[This Transcript is Unedited]
Hubert H. Humphrey Building
Room 705-A
200 Independence Avenue. S.W.
Washington, DC 20201
Call to Order, Introductions & Announcements - Simon Cohn, M.D., Jeff Blair
Discussion on Scope and Criteria for PMRI Terminologies
DR. COHN: 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, member of the full Committee, and the National Director for Health Information Policy for Kaiser Permanente. I want to welcome Subcommittee members, HHS staff, and others here in person, recognizing how bad the weather is out in Washington today. I also want to welcome all those on the internet and say you probably made the right choice listening in on the internet as opposed to trying to get in here today. I want to remind everyone to speak clearly and into the microphone.
This morning we will continue our discussion on scope and criteria that we will use for decisions on PMRI terminologies and next steps. Jeff Blair, our Vice Chair, who’s in charge of the PMRI areas will lead this discussion and the session, but obviously we want to start with introductions around the table and then around the room. I also as always want to remind those on the National Committee if as part of your introduction, if there are any issues coming before us today where there’s a conflict of interest, would you so publicly state and indicate that you need to recuse yourselves from either the discussion or any voting. With that, shall we do introductions? Jeff?
MR. BLAIR: Jeff Blair, I’m the Vice President of the Medical Records Institute, Vice Chair of the Subcommittee on Standards and Security. I’m a member of HIMSIS, HL7, AMIA, ASTM, and there’s nothing that I think I need to recuse myself of.
DR. HUFF: I’m Stan Huff, I’m with the University of Utah and Intermountain Health Care in Salt Lake City. I’m a former Board member of HL7, Co-chair of the LOINK Committee, former advisor to the SNOMED Editorial Board. I would probably need to recuse myself if we talk too much about LOINK. Other than that I think I’m ok.
DR. FITZMAURICE: Michael Fitzmaurice, Senior Science Advisor for Information Technology, Agency for Healthcare Research and Quality, liaison to the National Committee and staff to the Subcommittee on Standards and Security.
DR. STEINDEL: Steve Steindel, Senior Advisor for Standards and Vocabulary, Centers for Disease Control and Prevention, liaison to the full Committee and staff to the Subcommittee.
MS. HUMPHREYS: Betsy Humphreys, National Library of Medicine, staff to the Subcommittee.
DR. SUJANSKY: Walter Sujansky, independent consultant and advisor to the Subcommittee on terminology standards for patient medical record information.
MS. BEBEE: Suzie Bebee, Health Informatics Specialist, National Center for Health Statistics, CDC, staff to the Subcommittee.
MS. GREENBERG: Marjorie Greenberg, National Center for Health Statistics, CDC, and Executive Secretary to the Committee.
DR. ZUBELDIA: Kepa Zubeldia with Claredi. Member of NCPDP and X-12 and I Chair the Association for Electronic Health Care Transactions.
MS. SQUIRE: Marietta Squire, NCHS and staff to the Subcommittee.
MS. WILHIDE: Sheryl Wilhide, MAGNA Systems.
DR. COHN: Welcome all. Jeff do you have some opening remarks?
MR. BLAIR: This morning we’re going to be exploring the framework for PMRI terminologies, and the purpose of this is for us to get to the point where we are beginning to have some consensus on the criteria for selection of PMRI terminologies. This activity is within the context of the HIPAA directive to the NCVHS that the NCVHS will study uniform data standards for patient medical record information and electronic exchange of that information, and make recommendations to the Secretary. And we have made those initial recommendations in a report to the Secretary in August of 2000, where we set up the framework, the guiding principles for selecting PMRI standards, and a set of recommendations for accelerating the development of PMRI standards and their implementation.
In February of this year, it’s still this year, in February of this year we built upon that foundation and we produced the first phase of PMRI standard recommendations, those were for PMRI message format standards. And in August of this year we began our process of exploring what we should do in terms of recommending PMRI terminologies. Just so that we have some feeling for the dates and the frameworks and how this fits together, today we’re looking to wind up driving this to the point where we are deciding on what the criteria for the selection should be. The prerequisite for that is for us to agree on the framework or categories of PMRI terminologies.
We are expecting to send out a questionnaire to the terminology developers, which are derivatives of that criteria, by January 10th. That would give them five weeks to be able to fill in the questionnaires, the terminology developers, than our consultant, Walter Sujansky, will have five weeks to be able to compile, assemble, follow-up on the information that we derive from those questionnaires, and report to the Subcommittee by I think it’s March 24th or 25th. Then in May, we’ll have another hearing where we will be hearing from the users of PMRI terminologies, including the vendors and providers, and other users of the terminologies.
During June and July we’re going to be following up with any additional expert testimony that may be necessary to clarify any loose ends and also begin to develop our initial drafts of our recommendations to the Secretary on PMRI terminologies and it is expected, our target is for us to have those recommendations ready by September of this next year, September of 2003.
So that’s basically what we’re on track for. I might mention that before we ask Walter to step us through the update of the document that he has been working on, this is the second draft of the document to see if we agree on this framework so that we can go forward with the criteria, before we do that Steve Steindel has indicated that the Consolidated Health Initiative Project, CHI Project within the federal government, has been working on some other standards issues, terminology standards issues, so he has a document that he’d like to make us aware of, he’s handed that out, but Steve maybe you can take five minutes and sort of step us through that.
DR. STEINDEL: Thank you, Jeff. I just passed out that document for information to the Subcommittee as an information item, describing the results of a work group that just reported to the Consolidated Health Informatics Project. The charge to that work group was to look at the various terminology domains of importance to the electronic health record, prioritize those domains, not in terms of what’s important to the electronic health record but in terms of how quickly work groups could be established and present results to the CHI Council on recommendations for terminology in those areas.
The work group looked at various source documents, one of which was their own internally generated list of domains that was used for a budget. The other was work from Connecting for Health. And the third was notes taken from our Subcommittee hearing in August that was summarized. And one of the main problems of using these various sources was one of terminology, as one might expect, because people refer to similar domains by different names, and there was some looseness in interpreting whether they were the same or not.
But based on those source documents, we developed four priority domains, all of which we feel are important to the electronic health record. This is in terms of how quickly work groups could be established and how quickly results of those work groups could be presented. So the CHI was looking at a different charge than the charge to this Subcommittee. It’s presented as an information item of another set of domains that people are thinking of.
MR. BLAIR: Could you tell us what the charge is?
DR. STEINDEL: I was just about to go through that, Jeff. The first priority were those domains that we felt that terminologies were reasonably apparent and work groups could pick and select appropriate terminologies to recommend for CHI. Remember that the goal of CHI is the interchange of health care data among federal agencies, not within an agency, so it’s also a different scope than what this Subcommittee is looking at.
The second priority were domains that we think are very important to the electronic medical record, but terminologies may not quite be as mature as those in the first work group. I’d also like to point out these are assessments of the work group that was putting this together, these are not consensus ideas or anything, just as a working document.
The third priority were those domains that we felt that terminologies really, if they existed, existed in fashions that needed a lot of work, and in some cases the terminologies do not exist at all.
The last priority, and this came about within discussion of the work group, were terminologies that we felt that were essential for the electronic health record, but we realized that other groups maintained those terminologies already and they do exist. And they just should be adopted as federal interchange standards. An example of that, of course, is the various code sets associated with the HIPAA process.
Just quickly going down the various priorities, and we listed out sub-domains for information purposes. One charge, and one thing the work group realized it could not do, was exhaustively define what these domains are, and the work group also realizes that when specific groups are looking at the terminologies they may realize their terminologies or their charges overlap with some of the other work groups. It’s very hard to specifically define it. And what we’re hoping and what the work group was hoping in those cases was that the final result of the individual work groups looking at terminologies would probably be a similar set of terminologies, and then the Council could decide which would be the most appropriate.
In the first priority we listed demographics, diagnosis and problem lists, interventions and procedures, laboratory result contents, medications, and text based reports. If you notice, we did not mention laboratory result names, and I think many people who are in this room or are listening on the internet, realize that that’s probably a low hanging fruit and the reason that does not exist is because there are already bids and approved CHI recommendation, recommending lab LOINK for that purpose, so it did not need to be repeated. And I believe by the end of this week that should be actually an approved CHI, it’s in the final stages of approval.
In the second priority, we had encounters, history and physical, immunizations, population health, and population health.
In the third priority group we listed genes and proteins, multi-media, which we included various wave forms descriptions in that, nursing physiology and supplies. And I noted in the reading, supplies came up in our discussions. This usually does not appear in discussion of EHR terminologies, but I notice Walter included it in his document as well, and I was very pleased to see that. It’s something that most people do not consider and actually is very important, and we do not have good terminology in that area.
In terminologies used by other processes, we listed as group categories billing, and we noted that HIPAA was an area there. We listed chemicals, disability, scientific and fundamental and units. And we felt the terminologies do exist for all those areas.
And we are putting together actual work groups looking at six terminology domains, they probably will be formed sometime between now and the first of the year, and start their work hopefully soon after the first of the year. The domains that were selected for the work groups were demographics, interventions and procedures, laboratory result content, medications, encounters, and immunizations.
If there are any questions I’ll be happy to answer them.
DR. COHN: Jeff, can I ask a question? I was actually just curious about what you mean by encounters.
DR. STEINDEL: So are we.
DR. COHN: The reason I’m asking, obviously, on one level you sort of go geez, is it a short notice encounter, is it an urgent encounter, I don’t know that that’s what you mean is how to categorize encounters, and yet when I look at encounters I go well geez, an encounter is really diagnosis, intervention.
MS. HUMPHREYS: Our discussion was that that was the one that had the tentacles everywhere and it would be useful for people to start thinking about how they were going to define it in this group. But everyone’s reaction was very similar to yours.
DR. STEINDEL: And actually it was a reaction of the work group. When we were forming the various domains, encounters came to mind actually when we were discussing billing and the HIPAA transactions, because encounters is a very prominent feature in the billing transactions, and there was a distinct realization that the clinical encounter is different than the billing encounter and so we said ok, well we need to make it a domain. And then once we made it a domain we realized we could not really put our arms around that domain very well which is why it is not in priority one and why it is in priority two, where people need to define it better. And that was the reason why, as Betsy pointed out, it was selected as one of the first domains we were going to charge a work group with looking at, because we feel that the definitions that come from encounters will help drive the definitions and terminologies used for diagnosis and history and physical, because there’s obviously overlap.
DR. HUFF: There’s been a lot of work within HL7 on this. HL7 couldn’t get anywhere with transactions until it distinguished between a billing encounter versus the face to face, multiple face to face interactions with clinicians at a given facility, the difference between that and people being seen at different buildings within the same campus --
DR. COHN: Or versus telephone, versus internet, versus --
DR. HUFF: And so, this is, I don’t know that everything is done, but the ground has been plowed at least once in terms of trying to define the distinctions between all those things.
DR. STEINDEL: And that’s why it wound up in priority two and not priority three because we were aware of the HL7 work in the area.
DR. COHN: Would you go through again the list of things? You essentially list them by priority, but you were also taking another cut but what you feel is urgent enough that you’re establishing work groups.
DR. STEINDEL: It’s not exactly what was urgent enough, there was a deliberate reason. The Council wanted to select some low hanging fruit priorities, those in priority one, that we feel they are all urgent, in terms of the electronic health record. The priority grouping is in a sense how long and how much work the work groups looking at each domain will require and Council, just because we have not gone through the process, wanted to select domains from some easy areas and from some harder areas so we can get some experience on how long it would take, so it would be easier to assign work in the future.
DR. COHN: Could you again read the list?
DR. STEINDEL: I’d be happy to.
MR. BLAIR: Before you read the list, could I just get a little clarification because I’m trying to have an understanding of how we should understand this in relationship to what our goal and our mission is. If CHI is going about selecting terminologies and you’re on a faster track than we are, I think that we certainly wouldn’t want to be out of sync with what you’re doing. Is this something where we should consider modifying kind of our work plan or something? Could you give us some guidance on that?
MS. HUMPHREYS: There was a lot of discussion around the fact that no one involved in this CHI effort wishes to have the federal government come up with a set of recommendations which are going to be dissimilar from what the NCVHS would conclude is reasonable based on input and evaluation of what they’re hearing from the field or from this group. I think my sense about all of this is that the CHI wants to, in some sense while there is a very supportive environment for it within the federal government to move ahead, because I think we all would agree that federal adoption or even identification of the target, where the federal government is going, is something that’s going to really help adoption. But my sense that it’s going to be more of a going back and forth between the two efforts and that as you have testimony or whatever in your various areas, you can just bring in one of these groups, if in fact they’re far enough along to help you. And in other cases, you may get, depending on which is the priority or what you hear in some areas, because there’s a lot of areas here, it seems to me that you may get to them before the CHI is able to.
MR. BLAIR: So they’re kind of complementary.
MS. HUMPHREYS: I think that’s what it is, and I think that’s why it’s important for you obviously to know once these groups are established what their plans are and timelines. Steve would you agree with that assessment?
DR. STEINDEL: That’s why I looked at you to give it because that’s exactly what, there are two, I think, we have not really discussed the process of how this is going to happen within CHI, but one thing that the work group felt very strongly about was that there is a lot of private sector experts, that know terminology in these various domains, and we recommended to Council that the work groups be allowed to seek help from those experts in forming their decisions. And also, NCVHS and the Subcommittee is acting as our advisor, and I really feel, and I would recommend though it was not part of this report, but as the process becomes further, that we would seek your advice on the domains selected by various groups. So as Betsy said, I see this is as a very synergistic back and forth relationship.
MS. HUMPHREYS: In case there’s anyone listening in who doesn’t realize it, obviously the very first thing the CHI did was attempt to move through to approval in the federal agencies represented the recommendations that the NCVHS had made to the Secretary of HHS on February 20. That was kind of like, well what is the first good thing we could do, we could adopt these very sensible recommendations that have been reached based on a lot of thought and public input. So I think there’s every intention of trying to keep these things in sync.
DR. COHN: Remind me again when the work groups will have their first report out.
DR. STEINDEL: Who knows?
MS. HUMPHREYS: It will be varying, I think, by the groups.
DR. STEINDEL: It will be varying by the groups. What I would like to see, and this is what I mentioned at the Council meeting, is that in the first priority area, I’m looking at that as about a three month process. And the work groups probably won’t be formed until after January 1st. Now in the other areas, I cannot give an answer because it depends on what the work groups, the terminologies I don’t think are as clear and it depends on how much work the work groups feel they need to do. In some cases they may actually have to define the domain before they can start even looking at terminologies.
DR. COHN: Encounters might take awhile.
DR. STEINDEL: Actually, immunizations might as well.
MR. BLAIR: Do you feel as if the timeframe that we had set up until now, which leads to recommendations by September of this next year, is that early enough to meet your needs?
DR. STEINDEL: I think that’s very synergistic.
DR. COHN: Well let me make another comment also on something we need to think about as we look through our paper. To me, really, I mean speaking for myself, I think 75 percent of the value of what we’re trying to do, I think, may be setting up the framework and giving people the big picture of how everything fits together and what the requirement should be around that. And once again, I think if we do that right, it will cause everybody to sort of understand as we put the various pieces in. And it may be that at the end of the day, recognizing that the amount of horse power we have are the people that I look at around the room, all of whom also have full time jobs, and variable consultant, only one consultant not a group of them, how we actually put the various pieces inside that framework, we need all the help we can get. So I sort of look at this as a very valuable, and I sort of looked at your report last night and this morning, and said geez, this is a very valuable additional cut that gets down into sort of greater specificity, and might help put some meat on the bones, pardon the expression.
DR. STEINDEL: And Simon, my reaction right now to what you just said, was the main reason I passed this out to the group was we were looking at this group for the exact same reasons.
MR. BLAIR: Is Clem with us hopefully?
DR. COHN: No. He’ll have a chance to re-discuss it I’m sure.
MR. BLAIR: I think at this point it probably would be helpful if Walter steps us through the updated draft and --
DR. COHN: The one question I had for Steven Steindel, which I wanted him to go through, the domains that the work groups are immediately being formed on, and I just didn’t catch that?
MS. GREENBERG: What did you say, Simon?
DR. COHN: There was a list of domains that there are work groups that are immediately being formed, and I just wanted him to repeat that group of names if he would.
DR. STEINDEL: Demographics, interventions and procedures, in particular they’re interested in laboratory orders, laboratory result content, medications, encounters, and immunizations.
DR. COHN: Ok, thank you.
MS. GREENBERG: This may have already been said, I think in particular the CHI looks to the NCVHS for the external input and maybe the work groups as I understand it, maybe identify areas where they really would like to get some public testimony, although I think the work groups can individually identify, contact individuals consultants or experts as needed, any kind of organized way of trying to get input from the private sector needs to be done through this group, so I think that’s just part of the kind of interactive plan that we have. Of course, I know you’re already living in Washington but hopefully we could fit that in with your scheduled hearings that have already been set up.
DR. COHN: I was actually going to comment that I know I speak for Jeff when I say for him that he’d be happy to get a long term apartment for the next six months in Washington.
MR. BLAIR: Just about, that’s true, that’s true. Is there are any other comments before Walter begins?
DR. HUFF: I’d like to just make sure I understand the use of this document so that my comments are in context. I think, as I understand it, this document would be used obviously for the Committee --
MR. BLAIR: Are you talking about Steve’s document or Walter’s?
DR. HUFF: Walter’s. This document would be used to guide what the Committee’s doing in the Committee’s process. Do we intend to also send this document out as background for the, with the questionnaire? And would we at some point send this or some revised version of this to the Secretary to explain our process? Or is it only really just to guide the Committee?
MR. BLAIR: The way, and we could modify this as we go along, I was envisioning this as a working document that would enable us to come, that would be an enabler to help us come to consensus on what the criteria of selection should be that we could fold into a questionnaire as the first part of our information gathering activities. I felt before we came up with those criteria we needed at least to have a general consensus, at least a general consensus on how things related to each other, what the major categories might be, because we might have more than one set of criteria, there might be one set of criteria for a core, and another set of criteria for other terminologies where we might not be as demanding on selection.
In terms of when the questionnaire goes out, what I’d really like to do is prepare a draft questionnaire where the Subcommittee could take a quick look at that, we’re on a short time schedule to be able to do that so it might be that very first week in January before I send it out, and there would have to be a little bit of introductory information in it, but it was not my intent to provide this entire document. What are your thoughts on that? Do you feel comfortable with that or would you rather we do it a different way?
DR. HUFF: Oh, I think either way would work.
DR. COHN: Jeff, just let me check. Dr. McDonald just phoned in, Clem can you hear us?
DR. MCDONALD: Yes, at least I can hear those three voices that have spoken.
DR. COHN: We’ll make sure to speak into the microphone so you and others on the internet can hear us.
DR. HUFF: My other question is just one of process. Do we want to have Walter go through the document and review it and then start taking comments or do you want to go through this and take comments as we go or what’s your preference on process?
MR. BLAIR: My preference on process would be for Walter to step through, because Walter has pulled together comments from Betsy and yourself and Steve and Clem and myself and so it’s reflected in this document and he’ll be able to show us how the input from all of us have been assimilated and --
DR. MCDONALD: Does anyone have the sentence in the law that specifies how this fits? I think that could help guide us if we could have that in front of us?
MR. BLAIR: The sentence in the law?
DR. MCDONALD: The specific sentence that speaks to what this is.
MR. BLAIR: The NCVHS will --
DR. MCDONALD: It’s about HIPAA legislation about PMRI standards.
MR. BLAIR: If I repeat it, that the NCVHS should study issues related to the adoption of uniform data standards for patient medical record information and the electronic exchange of that information.
DR. MCDONALD: I thought it said something about terminology more specifically.
MR. BLAIR: No, it did not. It said uniform data standards for patient medical record information and the electronic exchange of that.
DR. MCDONALD: Just say it one more time.
MR. BLAIR: The NCVHS shall study issues related to the adoption of uniform data standards for patient medical record information and the electronic exchange of that information, and I would never write a sentence like that, that comes from --
DR. SUJANSKY: Clem, this is Walter, actually that sentence also appears in the document, it’s in the third sentence of the document itself that I sent to you yesterday, if you don’t want to write it down in its entirety right now.
MR. BLAIR: Clem, by any chance do you have a copy of the document that Steve Steindel --
DR. STEINDEL: Wait a minute, I have in front of me Walter’s document. Is this an earlier one or recent --
MR. BLAIR: What I’m about to mention is Steve just gave us a brief briefing on the work of the Consolidated Health Informatics, Initiative, the CHI group that Jared Adere heads up, and they’ve been working pretty intensively on like a parallel path for their identification of terminologies as standards and even though it’s a repeat, Steve would you like to just bring --
DR. COHN: I think that’s probably best done off line. We’re faxing a copy of this, I guess I would resent having to take the Subcommittee’s time to go through it again just for Clem’s benefit at this point. I apologize Clem, but I think we can handle that off line.
MR. BLAIR: But apparently it is a work that is being done in parallel in our efforts which could benefit us as we go through which is something that is maybe an additional dimension we should be aware of as we go forward. Walter?
DR. COHN: Walter, I actually have a question. Are you giving us a presentation or are you going through page by page on this?
DR. SUJANSKY: I’m going to give a presentation summarizing the information so we don’t have to go through it page by page. Obviously I’m prepared to discuss the details of the document, and I suspect we will, but I don’t intend on reading it or going through it in detail at the outset. In fact, what I’d like to do is this morning, is to do just that. First review the major changes in the document relative to the previous version, then to summarize where the document stands now in terms of the scope of recommendations its advocating or its mentioning at this point, and then open up for comments and discussions at that point. Then we can decide to make changes or additions or what not at that point.
MR. BLAIR: Clem, can you hear Walter ok?
DR. SUJANSKY: And then subsequent to that, time permitting, I’d like to begin going over the specific criteria that we’re considering for selecting terminology, specifically for the core, what’s now called the core terminology group, both technical criteria and some of the non-technical organizational, licensing and so forth, criteria, and see how far we can get with respect to that. This is all leading up to the next step in the process, which is as Jeff mentioned, send out a survey to terminology developers really seeking information regarding those criteria to move to the next step of the selection process and so forth. So assuming that’s consistent with everyone’s expectations and so forth, I’ll launch in then.
Clem, of course, you can’t see the slides that I’m presenting.
DR. MCDONALD: I can picture them.
DR. SUJANSKY: Actually, I guess the slides are not sent out over the internet are they? So I will obviously be going through these one by one and mentioning them verbally as well. The first part of it will be summarizing the changes between the two versions. And these changes were pursuant to lots of good comments I received from a lot of the Subcommittee members on the first draft.
The first change is that the core terminology has been recharacterized as a small group of tightly integrated terminologies, this is in section three of the document. This was in response to comments I received from a number of people, notably Clem and Steve Steindel and Betsy Humphreys, that a single terminology was probably not possible or practical, so that concept has been changed.
The second major change, one of nomenclature, is that this notion of groups has replaced layers in section three and 3.4, and again, many people rightly pointed out that the layer terminology that was in Jim Campbell’s original conceptualization is probably not the best way or appropriate way to represent the relationships among these sets of terminologies, and so there really isn’t a hierarchical or a layering relationship, so the language has been changed to reflect that.
Also, at the risk of confusing everyone, I changed the numbering of the non-core sets, and what was previously layer three, the HIPAA and Administrative terminologies, I’ve made group two. And again, this was in response to some feedback that this is really a more important set with respect to the goals of this project than the other group, the other group representing clinical terminologies that would not be included in the core, and this is also probably a more, a clearer group, more clearly delineated group, and so I’ve promoted it in some sense to be group two. I hope we can quickly adopt that change and that that doesn’t confuse people. And that’s in section 3.2.
And then as I mentioned, what used to be layer two is now group three, and consists of legacy, clinical terminologies, and that’s in section 3.3, and I’ve attempted to better define and characterize that group as well as motivate the need for that group in this draft of the document. That’s the next point, and that’s also in section 3.3.
The other changes include in section 4.2, I actually changed the content of that section and removed the list of preliminary candidates for the core group terminology. Again there were a number of comments that it was premature to list the candidates, and some of the candidates previously listed were not appropriate and so I thought it best at this point to not list any candidates, and to wait until we’ve conducted the survey, decided on criteria and so forth to generate a list of candidates, and a short list of candidates, and so forth. So instead in that section I’ve placed an initial list of the domains that we’d like to cover with the core terminology group, again to help guide the selection process and to help us start thinking about which terminologies we will require to cover those domains, and that’s in section 4.2.
And the last major change was a new section was added to discuss the role of the UMLS, and Betsy Humphreys had some very good suggestions about what the appropriate role should be and I agreed with those and I included a section discussing that in this version of the document.
There are a few other changes here and there. For example I added reference to the notion that whatever hierarchies are included in the core terminology group with the terminologies themselves, they probably will not be adequate for all data analysis needs, so the model needs to accommodate the addition of local or customized or task specific hierarchical relationships in some way. And a few other comments of that scope and nature.
So where does the document stand now? What is the model that we have this point? What is the scope of recommendations that will come out of this process as described in the draft document at this point? So let me review what that is.
The furthest document, the Subcommittee and the NCVHS will be making recommendations regarding first a standard terminology architecture, and again, the current proposal is the architecture as described in the document. Also the composition of the terminology groups, the three terminology groups. The degree of integration of terminologies in the core group, and perhaps some comments as to how that integration will be achieved and maintained as the maintenance process occurs. What the organizational model for governance, licensing and the ongoing maintenance of the core terminologies might be. What the relationship of the core group of terminologies to the other two groups should be. And what the relationship between the recommended terminology standards and message standards might be, and why such a relationship is important. All of these points are touched upon in the documents, and again some initial proposals regarding all of these points are made in the document, but the main point here is that the document proposes that the Subcommittee and the NCVHS made recommendations in these areas. So that defines a scope of the recommendations that will ultimately be made and we should come agreement on at least that today, hopefully.
So at this point I welcome comments and so forth on the current state of this document and what I just presented.
MR. BLAIR: May I solicit thoughts first from Betsy, because there was a significant change in the way we characterized this based on the comments you received from her and I wanted to make sure that she felt comfortable. Is there any other changes or alterations, Betsy, that you would like to see?
MS. HUMPHREYS: I thought it was, I like this version of the document. I mean I’m sure others have specific questions perhaps, but I thought it was fine and that Walter had done a very good job in addressing my comments and also others that he discussed with me had come from other people, at least those that he conveyed to me, so I like it.
MR. BLAIR: Other comments?
DR. MCDONALD: General comments about the whole document?
MR. BLAIR: Whatever, Clem.
DR. MCDONALD: Well, I think this is a big improvement, that there are some things that I think are tactically or maybe not optimal. We base it I think, what we’re saying in 3.2 and 3.3 is that we sort of a decision made --
MR. BLAIR: Clem, for my benefit, if you refer to the numbers, I don’t have it memorized by number. What are the sections?
DR. MCDONALD: I’m actually expressing a confused thought which maybe we should erase all that and let me try again. The starting out goal is to take a set of terminologies for something, actually now that I’m saying it I’m not sure whether that something is for adoption by the Department, I’m not sure what that something is. But it looks a little bit like on the way to taking this criteria we’re actually making judgment, that’s the part that worries me about the group two and group three. I think the group two works easier because for one thing it describes them and characterize them, but group three, it almost, once we put something into group three, am I making any sense? It just seems like we may be confusing, we made it a circular or a sharian(?) thing here.
MR. BLAIR: Clem, could I make a comment on that because at least the view that I’ve had is just slightly different from what you just stated. I think we’re using this direction of selecting terminologies almost as a tool or mechanism. I think it’s probably likely that we may be able to make selections or recommendations in the core and in maybe category three, but I think there’s going to be many more where the terminologies will not completely meet our criteria for selection, and the recommendations that we make to the terminology developers, so that they can begin to complete this architecture in the future, I think is at least as important, maybe even more important, than the initial selections that we might be able to make.
DR. MCDONALD: Well, I don’t disagree with that. I think what I’m actually trying to get clear is when we say we’re selecting terminology, I’m not sure if I really know for sure for what purpose exactly. Are we selecting for adoption by the Department? Are we selecting them for just to talk about to the press, to publicize? If we clarified that a little bit it might help. But getting back to the point about these three divisions, it may, are we saying these are the classes we will judge terminology to go in?
DR. HUFF: I’d like to comment on this part, too. I think we could add some more, but I thought Walter did a nice job in defining sort of the purpose of this and who would use it in the use cases, and we can add other use cases and we can add sort of some more intermediate, I think in the intermediate level requirements, section 2.1, he described how this would be used and expected users of it, I listed about five or six other things that we could put in there easily that would fill out who’s going to use it.
But the second part is, the same issue that you’re worried about was one I was worried about, the classification into these two groups, and what I thought is that I like the idea of a core, that’s clear, but I’m not sure that we’d gain anything by categorizing group two and group three, and that in fact what we might want to do is just say explicitly, if what we think is important is that there’s mappings between the core and CPT-4 and ICD-9, either rule basis or knowledge base, however complicated that is, just say what our intent is. Say the specific things we want to map and the specific capabilities we want it to have, and just say that. Then you don’t have to worry about whether it’s in group two or group one, just say we think it’s essential that we have an ability to map from the drugs here to the classifications that are needed for billing, say explicitly the things that we want to enable. Because I think, I wouldn’t want to lose a lot of time in groups arguing whether they’re in group two or group three, I would rather say well what it is we really want to enable. We want to enable this mapping between this terminology and this set of core terminologies.
DR. MCDONALD: That actually is I think in my point. I’m afraid it would just create friction by having those. I like what you just said to me.
MR. BLAIR: This is Jeff. The thing that concerns me about that is that if we come up with a criteria for the core, it’s going to be probably much more rigorous and demanding because of the functionality and the way the core terminology would be used, what we’d expect of that core terminology. And it would be probably inappropriate to use that same set of criteria for like category three, but if we don’t come up with a separate set of criteria for category three, then the message we’re sending out there is whatever you got out there is fine, there’s no need to improve it to improve clinical specificity, to improve the interoperability and the mapping between the terminologies, so that was I thought why we would want to --
DR. HUFF: I guess I don’t quite understand that, because we clearly need criteria for the core. The other things that we’re mapping to are not justified at all by their structure or their content. They’re justified because they’ve been mandated for other business uses and/or they’re so entrenched that we can’t change them. So we’re not adopting them at all, all we’re really saying is that given we have strong criteria for the core we need to define the other mappings and other kinds of relationships to these other terminologies because we can’t recreate the world. They’re essential and so we’re not adopting those other terminologies, we’re recognizing that they exist and that we need to have mappings to them to make the core more useful. So we have to have strong criteria for the core, we don’t have to have criteria for ICD-9 codes, they’re mandated for use. We don’t have to have any criteria for group two or group three, we just say they exist, we don’t need any criteria for them, we just say which mappings we think are essential to make this terminology more interoperable.
MR. BLAIR: Well maybe it’s because I can’t see it, but my understanding is that group two is basically the HIPAA identified, the billing codes, the reimbursement codes, and I absolutely agree with you that there’s no need for us to take time to develop criteria for selection of those, they’ve been selected. The ones that I was thinking of was the other clinical domain specific terminologies, such as the nursing code sets, such as probably the IEEE device code sets, although we specified those as message formats, but there’s codes involved with that, maybe DSM codes for behavioral, those were the ones that I was thinking of --
DR. MCDONALD: We don’t even know if some of those may not end up in the core.
DR. COHN: Betsy has a comment and then actually I have a comment.
MS. HUMPHREYS: I have two comments. One is to get back to Clem’s original thing about why are you recommending these. I thought I understood this perfectly clearly, Clem, so let’s see. It’s because this Committee has put forward a series of two reports. And the first report said hey, we’re going to use the following process and we’re going to recommend patient medical record standards, and we don’t think you need to issue regulations but we think there should be guidelines and the federal government should be early adopters, and it was a fine report and you guys produced it.
And then last February you sent in the first set of recommendations, so I feel that what you’re up to now is the next set of recommendations that go into more vocabulary, so I thought that was crystal clear as to why you were doing this, it was so you could send in your next set of recommendations. And at least at the moment it appears that this process is working quite well because I think we’re going to get HHS and the DOD and the VA to adopt your recommendations, your first set very soon, so hey, this sounds like it’s working, so keep up the good work. That’s why you’re doing it I think.
DR. MCDONALD: No, I’m glad to have it restated. I thought I knew, too.
MS. HUMPHREYS: Good, I hope we all do. On this other point, it seems to me that maybe what I’m hearing is that this document should be focused on the selection of the core and there should certainly be some statement, but maybe you don’t need to call it group two or group three, maybe it’s the issue that one of the things that has to happen with the core is that where it’s appropriate we’ve got to have functional mappings between this core and the HIPAA code sets because hey, the HIPAA code sets are mandated and they’re not going away tomorrow afternoon so in order for this to work well they have to be there. And then maybe not a group three but just an acknowledgement that other terminologies are out there that may not make it into the core and that there will have to be some accommodation for mapping them or for the fact that they may continue to be useful for local purposes. And maybe it’s this kind of elevation of it to group one, group two, group three, which is actually causing more problems, more than the actual substance of what it says in Walter’s paragraphs.
MR. BLAIR: Has Betsy captured your thoughts, Clem, and yours Stan, and if so --
DR. MCDONALD: I’m one for personally, and I think the more work you have out here the more we’ve got to wordsmith it, the more we can trip up on it. I would be sort of a compromise between what Betsy and Stan said, to still try to reduce the text, we’re asserting things that we don’t have to, to do our main job.
DR. COHN: Clem, can I comment? Because I think I fundamentally agree with you though I don’t know that ten pages is an overwhelming amount of material. But I guess what I’m hearing, and let me try to say, because I think I also agree, and once again this is a scope issue, so we need to think about it and it will have impact on what sort of survey we send out. Because one survey would be geez, are you a category one, are you a category three, this could get out to hundreds of thousands of different entities because everybody has their own code set. But really what I think we’re beginning to talk about is that we really want to focus on this core terminology.
We also have what used to be called group two, it’s really now called HIPAA mandated terminologies, and there are some characteristics that need to occur between the two to make sure everything connects well. And then there are probably a set of requirements that we believe that other terminologies that seek to be part of this whole need to meet which in turn has to do with good mapping. Is that right, Betsy, or am I overstating?
MS. HUMPHREYS: I would certainly, Clem is getting at the point that he doesn’t want to be, come at a second level of selection or as to whether you actually meet group three or not, I’m definitely with him and Stan on that one. I mean I think you’re going to have enough trouble selecting the core group without having to nominate people to group three. I think it’s more an issue that yes, these things exist and something is going to have to be done going forward and we’re not so stupid that we don’t understand that. But I don’t think you want to actually spend your time nominating group three. This could take forever.
MR. BLAIR: Just one piece here, because it seems as if we almost have consensus here, the only party that’s missing I think is Steve. Do you feel comfortable with this?
DR. ZUBELDIA: We are talking a lot about mapping, and there are some of the, especially in what is now here group three, and some in group two, perhaps don’t have a mapping to other terminologies. They’re so exclusive to their own domain that instead of having a mapping what we have to do is draw a bright line and say for this domain that is the terminology that we have to use because that’s what we use today. So we don’t need to center into always having a mapping --
DR. COHN: Well, Kepa, is that a core terminology that you’re describing?
DR. HUFF: I think that makes it a core terminology, the way I’m thinking about it.
DR. ZUBELDIA: It could be a core for administrative uses, not so much the HIPAA terminologies, but it would make it a core terminology.
MS. GREENBERG: That just confused me, core for administrative. I thought core specifically was for --
DR. ZUBELDIA: I’m looking at the CDT-2, the CDT-2 is used for administrative use, but in its domain it’s the only thing there is for that.
MS. GREENBERG: Well, I was thinking in that category for example of ICF, which is this classification of functional status, that it’s also sort of the only comprehensive classification of functional status. I had some issues with the original three layers in any event, although I like this better, I would agree that what this group should, this Subcommittee should be focusing on, just as Betsy said very well, is you did the message standards, now you’re saying now we’re going to do the terminologies, and I think what you’re looking at is these core terminologies that can be used for the primary documentation of clinical care, or should be used.
On the other hand, I’m a little concerned about, and I have to see how it gets written up, but the tone of this conversation that well we’re stuck with these classifications that HIPAA mandated, but if HIPAA hadn’t mandated them we wouldn’t have any use for them or need them anyway, and I think there’s a lack of appreciation for what classifications are used for in this discussion and I wouldn’t want that to get into the document. And that when you have all this very clinically specific data for a lot of purposes you’re still going to need classifications, you might need better classifications than those that exist now, you might have to classify --
DR. HUFF: Let me apologize up front, classifications are essential, that wasn’t my intent and I hope that wasn’t the message you got from that. I just mean that it’s a done deal, you don’t need to talk about it anymore. They’re essential, they’re mandated, essential in the sense that they were needed and they were created for a purpose and they do a good job of that, so that wasn’t my intent.
MS. GREENBERG: They’re not just mandated for HIPAA --
DR. HUFF: It’s not arbitrarily mandated, I mean it’s essential for the business to work, those classifications have to exist and so there was no intent on my part to impugn those terminologies. They’re essential and probably as good as you can image for the purpose that they were created for.
MS. GREENBERG: Well, I’m not sure about that. But they are, they’re not just for claims, however, they are for statistical purposes --
DR. MCDONALD: Could I just comment on this? I think you’re reacting to something that Stan didn’t mean, and they are HIPAA mandated, and that makes it easy to just say so that’s that. That’s the purpose, not saying they are only for that.
MS. GREENBERG: But actually like this language here under 3.2 that, because I think there is a lot of misunderstanding. There are people who feel like if we’re going to develop or name these clinically specific vocabularies, why can’t we just forget about going to more updated or more current classifications because we’re going to have, there is tremendous misunderstanding out there about that, and so I think that the language here, not necessarily the listing of what the terminologies are, but some language that is not just related to billing and that explains the need for mapping and why that need exists, and the need for classifications I think would be appropriate. And if you want to do that in text as opposed to grouping things, that’s ok with me.
DR. SUJANSKY: If I could jump in and make a few comments that maybe will help us move forward from this point, summarizing and consolidating all of your comments. A little background, the intent and the goals of characterizing both group two and group three sets are not to select terminologies that will help meet the objectives of standardized terminologies for patient medical record information unto themselves. The assumption is that terminologies in those other groups in effect will not meet those objectives. It’s the core group that will meet the objectives of interoperability, comparability and so forth, that are the ultimate goals of this.
The purpose of the other groups is to acknowledge that other important terminologies exist and will continue to exist and must be accommodated by whatever core terminology that is ultimately developed. Accommodated in the sense of they must be acknowledged and there must be some provisions to help those who are using those and will continue to use those to also use the core terminology and to adopt the core terminologies for clinically specific data capture and data analysis.
So a way to summarize is the core terminology is intended to provide the function that we’re trying to provide, a new function that doesn’t exist now. The other terminologies and relationships to those other terminologies are intended to promote adoptions of the core terminologies so that we can say well, yes, last year the government said you need to use these HIPAA mandates codes, and this is how these new terminology recommendations relate to that. That we acknowledge that that’s important, the HIPAA mandated terminologies are important and that mapping will ideally be available to those terminologies from the clinically specific terminology and it’s all going to fit together and it’s all going to make sense and work for you, so people don’t say well what’s this new terminology stuff you’re now telling us we need to do, or it’s the right thing to do and how does it fit with what you told us before. That’s why group two is important. And I’d like to keep the discussion of group two, we can characterize it differently, maybe not call it a group of terminologies, but we can just characterize the discussion as what is the relationship of the core group to these other terminologies and so forth.
Now as far as group three is concerned, just to finish my thought, I am amenable to really dropping group three as a group of terminologies that we’re going to select member terminologies for. I think that’s going to create all kinds of problems anyway in terms of who’s going to be included and excluded and why and so forth, and perhaps replace that with a discussion of a liaison type function between the developers of the core terminology and the other terminologies that are not in the core but are important. For example, the drug terminology, the commercial drug decisions support system and perhaps other terminology, to make it more information but to acknowledge that it is ultimately going to be important to interoperate with some of these existing terminologies in order for this core terminology that we’re envisioning to gain acceptance and adoption. And some mechanism needs to be in place to promote that interoperability, but it needn’t be selection of specific terminology within that group as part of this. How do people feel about that?
DR. COHN: Let’s ask, Steve was in line to try to make a comment here.
DR. MCDONALD: Whoever was just talking, I couldn’t hear as much.
DR. COHN: Walter was just trying to put it together here, we’ll come back to it after Steve talks.
DR. STEINDEL: I think Betsey did a very nice summary of the direction I think it should go and it characterized Clem’s thoughts and Stan’s thoughts, and I actually think Walter also just summarized it fairly well now, and it also represents some of the problems that I had, in that it’s very difficult. I like the idea of group three from a theoretical point of view but then I started looking at it from a practical point of view and said how can I put anything in it. And I decided that it’s very difficult to characterize a terminology as a group three terminology, to do it in any type of reproducible fashion and from that point of view I kind of started getting shaky. And I think what we’re looking at is the core terminology and the other terminologies --
MR. BLAIR: Which is the other clinical terminologies is what you’re saying.
DR. STEINDEL: No, I said the other terminologies, and I use those words specifically, and I think Walter just summarized that there really are two classes of the other terminologies. And one class is the administrative terminologies, however we want to characterize those. And the other class is the remainder of the clinical terminologies, some of which exist for specific purposes, some of which exist for historical purposes, whatever the case may be, and I think we should group those other terminologies as one group and in our discussion of it distinguish the two classes. Because both classes are going to be treated the same way. They’re going to have to interact with the core, map to the core, there has to be some type of relationship to the core, so what we do with the other terminologies is essentially the same, and we should group them and discuss them the same. At least that’s my view.
DR. COHN: So which should we discuss the same, the administrative and the other?
DR. STEINDEL: We should say there is a core terminology and we should define as Walter has done the characteristics of what we consider to be core terminology. And then there’s this group of other terminologies which do not fit in the core, one subset of the other terminologies is the administrative terminologies, what we’re now calling layer two. And the other subset of the other terminologies is what we have difficulty characterizing that we’ve been talking about as group three. And we’re now looking at this as one group and the function of what we do with the other terminologies with respect to the core terminologies is the same, whether they are administrative or non-administrative.
DR. COHN: I agree with part of what you said but don’t agree with other parts because I do think that there’s a value to specifically identifying the HIPAA mandated terminologies and showing how they all fit in. Now, Clem, are you hearing us as we talk?
DR. MCDONALD: Well, I heard all that. My concern is that if we, the reason I want it simpler is, if we set out a course where we’re going to make the perfect complete coverage of everything, all we’re going to do is attract a huge amount of contention and we won’t be able to get it done, so I just think we need to make it simple and get something done --
DR. COHN: I have an interesting proposal, and you guys can all boil me in hot oil if you want to, but once again, I think right now we’re talking about concepts. I think we’re sort of getting views of how of what we’re talking about here and we know we’re really focused on these clinical terminologies that put things together, so I’m not arguing with that. Now I have a little problem. Let me tell you sort of the reality as I know it.
The reality that I know is that geez, there really is not a strict line between administrative and clinical terminologies, and classifications and nomenclatures and all of this stuff. And geez, when I’m practicing clinical medicine, the truth be known I actually do look at ICD diagnoses and they help me sometimes take care of the patient, and I’ve even looked at Volume three procedures sometimes, and they actually help me take care of the patient. And similarly I actually sometimes look at NDC drug codes when I can figure out what exactly they mean, because they sometimes tell me what drugs the patient was on, which is clinically important to me.
And so I guess one of the things I’m wondering as we talk about all of this stuff, if maybe we are somehow, I mean I know that we are all sort of clinically focused and we’re talking about this core terminology, and once again, I just try this on for size, this may be a very bad idea, but maybe there is a core terminology that maybe at the real core is sort of these HIPAA mandated terminologies and classifications and nomenclatures, but the problem really is is that they aren’t alone enough, and we need extensions out to make things really clinically valid, useful, helpful, because obviously if ICD and CPT and all of these other things were enough we wouldn’t be having this conversation. So maybe really the terminology that we’re talking about starts with those mandated pieces, goes out a ring wider to really talk about all of the core additional, Stan may kill me on this one, and then we have further extensions --
DR. HUFF: I don’t think that’s a good way to characterize it. I think that’s a really bad way to go, Simon.
DR. COHN: Well, I’m just sort of thinking about reality as I know it.
DR. HUFF: I think your observations are accurate in that again, they’re extremely useful. But, and you know things about the patient by what was said there. But if you get into the details, you wouldn’t be able to take care of a patient, you wouldn’t be able to prescribe the drug, the one you want them to take by using those coding systems, and you wouldn’t be able to know, if you look at the details of the coding systems, if you take things at surface value, they look like they mean something that they don’t mean.
In other words, you look at a thing and it says asthma, but then you actually realize and read these coding books, you realize that that’s only asthma if you’re over 19, that there’s a different code if you have asthma if you’re pediatric age group or there are all of these nuances of meaning that are added by the coding handbooks that tell you when to code that particular code. All of those nuances are part of the meaning of those concepts, which makes them not appropriate for documentation of primary care or for message exchange.
Again, it’s absolutely essential that you know the relationship between the kind of asthma you mean when you’re documenting that this person has restrictive airway disease and know that there’s a difference though between that and the particular asthma code that you assign for purposes of either ICD billing or DRG grouping or for morbidity and mortality statistics. So I don’t think you can view those, I think they’re core in the sense that they’re essential. I think it just confuses things a bunch to think about them as core, as primary documentation of health care. I think that would be a bad thing to make into the thing.
DR. COHN: As I said, I’m just going to go out and stick my arm in hot oil, I don’t have to stick my whole body on the basis of your comments, Stan. I think Steve Steindel has an arm up.
DR. STEINDEL: Simon, if I can comment, I think it’s a very useful discussion, but as Walter pointed out at some point in the document, what we’re really talking about here is an information model, not the terminology itself. We can discuss an information model starting from the clinical documentation and working out to the administrative documentation, we can discuss an information model starting from the administrative and working out to the clinical. But I think that’s a totally separate discussion from the purpose in which we’re going on this particular recommendation of terminology for the clinical patient record. And I think when we look at it from that point of view, the terminologies we’re looking to recommend are those that describe clinical situations, and those are the ones that should be recommended as part of the core. The administrative terminologies don’t do that good a job. While they may be helpful in assessing clinical situations, don’t do that good a job of describing the clinical situation.
MR. BLAIR: May I interject at this point? Because as I’ve listened --
DR. COHN: I pull back my comments, you never heard it.
MR. BLAIR: I think, I’m going to make an attempt to see if I can articulate what I hope we could all agree on because from a pragmatic standpoint, we don’t have to do everything in today’s session in terms of defining this framework. We only need to do enough that we could agree on our next step and that it be generally useful for us as we go forward. I think I’ve heard a few things. Number one is that the only area where we need to define criteria for selection is for the core. Does everybody agree with that? Ok, so we have that agreement.
Number two is that even though we won’t define a criteria for selection or do any selection for as Steve would say "the other terminologies" from the standpoint of eventual recommendations next summer, we probably will wind up indicating some guidance or some recommendations related to how UMLS will facilitate mapping to those terminologies that are not in the core. Is that correct? Yes, ok.
So I think if those two statements are correct, then the framework of the other terminologies could be defined in a very general way, just a very rough way without us giving examples or fleshing that out, and that I think, unless you all tell me that we still need to thrash this thing out some more, I think that we could go on to examining and trying to agree on what the criteria should be for the core.
DR. HUFF: Well, I think there’s another issue that I want to bring up before we get to that, and I’m already repenting my harsh treatment of Simon. The thing that occurs to me is that maybe there’s a better way. Saying core, maybe we need to think about core clinical terminology and core administrative terminologies or something, because it makes it sound like the other things aren’t right in some way, or that they’re peripheral to the purpose and intent, and in fact I think they’re essential. And so maybe there’s sort of a core clinical versus core administrative or something like that that would make it clearer to people that they’re just serving a different purpose, they’re not any less important. We’re not going to select them, even though we’re not selecting them if we call these core clinical terminologies or something instead, so that it was clear to people that it didn’t mean that the other things were not core.
PARTICIPANT: That’s good, why don’t we call it group two?
DR. HUFF: But anyway, I think there’s bigger fish to fry than this.
DR. ZUBELDIA: Let me add to that, because I think that we can address it very simply by saying core terminologies that have not yet been adopted as standard under HIPAA. We already have some that have been adopted as standard, terminologies that are standard under HIPAA, there’s some that have been adopted, ICD-9, CPT, HCPCS and so on.
MR. BLAIR: They’re not considered clinical in this sense.
DR. ZUBELDIA: We don’t need to adopt them again, and all we need to do is focus on terminologies that have not yet been adopted --
MS. HUMPHREYS: Yes, but you have to very, you really have to distinguish between your activity here as at least based on your previous report and recommendation leading to something that might be adopted in the government as guidelines but not regulation. If you were to relate this to the fact that these are things that haven’t been adopted for HIPAA yet, then you will cause great consternation because people will feel that you’re about to recommend them as a HIPAA code set, so you don’t want to get into that game.
DR. ZUBELDIA: Perhaps we don’t even need to make a distinction because we can say that our work is not terminologies that have already been adopted by the Secretary, this is new stuff, for clearly core terminologies, we’re not trying to adopt or discuss things that have already been adopted.
MR. BLAIR: But we already wind up having what the law has given us as a mandate and we’re looking at PMRI, for whether that’s good or not, we’ve been sticking with that. So I think that the PMRI terminology should be the phrase that we’re looking at here for the core.
DR. COHN: I think we actually have a majority sticking their hand into hot oil with me at this point.
DR. HUFF: I think it’s me that’s being fried now. I think the second, actually the most important issue that I wanted to see is first of all to recognize and compliment Walter on this new section on the UMLS and it’s potential role. The thing that I would like to do, though, is in fact strengthen and go beyond that. My thought is in fact the only reasonable way to go forward with this is to say that the core terminologies are going to be part of the UMLS metathesaurus and that the National Library of Medicine would have charge of taking those and in fact that they’re willing participation, willing if not eager, to in fact do, make structural changes or do other things that might, and I’m not even sure what those things might be, but if there are any that they would be willing to make those changes to make some subset of the UMLS, these core terminologies usable for clinical purposes.
And so what I’m really saying is, is that we state in this, we state as our default position that the way that this would be done is that the core terminologies would become part of the UMLS metathesaurus, that the National Library of Medicine would have responsibility for doing that, that as core terminologies that we adopt, they would have the responsibility for contract to make sure that they were available in the metathesaurus, so I like what is said here. I want to go even further, I want to essentially adopt that as the default position of how this would be done and that would allow us even to say that, or make suggestions that, additional funding and support for the National Library should be appropriate to support this cause and this purpose.
MR. BLAIR: Any objection to that?
DR. COHN: I think what you’re saying is relatively non-controversial in this group, I think we’ve actually already gone on record as supporting that and I think we all think that the UMLS is potentially a low-cost distribution modality for terminologies in general, which I think is once again aligned with I think your recommendations. Do I see anybody being upset about that particular recommendation?
I have another issue I wanted to bring up but if there’s anything on this one before we move to something else, Michael?
DR. FITZMAURICE: Only as a recommendation in a document that has scope and criteria, I think that may well be one of our recommendations at the end, but we should reach it as opposed to starting off with it.
DR. HUFF: The reason I think it’s useful because if that is our stance, I think during public testimony it would be good to have other people say, if there are other viable things to get their comment on that, or endorse it, so maybe we don’t need to do that to get public comment, but that’s why I thought it was important to have in this document and have as an underlying, and the other reason is that we could then ask other questions that would say how easy is it going to be to incorporate your terminology, when we ask for public testimony, how easy is it going to be for you to put your stuff in the UMLS metathesaurus and how would it integrate with the existing other things that we think are going to be in the core, etc. It puts it in the context where you can ask more specific questions and they can give more specific answers relative to that.
DR. FITZMAURICE: Well usually when we ask for testimony we give them a set of questions, so we could give them just those questions for testimony.
DR. COHN: Let me ask another slightly different question because obviously we’re talking about scope and criteria, and Jeff clearly wants to move into the criteria discussion at this point, and I just have a question about the scope questions. Which is is that it’s unclear to me where the edges are to all of this, and I’ve certainly seen a list, knowing that everybody sort of defines PMRI a little differently, we’ve all just now talked about orderables and supplies and all of this stuff and we could certainly get, I’m trying to think of what potentially couldn’t be within scope on this discussion, and knowing that everybody seems to define these areas somewhat differently, is there a need, what to do we need to do to get a little clearer about what might be in or out beyond use cases, or is that the way that we’re going to do it, and how many pages of use cases are we going to create?
DR. SUJANSKY: When you say scope I understand you to mean the scope of the domain coverage of the core terminology group.
DR. COHN: That actually is what I was referring to, and I’m actually sort of, I keep sensing that in all of our discussions, be it CHI with their tabular list or our views where we sort of talk about this supporting patient care and other things like this, that somehow everybody means a little something different when we’re talking about this one. I’ve actually been sort of scratching my head and I hate to use the term, it start with m, it’s called models, but I’m just wondering how we’re going to, if we have to send out a questionnaire, how we’re even going to know who to send them out to.
MR. BLAIR: This would be my suggestion. I’m not sure this is right but maybe. I think that at least the first or second cut on the scope has been derived in Walter’s document. I think in terms of what would be in the core, from the discussion we’ve had today, it sounds like, from what I’m hearing, that the core are those terminologies that meet the criteria for the core, and that if we start to go through the criteria, that’s when our boundaries will be defined as we come to consensus on what the criteria should be for core terminologies.
DR. SUJANSKY: Well, actually Jeff, I don’t disagree with you, but I do agree with Simon that we need to avoid kind of a circular definition with the criteria, because on the criteria is good domain coverage in its defined scope, and so we need to define that scope before we can apply that criteria. For example, do we want to cover the domain of genomic information? Do we want to cover dental information? I think we need to make decisions on points like that before we can apply the criteria, although I do agree that we can define much of the criteria before we’ve specified which domain areas ultimately we’re going to address.
DR. MCDONALD: I always worry about being too specific or too, as you’ve heard before Walter. I almost think we ought to, we’re going to get feedback, we’re going to have people saying we ought to include this and that, and I think the scope, if we get it too well defined people are going to be complaining that we don’t want to hear what they’re saying. I guess I would not want it to be explicit about the scope to exclude or include those things. Plus we’re going to end up going cherry picking, and we should explicitly say this is just the first round.
MR. BLAIR: And I’ve actually changed my view on this as I’ve listened to the conversation here because Clem, I think what you’re saying is that let the criteria itself define who gets picked into the core. If we agree on what the criteria should be and send the questionnaires out to all candidates, we’ll wind up seeing who meets the criteria and that process will define then who is within the core.
DR. MCDONALD: I think so. Some of these things, like, when you try to nail it all down at the beginning --
DR. COHN: Clem, I basically agree with what you are saying except there are two issues. One is is that I just sort of wonder who we’re going to send this out to, or do we post it on the web and whoever feels it’s appropriate for them respond. That’s sort of question number one. And question number two, even though I think at the end of the day what you’re describing is right for the first phase, that approach will not give us a very good indication of what is ready yet, but needs to be there, I wouldn’t say to complete the core, but sort of help us maybe identify things that need additional efforts. I did notice that Steve Steindel’s activity did begin to try to sort of parse that out of what needs additional work.
DR. MCDONALD: If the scope of this document is to do, both pick some things to get started and to figure out what all has to be done, that’s more than, that’s more work. I think that may be more burden that we have to do, but I don’t object to it.
DR. HUFF: Yes, I guess what I saw was that it would be maybe sort of ill-defined but what you would do, I guess I view the core as a slowly or maybe fast increasing set of things, then maybe forever increasing, so it’s not sort of going to be done next year or even in two years, that we’re going to come up with a list of things and it’s going to say the kind of things that Walter has enumerated and actually enumerated in the CHI document, too, that says we need drugs, we need diagnoses, we need other things, and we’re going to prioritize that list and we’re going to start on what we think is the most important. And as we start on that one you’ll get input from HL7 and from public health and from clinical trials people and other people who say oh, for diagnoses in my context, this is what I need, and that will help define it as we go, and so it’s part of the process. So I think if we can put up names that are sort of meaningful as a useful way of prioritizing them and then start the work, it will sort of self define by the people who we anticipate using it and they’ll give us input and they’ll send lots of email to Betsy and everybody about what’s not there and that sort of stuff.
DR. MCDONALD: Well, that suggests that Walter it would be good to get something in the document explicitly stating that this is a step wise process and it’s not a one time winner take all.
DR. SUJANSKY: As far as the domains?
DR. MCDONALD: As part of the process itself. We’re not defining, that we’re going to define a phase I, and they’re will be a phase II, and that has two advantages. It reduces the sort of the struggle to get onto the box car before it leaves, or onto the train before it leaves, but that’s the truth. This is an iterative ongoing process, this has to be.
MR. BLAIR: Are we at a stage where everyone feels comfortable for us to maybe have Walter begin to take us through the criteria?
DR. MCDONALD: Walter, the list of categories that you describe as domain coverage? Would you be open to some suggestions to organize those slightly differently?
DR. SUJANSKY: Absolutely. I’m open to any input.
DR. COHN: Clem, we apologize, we know you came on late, we do know you are I think in the same time zone at this point. I think many of us here feel that there’s a need for a break. Why don’t we adjourn for ten minutes and then we’ll come on back and continue the discussion?
DR. MCDONALD: What time are you coming back, what time do you have now?
DR. COHN: Ten to 11. So we’ll start back at 11:00.
DR. MCDONALD: Can I hang up and call back in?
MR. BLAIR: Why don’t you just leave it open in case we have difficulty reconnecting?
[Brief break.]
DR. COHN: We’re going to get started here, could we have every body come back? Dr. McDonald, are you on the phone?
DR. MCDONALD: I am on.
MR. BLAIR: Now, one of the things, is there anybody that needs to leave before 12:30?
DR. MCDONALD: I do. I have to, that’s my problem, there’s a meeting I have to go to.
MR. BLAIR: What time do you need to leave, Clem? Because what I want to do is I want to make sure that if anybody needs to leave early that we’ve heard any of their suggestions and comments before you leave.
DR. MCDONALD: I think you did. I mean I think Walter did a nice job on this.
MR. BLAIR: Stan? You have to leave when?
DR. HUFF: If we’re done by 12:30 I should be fine, or maybe close to fine.
MR. BLAIR: Then Walter, please take us through the criteria.
DR. SUJANSKY: Ok. So this next part we’ll be discussion specific criteria for selection for the core group terminologies and prioritizing the criteria perhaps and discussing which ones are appropriate criteria or not appropriate and what the criteria should be.
Now I’d like to preface this by reminding people about two things. One is what are these criteria for? And what is the goal, again, for the terminologies that are selected?
DR. COHN: Walter, can you get very close to the microphone because my bet is that we’re in a situation where Clem and those on the internet are probably still not hearing you.
DR. SUJANSKY: Can you hear me now, Clem?
DR. MCDONALD: Yes, you weren’t too bad just a minute ago, either.
DR. SUJANSKY: So again, reminding folks that the purpose of the terminologies is to support interoperability and data comparability and that the criteria we’ll be discussing are relevant to that. And secondly, at this point, I think we should choose criteria that we think are honestly the best ones to support those requirements and those objectives, ignoring for the moment how many terminologies are out there that meet them and can meet them and so forth. We can discuss that later.
DR. MCDONALD: Is this in your document Walter?
DR. SUJANSKY: There’s a list of criteria there in one of the sections.
MS. HUMPHREYS: 4.3 I think.
DR. SUJANSKY: 4.3.1, 4.3 and there’s two subsections there and we’ll see if we can get to the non-technical ones, but we’re going to start with the technical ones. The slide I’m showing now is just saying that basically.
So diving right in, the other thing I’d like to do, I’d propose doing is to, as I mention in the document, to try to divide the criteria between ones that are "quintessential" meaning that without them the terminology can not meet the goals of interoperability and data comparability from ones that are desired that are not essential. So an example of a quintessential is non-ambiguity, if there’s a term that can mean or a code that can mean multiple things, that for obvious reasons really impedes interoperability and comparability of data, if you don’t know what a code really means, because it has multiple meanings, that’s a show stopper if you will.
DR. MCDONALD: Do you include a class definition as ambiguous or non-ambiguous?
DR. HUFF: Well, the way Jim Simino(?) talks about that it’s non-ambiguous.
DR. MCDONALD: Are these Jim’s, are these his desiderata?
DR. SUJANSKY: They include his desiderata but I believe also include some others.
DR. MCDONALD: I think it would be good to cite the sources of these things, if you get a chance.
MR. BLAIR: Clem, this is actually a derivation, if you recall we originally had our criteria for selection of PMRI standards in our report from two years ago, and then we did a derivative of that for the message format standards, and then on August 28th when you were here I wound up doing an expansion of that building off of Jim Simino’s desiderata and the CPRI work that Chris Shute(?) and Simon did and then become an ASTM standard, so we have this very expanded piece in that document --
DR. MCDONALD: All I’m saying is I think that in any kind of scientific document it’s good to cite from which things were derived.
MR. BLAIR: And what Walter has here is kind of an abbreviated version of that so when we develop questions we’ll probably expand upon these. If we just sort of agree on the general criteria that Walter is --
DR. MCDONALD: I’m not disagreeing with the criteria, I’m just arguing to the scientific principle of citing sources.
DR. SUJANSKY: Just to cite, and I certainly will, I’ll put a citation in the final documents, there’s probably a number of places where I should do that.
So there’s these quintessential criteria and there’s also the desirable but not essential ones, which enhance the value of a terminology but are not essential for interoperability and data comparability. An example of that might be formal concept definitions using description logic.
So let’s dive right in, we’ll go through the criteria, and again, what we want to do is consensus for each one on whether it’s useful criteria at all, and if it is whether it’s quintessential or desirable.
The last thing that I’d like to say is these criteria were chosen because, the ones we’re going to go over, because they were ones that are in the desiderata and are considered useful, but also they’re objective and verifiable, and I think that’s an important element, so something like meaningless identifiers, or ability to have multiple hierarchies is an example of an objective one. Something that’s less objective might be ability to map to reimbursement codes, and so we have to be careful how we collect information about something like that so we get useful information.
DR. MCDONALD: Would it be ok to talk about, I’ve got to leave at 12:00, to talk about things that, 11:40 probably, things that we maybe should think about the action, at first?
DR. SUJANSKY: Sure.
DR. MCDONALD: One of the things was cost or maybe a better way of saying intellectual, I know this is controversial --
MS. HUMPHREYS: That’s a whole other section, Clem.
MR. BLAIR: But that’s ok, maybe we can capture his thoughts --
MS. HUMPHREYS: Because he has to go.
DR. MCDONALD: Well, not encumbered usage of it more or less is what we’re after, so that people get the, I can’t say this very well but it’s kind of a cost/non encumbered --
MS. HUMPHREYS: I guess my questions would be if you go to section 4.3.2, are the characteristics that you’re concerned about represented there or not? Those issues were separated from the technical ones at the beginning.
DR. MCDONALD: Well, this is talking about funny mechanisms, it doesn’t talk about, it doesn’t really talk about the cost of non-encumbered, for in particular the IT control issue --
DR. SUJANSKY: How would you characterize the actual criteria then Clem?
DR. MCDONALD: Well, what you want to have is minimal cost and maximum reusability.
MS. HUMPHREYS: Well, I thought that that was what you said in the first one there, intellectual property and licensing terms for the core terminology that allow the widest use while preventing, etc.
DR. COHN: I thought that said it very well, Clem.
DR. MCDONALD: Not a hint about cost.
DR. COHN: Well, Clem, what would you say the cost, what issue, do you want to put a dollar figure here?
DR. MCDONALD: Well, the preferred one would be per use, but I’m not an absolutist.
DR. SUJANSKY: I think that’s an important point, there is no absolute number. We’ll certainly inquire about --
DR. MCDONALD: And I think the other thing that might be worth saying is as a criteria somewhere is usage experience, that is it’s not just an inventive, there has been, someone is using it.
DR. COHN: Usage experience sounds very good.
DR. STEINDEL: Walter, in line with that I think there’s a set of demographic type characteristics like where is this being used, usage history, think of some more that we’re missing from this, just describing how the product is being used and where.
MR. BLAIR: That falls under the area of market acceptance. We wound up expanding that.
DR. MCDONALD: The other thing that’s not exactly a question about an issue, I just want to make sure that the way these particular vocabulary lists look, it looks like they’re talking about atomic concepts, and would you, what would you do with a DSM 36 questionnaire question? Is that a vocabulary issue or is that --
MR. BLAIR: I’m sorry, Clem, what?
DR. MCDONALD: The vocabulary, focuses on the atomic things, and I’m just worried that we may exclude the dimension of variable or a question, especially I think about the survey questions.
DR. SUJANSKY: Questions like pain scales and so forth?
DR. MCDONALD: Pain scales, depression, Hamilton depressions, but the content is actually a sentence.
DR. HUFF: Yes, I think we want to include that in scope.
DR. SUJANSKY: We can specifically ask about that --
DR. MCDONALD: I’m not saying that we will care if it’s included, but just in terms of the criteria --
DR. COHN: Well, Clem, what do you do with like HEDIS criteria, is that in scope, too?
DR. MCDONALD: HEDIS criteria are for the, no, well, let me just assert no but then we can argue about it. I’m talking about those things that are applied are data captured from patients by all these other things, for patients medical record data. HEDIS is a cross patient method.
MS. GREENBERG: Assessment tools?
DR. MCDONALD: They’re not much different than a serum potassium except we ask it by asking a question. The last two week have you had much, felt like blue or crying, that’s I think in the Hamilton depression or the ASM 36, they’re all phrased, but they’re not just a single, they’re not a noun phrase, a noun in a phrase.
DR. STEINDEL: Well, I think that’s covered under the question where they’re describing their domain coverage with respect to scope, and they can answer us if they do that.
DR. MCDONALD: -- open to that level of thing --
MS. HUMPHREYS: And I think that you make a good point, Clem, because I think you could read this and not assume that that was considered in it and I figure there’s probably some way of getting it at in this domain business.
DR. MCDONALD: Well, when you’re talking about, sometimes that’s the question, and that’s one way to basically pass onto history and then you get a list of answers. But within a lot of this the formal survey questions, it’s a fairly elaborate critter, but it does mean one concept when you get done with it.
DR. SUJANSKY: Right, and I think the critical element there will be how standard are questions like that and questionnaires like that.
DR. MCDONALD: Well, they’re very, very standard, some of them.
DR. SUJANSKY: Some of them, and those are the ones we should focus on.
DR. MCDONALD: Maybe if you could open a dialogue of this text we’ll conceive of that being something that would be in it, in a vocabulary, in an expanded thing called a vocabulary.
DR. SUJANSKY: From a practical point of view those are important and we may need to relax some of the other technical criteria, like concept orientation at the extreme to accommodate that.
MS. HUMPHREYS: You know, it’s an interesting thing, the issue of concept orientation, you can define something like that as a concept but the lack of concept orientation to me is a real show stopper because I can tell you that the whole reason we went to, originally to the concept orientation in the metathesaurus, which I think is probably the first large scale effort in this area that ever had concept orientation, was because it was impossible to map between disparate things without it. And since we know we have to map this thing to these various HIPAA code sets, you basically cannot map a cluster of concepts to another cluster from another source, so you’ve got to have that or the mapping just goes right out the window. That’s where we ended up with it to begin with actually.
DR. SUJANSKY: I think we should consider --
MS. HUMPHREYS: That’s not against what he’s saying though, that’s not in contract with what he’s saying.
DR. MCDONALD: I just want to remind people that if we collected all nouns in the world we would be done.
DR. SUJANSKY: Clem, if you can, this area I must confess of terminologies, these questionnaires and surveys and so forth that are standard, I am not as familiar with, so if you can direct me in subsequent emails and so forth to specific questionnaires and specific concepts in this area that you’re thinking of, then that will help us ask questions about that.
MS. HUMPHREYS: You know, Clem, another approach to doing it, and I’m not saying that this is better, is just that basically the charge to the Committee is to make recommendations for standards for patient medical record information. And clearly the use of these standard questions is an area where even broader use of the standard questionnaires would be highly desirable and would improve all of the things that we’re engaged in trying to get to these standards for anyway. But in some sense, you could divide the problem, and say that that’s another problem and we actually have to focus on making recommendations about are there standard questionnaires that should be part of this. I mean I don’t know whether that’s the right way to go but you could do that.
DR. MCDONALD: We could encourage that to be used more widely, but I think when you get down into really dissecting this there’s not as much difference between the question, history, present illness or history, past history, or surgical history, than there are on these questionnaires. They just tend to have been more thoroughly, the sociologists have been more intense about studying the validity and the reliability.
DR. HUFF: In response to you, Walter, there are several of these things that are already in LOINK, and then there’s a nice article that was published by Thomas White in JAMIA about surveys and how you version surveys, and basically how you maintain concept permanence in surveys, and I don’t know, it was published sometime in the last six months, which was a very nice article and gives you good background on this whole subject area.
MS. GREENBERG: I think the important thing is that any tools or diagnostic tools or whatever that are used, whether they be laboratory tests of whether they be questionnaires or whatever, have got to be able to one way or the other, I’m not quite sure which the mapping goes, but they’ve got to map to the terminologies, the core terminologies, you have to be able to capture at least the answers.
DR. MCDONALD: You’re absolutely right, and what I’m trying to emphasize though is there’s a question and an answer in a lot of things, you could think of glucose, serum glucose is the question answer 130(?).
MS. GREENBERG: No, I hear what you’re saying, but I mean there are just massive numbers of diagnostic tools, questionnaires, whatever used throughout the clinical world.
DR. SUJANSKY: I do think we need to be circumspect in approaching this area because there are instances of these things that actually create problems, you know if you have a concept that specifies it’s the answer to the question do you have nausea or vomiting, and then later you want, and someone says yes, then later you want to do some kind of analysis on all patients who had vomiting, you really don’t know, there’s no way to map to that definitively, so we have to be careful that --
DR. MCDONALD: -- hard as any cases, they could also have, I just wanted to shake up the thinking a little bit as you were realizing the spectrum here.
DR. HUFF: All I understood you to say is that we recognize that when we talk about these terminologies, and especially concept orientation, the concept might not be a simple noun phrase, it might be a whole question, and we should conceive of that at least as being a part of a control terminology, and how many of them we include, I mean that’s just like all of the other things, how many other things we include is sort of an open question at this point, too. I think all we’re saying here is there’s, we want to conceive that a concept in fact could be a sentence, which is a question as part of a standardized survey and that should be in scope of our discussions.
DR. COHN: Clem, I also think you’re sort of stretching the envelope on where does pre-coordination end, and I’ll think about this one a little bit. A sentence, is it a paragraph, is it --
DR. MCDONALD: If it’s formerly, you can take, it’s a complicated thing --
DR. COHN: I agree, I think we should explore it and sort of see where we come out on it.
DR. MCDONALD: I’m completely certain that we should do this, I think there’s no question to paint, if you just say does this hurt, and you’re going to ask about pain, and collect data on it, you have to have some formalism, and we now have to by regulation record pain on every patient visit. And it’s not all in the atomic answers, a lot of the qualification of what the concept is, but what you get out of this thing, is how you say the question. And they’re out there and they’re being used. Glassco(?), Comoscore(?), you just keep on going, you’re going to find them all over the place.
DR. STEINDEL: I’m getting a little disturbed of the drift of this conversation because if we’re going to go down and look at the nitty-gritty of each one of the terminologies and how they handle each little clinical concept and say is it adequate or inadequate, we’re never going to get done. I think the criteria should be are they handling this in a broad sense --
MR. BLAIR: Or how do they handle this?
DR. STEINDEL: Or how, and we’re not going to worry about, we already know, I think everybody in this room knows that any particular terminology we look at, the first place we look we’re going to find deficiency. That’s been the case on every one of us, and to use that as a criteria is not going to move us forward, but we’re looking for characteristics of the terminology that allow it to move forward. And so we need to ask if the characteristic exists.
MR. BLAIR: Could I say this with respect to Steve’s comment? Is that what Walter is taking you through right now is a generalized list of the criteria and in a week or two, if you agree on the generalized list we’ll get to you an expanded list of questions that would generate information about each of the criteria that you could all look at and then we’ll move that into the questionnaire, and then you’ll get a chance to look at the questionnaire before it goes out. So if we could just see if we’re comfortable with these generalized set of criteria.
DR. MCDONALD: I think this looks good. I don’t have any, the only question I have is the multi-hierarchy, does that mean that the vocabulary comes with them or would permit them, because I don’t know, I think you could do hierarchies on top of any vocabulary.
DR. SUJANSKY: It means it would permit them.
DR. STEINDEL: In response to what Clem just asked, which I think is a very good question, but when we do send this out, and this is typical of the government, we should send it out without giving them any idea whether we consider this essential or a good thing to have.
MR. BLAIR: And we did that, we could phrase the questions just like how do you do this, or what’s your method, or --
DR. STEINDEL: And then when we meet as a Committee then we know well this is quintessential and if they said they don’t do it, it’s thumbs down.
MS. HUMPHREYS: The one thing that I had here was the last category here says representation of context information, for example, past medical history, family history. Well clearly we have to represent context information in the electronic medical record or the patient record. And we probably, we may even want a control vocabulary for what are the primary contract context elements within a patient record, I mean I’m not sure we’d agree on that list right now but maybe we need to work toward that. But I guess my question is, I don’t think it’s an actual requirement, for example, for a medication vocabulary, that imbedded in it is the representation of whether this some way within that vocabulary of indicating whether the person used to take the medication or they’re currently on it. I mean I don’t think that that is a, that this is a requirement that you want to apply to each vocabulary to evaluate.
DR. HUFF: I think that’s a requirement for either a messaging system or an electronic medical record system, is the responsibility of data representation, not of the terminology itself.
DR. SUJANSKY: Ok, so you would characterize that one as neither quintessential nor desirable?
MS. HUMPHREYS: Yes, I would, because I think it’s --
DR. HUFF: I think it’s important but not relevant, not relevant in this particular list.
DR. SUJANSKY: Ok, so that’s one down. Is there consensus on that? Ok, I’ll take the absence of dissent as consensus.
DR. STEINDEL: In response I think to something I heard with Betsy’s comment just now, and I think there are going to be some, there is a good possibility that there will be terminologies that we will select for core terminologies that don’t meet all these criteria, because it’s not appropriate. And we will be asking all the terminology vendors these questions.
DR. SUJANSKY: That’s why we’re dividing them between quintessential criteria and valuable but not quintessential. So in other words, some of these we may deem to be important enough that they rule out a terminology for inclusion in the core if it doesn’t demonstrate that others, all other things being equal, it may qualify it over another terminology with the same domain coverage, etc., but it’s not onto itself quintessential.
MS. HUMPHREYS: This gets to Jeff’s point of view about interaction with the vocabulary developers. The American Nurses Association actually had a very good process for this. People would come in and say I have a good nursing terminology, but the American Nurses Association wouldn’t indicate that they thought it was a good terminology for use if it didn’t have things like meaningless identifiers and some of this other stuff, which they thought was, so they would sort of say well, you know, you’re covering some concepts that are of interest here and this might be valuable, but you need to go away and improve this thing or we will not endorse it. So I think that you’re asking the questions, you’re going to make the choices, but I mean obviously if we all found the most brilliant terminology in the world and you thought this is it, we love this, and all they had to do is generate a random number to have a meaningless identifier and store their contextual identifier as an attribute of the concept rather than as the unique identifier, I mean we’d tell them to go away and do it over a weekend and bring the thing back. I’m not saying that necessarily is going to happen but some of these are very imminently fixable if people have a desire to fix them and others are you could fix them but it would take a tremendous amount of work if you already have a large extent terminology and it doesn’t have something like a concept orientation, this is a big job.
DR. HUFF: Do we want to even go beyond what, I think there are some things that really are required as quintessential as you described. Do we want to go even further for the ones that are non-essential, have some, express some priority or waiting for these individual characteristics?
DR. SUJANSKY: Actually what I’d like to do is go through and just to make sure we’re all on the same page, because there may be different levels of understanding in the room about what these criteria mean, at the risk of boring those of you who are experts in this area, to go through them and talk about the value of each one with respect to the requirements, at least briefly, and then do what you propose, Stan, is to say well what are the priorities, which ones are essential and in those categories, what are the priorities.
MR. BLAIR: Could I suggest --
DR. SUJANSKY: We have an hour left, so --
MR. BLAIR: Right, and that is Stan, I think it is useful to divide things into quintessential or desirable, and for us to identify those things, however, I really feel like as we go through this process, we don’t go through it with the idea that if a particular terminology fails to meet a quintessential element that means that we will not select it, so I think that we, from a standpoint of gathering information, it’s useful for us to do that, but I wouldn’t want us to lock ourselves in before hand, to not have flexibility after Walter analyzes all the results and we hear from vendors and everyone else, we still may wind up having some flexibility.
DR. FITZMAURICE: This is not the final report.
MR. BLAIR: This is not the final report, and the other thing that I found the last time we went through this, is that it’s probably going to be more useful when we wind up giving you the specific questions that flesh out these concepts in greater detail. You may wind up looking at a sub-question and separate that out in terms of whether it’s quintessential or not as opposed to the concept as a whole. Does that make sense?
DR. COHN: I think I agree with what you’re saying but I am a little struggling with if something is quintessential and we say well we’ll accept it anyway, I think we may be using the wrong words here.
DR. SUJANSKY: That’s why we should pick the criteria for that category very carefully.
MS. HUMPHREYS: It seems to me at least we could go through and say do we think it’s essential and then we’ll find out later if we really think it’s essential and that’s fine.
MR. BLAIR: What we might do is just have a scale of one through four, with one being the highest weight or most desirable or most important and four being recognized but of lesser --
DR. HUFF: I think another way, when I look at these I think the only two of them that I see as sort of, well, there’s some overlap between them, I mean if you don’t have concept permanence, I don’t think you can, well anyway, concept orientation and concept permanence may be the only two things I think are "essential." As Betsy described, you couldn’t make the metathesaurus, you can’t do the mappings if it’s not concept oriented.
MS. HUMPHREYS: And I guess in this day and age, although I was just talking to Simon about a serious case of this, the lack of it, if we go, we say meaningless identifiers, which I certainly think are highly desirable, but I do feel that permanent identifiers are essential. We don’t say that, but the notion of the reuse of identifiers, I think everybody thinks this is a terrible thing, even though people unfortunately are still doing it.
DR. SUJANSKY: But that’s what I mean by concept permanent, so what I’d like to do is go through, right there there’s just a bullet point with one word or a couple of words, what I’d like to go through and talk about what is meant by those things so that we can --
MS. HUMPHREYS: I would say concept permanence is two things, and I would agree with you that the permanent identifier is one of them, but the other is, that is you never reuse an identifier, but the other thing is that the concept never disappears. It may be annotated to be shown as no longer useful or deprecated or whatever, but it doesn’t sort of just miraculously disappear from the system.
DR. MCDONALD: Well, those would be nice things to say explicitly.
MS. HUMPHREYS: Those are two different aspects of what we mean by concept permanent, I guess if you did that then you would have the identifier issue in under concept permanence, is kind of like --
DR. SUJANSKY: There’s actually a third one in that is that the meaning of the concept, not only does the concept identifier not change and the concept identifier not disappear, but the meaning of the concept identifier doesn’t change.
MS. HUMPHREYS: Well, yes, that’s what I thought we all understood by concept permanence, because we were talking about it, that’s what the name connotes.
DR. SUJANSKY: Again, since there are additional different meanings to these things that we’re now discussing, maybe we can again go through the slides here, which does drill down a little bit deeper on each one of these, and make sure we’re on the same page, so in the next version the full meaning can be included.
DR. MCDONALD: Walter, have you captured the other two ideas already, that Betsy said about the identifier, the thing about the identifier?
DR. SUJANSKY: I have in the slides that I’m presenting.
DR. MCDONALD: They look pretty good from here.
DR. SUJANSKY: So the first one is concept orientation and informally defined as elements of the terminology are coded concepts with possibly multiple synonymous text representations and hierarchical or definitional relationships to other coded concepts. So basically this just means that they’re not terms, they’re concepts, and there can be multiple terms that are used to represent the meaning of that concept. And the value of this is that it supports explicit synonymy, which is important, and it enables other desirable properties, like meaningless identifiers and concept permanence. I think this is a quintessential feature.
Do we have agreement on that? Ok, let’s just go right onto the next one. -- have changed if the meaning of a concept needs to be changed or refined, a new coded concept is introduced, no retired codes are reused. Does that capture your meaning Betsy?
MS. HUMPHREYS: Yes.
DR. MCDONALD: I think the idea about retain that code with the fact that it’s retired, that may not be in there.
DR. SUJANSKY: Ok.
DR. MCDONALD: So people can find out what they thought they meant last year.
DR. SUJANSKY: Ok, right. So codes never disappear. That’s a good point. And the validity of this of course is that it supports longitudinal data analysis across versions of the terminology.
DR. COHN: Ok, let me just ask an additional question on this one, and that’s, I’m just wondering if, it seems to me that we’re obviously requiring everyone to have rules around all of this. The question is whether one needs to understand here is what the rules are, because what we see is in certain different terminologies, but then they say yes, what we do is we get rid of this, we leave it off for five years, or we leave it off for 30 years, and then we reuse it. And the question is, is that an absolute no no, is that something we need to understand? The only reason I bring this up is that if for example in many messages the actual date of the terminology sort of goes along often times with everything, and so you actually know --
MS. HUMPHREYS: Yes, if you’ve got your coded longitudinal patient database then you have certainly complicated the issue of aggregation of data when you discover that it used to mean bi-polar personality disorder, there’s an ICD example of this, the code used to be bi-polar and now it’s something like meningitis. So you don’t, and granted there was a 40 year, or 30 year, or 20 year distribution between those two items, but in the machine readable version, in the world we’re in today, I think we can fast forward and expect that somebody may have a database with both of those data codes.
DR. MCDONALD: Go out 30 years.
DR. COHN: I guess what I’m questioning on this one is that the question not be a yes or no, but if they have rules we need to find out what they are.
MS. HUMPHREYS: I would agree with that.
DR. SUJANSKY: We can’t say do you have concept permanence.
MS. HUMPHREYS: No, you have to say how do you handle these issues and problems. And obviously there is no perfect way of doing it, I mean I do not know of a single terminology that deals with this in a rigorous enough way that we would all agree that perfection had been achieved. Maybe in the future.
DR. SUJANSKY: I think as Jeff pointed out, in addition to being sort of criteria for selection, these are also prescriptive criteria in a sense that may direct the future evolution of certain terminologies that are either hard shows and are being considered, or may be considered.
MS. HUMPHREYS: You skipped comprehensive domain coverage at the beginning.
DR. SUJANSKY: Yes, I reordered these.
MS. HUMPHREYS: And that one to me is definitely not quintessential because if we have a good vocabulary we could always expand its coverage and its domain.
PARTICIPANT: They need to describe what they’re domain coverage is.
MR. BLAIR: Exactly.
DR. SUJANSKY: Ok, agreed.
DR. HUFF: I mean in the question there, you could even ask them to quantify, how many drug names do you have, how many body parts do you have, how many living organisms --
PARTICIPANT: Your terminology, that is.
DR. HUFF: Yes.
DR. SUJANSKY: Sorry, that’s what I intended to say.
DR. MCDONALD: I’ve got to excuse myself, I had an obligatory that’s a problem presenting that I have to go to. I’m sure you’ll manage.
MR. BLAIR: Clem, thank you very much. Do you feel generally comfortable if we go forward?
DR. MCDONALD: I feel comfortable most of the time. I’m not worried, you’ve got a lot of smart people around the table.
MR. BLAIR: Thank you. We’ll get you the questions I hope, Walter do you think we can have the questions to these folks by next Wednesday? For them to edit?
DR. SUJANSKY: It’s likely.
DR. MCDONALD: I’ll get it when I get them.
MR. BLAIR: We’ll try, best effort.
DR. SUJANSKY: How’s that for non-committal? Alright, meaningless identifiers. So here, the unique codes used to identify concepts in the terminology are unrelated to the meaning of the concepts or to their locations in the concept hierarchy. That’s what we mean by that. The value, this enables multi hierarchies, unlimited number of siblings, in certain cases, and reclassification if necessary. So, and there’s some examples here.
The counter example, for those of you who may be familiar with this one, is one of the problems with ICD-9, that you can’t have multiple classification if you’re hierarchical position is represented in the code itself. So this is an interesting one to consider whether this is quintessential --
DR. HUFF: There’s intermediate ground, and what I mean is that if there’s one, I mean one way to think about it, one rational way to do this where you don’t have meaningless identifiers, is to assume that hierarchies are not equal. In fact there’s one hierarchy that’s predominant in that you give it a hierarchical code but then you treat it as a meaningless identifier thereafter. And so what it means is that you have a computationally efficient way to do this so, anyway I think this is very important but I don’t think it’s quintessential.
DR. SUJANSKY: It can be compensated by explicit relationships, explicit hierarchical relationships.
DR. HUFF: So essentially you represent one hierarchical relationship in your identifying or represent others as explicit relationships.
MS. HUMPHREYS: I think the issue here is what is important, is what we’ve said before, which is the permanent, the identifiers have to be permanent, so if in fact you can’t figure out a way to make them permanent and also make them carry meaning, then it becomes quintessential. That’s why I believe that you need to separate out meaningless and permanent identifiers, because I would say that permanent identifiers are quintessential even though I wouldn’t say that meaningless identifiers are quintessential.
DR. SUJANSKY: Would you say we’ve done that, the concept permanence handles the permanent identifiers, so this is desirable but we have a way to work around it.
DR. FITZMAURICE: You might cross reference, though, because people won’t understand it. They’ll see it one place but they won’t associate it here.
MS. HUMPHREYS: Well, it depends on how you put your questions together. It’s sort of like what are the characteristics of your identifiers, do they carry meaning, are they permanent? So I have a feeling that you’re going to end up asking those questions around the identifiers, even though the reason, they split into these other functional things.
DR. SUJANSKY: Non-redundancy, each concept is represented by just one code in the terminology. So another counter example, kind of from the old, these aren’t the actual codes but from the old mesh world, there were some problems with this.
MS. HUMPHREYS: No, that’s not a good example. The reason why it’s not a good example is because the actual identifier, those numbers were never the identifiers, it’s just that people saw them and said oh, you have two identifiers, but in the back room they were never the identifiers for the mesh thing.
DR. HUFF: There’s lot of these in PTEXT(?) that I can give you as examples, though.
DR. SUJANSKY: Basically this occurred because there wasn’t multi hierarchy, so when the same concept happened in PTEXT, the same concept had to be put in a different place you actually had to create a new concept that meant the same thing.
MS. HUMPHREYS: The interesting thing was that in the context of the medical subject headings, we published the hierarchical numbers because they affected people when they did searches, and for many years we did not publish the actual identifier, which was a separate thing.
DR. SUJANSKY: Ok, yes, I understand. I had always thought there wasn’t an identifier, until the time it was published, but now you’re informing me that that always existed, which I wasn’t aware of. So the importance of this is that it prevents basically false negative query results, if not all the codes are known at the time you’re doing analysis and it also necessitates more complex queries even if you do know all the codes. So it’s important to data analysis, to making data analysis reliable. Is it possible to do reliable data analysis with redundant codes? Yes. It’s possible. So --
DR. HUFF: Again, I think this is highly desirable but not quintessential I guess.
DR. SUJANSKY: Alright, agreement on that. Non-ambiguity. Each code in the terminology has a unique meaning, this is sort of the inverse of non-redundancy. So a counter example, this is seen in some terminologies, again, the code is made up, the same code and the same text is used to represent two things.
DR. HUFF: And again, this is the one that Clem was talking about, in that it’s ok to have a term that means Hepatitis, which is a class to mean all of those, or even viral Hepatitis, which subsumes Hepatitis A, Hepatitis B, this is not saying that you can’t have class names that subsume a lot of meaning, it’s that --
DR. SUJANSKY: This is saying more about this, this is intended to talk about this, not the class name.
MS. HUMPHREYS: Yes, I think it would be, and obviously you need to be able to, as Clem was saying, represent classes of concepts, but that’s different from this type of thing where you basically can’t tell when you look at the thing whether the person who used the code, you potentially don’t know whether they actually meant --
DR. HUFF: The person took digoccin(?) or they had a digoccin level measured, or --
MS. HUMPHREYS: Exactly.
DR. SUJANSKY: Now, theoretically, from, there’s lot of context and other information, and that’s all standardized and so forth, you can tell, but it may be quite a while before all those elements are standardized, so in the absence of that I feel that this is a quintessential one.
MS. HUMPHREYS: I do, myself, too, because I think that the problem is, I mean I don’t have any problem that somebody, I mean I might not like it, but I don’t have any problems if somebody uses the name of a substance for the name of the test for the substance, as long as that’s all it means in that system. I mean we can handle that, but the issue of having it mean both things, in the same system, I would agree with you.
DR. COHN: Let me ask you sort of pursuing that a little bit, because this is obviously one of the things that’s sort of in the back of my mind as an issue, which is this whole issue of context, otherwise known as the information model, which we’ve down played in our paper as an issue, or as more of a longer term issue, but is the same issue here of you have a diagnoses where the field says present diagnosis and then another field says past history of, are we talking about the same thing here?
MS. HUMPHREYS: No, I don’t see it.
DR. COHN: So that’s really different. Explain to me the dividing line.
MS. HUMPHREYS: Well to me it’s very different, it’s like if we say we have a disease, we all know it’s asthma, or whatever it’s called, and the question is where do we put it in the record so that we knew whether it was my mother’s asthma, my asthma, my previous asthma, my current asthma, but we know that asthma is asthma. In this particular case, or let’s say serum sodium, if we’re looking at this term serum sodium and this code, and we actually cannot tell because of the inherent ambiguity in the vocabulary source, you go and look there and it says this means either the substance or the test for the substance, then the vocabulary itself has such an inherent level of ambiguity that we will never, the vocabulary itself is ambiguous, as opposed to the fact, and we have many examples of this in the UMLS, where one vocabulary means the test and the other one means the substance. Well, that’s ok, at least within each, I mean we might like one approach to naming better than the other, but at least within that system it is not ambiguous, they always mean the test.
DR. SUJANSKY: I think the point you’re raising Simon is a good one that the full meaning of a code in the context of that patient’s care and documentation of that patient’s medical history and so forth, does go beyond having non-ambiguous concepts under themselves, and it does require representation of context in the big picture, so you could say non-ambiguity of concepts is a necessary but not sufficient criteria for non-ambiguity of the full, of all the medical information that’s being transmitted or analyzed.
DR. COHN: Betsy, I basically agree with what you’re saying, I’m just saying, it’s a hard time to know exactly where the line is.
MS. HUMPHREYS: There’s a classic example that we just learned, and that is this gene ontology, which is supposed to annotate gene function, and we discovered that in this vocabulary, they have a system, it’s just like the test issue. It’s just like this issue. They have a systematic approach that the enzyme activity, which is all they’re concerned about, they’re vocabulary does not name the enzyme, but it uses the name of the enzyme to mean that enzyme activity. So we said oh my goodness, we thought that the procedure assay one was the main one that we ever had to deal with and we discovered no, we have another systematic one, it’s over on the gene side of the house. So we now have to be generating an alternate name that says activity at the end of this just so we have one unambiguous name in the UMLS.
DR. SUJANSKY: Ok, so we’re agreed that this is quintessential. Multi-hierarchies. A coded concept, may be a child of more than one other coded concept in the hierarchy. This is valuable because it prevents redundancy when concepts require multiple classification, we saw a counter example to that, and it allows local additions to a hierarchy, this is to your very good point about the need for that. So in one way or another, you can create a multi hierarchy whether it’s, and ideally it’s through explicit hierarchical relationships, actually that’s probably the only way you can create a multiple hierarchy as opposed to wrapping it up in the code. So I think we know what this means. How important is it to the extent that it enables local additions of, or local hierarchical additions and, how important is that, first of all?
DR. HUFF: In the context of what we’re doing, I mean saying that we’re going to put it in the metathesaurus, the metathesaurus allows it, so if the source terminology didn’t, we would have the ability, based on what we’ve agreed to, to add them. So I think this is desirable in the sense that if the source terminology already does it you just have that much more value in the terminology. I don’t think it’s quintessential in the source.
MS. HUMPHREYS: No, I don’t think so either, because I think that given things that are quintessential, like permanent identifiers, and a concept orientation, if I care enough to do it, I can take anyone’s vocabulary and create my own multi hierarchies for it.
DR. COHN: Walter, didn’t we already have one that said hierarchies and we said that the terminology needed to be capable of? Wasn’t that a --
DR. SUJANSKY: Well that came up, actually Clem asked about that in regard to this criterion, that’s why we started going around --
DR. COHN: So it’s basically capable of then, probably.
DR. SUJANSKY: So this is desirable but not quintessential, and the main point is that there’s capability to have multi hierarchies. Actually I guess, it’s not one of the criteria that I listed but maybe we can briefly discuss the importance of a hierarchy in the terminology itself. I guess this is kind of a domain coverage point. The richer and more extensive a hierarchy the terminology has, the better. Is that true?
MS. HUMPHREYS: I would say that it’s desirable and it’s extremely desirable because if you are maintaining a vocabulary and you do not have some sort of a hierarchical arrangement, it is extremely difficult to maintain the concept of non-redundancy. Just for you managing your vocabulary, it’s very difficult to keep track once you’re up in, you’ve got 20,000 items, or 100,000 items, and you have no logical grouping of these items, it’s pretty hard to avoid introducing redundancy.
DR. SUJANSKY: But I also mean as a valuable element of the terminology unto itself to support things like data analysis, aggregation --
MS. HUMPHREYS: Data entry, the selection of the appropriate term, yes, it’s, once you get beyond a certain size, I could image if we agreed that we had our special control vocabulary that only covered certain demographic characteristics or whatever, and there were literally only ten or 15 of them that we could use and undifferentiated list, but --
DR. SUJANSKY: It’s desirable. Ok, so we’ll call it out --
DR. STEINDEL: Well, it’s more than highly desirable, but you can’t, I don’t think we can make it quintessential, and a good example is LOINK, which is non-hierarchical, but CHI told LOINK that they better make it hierarchical.
DR. SUJANSKY: And you can infer a hierarchy because of the fields and so forth. That’s a good example, I was just going to say that.
DR. STEINDEL: Yes, but it’s not, we don’t think it was, we can use LOINK without LOINK becoming hierarchical, but we will make it hierarchical for our own use and we did at CDC.
DR. SUJANSKY: Betsy, will RX-NORM have explicit aggregations of medications, things like ace inhibitors and --
MS. HUMPHREYS: Yes, it has, it’s aggregated, there are various ways. There are the active ingredients, there are a variety of different related that are related to things.
DR. COHN: Walter, I need to do a time check with you at this point. We have I think about, we have 30 minutes left for the meeting, we actually have to quickly lead through the draft statement of work for ICD-10 at the end, so you have about 20 minutes, 23 minutes, I just wanted you to manage your time appropriately, so that we can get through this.
DR. SUJANSKY: I’m getting close to the end here, of the technical criteria, anyway. Thank you, Simon. Formal concept definitions, description logic.
MS. HUMPHREYS: Desirable.
DR. SUJANSKY: I knew you were going to say that. And I agree. There’s consensus on that. Infrastructure for collaborative development.
MS. HUMPHREYS: Desirable.
DR. SUJANSKY: And then some others that I didn’t have time to put on their own slides. Explicit version identifiers. This is an interesting one because if you have concept permanence, maybe you don’t need this.
MS. HUMPHREYS: Oh, yes you do.
DR. COHN: We certainly need one or the other.
MS. HUMPHREYS: You probably need both. We’ve been trying to work through these areas conceptually about what happens to people when they get, supposing we fast forward and we have what this desirable thing and we have the UMLS and we have this core set of clinical content and they’re all labeled and identified and they’re being updated and the UMLS is coming at you four times a year so that you can keep up to date and everything is wonderful, you’re going to, the explicit version identifiers basically, it’s desirable. It’s desirable because we’ll actually create them, putting the thing in the UMLS, but if it came with it, that probably would lead to less potential problems for everyone. But it will have to be there in the UMLS going in, or none of us will be able to update anything properly.
DR. SUJANSKY: So you’re referring to then if basically the knowledge of which version a particular concept, which version of the terminology or particular concept first appeared in, appears in, and when it was retired and when it was retired and so forth. That type of information.
MS. HUMPHREYS: Yes, and I guess what I’m saying is that where the management of the connections and the integration and the mapping and whatever, if each vocabulary itself had its own concept of when I release a new version, it has a discreet identifier and we can all know what that is, that’s probably highly desirable, but it’s not essential, because as I say if we pause it that at least one mechanism for the distribution and use of this, or perhaps the primary one, or the one that has high status, is through the UMLS, then basically we will give these things version identifiers when we incorporate them.
DR. COHN: Well, Betsy, I think you’re saying that this is critical, it’s just a question of who’s doing it. Actually I’d say that this is probably more important almost than concept permanence personally, because I think you can have, I mean this is like probably the critical issue, it probably needs to be linked to that piece that has to do with concept permanence, because this is another piece of it.
DR. SUJANSKY: Are you saying that it’s quintessential?
DR. FITZMAURICE: Suppose you have a large database, and you are searching for a concept and it’s not there so you can assume well none of the patients had this, but then you find out later that it was a version that didn’t have that concept in it. That’s why you need to know the version.
DR. ZUBELDIA: If you’re exchanging data between two parties and one from a newer release is going to send to somebody that’s on another release, you know it’s not going to work. You have to know this.
DR. SUJANSKY: Presentation of context information, we covered that one. Availability of methods into tools to ease local implementation and incorporation of subsequent updates.
SUBCOMMITTEE: Desirable.
DR. SUJANSKY: support for local customization and subsequent merging with updated standard.
SUBCOMMITTEE: Desirable.
DR. SUJANSKY: Especially since nobody does that.
DR. COHN: Could we add one to this because it feels like it’s somehow related to this particular screen which has to do with, I mean especially with the version identifiers and local customization, this whole issue of change sets and the existence of change sets that shows you what the difference is between the versions. And I think that that’s a critical piece in terms of all of this also.
DR. SUJANSKY: So explicit, --
DR. STEINDEL: Walter, you could make that as part of the tools.
MS. HUMPHREYS: Yes, well, but I would say that the delivery of an explicitly designated change set, what is explicitly defined in some way, what is different between this one and the previous one, is a higher, they’re perhaps both desirable but that one is much more highly desirable.
DR. FITZMAURICE: It’s like having changes between two documents, you want to know the difference between the two.
DR. SUJANSKY: So, just to be sure we’re clear on this then, the precise criteria that we’re talking about is with each new version, there’s explicit change set.
MS. BEBEE: Does the idea of mapping need to be included in the criteria in some way, if it’s an important part?
MS. HUMPHREYS: You mean whether they exist already?
MS. BEBEE: Yes, I mean if not in this area of criteria, but a question on the questionnaire?
DR. STEINDEL: I think that’s part of the demographics we need for the terminologies, that we should ask what maps do you have.
MS. GREENBERG: And also, how do you do them, I would that you’re going to ask if they have mappings to classifications, should ask if the developer of the classification has validated the maps. A lot of people have maps, but --
MS. HUMPHREYS: I would say that the general rule is that people have maps and one of them have been validated.
MS. GREENBERG: So we can just assume that?
MS. HUMPHREYS: Well, no, but I mean I think that therefore one has to say that you can’t, you want to know if they exist, I agree with, I think we should ask the question, but it’s not any type of a show stopper because my view of no matter what we pick they’ve all got to be reviewed and probably substantially revised and upgraded because I don’t think, I mean there might be a case somewhere and some of them that we have in the UMLS have actually been validated because in fact people actually looked at how we mapped their vocabulary to something in the UMLS and they said yes, that’s right, but there are very few in the real functional world of HIPAA classifications or whatever that I’m aware of that have been fully validated.
DR. COHN: And the question is validated by who?
MS. HUMPHREYS: Exactly, and for what purpose.
DR. SUJANSKY: Ok, we’ll ask a question about that. That’s a good point. Alright I think we’ve covered all the technical ones. Now we’ve got about 15 minutes left to discuss the non-technical criteria that we’ll be asking questions about and I’ve divided these into three sections. Ownership and licensing, and this includes things like intellectual property and cost and provisions for distribution, redistribution and things like that.
Secondly, maintenance processes, and these are aside from the technical maintenance processes about concept permanence and so forth, what is the, how many people are involved in updating this. How is their input received and integrated and processed and so forth? How frequently is the terminology updated, things like that. I don’t know, again, if the Subcommittee has any comments at this point about criteria, for example, no terminology that’s updated less frequently than annually should be included, things like that would be useful at this point. I’m sorry, Jeff?
MR. BLAIR: Usually these were on a scale, there was not a black and white on any of the general ones.
DR. SUJANSKY: And then criteria related to viability and funding, and this ones kind of a grayer area, but something to the affect of how do we know you’re going to be around in five years and still doing what you’re doing.
DR. STEINDEL: The descriptions that you put around these in your document I thought were very nice. It’s enough that we could do a draft and look at the specific questions.
DR. SUJANSKY: Are there any other "non-technical" criteria that --
MR. BLAIR: The ones that we’ve had before that we might consider pulling from our criteria for selection of PMRI standards, I think I can remember about five of them that are general, and this is not mandatory, it was because it was inherited, it’s in the law that we have to give preference to a standard that is accredited as an ANSI accredited standard, and in the terminologies there’s only one and that’s not going to weight very much but I think that that’s something that should be in there.
DR. COHN: Well, Jeff, I actually want to clarify because that’s actually not a criteria for code sets, which I presume terminologies, last time I saw, were code sets. That’s only a requirement for the actual message format standards.
MR. BLAIR: Ok, well we had it in the overall back two years, but that’s fine, if you, if the group doesn’t want to include that it doesn’t matter.
DR. COHN: Well that was because we were talking about the message format standards specifically is whether or not it was there.
MR. BLAIR: It was very useful to that, yes.
DR. HUFF: I didn’t know there were any terminologies who’s content was actually ANSI accredited.
MR. BLAIR: You know what? What I think is the important thing out of that, which we might ask, is what is the process for managing their standard, including different codes? Is it consensus based, is it balloted, what is the criteria for membership, those types of questions, because that kind of gets the same thing that was intended originally when the HIPAA law wound up saying --
MS. HUMPHREYS: The exclusion, the wording of the actual HIPAA Administration section, Simplification section, to divorce this notion of the actual accredited consensus standard process from the code sets, was deliberate because everyone felt that the notion of balloting before we add the drug that’s the cure for AIDS or something is ridiculous. So I think your set of questions is a much more complete set. I mean how do you accept the kinds of things that we did when we described what’s an open process on the, or tried to, which is kind of like who gets to submit requirements, how do they hear back from you, how do you post what you do with all these suggestions, how do you handle it so that everyone who has a legitimate need is given a fair hearing in terms of getting their requirements met, rather than focusing on the standard aspects of it.
MR. BLAIR: The other items are some kind of an indication of the degree to which they are technology and vendor independent. And another one is, and you may already have it up there, probably do, something to indicate that the licensing and implementation is relatively low in cost, something to get at.
DR. ZUBELDIA: That’s what Clem was talking about when he mentioned free or low-cost, it’s not the ownership or licensing terms, it’s the actual cost.
MS. HUMPHREYS: I do think that a lot of people would say that if you have to go climb too many mountains or run through too many tunnels or hoops to get it, even if in the ends it is free to you, that that is a problem as well.
MR. BLAIR: Exactly, and that’s why I was specifying --
MS. HUMPHREYS: There are two issues, the issue of is it simple to get it --
MR. BLAIR: That’s why I was specifying the acquisition cost, whether it’s a license or whatever to acquire it, and then the other piece is the cost of implementation, because if you have to acquire somebody’s software to be able to support this particular code set, then that should be somehow captured.
MS HUMPHREYS: And I think what we were just saying is there really are three things. There’s the technology independence or that there’s the issue of what is the actual cost of the thing itself, and then there is the issue of what is the level of administrative hassle it is to get it, irrespective of what the cost is because --
DR. COHN: I would certainly agree about the administrative hassle piece.
MS. HUMPHREYS: Some of us who have negotiated for license agreements know that this can be endless.
DR. COHN: The piece I would make about, the point I would make actually about the cost of the license and things like this, I think we need to ask, just so we have that information, but recognizing that there are many ways to deal with that, as we’ve seen which includes government negotiating national licenses, etc., that I don’t think that the cost is per se a, it’s a piece of information, and I think that’s really --
MS. HUMPHREYS: Well, all of this it seems to me at this point is a piece of information, I mean when you’re asking the questions of people.
DR. COHN: Well, Clem thought it was a little more than a piece of information.
MR. BLAIR: We may have to ask two or three questions to be able to get at the whole concept of how it --
MS. HUMPHREYS: How easy it is to get.
DR. HUFF: I agree with the list of these things, I think that we’ll have some more interesting discussion when we talk about the weighting of them. And what I mean by that is that in this list of things, I think that one of the things that I would say is really, in this context is almost quintessential, is that these things be unencumbered so that people will have the creativity to incorporate them in knowledge basis, make new representations, new hierarchies, any new derived products, so that the licensing is not encumbering either the creativity or the ability of people to make useful by-products from this.
MR. BLAIR: Derivative works?
DR. SUJANSKY: What about reselling for example? If I take a terminology and I put some thin shell of tools around it and now I resell that terminology, is that --
MR. BLAIR: Well, we could ask what limits they have on either derivative or --
DR. HUFF: No, I think that should be fine. That should be allowed. Again, and you already covered it, you don’t want them to then have the ability to make new codes so that they’re not just redistributing but are in fact making a new terminology, then that’s the only thing you’re trying to prevent is the needless proliferation of slight modifications on the standard.
MR. BLAIR: Now the other areas in general which I think we’ve covered to some extent but I’m going mention them in case somebody wants to add additional thoughts or questions on them. One is market acceptance, and I think you’ve gotten to that to some degree because we’d be asking general questions about how many users do you have, how many licenses do you have, are they implemented, and we could get a whole bunch of things to get at the issue of market acceptance. What was the other one, areas to be added to the general questions? If not then we might just take a minute to try to indicate what Walter and I might be doing to move the process along during these next couple of weeks and what we’d be doing to get you the questions? Are we ready for that or did you have other questions or issues? Did I hear a pause?
DR. SUJANSKY: I think it’s self evident, we’ve got a lot of work.
DR. COHN: Walter, why don’t you go through it quickly and then we’ll move on to the next?
DR. SUJANSKY: I’m sorry, Simon?
DR. COHN: Why don’t you move through quickly what you’re anticipating, Jeff were you going to do it?
DR. SUJANSKY: Do you want to comment on this, Jeff? Basically my understanding is we’ll be, this input and our discussion of the criteria, both technical and otherwise, will be reformulated essentially as a set of questions that will be candidate questions for the survey that will be sent to the terminology developers who will be surveyed. And then that will be, that candidate list of questions will be circulated in the Subcommittee sometime next week, and comments will be requested by no later than the 20th, which is next Friday --
MS. GREENBERG: These are your suggestions to us for the questionnaire that Jeff is developing?
DR. SUJANSKY: Our deadlines are soft, yours are firm.
MR. BLAIR: Unfortunately, we’re bounded by our meeting dates and as they flow back through the process, that’s what’s kind of driving our tightness right now for the next month in terms of, so Walter will get to you the questions for your review sometime next week. We would need you to take a look at them as soon as you get them and return that to Walter by the 20th, which is next Friday, so you may only have two or three days to critique the questions.
DR. SUJANSKY: If it takes you longer than that to go through the questions, there’s something wrong with the questionnaire.
MR. BLAIR: So it is tight. And then the other piece is that I will go ahead and fold those into a questionnaire, a draft questionnaire, which I would get to you by, let’s see, by January the 6th, and I would need your response, critiques on the questionnaire probably by the 7th or 8th of January, so that we can get it distributed to all of the terminology developers by the 10th. So that is kind of tight, but --
DR. HUFF: A separate point is that if Walter, if you’ll give me a copy of the electronic form, I have other small edits on this document and I would reflect those in the documents and send it back to you for the next version of this document.
DR. SUJANSKY: You’re referring not to the slides but to the actual document.
DR. HUFF: Yes.
DR. SUJANSKY: Ok, sure.
MS. GREENBERG: Are you going to be distributing this questionnaire by email?
MR. BLAIR: Yes. As we had discussed, I will distribute the questionnaire by email and it will come back to me by email.
MS. GREENBERG: Because I was going to say if you needed to send it out by snail mail or FEDEX or something like that, we could facilitate that.
MR. BLAIR: Oh, thank you --
MS. GREENBERG: If you’re just going to email --
MR. BLAIR: I think we’re going to have a short list because HL7 has a list of registered terminology developers which I think is about 20 or 30 and I was going to wind up supplementing that with the terminology developers from the inventory of standards, so I think at the most we would have 40 terminology developers and so it’s manageable.
DR. COHN: Well, I guess let’s talk about that process for a second. I think first Jeff it would be useful if maybe when you send us the questions you indicate the developers that you’re intending to send it to. Second, given that we have sort of aborted the issue of domain, which I think is very appropriate, we need to have it up on the web-site so that a developer that we had not anticipated could actually complete it and return it to us.
MR. BLAIR: I’m happy to do that. Marjorie, are we, is that within regulations to do that?
DR. COHN: Is there a way for us to put it as, if people want to fill it out? How should we go through that process only because we haven’t really --
MS. GREENBERG: I’d rather not, actually post the questionnaire on the web-site. I think what we could post is something saying the Subcommittee is selecting information on technologies and if you have not been contacted --
DR. COHN: And feel that you are appropriate, ok.
MS. GREENBERG: Then contact me, or contact Jeff, however you want to do it, Jeff.
DR. SUJANSKY: That’s all I have.
MR. BLAIR: Walter, I want to thank you very, very much. I think you did an excellent job and you really helped us pull together this stuff in a manner we can move forward. Believe it or not we made the deadline.
DR. COHN: And nine minutes to spare.
[Applause.]
DR. COHN: If it’s ok, I think what we’ll do is to quickly go through the draft statement of work for the ICD-10-CM and PCS cost/benefit study, just to make sure everyone’s ok on it, it is a last reading before we move forward.
This draft statement of work, and I will just read through it. Certainly if any of you have issues about it let us know, but this is the assessment on the impact on the health care industry of moving to ICD-10-CM and ICD-10-PCS. The first piece is to assess the full range of implications on current users of ICD-9-CM, Volume one and two, to implement ICD-10-CM, including costs and benefits, quantifiable and non-quantifiable of the change.
And under that we have bullets, identification of affected entities and degree to which they will be impacted. The second bullet is identify costs, including opportunity costs, associated with transition, including, but not limited to, information system changes, rate negotiation, recalculation of reimbursement methodologies, training, changes to forms, etc.
The third is timing of transition, including impact of timing options, on cost and benefits, potential return on investment, and interaction with other major health information implementation tasks. The next bullet is identify immediate and future costs and benefits of improved data for, but not limited to, patient safety, outcomes analysis, reimbursement, disease management, utilization review and health statistics. A final is affected entities, including NCHS, CMS, and other third party payers and providers should be contacted to provide input to the contractor about their own assessments of cost and benefits within their organizations.
Any comments about that first piece of work? Ok.
The second piece of work is to assess the full range of implications of replacing current uses of ICD-9-CM, Volume 3, with ICD-10-PCS in the hospital inpatient setting, including costs and benefits, both quantifiable and non-quantifiable, of the change. And this is basically the same list of bullets, noting that the analysis should include all current uses of ICD-9-CM, Volume 3.
The final piece of work is to assess the full range of implications required to implement ICD-10-CM and PCS simultaneously versus separately. And the issues here are the same as those identified for ICD-10-CM and PCS, and identify any cost savings that may result from simultaneous implementation of both code sets.
Everyone ok on that? Kepa?
DR. ZUBELDIA: Are we biasing this last bullet by saying identify cost savings that may result from simultaneous implementation or should we say identify cost savings or expenses?
DR. COHN: I think that’s sort of a friendly amendment, so basically identify cost saving or additional costs that may result from simultaneous implementation of both code sets. John?
JOHN: Probably just simpler to take out that dash and put in an or, cost or saving.
DR. COHN: Ok, identify costs or savings. Thank you. Well, it’s nice to have you participating, John.
Then finally, the comment here is the contractor will (1) review written testimony, transcripts and minutes of previous NCVHS code set hearings and relevant minutes of the ICD-9-CM Coordination and Maintenance Committee meetings. Yes, that’s a fair amount of work to look at. And then (2) use scenario development and other approaches for estimating costs and benefits, where firm estimates are not available.
PARTICIPANT: And they’re going to have this done by May?
DR. COHN: Ok. Any comments or thoughts about this one other than, God speed, exactly, God speed.
MR. BLAIR: Is there any way for us to get this to Clem so that he can comment on it?
DR. COHN: It doesn’t matter, I mean certainly if he has any wordsmith changes he can send it back to Marjorie. I think at this point I think they’re going to be going off and try to find somebody.
MS. GREENBERG: It won’t be awarded by tomorrow.
DR. COHN: I think if we’re in a situation where he has --
PARTICIPANT: You could get it awarded very, very quickly now. You’re doing it on a task order, right?
MS. GREENBERG: Right, but it still takes a few weeks.
DR. COHN: I think we’re in a situation because of the timeframe that certainly if there is wordsmithing to this, I don’t think anybody would argue, but I think if we’re being, if there are major changes in scope being discussed, I think that that’s a whole other discussion, and I guess I should look at everyone here, since we do have a quorum of the Subcommittee, is everyone comfortable with us moving forward with this one? I actually would like a formal motion.
MR. BLAIR: I’d like to make a motion that we accept this particular draft of the cost/benefit engagement.
PARTICIPANT: I second.
DR. COHN: Ok, any discussion? All in favor?
SUBCOMMITTEE: Aye.
DR. COHN: Any opposed? Ok.
MS. GREENBERG: I’m glad we found something that the Subcommittee can agree on.
DR. COHN: Yes, that’s right, it’s taken us a while on this one.
Now just for everyone’s information, upcoming hearing dates for the Subcommittee are January 29th and 30th, just right after the NHII hearings as I understand it. The next after that will be May 25th and 26th and then we’ll have another session on May 21st and 22nd. We will I guess shortly probably query the members regarding hearing dates for the last half of the year. I don’t think I have much else to say other than that.
DR. FITZMAURICE: Excuse, Simon did you say it was going to be May 25th and 26th and May 21st and 22nd?
DR. COHN: No, I said March 25th and 26th, and then May 21st and 22nd.
Any other issues coming before the Subcommittee? Obviously for next year, this is going to be a very important year for the HIPAA administrative and financial transactions implementation. We will be spending time on the 30th talking about that with WEDI and other groups, and I’m hopeful that we will be able to include any issues that come up or major initiatives within the hearing data schedule, but I do want to warn everyone that if there are issues that seem to be particularly of concern, we may have to schedule some other hearings during that time to investigate them. My hope is is that this will all move forward, but I think we need to stand ready to provide leadership if needed over the next ten months.
Ok, any final comments, questions? With that we will be adjourned. Thank you.
[Whereupon, at 12:30 p.m., the meeting was adjourned.]