality improvement if the development costs can be shared.
DR. RUCKER: I think the heart of it is also that when you do the inpatient order entry, one of the problems is that the target order bowls all have different names. Skip the physician name, but, you know, the thing that you might call the servicemaster or, you know, your sort of charge catalog, I mean, those are not exactly the same thing, but, you know, every hospital probably going back to the 1960s, when they did their first billing system has -- you know, some hospital groups -- and I don't know if Intermountain Health Care has unified their servicemasters. Some have but the actual target orderable is highly unique for better or worse. Maybe that is an NCVHS type of project to say we are going to have a uniform servicemaster, as well as some of these other standards, but I think it is just worth remembering that that is a very tricky issue that I think is lost on most clinicians, who are really thinking of the more clinical display vocabularies or vocabularies based on etiology.
MR. HUNGATE: Okay.
Eduardo.
DR. ORTIZ: This is coming from a committee member, not from my VA hat. I was just going to add to yours is -- and I don't know, since you are the chair, maybe this isn't the right thing, but posing to the panel members to say, you know, not only what does it take, but in a way what are your recommendations to us as a committee that we should do. You know, what should we be doing, like maybe what recommendations should we be making to the full committee or to the Secretary in order to move this forward because this is obviously a very big challenge.
I know that the VA and Partners Health Care and other groups have been talking about this because of the fact that this is such a daunting challenge that it takes so much work and effort to develop these decision support tools, especially in inpatient settings and to make them where you could basically just share them with each other.
We have just gotten in some early exploratory talks. I am not sure how we are going to do this because we think it is a good thing to do. So, anything that you guys, Stan or any of the panel members have in terms of recommending ways that we can move forward on this would be appreciated.
MR. HUNGATE: Thank you for the question.
DR. GEERLOFS: This whole issue of -- I mentioned earlier the notion of the EHR certification group and this is something I know that they are struggling with, this whole issue of to what extent does a, quote, standard EHR have at its core a standard vocabulary because where it all starts are, you know, obviously, the vocabularies that are used internally and then EHR has need of a number of different vocabularies. You know, you have to have the medication list, if you will. You have to have the vocabulary that you are using for a problem list, which it can't necessarily be ICD-9s because, well, I am not sure how -- I need to go into this, but, for example, if you are going to do structured charting, if you are going to create notes in such a way that concepts of that note can be captured, you need a fundamental underlying dictionary.
Large numbers of vendors do this using completely either proprietary or force their customers to sort of make up these lists themselves and there are a smaller number of vendors, ourselves included, who are using standardized vocabularies. We have to use MediComp's MedCin(?), which is the same vocabulary that DOD is currently using.
The challenge comes when -- and there are a number of these out there. So, there is SNOMED. You know, there is probably three or four tools that could possibly be used. One of the great concerns, frankly, is when one of them is legislated over another. Of course, a lot of what is happening going on right now is all of the vendors of vocabularies, certainly MediComp, is busily mapping all of their concepts to SNOMED and, you know, to basically to LOINC and to basically all of the other standards out there.
We happen to think that MedCin is the -- well, let me just say from a vendor's perspective, we have got to serve two masters. One master is going to be this ever-increasing need to be able to both import and export data in ways that other systems and payers can understand. But the other master is the clinician using the system and it turns out that a lot of these vocabularies, SNOMED being a great example, were written for a purpose other than real time use by a clinician at the point of care, which has created a tremendous burden on vendors trying to figure out ways of making it user friendly.
Some people are doing it or getting there. They are chipping away at it, but it is not easy. We happen to have chosen MedCin because it came at this whole thing from the perspective of the clinician, making it easy for the clinician and then from there being able to either map out to other vocabularies or at least it is standardized.
So, this is a really big, complex issue but I do believe that as we move towards a commoditized EHR, if you will, that standard vocabularies within the EHRs and not really supporting EHRs, whose direction it is to force customers to build their own vocabularies, is the right direction.
DR. RUCKER: I think it is also worth pointing out that the vocabularies, they tend not to have a lot of operational terms in them. Right? I think there was the comment made about EMR is a poorer choice of words because it is sort of paving the cow path. But, you know, in an automated world, you know, the terms are not just, you know, diagnoses and symptoms and things. They are really -- you know, there are transaction terms like "deliver this," "evaluate that," "move this," you know, "cut here" type of things. I don't know. I mean, maybe I am missing one, but I don't think there is a lot out there in vocabulary on operational parts of health care. So then you are left with, you know, almost missing the meat on these things that you have to sort of then put around the system with some other type of semantics and you see the struggle on things like the guideline, you know, the guideline interchange format, the GLIF format, take Arden(?) and the rules sort of beyond the curly bracket problem.
I mean this is a heavily researched area, this sort of guideline sharing. I don't think we have a vocabulary of terms the way, let's say, a work flow engine might architect things. There is, you know, B-PEL(?), the business processing engineering language. We probably need some things like that around health care.
I would just point out that that is different than most of the things people talk about when they talk about medical term dictionaries.
DR. GEERLOFS: Although, and, again, I am not trying to put a pitch in for MediComp, but the reason we chose to go with them is they actually do have a whole semantic sort of rules engine around it. They understand meaning and interrelationships, which can create -- basically supports the ability, for example, to do something called intelligent prompting, where you can do differential diagnoses based on the symptoms, the physicals that you put in. So, some are getting closer than others, but I think that what is really important is to not just settle on one, which had a really deep understanding of all the implications.
MS. MC CALL: This is Carol. I want to ask a question but before I do that I want to set a reasonable amount of context here. As I do that, just keep in mind the question, which will be the following. Do you think any of this is possible? Okay? And maybe state it a little bit more. Is this anything that you hear being discussed? So, keep that in mind.
Here are some concepts that I want to bring in. It has to do with something Stan said, creating -- he used the term "market." Can we, in fact, have a market for certain types of discovery, certain things need to move from art to science and it is almost as though you need to create a market if you want to actually transfer from one organization to another, if you want to have some sort of shared learning that is going on.
So, think of that as a market. When you think about that, there is an entire movement of open source. So, when you take the concept of open source, not necessarily the hard technology of that, but the concept of it is that this is -- it is a public good and so those can be done in either completely open or quasi open environments. Take also that idea of what is happening in the gaming industry and these massively multi-player games and there is a concept that is, I guess, sometimes captured with the phrase "harnessing the hive." And harnessing the hive is a way for people, not just the programmers but there is an entire gradient or distribution of sophistication.
Some of the people that play the game are able to customize it, not unlike what we have been talking about here, but other people that play the game at the other end, the more sophisticated end actually can come in and do some real almost programming. Hang on one second, Michael.
Then another concept to bring in has to do with things you may have heard about with foksonomies(?), tags. I can tag something, things that are happening in a flicker and those types of things. So, as we wrestle with vocabularies, MedCin, SNOMED, what is the right thing to call a thing, the idea is that their could be a standard piece and it is, you know, a handful of vocabularies, but there is also another piece that is user driven. It is harnessed by the hive. It is created by the hive. It has come from an open source type of mentality. So, all of that is context.
Now the question then is is anybody talking about this? The technology exists.
DR. FITZMAURICE: I just wanted to have a clarifying -- by harnessing the hive, do you mean harnessing the beehive?
MS. MC CALL: It is the -- harnessing the hive is just kind of code speak for -- h-i-v-e.
DR. HUFF: The thing that is going on and I think this speaks to some of the question is beyond making messaging standards, there is a set of people and the VA has been at the head of this, Ken Rubin and others that are working with the VA have been at the head of saying let's standardize services and what that means is that I could build an application. So, in this example there could be an Intermountain Health Care order entry and if I adhere to the interface and pay attention to the shape of the interface there, I could talk to any body's back end and the VA could have a back end that does all of the order processing and that sort of stuff.
Then what that allows basically is that then like the Department of Defense can adhere to that same interface. They can have an entirely different application and because it adheres to that service interface, it can again, you know, do VA orders and then because, you know, the Cerner(?) back end, for instance, maybe adheres to the back end service interface, then the Department of Defense can talk to a Cerner back end. So, you get into this situation where you can have all kinds of different applications that are adhering to that same service interface and you have got applications that are adhering to a certain interface on the front end and you have got providers of service interfaces on the back end that adhere to that same interface.
You know, these slides are actually part of my slide set but I was already over time. So, I quit talking. But the idea -- and this is a revolution -- this is what has the potential to truly commoditize and revolutionize the way that we build these software modules because what it means on the application side is it means that I can have a one person or a five person company that can build an exquisitely smart, good thing to do who knows what, you know, TPN, you know, to concoct TPN solutions or to schedule my patients or something and I can do that because I don't have to worry about the back end. I have a set of services that say that once I have got the data, I know how to persist the data and so I can create these pluggable modules and I have created a marketplace because people are adhering to these interfaces and I can plug and play with anybody's application.
On the back end, you know, it is probably -- again, on the back end it is probably going to be a select set of people who basically win that back on the back end because it is going to be purely performance, cost performance, you know, adherence to the interface and become more and more a structured set of functionality as defined by the functional specifications that people -- but it truly then has in this, none of these interfaces work if the terminology and things aren't standardized, but to do this, you have got to do more than that. You have to standardize these API to services and make these services available but now it is an entirely different thing because I mean there are so many situations now where you go through installing a system and because the applications are so tied to the back end, you can never change, except in lock step with what one vendor is going. So, your ability to share or the ability to have innovation in a small area is really tied strongly to that sort of -- of that architecture.
This, whether it works or not, I don't know, but I really think it will and I think if we get to that -- No. 1, if we get to the standards and No. 2, we get an understanding in the marketplace of the value and that these things really get implemented, I think it will dramatically change the price, the cost of these kind of systems and make it so much more possible to share good innovation in small areas, much more -- much easier than it is today.
DR. GEERLOFS: I really want to reinforce this message. One of my favorite books in the last couple of years is The Innovator's Dilemma, you know, by Clayton Christianson, who started the notion of disruptive technology, but what he talks about is the normal life cycle of a product. I keep mentioning the term "commodity." In the ambulatory world, not in the hospital world, but I think in the ambulatory world, as I have mentioned, I think we are moving toward the commoditization. The nature of commoditization is that companies initially in order to innovate, innovate through functionality. You know, they have got great products, but as they commoditize, the functionality isn't the innovation. The innovation now is efficiency of updating the product, maintaining the product and so there is a very natural tendency towards compartmentalizing or making products more object oriented, if you will, such that third parties can -- in effect, innovative small companies can help innovate the functionality of the product.
So, what we are seeing in our own company is that the very next version of Touch Work, Version 11, coming out next year, has moved to an object oriented paradigm and our initial motivation to do it is not what I have just said. Our initial motivation is how can we -- the pace of innovation is happening so rapidly, how can we keep up and we can't keep up with the monolithic products. I mean, it will cost us a couple million dollars to test our products every time we do a forward aggression test, when we bring out a new version.
So, by compartmentalizing the product into objects if you will or plug-ins, we can now build better plug-ins and give them to our customers off cycle. The testing is much less, et cetera. But what we are starting to talk about is exactly the same notion, which is, hey, let's just publish the API and create a market out there for certain of the types of plug-ins so that third party companies can build them and, boom, that does create a whole new market. So, I think you have heard from two -- I think it is fascinating actually that, you know, two of us are thinking very much along the same lines and I think this is something that is a --
MR. HUNGATE: That is encouraging that two of you are thinking --
MS. GREENBERG: Can you tell what those initials are?
MR. HUNGATE: What do the initials API stand for?
PARTICIPANT: Application Programming Interface.
DR. GEERLOFS: Think of it as the sort of the back end of the product is where you store all the data and the front end of the product may be where you have the user interface and some of the roles around how that works and an API -- so that the front end could treat the back end as a black box. All it knows is a very sort of narrow set of commands that allow us to talk to what, in effect, is a very complex back end, but it completely simplifies the view of it to that front end.
DR. HUFF: You know, a good analogy that Clem has often used is, you know, just plugs between stereo equipment and instruments. It defines how you can connect, you know, any instrument to any amplifier and it just -- you know, it is a software configuration rather than a hardware configuration, but it says how I can talk to another piece of software in a consistent, well understood way. That is all an API is.
MR. HUNGATE: In the context of our discussion on quality and the decision to buy and EHR, do you have to know what this interface is going to be before you decide what an EHR is for you?
DR. HUFF: Well, these things don't exist yet. Excuse me. They do exist. They are different for every -- if you are doing actually the vendor's applications now, these are entirely different shapes from every vendor and for every application. So, there is nothing that plugs and plays with anything else.
MS. MC CALL: So, the breakthrough here is to say, you know, a lot of this coming together is handled through the interface layer, the API, and that that should be standardized. I want to make sure that I understand it because I think it is an important concept. Now, as I get standardized, is that also where we were talking about vocabularies and I was trying to get a sense of how structured the world needed to be, but also how nimble it could be.
So, are a lot of those translations happening in the interface in your slide?
DR. HUFF: You could do the terminology translations in the middle piece. I mean, one of the reasons that open source software hasn't taken off in medicine is because of the terminology in the data modeling.
MS. MC CALL: That is right.
DR. HUFF: Because, you know, they have all these great widgets where open source as far as the software would work just fine, but if the drug names aren't actually the same or if there isn't some common understanding or common mapping, you know, if the guy up here on the top is talking French and the back end is talking -- in spite of the fact that they can hear each other across the line, you know, they are not going to be able to perform the service you are asking for.
MS. MC CALL: And I have heard -- this is critical and the reason it is is that we heard from Peter earlier that adoption is key, that there is a lot of low hanging fruit, if you can just get adoption up. We also heard that one of the things to watch out for if we are looking for quality -- we heard yesterday about numerator, denominator, data quality, integrity, those types of issues. So, if we get a tower of babel going, then we pay a heavy price down the road. So, it begs the issue of if you need perfect data and perfect data is a barrier to adoption, then how do we thread that. So, it sounds like that this type of thing, this type of API that does a lot of that translation is a key component. Is that a fair assumption?
DR. GEERLOFS: I think it certainly could be and remember also that all data is not created equal. I mean, this whole discussion of data and standards is obviously a huge discussion, but I think it often gets oversimplified in this notion that sort of all data is equal. The truth is that probably about 20 percent or maybe 10 percent of the data generated within practices is the stuff that really needs to be shared. So, that is one issue.
Also, I think that we are probably in many cases not going to end up with any one standard for any type of data. My guess is there will for some time be more than one standard. So, this translation engine -- in other words having a public domain just as an idea, engine to translate among the most common sort of general vocabularies around problems, you know, whether it is SNOMED or MedCin or ICD-9 or what have you. That is the kind of thing that could begin to matter.
But I really believe that this whole push towards standardization of the API is going to be market driven. I think about the data backs and all of these issues. Of all the experiences in the past, I really think what is going to happen is kind of what has happened, for example, with Photo Shop. You know, Photo Shop, in effect, published its API and a lot of its business because of the open API are third party companies that create plug-ins, specific plug-ins. So, I think that a very important discussion is going to be the balance between the market forces that can really drive us towards this and the gentle influence, you know, that can come from government, as opposed to the other approach, which is, you know, attempts to just sort of mandate it independent of market.
I think the reason it has taken so long is partly because that has been the approach. The market is starting to heat up now. So, I think you are going to see more and more opportunity for companies to be out there in leadership and start publishing some standards and some are going to take off and some aren't and before long, you know, who knows, we may be in a much better place.
MR. HUNGATE: All right. I want to give Dr. Rucker one minute here and I think we are done as a panel. Stan has got to go catch a plane.
MS. GREENBERG: I have a question for Stan, if I could.
MR. HUNGATE: Marjorie has a question for Stan.
MS. GREENBERG: To show my complete stupidity or at least elucidate something for me and that yesterday when we were in a different meeting -- we have been meeting continuously for about four days here -- I asked you, Stan, about the vocabulary that you used in the Intermountain system and the relationship with SNOMED, et cetera, and you said that you had sort of a local vocabulary, I guess, that you had built, but that it all mapped to SNOMED. Something like that, right?
DR. HUFF: I said 80 to 90 percent because I am in a hundred percent agreement with -- there is no terminology for orders or for orderable items or for order sets and there are all of these things that you need to make a working system that don't exist in LOINC or SNOMED or anything, that are just silly things like order statuses and process steps and --
MS. GREENBERG: Go to X12. They probably have all those.
DR. HUFF: They are in a different universe.
MS. GREENBERG: That is in a different world, yes.
But my question to you was is the terminology base or vocabulary base that you use, that you have developed at Intermountain, is it kind of an equivalent of something like MedCin?
DR. HUFF: Yes, in some ways, but -- well, MedCin is -- some of the terminologies we use in the health system are really almost identical to MedCin kind of terminology because what it is is a -- MedCin, you know, is truly organized for ease of use by clinicians and for using what they enter as -- to inference and be able to get quickly to an order set or, you know, to what you are doing. We have terminologies like that and that is, again, one of the -- some of the things that are in our terminology that are not in SNOMED or not in LOINC or other standard terminologies..
MS. GREENBERG: Because of the challenges -- and we have heard this and we are hearing it across the pond, et cetera, of just putting SNOMED as sort of the entry tool -- I am not even sure that it is intended to be that, but there are maybe a few -- everyone doesn't have to -- obviously, you started this a long time ago and I would think you would need to add terms locally, et cetera, but everyone doesn't have to start from scratch. I mean, there are some kind of interface or some entry level types of terminologies out there that could give people a head start. They wouldn't have to try building it from scratch.
DR. HUFF: And market forces are doing that, again, already, much more so in the outpatient environment than the inpatient environment. The outpatient environment I think, you know, the combination of people who are using MedCin within products, for instance, you know, with Logishin(?), Logishin has a consistent terminology across all of its implementations as well.
I don't know if anybody has sat down and tried to match up Logishin's terminology to MedCin terminology. There are only so many terminologists in the world. So, actually people are usually pretty familiar with what is going on in the other -- but that is happening much more quickly in the outpatient environment and consolidation of products in terminology than it is happening in the inpatient environment.
You just don't see commonality yet in the inpatient environment between these orderable items and things like that. People aren't sharing at that level yet in the inpatient environment.
DR. GEERLOFS: Just to comment, in the ambulatory environment -- I have not studied the inpatient, but in the ambulatory, we just hired a third party, a company called Apilon(?) to do a review for us because we wanted to create what we call an order concept dictionary. So, we would with all of our templates and order sets, et cetera, it would be based on a standard dictionary and we are looking for who out there could possibly do it.
Interestingly, we ended up -- they ended up coming back and saying that MedCin is the one out there and they actually do -- we have now done the analysis -- they do about 98 percent of everything we need around order terms. So, it kind of surprised us because we haven't been using that aspect of it. So, DOD must know something, I guess.
MR. HUNGATE: Don, any final comments?
DR. RUCKER: I would just throw out when you do these inpatient order systems it gets very complicated very quickly on the interfaces because you have to do all the things like security and log on. You have to have modules for drug checking, you know, drug dose checking, drug allergy checking, drug duplicate checking, extraordinarily tricky. You know, is that next sliding scale dose of insulin a duplicate? You know, what part of the opiate regime of, you know, the patient controlled analgesia and the long acting and the short acting opiates, are they are duplicate? You know, what is the dose checking?
I think that is why these things in the real world get real complicated and if anybody sort of wants a metric of complexity, I would throw out sort of two sort of touch points. One from a very simple point of view is to look at what Jenks(?) and Walters Clewers(?) are trying to do with just simple evidence based medicine and putting simple sort of evidence-based medicine guidelines in and talk with those folks about, you know, what they are trying to do with standardized vocabulary.
The other thing on the complicated side is, you know, they talk about thought experiments, you know, the Duncan(?) experiment, pick up any, you know, ACOG, ECOG, I don't know all the oncology group acronyms, pick up any hemo-therapy protocol, you know, a hundred plus pages, sit there and try to just stub it out in English or, you know, in visual bases, just take a random page and try to represent that protocol, just pick a random page, you know, as a random number generator and you have the cycles and, well, you know, the cell count was down, but now it is up and their body weight was down and their total body weight was up.
You know, the parameters passed on an API, who incorporate all those variables would be, you know, like a thousand -- you know, a thousand items. So, you have to -- you know, everybody has their own tricks on how to simplify the world, but I would just caution the committee that that is a little bit of a very complicated road that seems simpler on the front end than when you are deep in the middle of it.
MR. HUNGATE: Okay. Very good.
I think we need to take a ten minute break and thank the panel and we will come back and do a little workgroup discussion and then we will finish for the day.
DR. GEERLOFS: Thank you all. It was a pleasure.
MS. MC CALL: Thank you, Peter, so much.
[Brief recess.]
Agenda Item: Workgroup Discussion/Strategic Planning -- Next Steps
MR. HUNGATE: We have about 35 minutes to use as productively as we can. The first thing we need to do is agree that it is not going to take me 15 minutes to do any closing remarks. So, we will use the rest of the time for talking about what are our next steps.
I think that you have some other thoughts here, Carol?
MS. MC CALL: Before we get to next steps, I think it might be helpful if we just go around quickly and just each person, just a couple of observations and then kind of go into next steps. There is going to be a lot to digest, but just a couple of what were the key things that struck people out of what we have heard today.
MR. HUNGATE: Fine. We will start at the left and go around.
Michael.
DR. FITZMAURICE: One of the things I learned --and I apologize for being gone for a period in the middle of the day. I had to go to another conference and then back again -- is that templates can be useful in producing quality measures, maybe HEATIS(?) measures, other measures, ahead of the need to report them, rather than the expense of going back to chart review. But a lot of things are fraught with terminology and vocabulary and from the last set of speakers, I learned that decision support isn't as easy as it is when you are just talking from a slide.
I remember maybe six to eight years ago, Jim Simeno(?) and Stan Huff saying that they tried to exchange some Arden(?) syntax decision support modules between their two respective hospitals and they ran into trouble with the vocabulary, that they spent more time than anything making sure that the terms were the same and that their physicians agreed with what those rules were, based upon the terms that they had substituted.
It just was not easy to move from one system to another and a good part of that was the vocabulary.
MR. HUNGATE: Bill.
DR. SCANLON: Well, first, I think I felt more optimistic about the time horizons for us for moving forward today and I guess I am hoping that we didn't get with all due respect to the two panels, we didn't get the wrong impression from them because they could be sort of -- this is their world. They are under the lamppost and they are seeing sort of all this enthusiasm, but I am hoping it is true. I think some of the statistics that were cited suggest that it is. I mean, I think that is a very positive thing.
This afternoon left me a little more concerned because I think of us as external users or demanders. I mean, we are going to be saying to the people that have these systems give us this, give us that and then tomorrow we are going to change our mind and say give us something new. This goes back to what Justine has raised in the past, which is this issue of flexibility or being dynamic. I am not sure about how flexible and dynamic it is. That is kind of the reengineering, the rejiggering of your system when somebody comes up with a new thing that they want, how straightforward that is going to be.
Weight, blood pressure, hemoglobin A1C, these things are all going to pop out with no problem, but when we get down to some other lab values or something or maybe lab values is not a good case, but something else that comes out of a physical that is maybe sort of more difficult. That lessened my enthusiasm in this afternoon.
In that regard, I guess, there is a question of what can be done in terms of trying to encourage some level of flexibility being built into the systems. How do you define a level of flexibility? That is a more challenging task than defining a specific function, saying be able to do X. Instead what we are saying is be able to do the gamut and we can't tell you what the gamut is today.
MR. HUNGATE: Carol.
MS. MC CALL: I guess an echo on the optimism and I guess I am going to take what you said, Michael, about terminology and decision support both being hard. I think there is a third area. There are three main areas that I see that involve massive translation exercises. One is on vocabulary and learning either how to translate very quickly from one language to another or having to have everybody speak English, which is kind of what we are struggling with.
Decision support is also a translation where getting it back is a process that is hard and how do you actually share. Then there is a third area that I don't see anybody talking about and I think it is related to this broad domain of quality, which the context is the interface that Stan talks about is around the raw data and the indicators and, yet, to go from an indicator to a measure is also hard. You know, how often do I want to see a thing in a certain period of time and all of that and that is research before it is decision support.
So, the diagram show everything going into this warehouse. So, that is the end game and yet there is a whole bunch of processes around that in science and art. It is first art and then it is science and I think that there is a need for an API at that point. And it could be maybe an open environment of methodology and science and analytics, but -- and I think one of the speakers today and I think it was David Lansky said, you know, we really need to look at kind of an analytic level as well. I think that there are some contributions that we can make, but there is an API there about how to transfer discovery and knowledge and metrics and findings that is a technical issue and it is also a process issue. I don't see anybody really talking about that yet.
MR. HUNGATE: My morning take away was that the changes of cultures and the resistance to the changes are more serious issues than the specific content of the EHR, which I think is important and very relevant to our content. From the afternoon, I had an interesting short dialogue with Don Rucker just before he left and he said I am not sure all the emphasis on this vocabulary is appropriate and he expressed that as saying that it is hard to use those vocabularies, that they don't necessarily fit with the clinical work flow. So, it is not the clinicians that are looking for these vocabularies. It is people from someplace else. That is an important observation about what we have, I think, universally kind of accepted as a critical variable. But it is not variable, it is not critical from the clinician viewpoint. That is an important limit in what we are dealing with.
So, those are two things that I have picked up that are important, I think, to our agenda.
Deb.
MS. JACKSON: A sense of time, warp speed, how much things have changed in the last three to five years, I heard someone say. Well, that means the next two to three years will make such pivotal difference. Hearing things about the tipping point with -- the analogy I am thinking more about is this window, that we just seem to be so close to, you know, making a difference now when so much is in the design stage.
The participants, the speakers all sounded enthusiastic and ready to jump into it. At the same time, I am concerned about falling over the rocks as we are looking at the vision ahead. What I learned that I wasn't really familiar with was the difference in the ambulatory and the inpatient care. I am very taken by this one slide by Kibbe showing all this volume on the ambulatory and the outpatient where the money is at the other end. I would like to make sure that we are this concerned about developing the architecture that will help develop this horseless carriage in the EHR instead of something that is transformational and in really designing a structure that will make the kind of differences in the dynamics within the health care we would like to see.
DR. CARR: I think what struck me the most today was the complete package that delivers quality and it begins with -- I am turning into Bob, Bob is turning into me, very scary, but it does begin with some sense and roadmap of what the clinicians believe to be quality. This is what we want. The next thing is the electronic data capture and I think the part that I understood better than I have before is the integration of the patient medical record and another feature, of course, is the kind of reminders and the decision support.
So, by having the patient medical record so integrated into the flow of things, clearly rounded out a fragmented picture and having the targeted reminders ensured that what could be made better in the episode of care would happen.
The third element is data manipulation and aggregation, trend evaluation, data display and drill down. I think that without that, we have what we kind of have today is just millions of data elements floating around and a scramble to send them where they need to go when someone asks for them. But the context of really improving care was so palpable with the VA system.
Then finally the -- well, two more things. One is the utilization that you trend it, but then you give it to someone who is accountable and there are consequences. There are expectations going back to the vision. Their vision isn't to be at the average of CMS range. Their vision is to be better than the rest of the country. There are consequences and it sounds like they will have their own internal pay for performance.
Then finally, you know, the interoperability across the entire VA system was very compelling when you think about what it meant with Katrina. So, you know, I am kind of stymied with how we get there, but I can see where there is.
DR. HOLMES: I thought that the two panels for me illustrated the distinction between where this group goes in terms of taking on tasks, the morning session being very kind of general and policy oriented, as opposed to the afternoon group being very detailed.
Another distinction is that perhaps the morning group being more focused on population or the promise of population health; whereas the afternoon group being more focused on individual health outcomes. I continue to be very puzzled, i.e., confused about whether there is a connection between the two or how one would attempt to traverse the areas.
MS. GREENBERG: Thanks for giving us this opportunity. I always find it very interesting. Sometimes I wonder if I was at the same hearing or meeting, but I actually don't today. I agree with everything that everybody has said. I mean, they are all things that I was thinking. The first thing I had written down here was optimism. So, before anyone said anything.
Certainly, I just think in a year, maybe it is the people we are hearing from, but I don't think it is just that. I think there is momentum and I think that one -- you know, it is always risky to make a cause and effect statement, but the fact that the Department and the Federal Government has really kind of started to organize itself around this and try to push it even though they haven't, you know, put a lot of money into it, but I think, you know, may have -- at least it doesn't seem to have pushed people in the opposite direction. You know, I don't know how much that impact has been, but it is interesting to speculate on because we won't know.
People have been calling for leadership by the Department in this area for a number of years, going back to I remember when Secretary Sullivan was here. So, you know, it is time.
I like, I think, Bill and others I think mentioned it, too, felt less optimistic about not just flexibility but interoperability. Still, I find it kind of scary that we may not in the VA system but in our overall health care system or non-system, and I think that is what Don Detmer made a point about that we don't actually have a health care system in the U.S., that we could have a lot of, you know, really good, hopefully much better care going on in individual silos, but that these things won't necessarily talk to each other. They will at least within -- in those silos. I mean, we have had the system where the lab system didn't talk to the radiology system in the same hospital. So, that is at least I think that is really being addressed, but whether these will really be interoperable, even to address the issues of patient care, let alone, of course, to serve the needs of broader community care and population health.
So, I didn't hear a lot that encouraged me about that, but it just might have been that they didn't really focus on that. To me, the case for electronic health records in the -- for all theses things that like the IOM has pointed out is so -- you know, I think it was made more clearly during this week, maybe because Stan spoke twice or whatever, than I have ever heard it made because it wasn't just hypothesizing or assumptions. I mean, there was a lot of evidence, I felt, of just -- I don't know why the patients are putting up with all -- you know, if people really knew how many things you could improve, I don't know why they would put up with the current system because, you know, the elimination -- well, not the elimination but the reduction of errors of timing, of, you know, just everything, it just seemed to be very clearly, you know, displayed.
I really hope that this is getting into the literature, not just into the scientific literature, but into the consumer literature, et cetera. I still don't think, you know, you see much of it at all. You hear a lot about the hospital where they threw the electronic system out because it was driving them crazy. But I mean these are really successes every minute of the day that are making our care safer and people, hopefully, being healthier, reducing errors, you know, getting medications to people quicker, you know, all of this stuff and I don't think that is really well documented.
The fact that it should do that is getting out there, but the actual examples -- and I think if we can think of any way to try to encourage that, getting that more into the literature, more into the consumer literature, more into the -- you know, other literature, not just to influence physicians, but really to influence consumers.
I mean, if consumers know that one car would cut down their mortality rate in half or something, I mean, wouldn't they be buying that car? Probably.
But anyway having said all that because it makes the case so clearly, I come to the conclusion that -- with David Lansky, that the real role, I think, of the National Committee is to be doing what Dr. Brayler(?) asked the committee to do and that was to address the capacity of NHIN and EHRs in the measurement of population health and its role in improving population health. I am reading from Justine's document.
To me, it is -- of course, that is where I come from. I realize that, a population-based agency and function and discipline and everything else, but I really -- I am not sure that we can make as big a contribution in this sort of, you know, one person at a time health care environment because it seems to me that it is happening and, again, that is where to the need to really get this out there and document it. I think there are things we can do, but I think we have got to keep -- nobody else is looking at the population issues really. I don't think anyone else is. So, that is where I came out.
Oh, and as for vocabulary and terminology, what it has to be is sort of invisible to the clinicians but it is all about vocabulary and terminology because that is the only thing that will make things interoperable.
MR. HUNGATE: I don't disagree but it is not perceived as a benefit at the level where adoption is critical.
Go ahead, Susan. You get an option, too.
MS. KANAAN: Well, I would say two things. I always like the big picture talks. So, I was very interested in the contextual remarks that people made in that first panel, the basic system issues and thinking about our document, I think that we need to begin there, even if we end up saying we are not going to address those things. I think that we have to understand that those are the givens.
The other thing that I found very interesting and that for me is one of those kind of thought questions, like one of the speakers mentioned, is how can this value and principal of patient centricity really be implemented on a practical level. It seems like a very, very interesting challenge.
MR. HUNGATE: Eduardo.
DR. ORTIZ: I just jotted down a couple of things. The first thing that I got out of this was that it is really, really important if we are going to move forward with the quality agenda is that we have to be able to share data with each other. We don't do a good job of that and we have to find a way to do that.
The second thing was that we need to do a better job with determining what constitutes good quality beyond the traditional measures that we are using. So not that they weren't good measures, as David was saying, that those are important measures, but if you look at almost everybody, they are all doing the exact same thing. Beta blockers after MI, hemoglobin A1C and it is like we are getting in this thing where everybody is doing the exact same 10 or 15 or 20 or 30 measures, which are important, but we have to get beyond that. I think it is really important.
So, we need to figure out what really constitutes good quality. We need to figure out how do we collect it, how do we do it using these electronic tools so that we can be as efficient as possible.
The third thing was that we need to make sure that incentives are aligned to improve quality of care because that is a theme that keeps coming up over and over and as we saw when Stan was talking about his things, that a lot of times doing the right thing will actually hurt your bottom line because readmitting those patients actually generates a lot more revenue to the hospital than not readmitting them. Keeping them in the hospital longer because you are not managing them optimally for community acquired pneumonia actually generates more revenue.
Until incentives are aligned, it is going to be very difficult to get people to do the right thing.
The fourth thing that I got was that we really need to be able to share medical knowledge and decision support tools. That is such an important thing. You know, with the electronic health records you have got the benefit of having this data available 24 hours a day, seven days a week, anytime, anywhere, but you really get a lot of benefit from it when you can start building in these decision support tools that help you make better decisions, help you manage patients better and it is so complicated, it takes so much work. We have to learn to share it, but it is such a daunting challenge. It is not going to be easy, but we have to figure out a way to do that.
MR. HUNGATE: Gail.
DR. JANES: I would like to cherry pick from what everybody else has already said. Just a few things and they really do reiterate some of the things that have already been said and both of them relate back to standards, maybe because we have all worked so long and so hard around issues of standard vocabularies and standard terminologies, whether we think about HIPAA or some of the things going on within the CHI initiative. It goes on and on.
I tend to think positive. You know, well, we have made a lot of progress. We have worked hard on this, but when I hear talks like the ones that we heard today -- and admittedly I only heard the afternoon -- it kind of brings me up short. Two things struck me and both of them relate back to a more sobering view of standards and perhaps not so much belittling at any level how much we have gained from the work, but realizing as you sort of turn and you face forward and you realize how much work we still have to go to.
As I said, two issues. One was I found the discussion at the end very interesting, some of the discussion that was going back between David and Stan when we were talking about API and this issue of the back end and the front end and the fact that even though we have these standard terminologies that, in fact, these systems cannot talk to one another. That sort of brings us to some of the things Marjorie was talking about, you know, the scenario that I think a lot of people have cautioned us about and that is that despite all the movement around EHRs, that we are still at risk of building a series of silos, which are not interoperable, which cannot talk to each other, despite all the progress that we have made.
So, I thought the discussion about API and like a couple of people, I kind of struggled to get a sense of exactly what that was, but it was clear to me that this was a good thing. I thought that was very interesting and I also thought it was interesting that it got raised eyebrows from all of our speakers, but I definitely got the impression that this was something that was still being -- that was still very much sort of on the cuff, on the edge, something that people were talking about that hadn't been sort of probed that much.
So, that certainly raises questions about whether this is something we might want to think more about. So, the only other thing that occurred to me in listening to the discussions again relates back to standards and to some of the things that Eduardo was saying was I was very interested in sort of again as a contrast to the data that we saw, which I think we have all seen before about how much has been accomplished around the improvement of care, using quality measures and relying heavily on some of the modes of standardization that have been pursued.
Also, the discussion -- and I can't remember who made the comment about the limitations in some of the vocabularies and the terminologies in the fact that at this point they don't really allow us to describe transactions more complex -- to describe some of the more complex activities of medical care particularly well.
Again, as Eduardo was saying, I mean, at some point if you think about the quality measures that we have now, HEATIS measures and the like, a lot of them are switches. They are dichotomous measures. You know, was something -- you describe something in a standardized way and you ask was it done or was it not, yes, no. At some point we are going to want to move beyond that. I was thinking about what David was saying about actually being able to sort of delve into the area of complex protocols and begin to sort of track that.
The message that at least I thought I was hearing was was that the terminologies and the vocabularies that we have now are not well suited to, in fact, move us perhaps into that next realm of quality assessment. Again, I thought that was very interesting and perhaps a bit sobering.
MR. HUNGATE: Okay. We don't have much time. I think that we are not going to get very far in a discussion of next steps here in seven minutes. So, I think we need general agreement to do two things. First is individual work, to go back and think about what would I like to know next in the context of this arena that we are talking about because I think that is the first place to start in terms of each of our heads understanding of what we have heard so far and what we feel like we need to know next.
Then we need to put two or three down in a list and then let's arrange a way to exchange that, which is probably through e-mail, and then use that base of information for a conference call, not trying to do any judgment at that level, but just gathering ideas. Does that make sense? Is that an agreeable approach?
MS. MC CALL: Yes. I would like to add one thing to -- some homework. One thing I have found very valuable is to get not only transcripts and I would also like to get a complete packet of all the materials electronically, which would be very helpful. It allows me to reconstruct and also Susan --
MR. HUNGATE: Transcripts are very valuable.
MS. MC CALL: Susan has also done an excellent job in the past of synthesizing and trying to sum up, losing some of the transcript grain, but still trying to keep the common ideas. I have found that really valuable.
So, if those could be materials that could come into kind of the homework, I would find that -- both of those really good. Then I would like for us to put down what we think would be, you know, maybe more refined thoughts but also what we think possible next actions could be and then come back together in a group and literally plan what we want to do next, share it and then see how we want to proceed.
MR. HUNGATE: Marjorie.
MS. GREENBERG: Well, you know, we have our new policy that we only do minutes or summaries of the meeting when the committee wants it. I assume that you do want that and, of course, Susan often does our minutes for us anyway.
MR. HUNGATE: We depend heavily on Susan.
MS. GREENBERG: So, we will have that. We will do that.
The other thing is I get back to I think it was Bill, who said yesterday that maybe we know more than -- we have got more information than we thought we did. I just think everyone should really -- and maybe this is what you were saying, too -- but should really think about are there really other people we need to hear from? Are there, you know, site visits we need to make? Are there things we need to read? Or is the next step maybe pulling this altogether not with a bow, but in a way that we can then really take out to people in some form and possibly, you know, the regional hearing approach or -- you know, I am saying this knowing what my budget is, but, nonetheless, but, you know, I -- but if there are holes in the information, then I think we need to identify those sooner rather later, schedule a hearing, but just think about that. I mean, I would appreciate even from our planning point if we could really be thinking about --
DR. CARR: Well, I would like to hear from Michael, from AHRQ, because I know at the meeting two weeks ago, I know Carolyn is speaking before the AHIC group and I think there is -- you know, what we are touching on sort of is an interface between AHRQ and Standards and Security. It is sort of an area that touches both of them. I would want to have an understanding of what each is doing and, you know, is there truly a space that isn't being addressed or if there is, is there something that we could partner with AHRQ or whatever.
DR. FITZMAURICE: I would like to offer maybe five thoughts because you heard my thoughts first about listening to the testimony, but listening to you, I got better thoughts than I had before. So, I am kind of echoing back your thoughts.
The first thing if the Secretary were to come in and say, all right, what did you learn from today? What do you have to tell me? I would say, well, first, that there needs to be some focused research on what are some good quality measures, how do we know that they are good? Is it a consensus? Do we have studies that consumers want these measures?
The second thought is how well do these measures that we consider good scale up to population health measures? The third thought, what is it that the consumers want and do higher quality measures give it to them? You know, if I were to go into the hospital to see a doctor, what I want to do is get up on my feet faster. I want to get back to work sooner. I want to climb the stairs better.
Do these quality measures, getting a foot exam if you are diabetic, do they help me do that? I think that it probably does. But does it give the consumer what the consumer wants out of health care? That ignores the prevention. I don't want to be sick in the first place, but we are talking about quality health care being delivered to sick people.
Fourth, I guess the one recommendation I would give back to the Secretary is that the National Health Information Networks and the standard harmonization contracts and the CCHIP contracts should emphasize terminologies and vocabularies for interoperability solutions. The question I would say should there be a lot of mapping between all the terminologies that are out there or should there be some focus on single choices for vocabularies for specific functions?
I don't know the answer to that, but you can spend an awful lot of time pursuing both ends and what works in a pilot. So, let's get some pilots to test these thoughts and get some recommendations back from these contracts.
The final thought is what incentives could be developed and put into place that would improve systems that produce better quality of care measures and better patient outcomes. It is the patient outcomes that we want. We think the quality of care measures are the way to get there. The greater percentage of people who get quality of care, the better we think patient outcomes are.
There are studies that show for a lot of the quality measures that is true. Let's focus on them. Let's find out what it is we know and what it is we don't know. So, it is research. It is giving direction to the existing contracts for interoperable solutions and it is finding ways to make it happen through incentives.
MR. HUNGATE: Okay. Thank you.
I have got a plane that is pushing me a little bit at this point. I think we need a conference call the second week in December. I don't think we want to do the polling here. We want Cynthia to do that through the Internet of the week, the 5th through the 9th of December.
This is a two hour conference call, I would guess. I would like to fuel that with the three wishes, which will involve maybe next steps, but it will also just involve content that you feel needs to be pursued as the starting point for that.
Anything that you can add to that, fine and getting the information out from this meeting to everyone with a brief summary to the extent that you can get it done in two weeks. Not a chance?
MS. KANAAN: I am afraid that won't happen. It will be a month. We won't even have the transcript in two weeks and I have another --
MR. HUNGATE: Let's get the slides out and that will be the content, I think, for that meeting. Now, I think as we individually think about that, there will also be suggestions that occur to you for how we ought to spend the time in that two hours. Feel free to share them in the context of the e-mail communication.
I think that --
MS. MC CALL: Actually I would strongly encourage sharing them in advance so that what we have is an opportunity to digest what other people have thought prior to coming on to a call so that we can talk as opposed to --
MR. HUNGATE: Let's talk the question of mechanics. I am under the belief that Cynthia Sidney maintains the best mailing list, most reliable mailing list of the content of our group. So, I would suggest that we work through her in the communication with each other. We could each have our own mailing list and broadcast to everyone, but I don't think we should do that. I think we should put our information to Cynthia and have her send it out to the group.
MS. MC CALL: There is a practical reality that once she sets the pace, all you have to do is reply to all.
MR. HUNGATE: So, maybe what we want you to do is send an invitation to all of us to send this information out and then we have got the list and we will always reply to all with our list of three and whatever else we want to communicate. Is that okay? Everybody okay with that?
MS. JACKS0N: So that this doesn't come through kind of piecemeal, are we looking at trying to get the responses in within a week or so, all to Cynthia, and then she will do a composite and then send what --
MR. HUNGATE: No.
MS. MC CALL: There are three things that we need to do. One is we need to poll for a date. And that is going to be done by Cynthia. Cynthia is also going to send out electronic material from this meeting. That is No. 2. No. 3, as we each individually are done with whatever it is that we would like to share, things around -- and I love what you did, Michael, in terms of if I had to make a set of recommendations today, what would they be. That is one way to look at it.
Another is to say what would we say our findings to date were. That is another way. There may be other ways. Share what you have through the entire group when you have it because that actually is a greater service to folks. It gives them more time to digest it and we will have a better quality discussion.
MR. HUNGATE: Make sense to everyone?
MS. KANAAN: Would you all like me to send you Michael's five --
DR. FITZMAURICE: I will send it.
MS. KANAAN: You will do it. That is even better.
MR. HUNGATE: All right. We are done.
[Whereupon, at 4:35 p.m., the meeting was concluded.]