[This Transcript is Unedited]
Hubert H. Humphrey Building
300 C Street, N.W.
Room 705A
Washington, D.C.
Call to Order and Introductions
Presentation on PMRI Terminologies: Dr. Walter Sujansky
DR. COHN: Would everyone please be seated? We will be starting momentarily.
Good morning. I want to call this meeting to order. This is the second day of two days of hearings of the Subcommittee on Standards and Security of the National Committee on Vital and Health Statistics. The committee is the main public advisory committee to the U.S. Department of Health and Human Services on national health information policy. I am Simon Cohn, Chairman of the subcommittee and the national director for health information policy for Kaiser Permanente.
I want to welcome fellow subcommittee members, other HHS staff and others here in person. I also want to welcome those listening in on the Internet. I want to remind all those in the room here today to make sure to speak clearly and into the microphone, so those on the Internet can hear our deliberations.
This morning we will discuss testimony so far on patient medical record information standards and terminologies and next steps. We have asked Jeff Blair, the Vice Chair, to lead this discussion. As we commented yesterday, we also want to welcome Walt Sujansky, who is our newly associated consultant, to help us with the process of medical terminologies.
With that, let's now have introductions around the table and around the room. For those on the subcommittee, I would ask, as we always do, if you would note in your introduction if there are any issues coming before the subcommittee today for which you need to recuse yourself.
With that, Jeff, would you like to start with the introductions?
MR. BLAIR: I'm Jeff Blair, Vice President of the Medical Records Institute and Vice Chair of the Subcommittee on Standards and Security.
I don't believe I need to recuse myself from anything, but I do want to identify the standards development organizations that I am a member of, or professional associations that may relate to this discussion. I am a member of HL-7, ASTM, ANSE, HISB. Professional associations that might have an interest in this include HIMSS -- and I have probably forgotten some, but that gives you a flavor of that.
DR. AUGUSTINE: My name is Brady Augustine, Gambro Health Care, statistician, medical economist. I have no conflicts, but people say I am conflicted.
DR. FITZMAURICE: I am Michael Fitzmaurice, Senior Science Advisor for Information Technology to the Agency for Health Care Research and Quality. I am liaison to the National Committee and staff to the Subcommittee on Standards and Security.
DR. STEINDEL: I'm Steve Steindel, Centers for Disease Control and Prevention, Senior Advisor for Standards and Vocabulary, and staff to the subcommittee.
DR. GRAHAM: I'm Gail Graham, Director of Health Informatics for the Department of Veterans Affairs, and representing Dr. Robert Kolodner today.
DR. SUJANSKY: I'm Walter Sujansky. I am an independent consultant. I will be working with the subcommittee on recommendations regarding medical terminologies for patient medical record information.
MS. BEEBE: Susie Beebe of the National Center for Health Statistics, CDC, and staff to the subcommittee.
MR. BLAIR: Are there any others?
DR. BERGLUND: David Berglund, National Center for Health Statistics.
MS. WILLIAMSON: Michelle Williamson, National Center for Health Statistics, CDC.
MR. HASTEN: Philip Hasten from the Health Care Services Board of the Institute of Medicine. I am here as a project director of the patient safety data standards project.
MR. ZEROKOWSKY: Art Zerokowsky, Center for Devices, FDA.
DR. MITCHELL: Corey Mitchell, College of American Pathologists.
DR. MC LAUGHLIN: I am Mark McLaughlin with McKesson.
MS. BOWMAN: Sue Proffitt Bowman from the American Health Information Management Association.
MS. SECHAIN: Nellie Leon Sechain, American Hospital Association.
DR. COHN: Marjorie, would you like to introduce yourself?
DR. GREENBERG: Marjorie Greenberg, National Center for Health Statistics, CDC, and executive secretary to the committee.
DR. COHN: Good morning. With the introductions, Jeff, we are turning the session over to you and Walt to help us move through this information. Do you have a couple of introductory comments you would like to make? I will turn this session over to you.
MR. BLAIR: Thank you, everyone. Let me just say a few words to put this in perspective for those folks who don't have a background in how we are proceeding.
The Health Insurance Portability and Accountability Act in its administrative simplification provisions directed the NCVHS to study issues related to the adoption of uniform data standards for patient medical record information and the electronic exchange of that information, and make recommendations to the Secretary by August of 2000. That report is on the website.
That report wound up setting up a philosophical framework which we are following. It set forth criteria for evaluation and selection of PMRI standards. It set forth ten recommendations, which I won't go through, but basically that is the foundation upon which the subcommittee is acting to select PMRI standards. The first phase of that then followed in February of this year, when the subcommittee made recommendations on PMRI message format standards. That was February 27 of this year.
On August 28, we formally began the process of evaluating and seeing what recommendations or selection of PMRI medical terminologies. We have, I would say, four different open issues that we are dealing with as of this morning. Let me just mention them, then we will decide which ones we will address this morning and how we should proceed.
The first is, I created a matrix with the help of Steve Steindel and Susan Beebe to summarize the testimony. We had 14 testifiers on August 28 to help us define the scope and the criteria for selection of PMRI standards, and tried to come up with some type of a decision framework to assist us in our decision making process. That was sent out to the subcommittee members and staff a week ago. Susie, I believe you have made sure that all that is in our packets as well?
MS. BEEBE: Yes.
MR. BLAIR: As a result of that, there are four different issues in that track which deal with first, what kind of an approach or strategic direction should we take for the scope. We would have to really resolve that before we could go on to the other issues in terms of prioritizing which terminologies would be in or out of scope and what would be our criteria for selection, because the criteria selection might be different for different groups.
Now, that is the framework. In a moment, I'm going to ask Walter -- Walter has given some thought about how we would proceed on that. But let's make life more interesting. To make life more interesting, there are several other things that have happened in the background here. One of them is that there was a meeting yesterday between the NLM and the College of American Pathologists.
We have a representative here from CAP. Steve Steindel, I think you have some information, but I don't know how much we can publicly indicate. So first of all, let me ask Steve if you could -- Steve, or --
DR. STEINDEL: Yes, thank you, Jeff. Yesterday at the bequest of the American College of American Pathologists, the National Library of Medicine met with a group from the College. I don't really know who was representing the college at that meeting. Last week, Betsy Humphries sent an e-mail to the federal negotiating team, advising us that the National Library of Medicine was very responsive to what the College had to say. She indicated that several of our objections were being addressed, and that in the opinion of the National Library of Medicine, it paid to reopen the negotiations. They are reopening the negotiations on a time-limited basis with the end date of January 15. So by January 15, there will either be an agreement in place or we will go ahead with other steps, the federal government will go ahead with other steps.
That is all I know. There will be a phone call among the federal negotiating team tomorrow, and we should have more details at that time. But I think this is very interesting, that the negotiations have been reopened.
MR. BLAIR: We have a representative here from the College of American Pathologists. Do you choose to make any comments to clarify or to add to Steve's comments? If so, come to a microphone, please.
PARTICIPANT: Hi. I actually can't make any more statements about it at this time.
MR. BLAIR: Okay, that's fine.
DR. FITZMAURICE: Can you verify that what Steve said is true?
PARTICIPANT: I cannot, personally.
MR. BLAIR: All right. There are two other items. Those folks who can see can see that Clem McDonald and Kepa Zubeldia of our subcommittee are not here right now. That means we have a bear forum, and we will not have the benefit of their comments. Susie, have you received Clem's comments?
MS. BEEBE: Yes, they are distributed.
MR. BLAIR: Great. So at least we will have the benefit of Clem's comments with respect to the matrix and to the discussion framework. One other item before we go to Walter's suggestions on what we might do to address issue number one. Simon, do you want to include in our open discussions right now the consideration of an additional meeting date in April, or is that something you want to do outside of the meeting?
DR. COHN: I think that probably the issue that needs to be discussed is the work plan, which I saw a copy of yesterday, but somehow didn't receive one at the end of the day. But I understand that there is a work plan, and I think we need to review the various parts of the work plan to make sure that we are comfortable with those being the right steps. I think that at that point should help guide us as to the need for additional hearing dates. That may not be resolved today.
MR. BLAIR: And since we just heard about the SNOMED situation, we won't know the answer until January.
DR. COHN: Let me make a comment or two. Obviously we are in open session at this point. My understanding is that there was a meeting, but any comments or any conclusions to be reached from that are very premature at this point.
I find myself also, as I reflect on the work of the subcommittee -- I guess I would be looking to Walter to see if he can help us think through this stuff. But I'm not sure that the existence or the non-existence of the contract necessarily either helps or hurts us. It is basically still an issue of recognizing that the industry may have access to a low-cost license. The question is, how should it be used, what domain should it function in, or should there be other domains; how can the industry and health care generally make maximum use of PMRI terminologies, and what does the federal government and the subcommittee need to do to assist the industry and the country with that, to help in the range of administrative simplification.
So I think there is a great hope that we can somehow go to sleep one night, and wake up and the problem is solved the next day. I don't think this is the full-scale solution. This is just a piece.
DR. STEINDEL: I agree.
DR. COHN: Steve Steindel has his hand raised.
DR. STEINDEL: Jeff, I have to echo what Simon said. I think this is the beginning. The work of this subcommittee is still very important in identifying those domains that we need to cover, we need prioritization on. Those of us in the federal system who have looked at SNOMED would appreciate those types of comments greatly.
MR. BLAIR: Great. I think that there is a consensus among the three of us, if Brady chimes in?
DR. AUGUSTINE: Oh, yes.
MR. BLAIR: At this point, -- you have your hand up, Marjorie. I overlooked you.
DR. GREENBERG: I know, I was waving at you. I just wanted to add that, as you know, I just returned from the annual meeting of WHO collaborating centers to the family of international classifications. There really is a very keen interest in how we go forward with vocabulary development and use in the United States and internationally. I hope that we will keep that on our radar screen.
As you know, certainly SNOMED International is either being used or being considered in various countries around the world. There is a lot of interest and hope that whatever solutions we come up with will be in an international context. There are other countries who are interested in working with us, certainly the U.K., Australia as two other major English-speaking collaborating centers and countries.
So I just want to keep that on the table. I don't know what the nature of the negotiations are. I know that in the past, we were looking fairly specifically at uses in the U.S., but I hope that we can keep a broad perspective as to solutions and recommendations that will work internationally as well.
MR. BLAIR: Then let me do this.
DR. GREENBERG: Steve wants to react to that.
DR. STEINDEL: A very quick comment on the last comment that you made. Unfortunately, the sole source justification we published in the Federal Register was limited to strictly U.S. use.
DR. GREENBERG: I realize that.
MR. BLAIR: Let me make the following suggestion on how we proceed, and if there are no objections, then we will do it in that manner. I am going to ask if Walter would, for the benefit not only of those on the Internet, but maybe for those who are here in the room that may not have had copies of the discussion framework, if you could, Walter, review for us at least the first issue with the division of those categories, before you begin to wind up making your suggestions on how we proceed, so at least we have a level set on that. Then I am assuming that as that discussion begins, we could begin to then bring in what Clem has sent us as part of our discussions. Is that okay with everyone, if we proceed that way? Walter?
Dr. Walter Sujansky
DR. SUJANSKY: Thank you, Jeff. As Jeff mentioned by way of background, the subcommittee solicited and collected testimony from a number of individuals, there may be 10 --
MR. BLAIR: Fourteen.
DR. SUJANSKY: -- and put them to a series of specific questions regarding the scope of terminologies that the subcommittee should be addressing, as well as the criteria for selecting the terminologies that were deemed to be in that scope. There was a lot of information collected, both in written and verbal testimony. Jeff and others in the subcommittee and the staff have done a very good job in summarizing complex and varied and voluminous testimony.
At this point, the first issue that we need to address is the approach and the method by which the subcommittee will organize and consider the terminologies, in order to select a strategy or a set of terminologies to be recommended as standards. The testimony, the input from the testifiers on August 28 fit into a number of buckets, a number of suggestions that they made about how to define the scope for the terminology effort. I'll go over that summary quickly here.
Essentially, it broke out into four or five categories. The first specified organizing the terminologies by clinical function and/or the kinds of messages that were addressed by the standardization effort. So this included suggestions about domain oriented groupings such as categorizing according to terminologies that address pharmacy terms, laboratory terms, diagnosis and so forth. There were also recommendations about functional organizations regarding the types of tasks that would be supported by the different types of terminologies, for example, nursing tasks versus physician tasks and so forth.
Jeff summarized the strengths and weaknesses of this approach, and I will go over those quickly. The strengths of this approach is that it is a pragmatic and efficient approach, because it develops terminologies that directly address specific functions and uses. It leads to terminologies that fit nicely into a reference information model designed to support messages. It provides messages that are easily stored and retrieved. It appears to be less expensive than a strategy that includes a reference terminology, which will be referred to later.
The weaknesses of the approach according to the document prepared by Jeff are not well suited for decision-support applications that need to use and compare data for multiple domains. It does not appear to address the need for clinically specific outcomes data to improve clinical processes when compared to a multi-axial reference terminology. It does not address the need for clinically specific data to advance clinical research, and interlocking individual terminologies may prove more expensive in the long run than mapping to a single reference terminology.
The second approach has been referred to in some of the comments. The first approach entails organizing the terminologies with respect to a convergent or reference terminology that is the core or first layer. A number of the testifiers advocated this approach, including Kent Spackman and Jim Campbell, Peter Elkin and Colin Price.
MR. BLAIR: Just to be consistent, Walter, since you are mentioning those in that second category, maybe you could go back and touch on the names of the individuals in the first category.
DR. SUJANSKY: Sure. That first category, organizing by clinical functions and/or messages -- and I don't know if this list is comprehensive, everyone who advocated this, but assuming it is, that list included Mark Overhage, Stuart Nelson, David Laroux and Steve Brown.
Again, returning to the second classification, with respect to a convergent or reference terminology as the first layer, this entailed some variation of specifying a single relatively comprehensive quote-unquote core terminology that would be used as an initial primary terminology for accessing and analyzing clinical data, clinically specific data.
The idea here is that there would be other layers of terminology that might include more limited and perhaps functionally specific terminologies that were related to things like bar coding or certain clinical functions such as nursing or emergency departments, those that exist at a separate layer. Additionally, there would be a separate layer for more administration types of terminologies that are used for billing purposes or administrative reporting for disease surveillance and so forth, that were pre-existing terminologies that were required for administrative functionalities. In the center of that would be a core reference terminology that was quite comprehensive.
The strengths of this approach is that it provides a unifying mechanisms for all terminologies. It facilitates clinical decision support to improve quality of care and patient safety, improves clinical outcomes data to facilitate improvements in clinical processes, does this more effectively with lower costs relative to functionally specific terminologies, and it has the potential to improve the accuracy and lower the costs of clinical research.
The weaknesses are the relatively high cost of developing and maintaining a convergent terminology. We have received some information relative to that to date, although inconclusive. There is limited acceptance by the marketplace at this time of candidate convergent terminologies, the vast number of them that exist. This approach does not offer complete coverage for PMRI terminologies, due in part to development costs for additional domains. There are difficulties often encountered when mapping administrative terminologies to such a core reference terminology. Frankly, I think that comment may be extended to difficulties mapping core terminologies to any other terminologies that were developed independently. It creates a challenge of how to harmonize a terminology model with an information model.
DR. COHN: Walter, I don't mean to interrupt, but I will anyway. You mentioned the mapping issue. I saw a recent draft document that had talked about the amount of information lost that occurred with mapping. Do you think we have any data that we can bring to that?
I think that is going to turn out to be one of the central issues we may, as we move down the road, have to be thinking about, since by definition there is no such thing that I know of, of a single comprehensive terminology that handles all potential uses. that doesn't require any mapping. Maybe sometime in my lifetime, but I haven't seen that one yet. So do you have any figures on that? Or is that something you think we will be able to develop as we go along?
DR. SUJANSKY: I think there is some information on that in the literature. I believe some work in that area has been done with respect to mapping SNOMED to ICD-9, for example. And I believe there is some published literature available, or at a minimum some white papers, I don't recall. I recall reading something about that; I don't recall if it has been in the published literature. I don't know if there is a quantitative measure of how much information is lost per se, but that is something we can investigate.
DR. COHN: I want to bookmark that as an important issue later on. Steve Steindel raised his hand.
DR. STEINDEL: Just to address that a little bit, a lot of the testifiers at the session stated that the mapping process should be treated with as much rigor as the development of the vocabulary itself, because of is importance. So I think that can give us an estimate of the cost of mapping.
DR. COHN: I know the cost of mapping is significant, but hopefully once it is done, it becomes add-on year to year changes or moment to moment changes. I think my bigger issue was this issue of information loss, of how much are we sacrificing, and how do we deal with that issue. For example, ICD is a critical piece of all of this stuff, and we need to be aware of all of those parts of the framework. We can't ignore them. That relates both to administrative codes as well as potentially things on the fringes of either a small or large terminology. Once again, there is not a single terminology that is going to handle everything.
DR. STEINDEL: So what you are suggesting is that in the cost-benefit analysis for moving to ICD-10, we should also include the mapping. I am just waiting for somebody to go, no, no. That was a joke.
DR. COHN: I'm glad.
DR. STEINDEL: Intended to be a joke. I do think this is a very important issue to explore. It might be worthwhile holding a hearing session on getting views on that. As Walter pointed out -- and I am not aware that much of the literature that actually addresses the point you were making. I do know in the SNOMED editorial board, and David has been involved with the ICD-9 CM -- or it is just nine, isn't it, not CM itself?
MR. BLAIR: Walter, do you think this is an activity that you would be able to do some research on and report back to us at the next NCVHS meeting?
DR. SUJANSKY: Certainly, yes.
DR. STEINDEL: I think it is something we should explore. In many cases, SNOMED when they have found problems with mapping to ICD codes have just added an equivalent SNOMED code.
MR. BLAIR: Basically, what I have heard from both Simon and Steve here on this, there are two dimensions of this issue. One is the issue of the degree of missing data, and the other is the issue of cost. Did I leave anything out?
DR. COHN: Well, actually I don't think the cost was necessarily the central issue here.
MR. BLAIR: The missing data.
DR. COHN: The missing data. This is one of those key architectural issues. No matter what we do and no matter what goes on forward, I think it is something that we need to be looking at and providing some guidance. I think our responsibility is to figure out what ought to happen and try to mitigate barriers and issues. If people are trying to use clinical terminologies, be they one or many, we need to be looking into this issue of mapping to make sure that they can map. Marjorie, you have your hand raised. I knew if I talked long enough I'd get your hand up.
DR. GREENBERG: Just two things. I know we have a representative here from AHIMA. Sue has been very involved in the mapping process between SNOMED and ICD-9 CM. I don't know whether she wants to say anything in that regard.
The other thing is that at this meeting that I just referenced, we did establish an international work group on vocabularies and their relationships with the family of international classifications. The Australian center, I believe, will be taking the lead as part of this. The work group is actually chaired by the U.K., and the U.S. is going to participate. In fact, David will participate in the work group. I don't know if we have told him that. I haven't seen him since we got back. Australia I think is going to be taking the lead on mapping between SNOMED and ICD-10, so that is activities that I just wanted you to be aware of.
DR. COHN: Sure. I'm not suggesting that we solve the problem today. I am bookmarking this as probably something we are going to want to be looking into a lot further, in keeping with our initial recommendations in the year 2000, where we did identify this issue as something that needed to be handled. Certainly up until recently, we haven't even had the right terminologies to talk about mapping to, but this issue remains whether or not we have that one way or another.
MR. BLAIR: Can we go back? I think you have gone through the first two, and then you have the next three.
DR. SUJANSKY: The next approach is -- and by the way, these approaches, I believe, are not necessarily mutually exclusive. They represent conceptual buckets that the testimony can be put into. But I think you can combine some of these buckets or elements of some of these buckets.
Anyway, the next approach is the development of a web of terminologies blended between information models and terminology models. My reading of this recommendation is that there is a natural and necessary interaction between terminologies and message structures and medical record structures, for example. There is a certain amount of overlap in the information content of terminologies and message structures. That is important to recognize in order to achieve the objectives of PMRI standardization.
This is an approach that was articulated by Chris Shute, especially during the testimony. He advocated what he referred to as a web of terminologies that integrate information models and terminology models. I should note that I was not present at the testimony.
DR. COHN: Walter, just for clarification, are you saying there should be or there was?
DR. SUJANSKY: I'm sorry, there was.
DR. COHN: This was something that there should be better -- what verb were you using earlier on when you were describing this? This is just a recommendation from Dr. Shute?
DR. SUJANSKY: A recommendation that there ought to be.
DR. COHN: I thought you might have said that there was natural coordination between the models and the terminology models. That was why I was looking at you.
DR. SUJANSKY: An example of this kind of issue would be, if there were a term, for example, in a terminology that was family history of coronary artery disease, and in a patient record structure there were a field called family history diagnosis, and in that field there was a coded term called coronary artery disease. Now you have two ways of representing the same information, essentially, and that is always a problem with standardized medical records. So that is an instance of the kind of problem that Chris Shute was thinking about in advocating this approach.
The next approach is to retain the current representation for the scope of PMRI terminologies with modifications. So the current representation was a graphical categorization of many of the available terminologies that I won't describe right now, but it was presented to the testifiers as part of their information packets. I assume it can be made available to anyone who requests it.
MR. BLAIR: I think Susie, you passed out copies?
MS. BEEBE: Yes.
DR. SUJANSKY: The modifications that were recommended included adding and removing certain terminologies from certain categories or from consideration altogether. Some of the terminologies were considered to be message specific codes. For example, the deeds terminology for emergency department is more of a code set, more of a record structure, if you will, with some limited terminology specifications. So that was the recommendation to remove that from the consideration of scope and so forth.
This approach may be the fastest and simplest to characterize the scope of PMRI terminologies. The weakness is that it does not provide a strategic approach to unify, prioritize or address the needs for new PMRI terminologies. The last approach that was gleaned from the testimony was the endorsement of PMRI terminologies used by the government or the private sector. Presumably this includes things like ICD-9, CT-4, NDC and so forth. This particular approach was advocated by Captain Brian Kelly, Michael Beebe and Margaret Haber.
The strengths of this approach are that it avoids any objections from existing developers of PMRI terminologies, which is not to be minimized, I suppose. The weakness is that it offers no strategies or paths that unify, prioritize or address the needs for new PMRI terminologies. One might add there also that it doesn't address the need for clinically specified terminologies, which, the terminologies already used by the government or the private sector, standard terminologies already used, do not provide much clinical specificity. I think most of us would agree on that, although that isn't something that the testifiers necessarily said. That is the end of the summarization, I believe, of testimony in that particular area regarding the scope.
DR. COHN: Steve Steindel, you had your hand up?
DR. STEINDEL: Yes. I have a comment. This concerns the last category. My recollection, and I have looked back on my notes, and Jeff and I have discussed this briefly, is that both Brian Kelly and Margaret Haber were advocating a linkage between the terminologies as well. They did not feel the terminologies could exist de novo. They had to interrelate to each other and map to each other. So there are aspects -- while they were a little bit more pragmatic than the people in the first group, there is aspects of the first group's characteristics in the last group. Michael Beebe was the only one who advocated a specific terminology in the linkage.
MR. BLAIR: Let me just add that I think that if we look closely at that discussion framework, you might find other examples of where it falls short of being completely accurate or completely perfect and, as Walter suggested, not necessarily mutually exclusive. It was kind of intended to help us group our thoughts so that we could examine these things to see what is the right direction to go in.
DR. SUJANSKY: I think there is a very important pragmatic aspect in the last group. I just wanted to point out that there was attributes of the first group in people who presented in the last group.
MR. BLAIR: Thanks, Steve. Walter, did you want to go forward with your suggestions on how we proceed, or did you want to review Clem's suggestions on that first, or afterwards? How did you want to weave this in?
DR. SUJANSKY: It might be useful to -- I actually haven't reviewed Clem's suggestions yet, they just came in this morning. If you feel it would be useful to read that into the record, if you will.
MR. BLAIR: Why don't we do that? Why don't we read Clem's in, because maybe that will modify what you recommend we do.
DR. SUJANSKY: Susie, I'm a little concerned that in the copy I have, it is cut off on the second page, and I'd like to make sure we have the whole --
DR. FITZMAURICE: That is what Clem sent me. I edited throughout using mostly spellchecker, but I didn't want to edit that out, for fear that it could serve to jog his memory that there is another paragraph.
DR. SUJANSKY: In the interest of completeness of information that we have thus far, I will just read what Clem McDonald sent recently. It is a page and a half document. It addresses the organizing of the PMRI terminology within our scope. I quote, this is all HIPAA driven, so I think it would be important to cite the relevant sentences that authorize this work in the law verbatim as a starting point. I believe it would provide some focus and direction.
I would not have classified the speakers into so many categories. I believe there were really only two or possibly three themes. There were the pragmatists and those who proposed targeting low-hanging fruit, most of them hinting at a federated model, but perhaps not specifying that very clearly, without specifying the degree of federation. This would include all of the speakers under your first bullet, as well as those under the third bullet, Bach and Berglund and the folks under the fourth bullet, who said, endure the terminologies already used by the government and the private sector, which is by definition a federate and pragmatic approach.
Point B. There were those who argued for a more central authority and more unified vocabulary, with SNOMED as the central core, namely, Spackman, Campbell and Price.
Point C. Elkins and Shute were more alike in their messages than different, especially if you count the Elkins exposition from the audience. They both addressed the issues relating to the data model and the ties to vocabulary, but they did not espouse a major central coding system, but a tight federation or overarching modelling to keep everything.
I should confess that I am on the side of the pragmatic federate approach that exists at present, with the emergence of RX norm and the existence of ICD-10. CPT is likely to exist for some time, no matter what we decide. The statement that the first approach is not well suited for decision support overstates the negative. The first approach being the grouping of functionally specific terminologies.
Systems that do decision support are accustomed to dealing with some degree of mixed vocabularies, for example, ICD-9, CPT, interest vendors, drug codes, as well as their own local codes. They manage, because these codes do not overlap conceptual space. They can get by stating the logic in terms of the presence of a set of comes from one system or from another system, and/or they do internal mapping.
We have done decision supports under these circumstances for decades. The statement that a federation cannot solve clinical outcomes or advance clinical research really is a criticism of some of the individual code systems that might be included in the hierarchy. One might criticize ICD-9 as being not granular enough for some outcome measure or some advanced clinical research, but this assertion then depends upon what collection of codes is included in the federation.
For example, if Medra or SNOMED was used to represent diagnosis and findings in this federation, it would be hard to make the criticism stick. Further, the statement about outcomes in advanced research fails to consider that many outcomes are measured as quantitative results, for example, ejection fraction for cardiac output, vital capacity for pulmonary function and the number of hospitalizations. Other code standards, for example, HUGO gene names and evolving SNP nomenclatures exist and could join the federation to cover some of the advanced research vocabulary.
One could not necessarily presume that single or multi-axial solutions are the exclusive preserve of either of the models. A federated model could also have axial codes. The real question is between compositional and non-compositional, and that is a very complex question. That is the end of the communication.
I think some interesting points are raised there about the historical use of some of the existing terminologies in real-life decision support systems, and the apparent utility of those systems in practice today.
DR. FITZMAURICE: Walter, as you read it, my interpretation was that this supports pulling everybody together, having a party, everybody pooling their terminologies, their codes, and let's make the best of what we have. What seems to be unsaid is, let's also focus on mapping what we have rather than starting a core reference terminology and working outward.
Have I described -- is that logical, following this?
DR. SUJANSKY: You mean from Clem's comments?
DR. FITZMAURICE: From Clem's comments.
DR. SUJANSKY: Yes, I think to a certain extent, although I don't recall that he advocated mapping the terminologies to each other. But aside from that, I think your characterization is accurate.
But I think there is a big omission there, in the sense that there is a lot of overlap in meaning among the existing terminologies. It is one of the biggest problems of this federated approach. As we all know, there is no shortage of terminology standards, and that creates a problem. So by pursuing an approach that includes all or a lot of the existing terminologies without carving out specific areas or particular terminologies to cover specific areas, that still leaves a lot of room open for overlap and redundancy of representation of the same information, and so forth. I didn't see a solution to that in Clem's comments, but maybe it is not as large a problem.
DR. COHN: I guess I maybe saw this a little bit differently. First of all, I don't think anybody should be against synonyms, which is another word for redundancy. So I'm not sure I am against that.
As I looked at Clem's notes, and recognizing when the hearings were -- and Walter, I'm sorry you weren't there for the hearing, but I think there was a fair amount of posturing going on by various people testifying. So I find myself screening some of the comments made by the presenters.
I think that there may be a distinction going on that we are observing. This is something I have observed working with some of my research colleagues over the years, about whether or not one picks low-hanging fruit and deals pragmatically with things that one can take and make some use of, or one is looking constantly ten years on the horizon for perfection.
Now, of course, what becomes low-hanging fruit can vary from moment to moment. I think we are in a situation where right now, we are in the midst of maybe some transition and uncertainty about what is low-hanging fruit. Hence, you are hearing people talking about low-hanging fruit being XYZ, when it really could be A through Z, or one through seven or whatever. The question is what to do with that in the near term. Versus, some people that were looking far out in the horizon at the perfect united reference terminology that covers absolutely all concepts, et cetera.
I think probably the subcommittee needs to take a middle road on this one. I don't think we are going to do very well if we look ten years ahead. I think we probably need to be pragmatic, maybe not exactly the way Clem was describing it, but we need to do things that are pragmatic and useful in the near term, but obviously we need to make sure that what we do is extensible, and can move toward that grand united whole as we move forward.
I think we all have observed for many years, it isn't that there are so many standard terminologies; it is just that there are a lot of terminologies, and nobody knows which ones are standards and which ones aren't, so people use whatever, and they use their home grown and they use whatever. They also use the Webster Dictionary, which is yet another terminology that we need to deal with.
I think the question is going to be how we make progress in this odd environment right now. I guess as I look at the organization of PMRI terminologies, I recognize the uncertainties of the environment, whether that is really the key question or not. I guess I would address that to Jeff at this point. Do you think that is the key question, or do you think it is more a question of being light on our feet, and as the world gets a little clearer over the next couple of months, figuring out what the world looks like and figuring out how to make it maximum useful and deal with the issues that exist?
MR. BLAIR: I'm not sure that I have a hard grasp on your question. But let me try to reply in two ways. One is my own reply, and the other is to turn it over to Walter, because Walter I think has some suggestions on how we proceed.
The first way I would respond, honest to gosh, Simon and I did not discuss this beforehand, but my thoughts I think reflect very much the comments that you just made, but I would like to put it in slightly different words.
My feeling is that the first two categories that Walter just took us through reflected very important constituencies. They both address needs that are very real. I have a great appreciation for the pragmatist approach, whether Clem identifies himself with that or not. I feel as if it is important that we support, nurture, nourish, the advancement of the pragmatic solutions.
But I feel that the pragmatic solutions very often are based on what is available today and what business needs give the most pressure today. I think that on the National Committee on Vital and Health Statistics, we have a broader view than just what is important to a particular institution, a particular government agency or a particular professional association or a particular standards development organization.
Here is where I am going to mentally slide to the second area. It is consistent with what you have said, but I am going to phrase it somewhat differently. The nation has a major problem, where health care is almost 14 percent of our Gross Domestic Product, and we don't have the data to be able to manage it, to control it.
I go back to our NHII work and the paradigm shifts, and the fact that most other industries have moved into the information age, where they are able to improve quality and add additional services, and do so at lower costs. We don't have that information infrastructure in place to be able to do that.
The pragmatic approach I think is very good for immediate measurable results, but I'm not sure that it is creating the information infrastructure to move us to that new paradigm. I see that second one as more capable of helping us move to a broader information infrastructure. It may be a little bit more expensive, it may be a little more time consuming, it may be more difficult with mapping, it may involve more difficult political issues, but I think we have to do that as well.
So in my mind, I am not looking at category one or category two. I think we have to do both. In my mind, I know I'm very shy, but like Chris Matthews, what do you really think, I really think that Chris Shute wound up saying, which is why I put him in the second category, that there is a value to that information model and there is a value to terminology models. Somehow, I think that our approach in strategy should consider how do we get the best of both of those worlds. I think we need both.
I didn't necessarily want to come to this opinion so quickly, and maybe I will modify it as we go through our deliberations. Simon, does that kind of reflect what you were saying, just in different words?
DR. COHN: Yes. It is good that we are on the committee together. Brady, do you agree? I guess as I am sitting here mulling, we are talking about mapping as one of these big conundrum issues. Another one is this issue of how the terminology model and the other models, the information models, how do they all fit together.
What is it that we are trying to do? Are we trying to insure that a message in HL-7 can be sent? Is that our end goal here? Or are we really talking about decision support, which is a little more -- as I look at maybe what is going to decrease our rate of increase in spending from the current ten to 14 percent a year right now down to something a little more manageable, I have more hope that that might be a piece of the equation.
It is like different end points. I may be being -- I'm not an expert in decision support, but it occurs to me to be slightly different end points that we are seeking here. I see Steve Steindel raising his hand and looking worried.
DR. STEINDEL: I think you just made a very important point that I would ask the subcommittee and if necessary the full committee to endorse. That is, our role is not to endorse terminologies for use in messaging, but for the use in PMRI systems. I think that was our charge under the law, and I think we should make it very clear that we are not looking for the best terminologies to appear in an HL-7 message.
The PMRI standards that we wrote in a letter to the Secretary indeed are not in some cases suitable for HL-7 messaging, though they can be used for it. They more encompass what can be done inside an institution than exchanging between institutions, and HL-7 is designed for both.
So when we were looking at those recommendations with regard to the comprehensive health initiative for the government. We differentiated between those that were suitable for exchange between institutions and those that were suitable for exchange within institutions. I think we need to be aware of that distinction and focus on the selection for terminology outside of the realm of messaging.
MR. BLAIR: Walter, I have been using my dictaphone to capture significant points. Is it possible for you to capture those? You may be able to do a better job of that than I can.
DR. SUJANSKY: Sure.
DR. COHN: Walter, you see where we are going. Do you agree, disagree? One of the questions I have, I'm not a member of HL-7, I don't go to their meetings, but the question is, is there some sort of middle ground also? I don't know what that is.
MR. BLAIR: I don't think of it as a middle ground, because I don't necessarily think of it as a compromise. I'm not necessarily saying that HL-7 should compromise. What I am thinking of is that we really need to do both. Both address a different set of problems, both are focused on different types of things. They can contribute to each other, and they need to work together. So I don't think of it as a middle ground or a compromise; I think of it as both.
DR. COHN: The only reason I bring that up is that I know there is a lot of work going on in HL-7 around standards for electronic health records. This must be one of the central issues that they must be thinking about.
MR. BLAIR: They are just starting on that. A lot of it is being driven quite candidly by the time frame within England, with the national health system. They are trying to wind up coming up with a national health record in an ambulatory environment that they could wind up using. They thought it would be better if they did it within the forum of HL-7.
MS. BEEBE: I wanted to give a few of my ideas, and also ask you some of my questions. Talking about decision support versus messaging, I think Steve Brown said something interesting yesterday about the work that the FDA and his agency along with the NOM are doing with RX NORM.
One of the comments that he said was, don't worry about he decision support, because the computers can do that. So getting the terminology right is what I see working in that model, keeping in mind that the end result is that they do want to be able to use the terminology for the decision support level at the application level.
As far as HL-7, what I do know also, beyond the electronic health record that is going on there, there is mapping that is going on between the HIPAA X-12 transactions and the RIM. So I guess I say that as a suggestion that when we talk about mapping and all the effort that there will need to be, there will be SDOs and others doing that already that we could probably leverage. So that said, --
DR. SUJANSKY: Let me make a couple of comments on the various discussion points that have been raised. First of all, apropos of Clem's comments and some of the followup comments related to that, I think that it is important for the subcommittee to step back and think about, consider and articulate the purpose of standardizing clinically specific terminologies.
I have some thoughts on that that I can go over in a moment, but Clem seems to suggest, and I may be misinterpreting it, that there seems to be an implication that these terminologies are necessary for institution-specific decision support and reporting and so forth. He is making the case that for that, terminologies already exist. We are using them at our institution, and there are a number of other institutions that we are familiar with that have sophisticated decision support and reporting and surveillance and tracking and so forth systems in place. They use some terminologies for that, and it works. It has been demonstrated in the literature and elsewhere that these systems are effective and have positive return on investment and clinical outcomes and so forth.
So all that is happening at specific institutions already. So I think it may be useful for the subcommittee to consider, that is all well and good, but what is missing that requires a standardization effort of some sort. For example, this is happening at only a very few institutions that have the resources and the knowledge to develop their own sophisticated systems for doing this. The vendor community is not providing this on a wide scale yet. So why is that and what is the role of standards potentially to facilitate that.
The other aspect of that is, independent of decision support or any institution-specific functionality, it may be useful to consider the rule of terminology standards as well as messaging and so forth standards in the cross facility or super facility aggregation and sharing of information. This might be the most compelling area of need with respect to standards, where institutions, organizations that are doing things differently wish to either share information that can be processed by computer applications, or they wish to contribute their data to some pooling or aggregation of data, such that it can then be analyzed reasonably.
To the extent there is value in doing that, for clinically specific information, I would say there is really no terminologies or effective standards in place for doing that kind of thing today. That may be a compelling reason for standardizing terminologies and other areas of medical information representation. So I think -- and this is getting into what Jeff was alluding to earlier, my thoughts on how to organize the scope a little bit, how to set the context for the work of the subcommittee in this area. That is, to take one step back before we move forward and think about the context of this work a little bit. So if you don't mind, I'll put up a few slides to stimulate thought and stimulate discussion, and maybe put up a straw man approach that we can then discuss.
MR. BLAIR: Walter, I hope you won't be offended if I point out typos or mis-spellings.
DR. SUJANSKY: All right, feel free. The first slide is some opening thoughts. These aren't opening thoughts anymore, but there are some thoughts hopefully along the lines that I mentioned. Why patient medical record information standards? Again, this is not holy writ or anything, this is to stimulate discussion. These are some of my own opinions, and will hopefully stimulate discussion on this topic.
Personally, I think the reason is to standardize computerized access to patient medical record information data. I emphasize data. My good friend Lieman Lough likes to say, terminologies are not about medical knowledge bases or the organization of medical knowledge; they are about data. They are about facilitating access to data. Sometimes it is important to remember that when you are evaluating terminologies, qualities of terminologies and so forth. I agree with her on that point.
So again, the purpose is to standardize computerized access to PMRI data. On this slide, I mention three sub-points. That is, to standardize access across patients, across time and across providers and facilities. These are the things that in the absence of standards, the data will end up represented differently across patients, across time for the same patient going to different facilities or different providers, even the same provider across different points in time, and of course across providers and facilities using different systems and so forth. The data today is represented differently. Standardization is needed to allow computerized access to that whole body of data.
One view of the model is a graphical view of the model. It shows some patient databases at the bottom and expresses that clinicians, lab systems, devices and other sorts of information are all feeding these patient databases. They are various; it is not just one database, it is many different databases across patients, across time, across facilities.
So one view is that the need for a standardized models is to create a model, a view, an interface to this data that is standardized, that is uniform, so that other processes and applications can get at this data and do practically useful things with it. Those things include other information systems getting at it, so that they can share the information from one system to another and have it incorporated into that other system, shared, and computer processible after that is done, without manual intervention. Standards are important for that kind of sharing.
There is decision support. Standards may be very important for decision support to make decision support a widespread thing, automated, computable decision support, to make it a widespread thing, and not just confined to the Regenstrief Institute and Harvard and the LDS hospital and a few others.
MR. BLAIR: Could I just jump in to make sure that we are thinking of the same thing when we are talking about decision support, because it could be a broad or a narrow topic.
It could be something as simple as a drug to drug interaction, in which case existing coding systems like RX NORM might be just excellent in being able to accommodate drug to drug interactions for decision support. But if you wind up going to that patient record, where you are about to order a medication or a therapy or a lab test or whatever, let's say it is a drug and you are checking against allergies, or you are checking against whether somebody had rheumatic fever when they were a child, or you are checking against other symptoms and signs or items in the problem list, then you are starting to go across what are existing terminologies.
If you lose the clinical specificity in that, that is where I felt a great concern. I don't think -- I'm not sure, if Clem was here, he might wind up saying, existing coding systems can accommodate that with mapping or interlocking things, and to some degree they might, but I am a little bit concerned that as we begin to go forward with that, that they may be able to do so, but maybe not as well as if we had a convergent reference terminology to accommodate that.
DR. SUJANSKY: Again, you are raising the point of something that works in a particular facility and the selection, the choice of terminologies they make to implement their particular system, and maybe there are existing terminologies that support that now, versus a truly standardized representation that for example vendors can write decision support systems against, and know that 80 percent of the hospitals and clinics in the United States will be able to run those decision support rules as their system is compliant with some standard.
But I agree with you that there are different levels of decision support. At the limit they do require the standardization of a variety of clinical information, including allergies and past medical history, lab results and so forth. There is also the notion of guidelines as decision support types of vehicles that then get into findings and symptomatology and so forth, that is even more sophisticated.
The third category of applications that can benefit from standard PMRI models are what I call statistical inference. By that, I mean research, outcomes research, even clinical trials and so forth, where you are pooling a lot of data across patients, across time, across providers and facilities, and you need to have comparability of that data and aggregation of that data in order to draw valid statistical inferences. Obviously, the more data you have, the more power you have to draw inferences. So digging one level deeper than this high level model, personally I think it is useful to consider more detailed requirements for the selection of standards in terminologies and other areas of standardization. This level that I am articulating here I think has already been addressed in the July 2000 report to some extent, and has been referred to.
I think we all agree that it is good to do these things. But in terms of selecting specific terminologies that enable us to do these things, I think it is important to drill down one level deeper and think about what are the types of information that need to be standardized and so forth. So again, to stimulate discussion, I have prepared a few use cases or examples of the kind of standardization-enabled scenarios that maybe we can use as a touchstone to say, does this strategy provide a solution.
The first one is an instance of interoperability. It considers the documentation of a patient encounter in the emergency department, it is electronically submitted to the patient's primary care physician and automatically incorporated into the EMR of that physician. This is done without any manual intervention, two different systems, exchanging information. At the end of that, it is again electronically represented and processable. The diagnosis and the ED goes into the patient's problem list at the PCP. The meds prescribed in the ED goes into the medication list in the PCP system. Procedures performed in the ED go in the past medical history portion of the EMR. This is an example of the kind of interoperability that could be enabled by standards, and is not really possible today, given two systems that are developed independently.
DR. COHN: Can I just ask a question here? I am presuming that if we go further with this -- I am looking at this and I'm going, that is interesting, but clearly terminology standards are not anywhere near enough to do what you were just describing.
DR. SUJANSKY: Right, they are necessary but not sufficient.
DR. COHN: So if were ever to represent this side of this room, I presume we would have to drill down and talk about the other pieces that have to be in place also.
DR. SUJANSKY: Yes, and that is an important point. Terminology standards unto themselves are not sufficient to do most of the interesting things that patient medical record information standards are needed to do. So I think we do have to have a holistic approach to some extent.
Now, obviously it is more difficult and less practical and takes longer and so forth, but I think that has to be in the back of the subcommittee's mind.
DR. FITZMAURICE: Walter, is PMH patient medical history?
DR. SUJANSKY: Yes. Just stop me if you think this is not the right approach, to go through these and our time would be better spent moving on. There are about three or four of these that I have put into slides.
The second specific use case is for comparability and aggregation of data. So imagine you would like to do an outcome analysis with treatment with beta blockers versus ACE inhibitors in congestive heart failure patients over a two-year period. So you need to consider diagnoses and medications and the classes that those medications are in, and subjective and objective findings that are used to gauge the outcome of a CHF patient, including quantitative things like ejection fraction and echo cardiograph results and qualitative things like edema and shortness of breath, et cetera.
Ideally, all these things are available electronically in a standard format, and you can draw statistical inference.
DR. COHN: This is actually a very interesting one. There was a CMS QUISMIC indicator and study related to this that you might take a look at.
One of the big issues has to do among other things with contraindications to taking the medicines, which there was no way to identify that, other than by doing manual chart review, for example.
DR. SUJANSKY: Right, exactly.
DR. COHN: This would be an interesting case study to get this sort of data. CMS people basically had to go in and basically do manual chart reviews, because there was also no way to get the objection fractions, which was the other indicator here. So I just mention that to you as a source to more fully flesh this one out.
DR. SUJANSKY: Yes, if we use this kind of example in the future.
MR. BLAIR: Let me jump in. When I start to do these types of things, I always feel really nervous because of my lack of credentials as a clinician.
DR. GREENBERG: That doesn't stop any of us.
MR. BLAIR: Yes, I am willing to embarrass myself anyway. But I feel like this is an area that is extremely important. I almost feel as if while the example here is appropriate and relevant, it understates the importance of this. Going back to the fact that when we -- we are one of the few institutions that is in a position to try to make a difference in information infrastructure that can affect the cost of care and quality of care in the nation. When we look at outcomes, we are able to do some things with outcomes, but we are so constrained by limited by lack of data, cost of data, that the things that we do tend to be one to one correlations, like with the beta blockers.
Here is where I think in terms of the need for us to have a broader view than just the pragmatic view that exists today with messages. We take a look at clinical studies and outcomes studies, and we really want to make improvements in this. We need to be able to do this in a multifaceted manner, where you are looking at the age of the patient, where you are looking at other complications and other medications and other environmental conditions at the same time that you are looking at the efficacy of that particular beta blocker. That is where we are going to wind up with really meaningful important improvements in outcomes. That is where at least I am envisioning that that reference terminology begins to have power and importance to the nation. Now, did my simplistic non-clinical background make me -- did I say something silly or was I on target?
DR. COHN: Well, Jeff, you are probably right on one level. I think it is a question of what -- maybe this is an issue that we need to talk to the industry about, about where the real value is here.
One group of people would say that to do an outcomes analysis will be the savior, reduce health care expenses, be the answer for reducing health care expenditures in the next ten years. Another group would say, we do a lot of outcomes analyses already, and the big issue is implementing the learnings from it, which is really what decision support is all about.
So you get into a study like this, or even one like you described, and my thought is, that has been done four or five times already, probably in multiple different institutions. The problem is, when Simon Cohn goes and sees a patient, do I remember that information at the point of care, can I put that guideline into place on the specific patient, and population do what all these 80,000 guidelines that have been developed say.
Steve Steindel has a hand raised. He was also hugging Mike Fitzmaurice.
DR. STEINDEL: I was hugging Mike because of course his agency is very involved in doing this sort of stuff, and I think he would love to see better access to this data. I read his research activities thing every month, and I'm amazed at what they get.
But what I was going to comment on is, as Clem points out, we have been doing this type of decision analysis using the specific terminologies for a long time. The way we have been doing it is being very creative in writing our queries into the databases. The information that we get from those databases is how original we are and how good we are in writing those queries.
Likewise, in using the reference terminology for its support in decision support depends on how they modeled the concepts in those reference terminologies and what biases they used to do those models. If we can substitute multiple models in the reference terminologies freely, yes, we may have the ubiquitous answer that Jeff is envisioning, but in general we are looking at specific models that exist in the reference terminologies, that may or may not provide the information for a researcher.
I know CDC's investigations right now of SNOMED, which we have done a lot of, and we have commented on it at the editorial board meetings, SNOMED is geared to clinical situations, which is what it was designed for. Unfortunately in public health, that is usually not the type of queries we are interested in doing. So we would either have to redo some models and impose them on SNOMED for public health queries, or we would go in in any type of databases that are encoded in SNOMED and do exactly what Clem is talking about, and that is coding our queries into those databases and being creative that way.
So there is a balance between the two. I think that is what Chris is pointing out in his web of terminologies, that we do need multiple interlocking views of the various terminologies to get the maximum information from them. Even if we choose a convergent or a reference terminology, we probably still need these multiple interlocking views.
MR. BLAIR: Well said.
DR. FITZMAURICE: I just want to let Simon know that he has put his finger on the two most important concepts of our agency, the Agency for Health Care Research and Quality, not only developing good evidence for evidence based practice of medicine, but also translating research into practice. Getting that research into practice is just as important as good evidence.
DR. COHN: Brady has his hand up.
DR. AUGUSTINE: One other thing I wanted to point out here is, since most outcomes analysis is retrospective and observational as opposed to experimental, if you are going to compare outcomes, some type of risk and severity adjustment, which means you are going to need to have good patient information, good list of comorbids, things of that nature, which we don't really talk about here.
DR. SUJANSKY: Because it is not explicitly mentioned?
DR. AUGUSTINE: Yes.
DR. SUJANSKY: Comorbidities?
DR. AUGUSTINE: We are definitely going to need to look into risk and severity adjustments. You have seen many a physician profile report, and the first response you are always going to get, if you look into a hospital physician or whatever, they're going to say, my patients are sicker.
So for any type of outcomes analysis, you are going to need really good information from the medical record with regard to those other aspects, so you can do good risk and severity adjustments.
DR. SUJANSKY: I would absolutely agree. I guess implicit in this is availability to that level of clinically specific data, and maybe you are also arguing for improved terminologies that have that level of detail for comorbidities.
Some of these things can be captured in ICD-9 and so forth, but without the limit, the clinically specific information is not available.
But to address Steve's point also, I would say that no terminology that will ever be developed will be able to support any outcomes study that can be imagined. So just as today there are certain studies that ICD-9 supports, and studies are done against ICD-9 data, but what you can do with ICD-9 data is relatively limited. If SNOMED were adopted, or something much more clinically specific, that would enable a lot more kinds of studies -- and data were available that were coded, that would enable a lot more. But there would still be missing pieces and studies you couldn't do, and you would have to go to the charts and so forth. But there would be a process in place maybe to incorporate the input from that back into the terminology development process in order to enable those studies in the future.
DR. FITZMAURICE: I just want to add to what Walter said. If you have the terminology, the codes and the electronic patient record, you can conduct a study by putting flags on patients you are studying, so that the information is entered into the medical record. Sometimes when you do a chart review, the information you want is not there, was never put in there.
MR. BLAIR: Can I just simply ask, are we recording this session? Is there some recording of it for the purpose of minutes?
DR. COHN: Yes.
MR. BLAIR: Great, because there are so many good ideas that are coming out here, I want to make sure that Walter has this available to him.
DR. COHN: And it is also being broadcast over the Internet.
DR. GREENBERG: Recorded and transcribed.
DR. COHN: Brady had a comment, and Gail.
DR. AUGUSTINE: Some of these studies that are based on ICD-9 are giving us false positives. They are actually guiding the industry in the way we practice. Having a better system would give us better practices, since we wouldn't be guided by studies that may not be accurate, by the use of ICD-9 systems, because there is a lot of gaming.
That is a billing system in many ways. The accuracy, whereas it is very good, it is not perfect, whereas a clinical system better suited for research would give us a lot better information.
DR. GRAHAM: We at the Department of Veterans Affairs had similar problems with the ICD-9 too, and their clinical reminders and our decision support, differing views and differing rules between inpatient and outpatient. Ruleouts being coded as positive on inpatients and not on outpatients. It has been a real problem for us that we have had to accommodate for in our systems.
The other comment that I wanted to make is, for the example that is posed right now, what we do is a chart review. So our information is based on a very small subset. The inclusion of the vocabularies give most of us the ability to make decisions on a population and not on these small segregated subsets.
Not to mention the cost. I don't think I am alone in saying that we spend upwards of $20 million or more annually for chart review to get access data for this subset.
DR. AUGUSTINE: Consider HETIS. The money that is spent in the managed care industry, just doing chart reviews for their HETIS numbers as well.
DR. COHN: I do want to comment. Obviously I am a strong believer in the importance of clinical terminologies. But I also worry about us over-selling them a little too far. This is an area where it gets a little bit like quicksand, if we don't watch out.
Certainly one tends to ascribe all sorts of wondrous things to clinical terminologies. Yet, one would observe -- and actually, CMS is already doing this, they have commissioned and developed some very good risk adjustment models based on ICD, and rolled them up, because ICD is too granular. Similar at the level of DRG.
So you wind up in a situation where you actually have things that are very good risk adjusters, some of which are actually being implemented now for payment in 2004 for certain types of Medicare patients, based on our poor ICD, but are obviously superior. I would have to be convinced that this level of granularity would necessarily achieve better scores. Now, I could be convinced about that, but I would have to be convinced.
DR. AUGUSTINE: My belief is that what you are trying to do with these adjustments is, there is a amount of variability that you are trying to account for.
DR. COHN: Exactly.
DR. AUGUSTINE: The adjusters are accounting for different amounts of that, so you are really looking at the aspect of interest. Whereas there are adjusters now that I would qualify as being good with better data, you tend to make better decisions with better clinical terminology, the adjusters would be more accurate.
DR. COHN: I would be willing to be convinced of that. I just haven't seen the data yet to convince me. Steve Steindel.
DR. STEINDEL: Picking up on this a little bit with regard to risk adjusters, et cetera, and moving out of the area of ICD and billing, of course CDC's interest in the terminologies is population-based. Rolling out these types of systems, both from a population point of view and into a primary care setting for this ubiquitous thing we have been talking about called syndromic surveillance. We are hoping that we can get primary clinical information that triggers our surveillance systems.
In particular, the hot button right now is in the case of a bioterrorism event. There was a big discussion on it about a month ago in New York City, the first talks on syndromic surveillance and how to do it. One of the things that it involved is that we can't ask clinicians to input this information. It has to come from an electronic medical record system. We have to have good decision support on top of that to provide the warning bells and whistles that trigger syndromic surveillance.
One of the things that is involved with this, as Brady has pointed out, some of this involves risk adjustment. If you are dealing with the general population, the incidence of anthrax, even including the cases a year ago, is essentially non-existent in the general population. But the incidents where you do have an event occurring changes your risk adjustments and your syndromic surveillance.
So we need to look at flexibility in our terminology systems that involve this type of thing. You can even extend this into the non-surveillance arena, into general medicine. We are all aware, you have a 20-year-old healthy adult. Your risk adjustments that you are looking at for a cardiac event are much different than you have in a 50-plus-year old. So this is what can be feeding into these decision support systems, even outside of the ICD where we are familiar with it.
DR. GREENBERG: By the way, speaking of the fact that this is being recorded, I do suggest we take a break shortly.
MR. BLAIR: Oh, what a good idea.
DR. COHN: Walter, are you willing to have us take a break at this point?
DR. SUJANSKY: That's fine.
MR. BLAIR: A ten-minute break?
DR. GREENBERG: I don't want to interrupt Walter, if he just wants to get through his slides.
DR. COHN: You have one more?
DR. SUJANSKY: I have a few more, so it may be better --
DR. COHN: Why don't we take a break then?
DR. SUJANSKY: If you don't mind returning to this afterwards.
DR. COHN: Sounds cool. Ten-minute break.
(Brief recess.)
MR. BLAIR: Walter, would you like to continue?
DR. SUJANSKY: Yes. I've got about maybe five or ten more minutes. We can move ahead.
One note. We spent a lot of time and had a lot of good conversation about the use cases that I have presented, and the kinds of things that a standardized terminology would enable. My intent was not to go over the ground of why terminologies are good and so forth; I think we are all pretty familiar with that, and a lot of papers have been written on that, but to create a context of the kind of functionality that in the end, down the road -- to begin with the end in mind, so that we have a touchstone to reference when we think about the kinds of solutions we want to propose and recommendations we want to make, so that internally, before we do that, we can say, this is a good solution because it is on the road to enabling this, or this straw-man solution is not a good solution, because it doesn't enable the thing that we are really looking to do. So that was the point.
So in that spirit, I have got a couple more use cases. I'm just going to go over them very quickly and jump ahead. The next one is about surveillance. I think Steve talked about this a little bit. You want to monitor the instances of respiratory ailments seen in outpatient offices and emergency departments compared to normal trends.
Those of us who were at HL-7 heard an interesting description of the surveillance system they had active at the Winter Olympic Games in Salt Lake City that did exactly this, using HL-7. So I think it is useful to consider this kind of use case, and enabling this on a large scale across the country, and the role of terminology and other standards in doing that.
The last one is decision support, again a large area, as Jeff mentioned. A use case involving guidelines and alert rules that use specific clinical data, and that can be shared across electronic medical record systems, clinical data repositories, other clinical information systems.
This is a very sophisticated and difficult function, because I think it requires a standard patient record model that includes standard record structure and standard terminology, and models of context and negation and time, and all the things that are relevant to decision support at various levels.
Now, of course, there are less sophisticated decision support modalities such as drug-drug interaction checking and so forth, that don't require as sophisticated a system. But again, the point here is that it is not that you can create guidelines and alert rules and applications, because that has already been done. People have shown that you can do that. It is to do it more efficiently and on a wider scale, and to standardize the data model that those systems are using to access the data that drives the system, and to standardize it such that the systems become more common, more efficient to develop cheaper, and so forth. So anyway, those are uses cases. I'm sure there are others that are useful. It might be good if anyone on the subcommittee has suggestions for types of things that I haven't mentioned that are also good, and objectives that a standard terminology will enable, send me an e-mail and I will incorporate that into the discussion.
The next area that I was thinking about as I was preparing for the meeting and going over all the documents and discussions that have occurred on this, is, what is the meaning of the subcommittee's recommendations in this area, and how do we make it relevant.
Obviously, we are not going to answer this question this morning, but I think it is important to think about and consider as we move ahead. A couple of specific points. These are really high-level things, but the role of the government in the subcommittee versus industry, with industry meaning standards development organizations as well as specific vendors of information systems and so forth; how can we leverage the role of the government to enable the kinds of functionality that we described in the earlier slides, given that it is going to have to be industry that implements most of that functionality in the end, but what is the role of government in enabling that and fostering it and encouraging it and so forth, maybe mandating it, I don't know. More specific, what is the role of government funding, what kind of recommendations should the subcommittee make regarding funding; should it fund standard development organizations' efforts themselves, fund research, fund education; what is the best way to leverage the government abilities, capabilities.
The second one is, the government as a bully pulpit of sorts, quote-unquote; how can the government espouse a model for enabling these functionalities that industry will embrace and work to realize. Then government as a purchaser, how effective could the government be as a purchaser in fostering standards, encouraging standard, creating de facto standards, if such standards aren't actually going to be mandated. So that is the first element of making it relevant, what is the proper role of the subcommittee in that regard.
Then also, from a practical point of view, I think it is important to improve, extend, integrate existing terminologies without re-inventing the wheel from scratch. I think we all agree on that. Then from a practical point of view, how do you balance the existing stuff, which we are all saying is imperfect and inadequate in one way or another, how do you balance that with the practical reality of having to start there to get to the final point, and it is important for us to be practical in doing that. So I guess I would ask for comments at this point regarding -- perhaps going back to the beginning, the various input from the testifiers about the model, the approach, the way to categorize the scope of the terminologies. Considering the use cases and the long term objectives of standardization, if that sheds any light on the previous discussion, if that tends to favor one view over another of the scope and how to organize the scope. Or maybe not.
MR. BLAIR: Let me do a stab here. Jim Campbell did a representation. I know we put him in the second category, but within his representation, he referred to it as a first layer, and other people referred to it differently, but he referred that a first layer would be a convergent reference terminology. Steve, you pointed out very correctly that there probably won't be just one single core reference terminology. There may be interlocking models to meet many different purposes. So we may want to expand the concept of that core.
Then he had a second layer, which tended to be domains. In my mind, a lot of the progress within that second layer has been driven by folks that consider themselves driven by clinical functions or the needs for transactions in what we put in the first category. Is it possible to within Jim Campbell's model incorporate that first category within the second layer of Jim Campbell's model?
DR. SUJANSKY: Jeff, are you suggesting a model in which there is a single small set of core terminologies at the first level, what Jim Campbell called the first level, but can be functionally specific terminologies that were mentioned by some of the testifiers who advocated the first model, that those be included at the second level and mapped to the core terminology in some way?
MR. BLAIR: At least from the standpoint of us trying to move forward conceptually. We are going to have some other new members. I'm not going to mention who they are, but we are going to have some new members that will be joining us in the future, and they may have different views on this. But it would be nice if we could take advantage of this morning to see if we could begin to converge our thoughts to begin to move forward. So I am suggesting this as a possibility, where we would start to look at many of the terminology development works that have been referenced in that first category as being pretty much represented in the second layer of Jim Campbell's model. No? Yes?
DR. AUGUSTINE: The different terminologies that exist today would be in that second layer, administrative would be in the third layer.
MR. BLAIR: Yes. Simon and Steve, your thoughts?
DR. SUJANSKY: I think there is merit to that, for the following reasons, as a general concept. Those other terminologies will continue to exist and be used for some time, regardless of what the subcommittee concludes or frankly, what the government does in general. So a model that includes those will meet with more acceptance in general than a model that excludes those.
It is going to be impossible to tell someone, you can't use your emergency department terminology anymore, it is not standard, or if you are using it, then you're not standard in some way. It is better I think to include those, but put them at a different level of the standardization process, in a sense.
MR. BLAIR: Steve Steindel.
DR. STEINDEL: I'd like you to explain that statement you just made. If somebody is using a non-standard terminology, why shouldn't we -- and I am using the royal we in this sense -- tell them, you can't use it?
DR. SUJANSKY: Well, I don't think that will be practically achievable.
DR. COHN: Steve, I think the way I would interpret this, hopefully helping Walter here, is, I am using a terminology within my institution for my own purposes.
DR. STEINDEL: I think that is the American way.
DR. COHN: And sine when do you have the right to tell me what I am not using inside my own institution. If you are CMS and you are reimbursing me on the basis of things, you can certainly identify things. Or if I am sending things to the CDC or if there are other state or federal reporting mandates. But beyond that, I'm not sure we would get very far telling people internally what to use.
DR. STEINDEL: That is what I wanted enunciated. As long as we don't make any statements about, you can or can't use this within the walls of your institution, you can do whatever you want within the walls of your institution, but when you go outside the walls of your institution, you need to start looking at what the conformance to national norms are.
DR. COHN: I think the other case here also is, going back to your use cases, things like decision support and all of that, I think the industry is generally looking for direction.
DR. STEINDEL: Yes.
DR. COHN: Nobody in their right mind wants to make an investment in something as expensive as what we are talking about. Having implemented a terminology that requires as much expense and discomfort to everyone having to use it as the initial training and acclimation to that terminology would involve, be it local or national, and then be told, it is not being supported anymore, this is not the way the country is going, you are going to have to change yet again.
So I think even without making it required, I think it certainly provides a lot of guidance to the system and the industry. Brady has his hand up.
DR. AUGUSTINE: They gripe about changing their code sets and their local codes and their local terminology, but when all is said and done, they are happy, because they don't have to maintain that anymore, all those relationships and those definitions and everything.
DR. SUJANSKY: The other part of doing it this way is that at the same time, there is a core terminology that you are endorsing as the standard in an informal sense, that this is the reference terminology and it is relatively comprehensive and so forth. There is a model over time that the industry and the health care community and so forth, and we will gravitate towards that terminology. Given that there is a lot of overlap between that terminology and some of the more functionally specific terminologies, if we can call them that, there may be economic and functional incentives to move towards the core terminology as the one that is actually used over time.
This is what is always seen with standardization efforts. There is a self-reinforcing dynamic. There is a prevailing terminology, and vendors build more tools around that prevailing terminology, so more purchasers of systems gravitate towards that terminology because there are more tools and functionalities built around it, and so forth.
To the extent that the other terminologies don't have any special attributes that enable functionality that the core terminology doesn't enable, it becomes a natural force that compels people to embrace the core terminology.
MR. BLAIR: Could I add to that, because my thoughts are complementary to the idea that you enunciated, Steve, and Simon did as well. But it is from a different viewpoint.
My background is as a vendor of health care information systems. From that standpoint, if we set forth a framework for the evolution of clinically specific standards, including clinically specific terminologies, where the vendors can begin to plan their software development, knowing that that is the direction, they will then take several next steps built on this direction.
They are in the direction of trying to meet user needs, as well as possible, and they are competing with each other to do that. So if they wind up seeing that there are either message format standards and/or terminologies that enable them to produce additional functions for their customers, they will build those functions, whether it is outcomes or decision support or comparing electronic health records or whatever it is.
So in a sense, there is a partnership going on here. I don't think we need to mandate things or say they can't do things. If we wind up setting a direction which enables them to provide new functions, they will take the ball and run with it from there.
DR. COHN: That is certainly true. Going back to the model and the question of how we organize things, probably almost any model is fine, because we know we are going to change it as we move forward, anyway.
MR. BLAIR: You're saying that facetiously.
DR. COHN: No, I am actually saying that very seriously. We would accept a model preliminarily. We need to test it with customers, vendors and others. We will continue to refine and evolve that model. I am talking about the model that we would use to organize our PMRI terminologies as well as determine our scope. I certainly don't think on the basis of one set of hearings with a bunch of terminology experts, when you haven't heard from the industry and haven't heard from others that we are ready to put something in stone at this point. But I think it is a question of what helps us move forward.
The question I have in my own mind, in some ways Jim Campbell's model helps me. Once again, it may be convenient for us to think about it, not knowing whether this core that he describes is one terminology or multiple terminologies. But it is something that we consider to be a core.
I tend to think of -- that is obviously an area that we would like to see some standards being promoted around. Once again, I at this point don't know whether that is one or multiple terminologies, but whatever it is, one of the characteristics is that it needs to be relatively tightly integrated and mapped and hopefully not have a tremendous amount of information loss as you move across that.
But of course, knowing that we haven't for example decided whether supplies or -- exactly what is tightly in that, and the problem is that what is core in health care and in that decision making.
There is also what we describe as the outer layer of what we talked about, which is the administrative and billing -- so-called, the HIPAA standards. One question I wonder about is, obviously we have a lot of other terminologies that we have been talking about. We conveniently talk about the second layer.
To allow for comment, I'm not sure that we need to standardize the second layer. I would have to be convinced that there is a case around standardization of a lot of particular perspectives on things that are represented by those terminologies. But it becomes more that the issue is that we need to have a set of principles, if people are going to use them, how they play into things, that there needs to be good quality mappings that are agreed to by X, Y and Z, that needs to be maintained, all of those sorts of things, as opposed to saying, either these terminologies are in or out of the second layer.
In a sense, to me everybody is in the second layer. It is more like orbiting planets in many ways. Does what I am saying make some sort of sense?
MR. BLAIR: I am chomping at the bit because I am resonating so much to your ideas.
Here is my thought. In terms of the second layer, I tend to think of the HL-7 reference information model as an information model that could begin to provide a harmonization coordination among many of the different terminologies that are in that second layer. I tend to think at least right now, but it may become broader, I look at a reference terminology like SNOMED, but there may be others in the future as a core. So that terminology models or multiple terminology models at the core.
As I begin to look at that, I start to come to Chris Shute's visualization, only putting plurals in it. He was saying that there is a terminology model and an information model -- well, I think there may be more than one terminology model and more than one information model, but if we begin to put it into Jim Campbell's overall framework, I think that we have a way to start to begin to work together with these things, and start to resolve issues. This may help us as we begin to go down the path of our mission.
What do you think of that?
DR. COHN: I don't know. I think fundamentally you are probably right. I am struggling with how the information model piece fits into the terminology solar system. I tend to think it is not really part of the solar system, but almost another dimension.
It is like, when you stick data sets in with terminology models; how do you do all of this stuff together.
MR. BLAIR: Yes.
DR. COHN: I would need to hear more and understand more. I think it could be valuable, but it isn't quite the same thing.
MR. BLAIR: Yes.
DR. COHN: I think Steve Steindel will help in informing how this fits together.
DR. STEINDEL: I don't know if I am going to help or blur. I'll start off with my view of high-energy physics. That sounds sort of funny, but the way I look at it is, every time they build a bigger faster truck, and they hit the atom with a bigger faster truck, they knock off another particle, and they get a better picture of how the universe is put together.
I look at this picture of Jim Campbell's model and your analogy of the solar system similar to that. As we hit this model with bigger and bigger trucks, we are going to knock off other pieces.
While on the surface, I agree with Jim Campbell's model and the three layers of this intersect of convergent terminologies, one or more, and Jim actually listed three in that middle layer, one was SNOMED, another was LOINC, and if I take a look at my computer again, I will remember the third one, --
DR. GRAHAM: RX NORM.
MR. BLAIR: RX NORM, yes. So he had three, and two of them we would really not think of as convergent terminologies. Then he had his messaging middle layer, and then he had his administrative layer.
But the way I look at it is, if we hit this thing with a bigger truck, we will find something underneath the middle layer. To me, that is Chris Shute's picture of the blending between the terminology models and the information models running into a continuum, where we have to hang the information from the convergent terminologies, so that they relate to each other, and so that they can relate up. There has to be a fundamental model underneath all of that. That is the way I look at it. That is the way I look at the blending and the continuum going on.
I don't know if it is important for us in our decision making or anything like that to talk about the way these models relate to each other, but to realize that there are different roles that it plays, and different criteria for the different roles. The idea of enunciating criteria for the different layers is very important, because that hasn't been done. I notice in the HL-7 vocabulary special interest group, there is talk about creating a set of standards for terminology and what they should need, so other people are starting to think this way as well. Marjorie has her hand up.
DR. GREENBERG: I can remember the model less well than you can, but I remember the discussion around it. I am a little concerned about this discussion, because it seemed that there were -- maybe I am overhearing that you are embracing that model, because there were a lot of questions raised about it. I know I had questions about it as well, particularly the relationship between the second and the third. There were things in the third that I questioned why they weren't in the second. I don't have it with me now and can't really remember the detail, but I wasn't alone in some of those concerns, so I think we need to keep an open mind here.
DR. STEINDEL: Marjorie, I think that I totally agree with that statement. That is my feeling as well, and the feeling of a lot of people during the discussion, should we put this here, should we put this there. But I think a lot of people enunciated some of the basic idea behind Jim Campbell's model. There are different uses of terminologies and the way they relate to each other and the way we should think about them. Now, whether we put this terminology in the administrative layer or in the middle layer, and whether we put this in the convergent layer and the middle layer, I think that is for the profession to discuss. But I found this concept of the three layers was a useful way to conceptualize our discussion of terminologies.
DR. SUJANSKY: I think it may be useful to think about some of the differences, the fundamental differences between the layers. A couple that come to mind are, in the first layer, the reference layer, there is no overlap -- this could be a design criterion of it, in fact -- there is no overlap of content. There is no unrepresented overlap of content.
I think that is why there was the inclusion of SNOMED, LOINC and RX NORM, because SNOMED doesn't have a good vocabulary or good content for drug tests and medications, and is really looking outside. The combination of those things may create a relatively comprehensive non-overlapping terminology at the second layer, where you have a lot of different terminology for a lot of different purposes, including nursing and emergency departments and so forth.
I think there is a greta deal of overlap in the semantic content of those terminologies, and that will probably continue and should continue in this model of having two different layers. That is one way of distinguishing them.
MR. BLAIR: In the break, Carol Bickford who is here had some thoughts about the medical viewpoint that we have prevailed. Carol, this might be a good time for you to come to the microphone and express some of your thoughts.
DR. BICKFORD: Carol Bickford from the American Nurses Association. As you were doing the presentations and in the discussions that have been going on, I have been trying to figure out how I am going to take this information back to our work group, who has been the shepherd for the nursing languages recognition program, and taking this information back and helping the group move through this question.
As the discussion went on, it was very frustrating to me. I clearly saw the medical model of diagnosis and procedures, whereas others of us have a very different perspective of our health care initiatives and our relationships with our patients and our descriptions of our patients, and have done a lot of work to try and describe that. I don't see that richness being represented in the discussions that are going on.
DR. FITZMAURICE: Can you give us an example?
DR. BICKFORD: I'm thinking from the standpoint of how we have looked at the diagnosis and interventions, the outcomes and the relationships of all three of those concepts, how we have looked at the resourcing component and the context for our care delivery experiences, and the reflection of how the nursing management minimum data set supports that.
It sort of fits into where the core data, the core data elements as the center piece, and then we have the outside layer. It is like, how does this all fit together, and the model that is being presented may not fit for the rest of us in the world.
As you were doing the discussion about this core being the convergent terminology, many of our nursing concepts are integrated into that. Then they are at the second level or farther out, and it is like, hello, we have redundancy here. So how are we being discrete?
Should we be looking at it as being the core elements describing our clinical experience, and the moving to the next level being out from that? I don't know. I was just sitting there mulling this all over and being very frustrated, because it seemed to medical model focused.
DR. COHN: Maybe I can comment here. Carol, like you I struggle some with the overall concept here. I am thinking back to many years ago when we were initially trying to put together parts of SNOMED and all of this stuff. I think the view, which I tend to embrace, is that we are really trying to have a patient-centric model and a patient-centric core.
So I guess the question is, certainly if there are unique concepts that are patient-centric, that are discovered because of the unique role that nurses play in health care, I think they absolutely deserve to be part of the core. I think that somehow they need to be incorporated. But that maybe is a little different than -- one would observe at least as of this moment, and I know that the nurses are trying to consolidate and develop more of a unified taxonomy. But I struggle a lot, because there are obviously so many nursing code sets, many of which are probably more nurse-centric, in the role of the nurse and the health care, as opposed to the patient-centric piece.
So I am sitting here struggling, but that is how I tend to think about it. I think I for one would be excited to see that patient-centric piece brought forward to make sure that we really do get at part of the core.
DR. BICKFORD: Because our focus is patient-centric. We identify the actions related to that care around that patient, so it is patients. Other clinicians use our interventions and our outcomes.
DR. COHN: Marjorie?
DR. GREENBERG: I'm sure you all would be disappointed if I didn't chime in here with the topic of basically functional status and participation, environment, the components of the ICF. I think if you talk about convergent terminology in the three dimensions, I really don't think any of them, SNOMED included, really deals adequately with functional status. If you are looking at outcomes, I think these are issues obviously of a lot of importance to nursing terminologies as well. But if you are looking at outcomes and being patient-centric, you have got to look beyond diagnoses and symptoms and procedures and all of that.
DR. STEINDEL: Walter asked for a specific term, and we have a lot of discussions about this at the SNOMED editorial board meeting, which is where the nurses are, and where public health feel the same exact way as the nurses do, that we are not adequately represented. I can give them a term: depression.
DR. GREENBERG: What?
DR. STEINDEL: Depression. That was modeled very, very strongly from a clinical point of view. Nurses have many different other views of depression.
DR. GREENBERG: I thought you were saying that the nurses in public health get depressed.
DR. STEINDEL: Oh, they do get depressed, but that has nothing to do with the code example.
DR. SUJANSKY: I would say your point is very valid. There certainly are important patient-centric terms in the nursing terminologies, some of which already appear in other -- what you may consider medical terminologies such as SNOMED, others which do not yet appear, but should.
I don't think the model that we have been discussing of two levels excludes the possibility of having and adding patient-centric terms and concepts that perhaps today only exist in one or more of the nursing terminologies, adding that to a core terminology. There is nothing synonymous about core and medical, at least in the context that we have been discussing, and in any of our minds.
So at an abstract level, what you are advocating and what we have been discussing are not mutually exclusive. It is important to consider. I think that is true of any of the level two terminologies, if you will, whether they represent emergency department functionalities, nursing functionalities, pharmacy functionalities, et cetera. If there is an argument to include those in a core model, then they should certainly be added.
DR. COHN: But you do go back to one of your original precepts, which is that you try in the core not to have redundancy.
DR. SUJANSKY: Right, exactly.
DR. COHN: Which I think I would support as one of the basic principles we should try to have.
DR. SUJANSKY: Right, so as they are added, they are obviously incorporated into the model, so there is no redundancy. But there will always be discussions about, should this be added to the core terminology or not. Those discussions will always continue.
But I don't know that the model itself is bad because -- the two-level model or the three-level model is bad because there will remain some terms that aren't in the core terminology. I would hate to think that that characteristic of it made it unacceptable.
MR. BLAIR: Can I make just a brief suggestion here? There are so many models that we are talking about that at least I struggle sometimes to know which models we are talking about.
If we could refer to the notions within that second area, whether it is Jim Campbell's or anyone else's or Chris Shute's as a framework, that way we could separate those representations from the information model and the terminology model. Does that help others as well as me?
DR. SUJANSKY: The model I am referring to, if your comment is apropos, my use of the term model is one of the approaches for scoping the terminology issue, that approach being the one that was advocated by Campbell and Spackman and so forth. There are multiple levels, and in one of those is a core level, and there is also another level that includes what Kent Spackman called limited scope terminologies.
Actually, I did make a slide of something akin to this for discussion. In that middle layer core terminology there may be one or two or three or so with non-overlapping content and so forth. Then at the top level, Kent's model -- or maybe it was Jim Campbell's -- had a separate level three for what he called administrative terminologies, ones that are required for billing and reporting and so forth, but are not necessarily clinically relevant, but are obviously important for logistical purposes.
So I think a couple of important points about this slide are that applications, actual information system applications, can use different terminologies. There is no reason you can't continue to use a nursing system that uses one of the nursing terminologies, and have that continue to function just fine. But there should exist mappings from that terminology to the core terminology, because there will be other applications, and maybe some very important applications that operate against the core terminology.
DR. COHN: Susie had her hand up.
MS. BEEBE: I just wanted to refer us back to the first spreadsheet, scope question one. These layers are explained. It helps me to look back to see them. So when you were referring to Kent Spackman talking about his name limited versus use limited terminologies, and also the layers, are outlined.
DR. COHN: So, Jeff, do you want to go on?
MR. BLAIR: Well, actually, I didn't want to -- I was heading in a different direction than Susie's comment, so maybe if someone wants to follow up on what Susie is leading us towards?
MS. BEEBE: I was just trying to give some more information that we could look to, not really making a comment.
DR. COHN: And I don't think we are necessarily helped by being more specific on this at this point. I think sometimes as a high-level conceptual abstract, these things are useful. You drill down too far, you get into trouble, as we were just discovering about the nursing area. There are many issues having to do with the interfaces between all of these. Jeff?
MR. BLAIR: I have found this discussion this morning to be very helpful to me. There seems to be a number of points of agreement and consensus. Then there are a few points where I think we have some loose ends. I'd like to see if we could come to some possible agreement on those.
One of them is, Steve, you had suggested that it is possible that in that core, you saw in the core both an information model and a terminology model, whereas, Simon, you wound up saying that you felt as if Jim Campbell's representation might -- there might be a different dimension which would represent an information model. Is there any way we could --
DR. COHN: Let me try to represent my view a little better here. I have read Alan Rencter's articles for many years about the terminology model mapping to the information model and back and forth.
DR. GREENBERG: Any you suggest we all read?
DR. COHN: I was going to say, sometimes I feel like my hat size is not big enough for some of these conversations. But I still think that there is a significant outstanding issue about how these things fit together.
I think it is best referenced by one of your use cases, where you talk about -- there are pieces you are talking about; even though we are talking about the focus of this not necessarily being message format standards and transmission of data in them, but this is what we are talking about, when we talk about how does the information model a la the RIM deal with the terminology model.
It isn't immediately clear to me how these things work together. I can see the HL-7 using a core terminology, but I sure can't see in every field saying use ICD or SNOMED or LOINC or whatever, which would be one way for them to characterize it.
What you would want to do if you were HL-7, I would imagine, is, you would want to say, in this field you can use X, Y or Z at the core terminology. There is a specific set of terminology that you would use to answer that question in that field. Otherwise you wind up with this amazing headache around interoperability, I would imagine.
So how you take that and you leverage it. Once again, if others were here, I'm sure they would -- surely there has been a lot of thinking on this, probably much more recent than Alan Rencter, and we may want to talk to him about it.
I don't have an answer. Jeff, maybe you have an answer to this one. I see this as an outstanding issue that I think we need to investigate further.
MR. BLAIR: Yes. Let me ask this. Walter, is this something where you could pull together some information that could help the subcommittee, even beyond research? Maybe you could suggest two or three investigators that might want to testify to us, that could help us?
It seems as if this almost gets to the core issue of how do we get the information model and the terminology model to work together. If we could figure out at least conceptually how we do that, I think that may help a lot with our strategy and direction.
DR. SUJANSKY: I think the first thing to do is to look at the RIM and to think about what it says about terminology and d doesn't say about terminology.
I will look at it more closely with an eye towards that. My understanding at this point from what I know is that it says basically what Simon was saying; in this field you can use one of several terminologies, one of many terminologies, in fact, as long as you tell us which one you are using. That obviously provides a measure of standardization, that at least you know which terminology the term came from. But it certainly doesn't allow two applications that don't know much about each other, don't know that they are going to be able to use one of these many terminologies to interoperate.
One way to supplement what RIM and the HL-7 terminology efforts are already doing is to go beyond that and say HL-7 allows you to do this, but if you want to be compliant with the core terminology model, then not only will you use HL-7, but the core terminology compliant terminology for this particular field is X. So you can still be HL-7 compliant and not use X in that field, but if you want to be core terminology compliant, you have to use X in that field.
This obviously is just an idea off the top of my head, but it is one way in which there can be a complementary type of relationship between the HL-7 and the RIM, as well as a more proscriptive and tighter standard effort regarding terminology and the terminology that appears in HL-7 messages.
DR. COHN: Steve Steindel was going to make a comment which I look forward to, but I also would have to say that most recently, SNOMED has 235,000 -- ?
MR. BLAIR: 300,000.
DR. COHN: 300,000, I haven't looked recently, 300,000 concepts, saying that in this field you put some SNOMED concept in there. I don't think it quite solves the problem; you need to be a little more proscriptive than that.
But Steve, when I was talking about this one, it seemed like you were --
DR. STEINDEL: Walter's description of the RIM and its use of terminology is quite correct. It is very specific, and it says in this area of messaging, these vocabularies are allowed. It may be a very small set of vocabularies; it may even be one that HL-7 defines, depending on where it appears in the message structure, and it may be very broad.
What seems to be developing in the informatics community is, there used to be the talk that there was the terminology model and the information model, and never the twain shall meet. But now we are realizing that the clinical concepts are much more complex than that, and the twain is meeting. There is a blending, a continuum between the terminology and the information model.
How you describe that in a quote-unquote terminology -- people have not gotten a good grasp on how to do that. Chris was starting to reach for it when he was talking about his web of terminologies, and even Kent is starting to grasp at this.
Within SNOMED, as he is starting to look at concepts that exist in his findings axis, which he finds may not necessarily be findings, some of them may be disorders. The terminology model is not clean. There is overlaps between those various axes, and how does he handle this with respect to his terminology model, which is what he feels drive SNOMED. It is something we talk about in the continuum.
The introduction of the nursing terminologies into SNOMED, which SNOMED is trying to do, is a blurring of this terminology and information model, because similar concepts are used differently. So we are grasping with it. It is probably a good idea if we can find some people to give us some guidance on how to make recommendations for selections of terminology.
MR. BLAIR: You mean in the context of HL-7?
DR. STEINDEL: No, in the context of the patient medical record.
MR. BLAIR: I'm sorry, when you say patient medical record, we went back to how do we reconcile an information model and a terminology model?
DR. STEINDEL: Yes, because that is what we are going to be doing with that recommendation.
DR. COHN: I absolutely agree. It is interesting, your description of the discussions appear not to have changed much in the past couple of years that we are all grasping to have it do this.
My view is that although it may not be the primary requirement of our work, to make sure that it works well in HL-7 message formats, HL-7 message format standards and transmission, at the end of the day it better do a pretty good job there.
I have had enough conversations with leadership of HL-7 and others to know what the ongoing conversation is. The big problem getting in the way of interoperability in HL-7 messages is terminology. Now, I think maybe we need to understand that a little more. It is hard for me to imagine that putting the term SNOMED or ICD in every field is suddenly going to create interoperability in HL-7 messages. So maybe we need to find out from them a little more in that sort of discussion, what is very concrete as opposed to very abstract. Let's talk with HL-7 some to figure out what it is going to take around terminology, what sort of requirements are necessary so that we really do get interoperability in HL-7 messages, which is certainly one of the things that we are very concerned about.
MR. BLAIR: Simon? I didn't mean to cut you off.
DR. COHN: That's fine. This is what I think of as a proposal for December.
MR. BLAIR: I agree with that, but my thinking is, at least the way I am viewing this, is that that is part of the problem. I don't know whether that is 30 or 40 percent of the problem or 70, 80 percent of the problem, but in my mind, the business requirements and clinical requirements that have driven the development of transaction standards, which HL-7 has led extremely well, is I feel like only part of the scope that we have to focus on, and the other part of the scope that I think we have to make sure we keep in mind as we go forward on this.
So I am saying, in addition to what you just said, are the applications that have not yet been developed because we haven't had that clinically specific terminology to support applications for outcomes research, multifaceted clinical decision support, the ability to have concurrent decision support, all of those areas, I feel like we need to make sure that somehow we are representing in that scope that whole area of the emerging applications as well, so that our framework is broad enough to accommodate them, in addition to moving forward with message-driven terminologies.
DR. COHN: I see Steve Steindel has a comment here.
DR. STEINDEL: I think I agree with what you say, but I think we have to heed history in this area. That history comes from Colin Price, and what they have done in the national health system over about the last decade or more with the Reed codes.
I think it is very important to realize that when they developed version three of the Reed codes, which is a convergent terminology intended to do all these good things, support decision supporting, et cetera, the rate of acceptance versus version two of the Reed codes, which is basically just a pick list of terminology, was about zero. That is something we need to be aware of patient medical record systems, is how is this going to be used. That is an important component that we need to talk about.
SNOMED is starting to realize that as well, where they are talking about developing navigational hierarchies, used-based scenarios for SNOMED, in addition to the clinical-based scenarios for SNOMED. So I think this is a very important thing to think about when selecting PMRI terminologies; will people use them.
MR. BLAIR: Well, let me offer this thought, because I believe in being pragmatic, and I think your observation is correct. However, if you wind up saying in early 1950 how many people were using television, it was close to zero, just like you say, but the standards enabled vendors to be able to move forward, and within ten years there was a new environment.
I guess I feel like in addition to staying very closely focused on being pragmatic, because you wind up making sure that you don't wind up heading off into Utopia -- and I'm not suggesting that we head off into Utopia or nirvana, but I am winding up saying that there is an information infrastructure with terminology standards that I think we have to make sure is in place to support the new applications that HL-7 hasn't been able to address, at the same time that we are providing a framework where HL-7 can continue to advance and flourish.
DR. COHN: Let me comment. I'm not sure that any of us are really disagreeing with each other. I think it is this issue about pragmatic and near term versus longer term needs. I get very uncomfortable when we start looking too far out too quickly, only because I want to make sure at the get-go that we are meeting at least near term needs, in things that are pretty obvious.
Given the number of years that HL-7 has been working on an information model, trying to move to version three, the issue that we already know exists around interoperability of HL-7 is not going to be completely solved at least to my knowledge by version three, unless we get terminology straightened around. I think that is a very good -- and having seen that particular use case at least inferred in the overheads that Walter had come up with, I think that that provides us with something concrete that we should at least explore, not avoiding or otherwise missing these other issues.
The problem with asking for requirements around vaporware is just it is just that, vaporware. Until people start going off and producing things, it is hard -- the requirements are everything. You want everything until you actually develop the product. So I think it might give us a real pragmatic near term view to sit down and try to figure out can these things be made to work and be useful and helpful for HL-7, because part of it is almost a use case discussion.
MR. BLAIR: I'm not arguing with the use case discussion. I was just simply saying that that shouldn't be the only way we proceed.
DR. COHN: Sure, but the other observation of course is that things like decision support already exist. The big problem with decision support is not that they don't exist, it is that they haven't been diffused out into the industry. So once again, it is not the next generation; it is really more a question of how do we make these things move out there further, and what are the impediments, which I think is what you would observe also.
So there is a lot of stuff that we can take a look at that exists, that is practical, we can find out from people. I was thinking off to December already, about what we may want to do. I just thought that this issue with the information model was one that was very ripe at the moment and that we had better think about soon.
MR. BLAIR: The issue with the information model?
DR. COHN: I just think talking with the HL-7 to find out what they see as the issues and what they see as the requirements, and how might this all work in a way that really did support the interoperability, so that in two or three years, this doesn't become the excuse for why there isn't interoperability.
DR. SUJANSKY: I think it is worthwhile talking to HL-7 on that front. They will help to represent the vendor perspective also in a global sense that we may not get individual vendors.
But it is important to recognize as well, they have a slightly different view on standardization than perhaps the subcommittee does. They are very focused on backward compatibility, more so than perhaps we need to be, although it is important to consider that. But they are very focused on it. But I think it is worthwhile to talk to them, recognizing that their perspective is different.
DR. COHN: We have 20 minutes left on this session, so Jeff, how do you want to spend the time? Should we talk about next steps?
MR. BLAIR: I think that is exactly what we should do, yes. Here are some of my thoughts a little bit. Let me know if you all feel comfortable about this.
Walter, by the way, thank you so much for being with us. I am really pleased that you are available to help us. Do you feel as if you have enough information from this session plus the background information, where you could give us some information and guidance for our next meeting, which would probably be December, correct? Yes. So maybe we could get a better handle on what options there are to be able to have information models and terminology models begin to coordinate, harmonize, work together? What paths there are, what approaches there are?
That is one thing. The other piece that I think would be very helpful would be if you could take Jim Campbell's construct and see if there could be an additional step forward with that construct that may be helpful to the subcommittee, to visualize how we should move forward. Does the rest of the subcommittee feel as if those would be helpful for us, that those are the right things for us to ask Walter to do?
DR. COHN: Sounds good.
DR. FITZMAURICE: Jeff, I do want to raise an issue of the scope of work for what we want Walter to do. There is a scope of work that has specific details in it. Things can be fit into the details, but we probably should look for how we are going to proceed over the next six, nine months and see how Walter's time can help us get to the end result.
DR. GREENBERG: I think we have to recognize too, appreciate Walter coming in for these two days, and commit to getting our contractual relationship with him expeditiously completed. I think it is hard for him to commit to anything right now, until we have actually completed the contract. I did have a few things I wanted to say about the scope, also.
MR. BLAIR: Marjorie, are these things you wish to discuss in the context of the committee as a whole, or is this something that you and Michael and Walter need to work out?
DR. GREENBERG: I think we can probably --
DR. FITZMAURICE: I think we can work out the details, but I think the specifics is, is the scope of work that we have laid out for Walter consistent with how you and Simon see the needs of the committee going for the next six to nine months. If not, we'll change them.
DR. GREENBERG: Maybe we can meet afterwards briefly?
MR. BLAIR: Yes.
DR. GREENBERG: I did mention to Jeff at the break that any sort of systematic information collection really has to be done by the committee. It can't be done under contract, because of various clearance issues, et cetera.
The committee has in the past gathered information from -- that doesn't mean under contract we can't have analysis and going back and talking to people and gathering information, et cetera, but like references to questionnaires, et cetera, that is going to have to be done by the committee. It can't really be done under contract. So we can discuss the implications of that further.
DR. COHN: Marjorie, I'm sure that is something that can probably be worked out. I was just looking at the scope of work myself, and I guess I would have the following comments.
Number one, I know we are meeting in December and I know we are meeting in January, and I don't see necessarily pieces related to either of those meetings in terms of the scope of work. I guess I would also wonder whether we maybe need to hear sooner rather than later from the industry. I don't know what is Walter's work versus what is the work of the subcommittee, but I do think that we at the end of the day need to make sure with regular, timely interactions with the health care industry to make sure that we are on the right track, rather than waiting until the end of a process to ask them what they think. So we need to somehow do that. I don't know if that is an activity for Walter, except maybe to help us put together the particular session at a hearing or otherwise.
The other piece is, I know this is replicated on the basis of the work we did with the standards development activities and all of that. I guess I was just scratching my head a little bit, trying to figure out whether -- on this issue of gathering information from PMRI terminology developers -- and certainly I did hear what Marjorie was just saying, but I am trying to think of the significance of that in this work, about whether that has the same -- I know that we did that as part of the previous effort. Jeff, do you consider that to be a critical aspect here, or what?
MR. BLAIR: The feeling that I have right now was, I felt as if the subcommittee needed to get a little bit of a feeling on its own as to how we were going to step forward, the approach, the agenda for today in that discussion framework. For us to make a little bit of progress in that, I thought maybe that was where Walter could help us, and maybe come back with a synthesis of where we are now.
At that point, we could wind up asking for comment and critique, both from terminology developers and from the major message format standard developers as to any problems or issues with respect to the direction that we would like to take.
DR. COHN: So in other words, getting input from them for our proposed work plan?
MR. BLAIR: Yes. Does that make sense to you, Walter?
DR. SUJANSKY: Well, there is an issue of developing a work plan for my activity within the subcommittee and for the subcommittee. I would ask the subcommittee to consider if they want to develop a work plan for the next nine months at this point, or if it is something that they would prefer to proceed with on a month to month or interim basis and so forth, so that we can achieve some clarity about the time frame and the intensity of work that will be required over whatever time frame is decided and so forth, before we proceed with any other plans.
MR. BLAIR: Sure.
DR. SUJANSKY: There is an air of uncertainty that I think we need to resolve a little bit before moving forward.
MR. BLAIR: Yes, and I apologize for that. We made a general work plan, and then of course there has been other issues where we have tried to accommodate that, and we tried to maintain some flexibility, but that has left that uncertainty there. So let us try to see if we can close those gaps.
DR. SUJANSKY: Very good. I am specifically referring to things like, the work plan that is on the table now does include a fair bit of information gathering. To the extent that that is in or out of scope for what I can be involved with, that will obviously affect the work plan. So I think all those issues need to be resoled.
MR. BLAIR: Right. That was similar to what we did with the message format standards. But Marjorie has pointed out that from a technical standpoint, we have some issues as to exactly who does the questionnaire and how that is distributed. You will be in a position to do the analysis for it, but we have to modify apparently the work plan a little bit on that.
DR. COHN: Steve?
DR. STEINDEL: I have to talk to Marjorie to get a little clarification on that.
MR. BLAIR: What if we did this? Unfortunately, I wound up making airline reservations, where I really do need to leave at 12:30.
DR. COHN: That's fine.
MR. BLAIR: Would it be okay if we resolve some of these issues with respect to the scope of work for Walter by e-mail during the next several days? Marjorie, are you in the office, and Simon, are you in the office?
DR. COHN: Sure. Let me say, I think the number of hours and types of work are about right. It is a question of how we utilize you. I have no doubt that we are going to be utilizing you for this much time over the next X number of months. I just hope you have some flexibility in exactly what the work activities are.
DR. SUJANSKY: I certainly do.
MR. BLAIR: Great. Michael, I didn't ask you. Are you okay?
DR. FITZMAURICE: I'm out next week.
MR. BLAIR: But are you were during the next couple of days?
DR. FITZMAURICE: Yes.
MR. BLAIR: Good, so maybe tomorrow we can start to --
DR. FITZMAURICE: It may be useful for Simon, Jeff and I and Brady and anybody else who is on the staff to sit down and talk about what the issues are after the meeting, so we know the parameters of moving forward.
MR. BLAIR: That's fine. I think you can probably do that without me, because I need to go. But if you can meet afterwards and resolve these issues, that would be great.
DR. GREENBERG: I would think 10 or 15 minutes. I haven't seen my office for two weeks and need to get there myself.
MR. BLAIR: So are we targeted towards December for the next time when we meet to discuss these issues?
DR. COHN: Yes. We may have some brief discussion based on this discussion about next steps in the November breakout or whatever, but that that would be -- we will obviously spend a fair amount of time in November talking about next steps. I think we came up with a number of issues that we need to mull about, which may be a session during the hearing.
I think we are going to need some information from outside sources in December. Just to remind everybody, the next set of hearings are December --
DR. GREENBERG: 10th and 11th.
DR. COHN: -- 10th and 11th, thank you. Then dates have also been set for hearings during the first six months of 2003. Those will be January 29-30, March --
DR. GREENBERG: On this document, March 25-25 and May 21-22.
DR. COHN: Now, depending as we evolve the work plan, there has been some discussion about whether there needs to be yet another hearing. I would defer that discussion until we are a little clearer about what the specifics of the work plan are. But certainly we have those three dates already set up.
DR. GREENBERG: Are you planning to take testimony on December 10-11? Part of it is standards and security, right?
DR. COHN: Well, all of it is standards and security.
DR. GREENBERG: Not PMRI.
DR. COHN: A lot of the session on the 10th and 11th will be related to reports from the DSMOs, as well as perhaps a general discussion with interested parties on issues related to how we can smooth administrative simplification and make the whole process work a little bit better. But then we will hopefully have at least a half day to talk specifically around issues of PMRI next steps, and hopefully there will be some testimony at that point.
DR. GREENBERG: Did we discuss a time line for the best practices?
MR. BLAIR: If I may, Marjorie, since I need to go, I just wanted to summarize our discussion today, make sure that I have properly summarized it in terms of what we are going to be doing in going forward, and then maybe could continue that discussion with Simon, okay?
DR. GREENBERG: Sure.
MR. BLAIR: First of all, thank you, everybody, for participating in what I think turned out to be a very valuable and productive session. I think we were all up in the air, not knowing where we were going to go and what we were going to be able to do. The things that I think we did achieve today is, we seem to have come to consensus on a couple of issues. Number one is that when we looked at those different categories in the discussion framework, we seem to have come to the conclusion that category one and two both offer directions that need to be accommodated. So we didn't wind up saying it is one or the other; we seem to have come to the conclusion that both of those characteristics and attributes are needed.
You might almost summarize that by saying that we need both an information model and a terminology model, and that the work effort that came out of that is that we are looking to Walter to give us a little bit of guidance as to approaches or directions or ideas as to how we wind up accommodating that. That would include some things that Simon indicated, that probably some testifiers need to be brought in in December, that could give us some additional guidance on how that could be accommodated. The other piece that I think we came to accomplishments is that Walter pointed out that some of the vehicle to help us go forward is by using use cases, so that we could not veer too far from pragmatic real issues as we try to identify our direction and strategy for going forward into the future.
Then another thing that I think we also came to the conclusion on is that there are a number of other aspects that need to be represented in an expansion or next generation in Jim Campbell's model, including how do we make sure we accommodate things like a nursing perspective as well as a medical perspective a la patient focus, and that we include probably more than one single convergent terminology. I think that Walter might be able to take us to at least a next conceptualization of what that might be. Did I summarize these things properly? Is that okay? Yes? Then thank you all.
DR. COHN: Marjorie, what was your question or comment?
DR. GREENBERG: Just that I know the committee is expected to provide effective solutions for complying with the administrative simplification standards. If I recall, I guess we are going to get a report from CMS at the December meeting on their preliminary analysis or whatever of the ASCA submissions.
But obviously, if everyone has to start testing by April and implementing by October, providing these solutions needs to be timely, as opposed to --
DR. COHN: I absolutely agree. I think what we heard yesterday is that WEDI was going to be submitting to us their suggested list of what they believe they have in relations to best practices and publishable pieces.
I think we also just generally asked if others in the industry had things that would be useful. We also specified a fair amount of interest in the Utah Health Information Network HIPAA savings calculator that we somehow needed to create a link to or publicize the existence of, because we thought that was a useful piece. We saw there were two key issues. One was vendor, one had to do with education. So I think we were trying to move forward on both of those.
I'm sure that will be discussed probably further in November when we have some more data hopefully from CMS. I was expecting that we would, time permitting, have some time for us to discuss what the results of the analysis of the data that we received around the compliance delays was. So I am seeing that as sooner rather than later, not knowing what the schedule looks like in November, so hopefully we will have a chance to do that.
DR. GREENBERG: Okay.
DR. COHN: Are there other issues? We have hit 12:30. Are there any other issues coming before the --
MR. BLAIR: I guess I am away from the microphone, but Marjorie, who is the best person to initiate the resolution of the issues for the scope of the contract? Is that going to be you or Michael?
DR. COHN: Jeff, why don't we adjourn the meeting, and you can have a conversation with Marjorie about this. Thank you all. The meeting is adjourned.)
(Whereupon, the meeting was adjourned at 12:30 p.m.