[This Transcript is Unedited]
Washington Terrace Hotel
1515 Rhode Island Avenue, NW
Washington, D.C.
DR. COHN: Good morning. I want to call this meeting to order. This is the second day of three 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 body to the U.S. Health and Human Services on national health information policy.
I am Simon Cohn, National Director for Health Information Policy at Kaiser Permanente and Chairman of the committee. I want to welcome subcommittee members, HHS staff, and others here in person.
As was the case yesterday, we are actually not being broadcast on the Internet. My understanding this morning is, there seems to be some sort of a glitch that prevents transmissions for the next two weeks, so I think we might also assume that potentially tomorrow morning we will not be on the Internet, either.
However, we are recording the sessions, and there will be a transcript made. I believe it will be posted on the Internet after the website is back on. So I do still want to remind everyone to speak clearly and into the microphone. That includes questions from the audience. Pleas make sure to introduce yourself before you ask your question.
This morning we will be reviewing and updating as necessary the PMRI terminology recommendations letter, with an eye to bringing it to the full committee in September. Jeff will be leading the session, and obviously we have in attendance our consultant on this project, Walter Sujansky. We thank you for being with us today, as well as your hard work as you helped develop and helped us refine the letter.
This afternoon after lunch, we shift to an updated discussion of the federally sponsored ICD-10 cost impact analysis being undertaken by Rand. This will be primarily an update of work in progress. It will give us an opportunity to understand the progress being made, as well as understand the issues they are confronting as they begin to do their analysis and develop the overall cost impact report. We will also be hearing a report from the AHA and AHIMA on what they have done in the pilot study around use of ICD-10.
As mentioned yesterday, on Thursday morning we begin early, at 8:30. We begin with an update on claims attachment, NPRM, and from there, move into a discussion and update on the consolidated health care informatics initiative. From there, we will be talking about the HIPAA implementation, both receiving an update from the Department, and then followed by a panel of experts discussing issues as that implementation proceeds.
Hopefully we will also have some time to discuss subcommittee business, update meetings for the rest of the year, and topics for those meetings.
Hopefully there isn't anything else we are supposed to be doing over the next day and a half. Obviously, as I mentioned yesterday, I do want to emphasize that this is an open session. We do welcome comments and questions, as long as they are on topic and brief, from the audience. We will try to the best of our ability at the end of each session to provide an open period also, if there are additional comments that people want to express. Obviously, we also welcome e-mails and followup letters on any issues that may be before the subcommittee.
With that, let's have introductions around the table and then around the room. As is always the case, for those on the subcommittee, if there are issues coming before us today for which you need to publicly recuse yourself, please so indicate at this time.
Jeff?
MR. BLAIR: Jeff Blair, Medical Records Institute, Vice Chair of the subcommittee, and member of AMIA, ASTM, HL7, HMSIS, and there is nothing that I am aware of that I need to recuse myself from.
DR. STEINDEL: Steve Steindel, Centers for Disease Control and Prevention, liaison to the full committee and staff to the subcommittee.
DR. YASNOFF: Bill Yasnoff, Office of the Assistant Secretary for Planning and Evaluation, staff to the subcommittee, liaison to the NHII work group.
DR. GREENBERG: Marjorie Greenberg, National Center for Health Statistics, CDC and Executive Secretary to the committee.
DR. PICKETT: Donna Pickett, NCHS, CDC, staff to the subcommittee.
DR. SUJANSKY: Walter Sujansky, independent consultant and advisor to the NCVHS on MRI terminology standards. I have nothing I need to recuse myself for, but again, I would like to disclose other of my consulting engagements that may be related in any way to this NCVHS activity. First, I am currently engaged as a consultant by First Data Bank to assist with a market research project, which entails requirements analysis and documentation for a specific product feature. During the course of the NCVHS project, I have also been engaged by Amacore Incorporated, an EMR vendor that has licensed SNOMED CT for use in its products. In this capacity, I have also interacted with Health Language Incorporated, a vendor of terminology server tools, although I have not been under contract with that company. Several years ago I was also an employee of Appelon Incorporated, another vendor of terminology management tools. At this point, I have no vested interest in Appelon. Also, I was engaged by a company named Innovion, whose chief technology officer, Keith Campbell, is a member of the SNOMED editorial board, although Innovion is not a SNOMED licensee, and my work there was not related to SNOMED.
Thank you.
DR. HUMPHREYS: Betsy Humphreys, National Library of Medicine, staff to the subcommittee.
MS. AULD: Vivian Auld, National Library of Medicine, staff to the subcommittee.
MR. BURKE: John Burke, Office of the Secretary, staff to the subcommittee.
DR. FITZMAURICE: Michael Fitzmaurice, Agency for Health Care Research and Quality, liaison to the National Committee and staff to the subcommittee.
DR. BEBEE: Suzie Bebee, ASPE, staff to the subcommittee.
DR. MC DONALD: Clem McDonald, member of the subcommittee and member of the committee, from Indiana University and Regenstrief Institute, and chairman of the LOINC committee, which I would recuse myself from the discussion.
DR. HUFF: Stan Huff with Intermountain Health Care and the University of Utah in Salt Lake City. I am a member of HL7 and a co-chair of the vocabulary technical committee there. I am also a co-chair of the LOINC committee, and I would need to recuse myself from those issues if those come up today.
DR. SQUIRE: Marietta Squire, CDC, NCHS and staff to the subcommittee.
DR. BLUM: Amy Blum, NCHS, CDC.
DR. LINCOLN: Mike Lincoln, Department of Veterans Affairs and University of Utah Medical Center.
DR. GRIFFITH: Stan Griffith, Indian Health Service and staff to the subcommittee.
DR. TORAKASKY: Irene Torakasky, American College of Surgeons.
DR. J. FRASIER:: Judy Frasier, American Optometric Association.
DR. C. FRASIER: Chris Frasier, American Academy of Professional Coders.
DR. WILLIAMSON: Michelle Williamson, NCHS, CDC.
DR. MC NAMARA: Tim McNamara with Certa Corporation.
DR. WOLCOTT: Julie Wolcott, Institute of Medicine.
DR. FRIEDMAN: Maria Friedman, CMS, staff to the subcommittee.
DR. LESH: Kathy Lesh, the Kevver Company.
DR. ASHMON: Diane Ashmon, SNOMED International, College of American Pathologists.
DR. ECKERT: Karen Eckert with Medi-Span.
DR. SPACKMAN: Ken Spackman from Oregon Health and Science University, and I am chairman of the SNOMED International Editorial Board.
DR. KOMEZI: Said Komezi, medical informaticist, VHA.
DR. SABA: Virginia Saba, Georgetown University and the developer of the home health care classification system.
DR. LEVIN: Randy Levin, Food and Drug Administration.
DR. ALMONSON: Todd Almonson, American Hospital Association.
DR. ANCHISEN: Mary Lee Anchisen, American Hospital Association.
DR. BOWMAN: Sue Proffett Bowman, American Health Information Management Association.
DR. COHN: Before I hand the microphone and session over to Jeff, I do also need to publicly recuse myself, primarily for this afternoon. While I don't expect that we are necessarily going to be talking about CPT extensively, I do sit on the CPT editorial panel, and obviously if such discussions come up, I will be recusing myself from those discussions.
Also, just by way of information, though not as a public recusal, do need to indicate that as a large provider of health care, my employer, Kaiser Permanente, uses clinical terminologies, has been involved in the development of almost every terminology discussed today, yesterday and probably in the future, and I think probably has various contractual relationships, but also, almost every vendor that discussed clinical terminologies yesterday; if not all of them, I would be very surprised.
Jeff.
Agenda Item: Review and Update of PMRI
Terminology Recommendations
MR. BLAIR: Thank you, Simon. I am going to try to be briefer than I was yesterday, but we are in the process of the second phase of selecting PMRI standards. The message format standards were recommended about a year and a half ago, and formally adopted, been announced by the Secretary in March of this year on behalf of the consolidated health care informatics initiative.
The PMRI terminology standards is a process that began about a year ago. At that point, we wound up asking experts to give us guidance on how to modify the guiding principles for the selection of PMRI standards from August of 2000. We received that guidance, we folded that into a questionnaire, which was sent to all organizations that indicated that they felt they wanted to be considered as a PMRI terminology. There were 44 of them in total.
The first results of compiling the responses to that questionnaire were presented by Walter Sujansky to the subcommittee on March 23 or 24 of this year. We then decided that we wanted to make sure that we were basing our judgment on correct information from the terminology developers, with consistent criteria for interpreting that data. So there has been a second and a third version of that analysis, which has gotten additional critiques from the terminology developers. We wanted the terminology developers to look at the way we analyzed the questionnaire that they submitted, and make sure that our interpretations were correct and consistent.
We are now about to begin a review of the second draft of the PMRI terminology recommendation letter, which we hope and intend will be approved by the full committee in late September, and then be able to be sent to the Secretary of Health and Human Services.
Now, Walter, were you able to set this up where it is on a display where everybody can see it?
DR. SUJANSKY: Yes. I can project it whenever we are ready to do that.
MR. BLAIR: Great, and everybody has copies, so we are all singing from the same letter.
I think it probably would be helpful if, Walter, as you take us through this second draft of the letter, the first few sections probably have less controversy or less concern on the part of many. They tend to be the introduction and the background. So maybe if you read us through those sections, and then at the end of teach of those sections, you then ask for any comments or critiques or questions. That would maybe get us through maybe a little bit faster, to the section where we are actually making our recommendations in the back. I think that is where we will have our greatest interest.
For those of you in the audience, you should be aware that the letter is organized in a manner where we have specific recommendations for terminologies that would be part of the core set of PMRI terminologies. In other words, they would serve as reference terminologies.
In addition to that, we have also identified in the letter how those terminologies would be mapped at the National Library of Medicine to other clinical terminologies, as well as to the first set of terminologies identified by the HIPAA financial administrative transactions. Then in addition to that, we indicate a section which we refer to as important, related terminologies. Those are identified. Then beyond that, there is a section which discusses areas that we consider to be important, but we couldn't make a determination at this point, so we are noting it for further interest and attention.
With that, we turn it over to Dr. Walter Sujansky, who will take us through the letter.
DR. SUJANSKY: Thank you, Jeff. By way of just a little more background about the letter itself, it should be noted, this is a second draft of the letter. There was an initial draft a few weeks ago that was reviewed by the subcommittee on a conference call in late July or so. That call is a matter of public record, I believe, and the transcript of that call is available. Comments were made that led to some changes; that led to the second draft, which was distributed to the subcommittee electronically late last week, and is also available in the back of the room right now for anyone who is here and doesn't have a copy of it yet.
So we will proceed. Again, we are going to try to work through these first few parts pretty rapidly before we get to the recommendations. I will go ahead and read each of these paragraphs. There have been a few changes to the previous one. We will proceed.
Dear Secretary Thompson. As part of its responsibilities under the Health Insurance Portability and Accountability Act of 1996, HIPAA, the National Committee on Vital and Health Statistics, NCVHS, was called upon to, "study the issues related to the adoption of uniform data standards for patient medical record information and the electronic exchange of such information".
NCVHS studied these issues, prepared a comprehensive report on PMRI standards, and presented it to the HHS Data Council on August 9, 2000. This report provided a framework for PMRI standards and a set of guiding principles for the selection of these standards. The NCVHS then used these guiding principles to evaluate and select PMRI message format standards. The PMRI message format standards then became the first set of specific PMRI standards that were recommended to the Secreary of HHS on February 27, 2002.
This letter will recommend the second set of PMRI standards, based on the guiding principles mentioned above. The second set will be the core set of PMRI terminology standards. They will complement the PMRI message format standards. Standards for PMRI message formats and terminologies are important, because they will facilitate significant improvements in the quality of patient care, promote patient safety, control rising health care costs, enhance the productivity of clinical research, and strengthen the nation's ability to identify and respond to health care emergencies. They are critical to the creation of a national health information infrastructure. Additionally, PMRI standards will enable government to provide higher quality and more efficient health care within its own delivery systems through the consolidated health informatics initiative.
Any comments so far? Michael.
DR. FITZMAURICE: There is one thought that is not in there. That is that the Secretary did in fact adopt the recommended standards. So I would change the last sentence, which says the PMRI message format standards, to read, on February 27, 2002, the selected PMRI message format standards -- I would strike then -- making the first set of specific PMRI standards to be -- I would strike and were -- to be recommended to the Secretary of HHS. I would add a sentence, you adopted them on March 21, 2003 for HHS use. I can supply those words later on if it sounds all right.
DR. SUJANSKY: Great.
DR. HUFF: In the second paragraph, I had the thought that we might at some point in the future adopt other terminologies to be part of the core. So the last sentence of the second paragraph, you modify maybe to say, the second set will be the first set of core, or give some indication that this may not be the last set of core terminologies we adopt ever in our -- just a thought. I don't know if that is important.
The second thing is maybe a minor point. It is really the use of these standards that will bring all of those benefits that are described in the last paragraph. So I guess my preference would be to say something like, when the standards for PMRI message formats and terminologies are in widespread use, they will do all these things.
DR. SUJANSKY: Maybe here we can do a little wordsmithing on the fly. Adoption of standards for PMRI message formats and terminologies are important, because this will facilitate significant improvement, et cetera. Do you want to say use?
DR. HUFF: Well, adoption can be ambiguous, whether by adoption we all mean that we are going to use them, or whether adoption means that they are actually in use. So I like the use word better.
DR. COHN: Actually, I think adoption, only because the Secretary can't -- he can certainly adopt them.
DR. HUFF: I know, but these benefits only accrue if people use them.
DR. HUFF: How about adoption and use?
DR. COHN: Marjorie and then Steve.
DR. GREENBERG: This is just stylistic, but it just struck me that since this letter, and I guess it goes along with what Mike proposed, since this letter is already to the Secretary, I'm not sure if we should be referring to him in the third person,or say something about, which were recommended to you on February 27, and then say, we were gratified by your adoption of these standards on March something.
DR. SUJANSKY: Right.
DR. COHN: I think that is a friendly amendment.
DR. GREENBERG: What?
DR. COHN: I think that is a friendly amendment.
DR. GREENBERG: It is friendly, yes, trying to be friendly.
DR. COHN: Steve.
DR. STEINDEL: Just to pick up on what Stan said initially about, the second set will be the core set, we do talk later in the letter about augmenting the core. So I think it is appropriate to say the initial core set of terminologies, something like that. We do actually at the end of the letter refer to augmenting that set.
So just to be consistent with its use, the last sentence in that third paragraph, where it says --
DR. SUJANSKY: Which paragraph? This one?
DR. STEINDEL: Yes. The last sentence says, additionally, PMRI standards, I would just say, additionally, use of PMRI standards will enable the government to provide higher quality.
DR. SUJANSKY: So just add initial?
DR. FITZMAURICE: Initial set of core PMRI terminologies.
DR. SUJANSKY: Well, elsewhere we use this phrase, core set of PMRI terminology standards. I don't know if we just want to keep that together.
MR. BLAIR: The only thing is, if I might say, I think it may be confusing if we introduce the concept right here, because we are winding up saying this is the second set of specific recommendations for PMRI standards. Then we wind up saying this is the initial set of the second set. I do think that we do show how it is augmented when we get to the specific recommendations, that this won't be the end.
Stan, do you still feel strongly that we make that point up front?
DR. FITZMAURICE: We are making the point one way or the other.
DR. HUFF: I don't know that -- as long as we are all in agreement and we say it later, I don't know that it has to happen right here.
DR. HUMPHREYS: It seems to me that the point in this paragraph is that your second set of recommendations are covering terminologies.
DR. SUJANSKY: Correct.
DR. HUMPHREYS: So it seems to me that you are really saying, this letter will recommend the second set of PMRI standards, or a second set of specific PMRI standards, and something about the fact that these are terminologies and go on, because you get to this core business later on.
You don't even have to mention that this is a core or not in the second paragraph. You are basically saying, we recommended message standards before, and now we are recommending terminologies.
DR. SUJANSKY: So we can leave the first sentence basically, or we can say, recommend the second part of specific PMRI standards, the part will address PMRI terminologies.
DR. HUMPHREYS: A second set. This letter will recommend a second set of specific PMRI standards, or terminologies or something, because you don't know how many recommendations are going to come in the future.
DR. SUJANSKY: Right.
DR. MC DONALD: Can I make a suggestion? I think you can do it fairly easily if you say, the second set consists of terminology standards that will complement the PMRI message format standards. You don't have to -- the first sentence still works.
DR. SUJANSKY: Say that again, Clem? What should the second sentence read?
DR. MC DONALD: I didn't write it down. I saw Betsy writing it down.
DR. HUMPHREYS: This second set consists of terminology standards that will complement the PMRI message format standards.
MR. BLAIR: I think we have to refer to them as PMRI terminology standards, so they don't get confused with the financial administrative.
DR. FITZMAURICE: But it does take away from the notion of core.
DR. HUMPHREYS: That comes up in big set later.
DR. FITZMAURICE: We could say that this is the initial core set.
DR. SUJANSKY: I'm concerned that we are getting bogged down in this. I think I understand the intent here. Why don't we move on, since we are unable to resolve this quickly, and then we can discuss this more later or on another call.
DR. FITZMAURICE: Good idea.
DR. HUMPHREYS: Just in the third paragraph, I was thinking we might broaden it very slightly to say, to respond to health emergencies, rather than to respond to health care emergencies.
DR. SUJANSKY: Scanning, scanning. Okay, I've got it. I'm fine with that. Background information for this area I just described, how we got here.
DR. MC DONALD: That was the non-controversial section.
DR. SUJANSKY: Hopefully this will be less. Process to select PMRI terminology standards. The NCVHS commenced the process of selecting PMRI terminology standards in August 2002 by soliciting input from medical terminology experts on the appropriate scope and criteria for selection of PMRI terminologies.
Based on the input received and assessed by the NCVHS Subcommittee on Standards and Security, a working document was prepared that specified the scope of the terminology selection activity and the criteria that would be applied in the selection process.
Subsequently, the developers of 46 candidate health care terminologies were invited to complete a comprehensive questionnaire describing various aspects of their terminologies, including one, content and technical features, two, extent of usage and market acceptance, and three, matters of licensing and intellectual property. Rigorous analyses of these data with respect to the previously agreed upon selection criteria were included in the initial version of the NCVHS PMRI terminology analysis report. This report narrowed the pool of candidate terminologies to nine.
We are looking at whether that is nine or ten. There is still a little open question there, just as a comment. Obviously that will be decided by the time this is sent.
Three separate versions of this analysis were prepared, progressively incorporating data from the questionnaires, comments from members of the NCVHS, and feedback from the terminology developers themselves. Written and oral testimony was then provided to the NCVHS by users of the candidate terminologies to ascertain the utility and limitations of the terminologies in practice. This testimony was assembled and analyzed in a companion NCVHS PMRI terminology user report. Some of the information collected through all of these activities was employed by the NCVHS to arrive at the recommendations presented herein.
Are we all on the same page there? Presumably we will add something about the testimony we heard yesterday to that.
Guiding principles used as criteria for selection. As in its previous recommendations on message format standards, the committee emphasized the following general criteria derived from the PMRI guiding principles in formulating recommendations for the PMRI terminology standards. One, the extent to which the standard enables interoperability between information systems, two, the ability of the standard to facilitate comparability of data, three, the aspects of the standard that support data quality, accountability and integrity and four, the degree of market acceptance of the standard.
The criterion of quote market acceptance was broadly interpreted to include widespread use and/or strong user support for a new PMRI terminology during NCVHS hearings. This allowed for the appropriate consideration of recently developed terminologies.
Any comments on that section?
DR. FITZMAURICE: Excellent.
DR. SUJANSKY: Recommendations encourage HHS guidance and incentives rather than mandates. NCVHS recommends that HHS set forth guidance for industry use of PMRI terminology standards, rather than create new federal regulations. NCVHS also recommends that the Secretary direct government agencies to follow this guidance by becoming early adopters of PMRI terminology standards by accelerating the process envisioned for the incorporation of standard terminology by the consolidated health care informatics initiative. This action will serve as an example and incentive to the entire health care industry.
Perhaps we should say these actions, a minor point that comes to mind.
If there are no comments there, we'll get to the heart of things. Again, let me remind everyone, this is a draft letter. That is why we are working on it right now, and we are here to discuss it.
DR. MC DONALD: Just to clarify the schedule though, is this the penultimate discussion or just one of many?
MR. BLAIR: We are planning on having a third draft, which would be sent electronically to the subcommittee, where we could have one more edit before it is presented to the full committee.
DR. COHN: Actually, unless we are having a conference call, Jeff, I don't think we have scheduled at this point, sometime within the next ten days, to meet deadlines for going to the full committee. I think you should consider this effectively the penultimate discussion, and only wordsmithing changes after this discussion would be accepted, unless there is something that turns up during the discussion to be irreconcilable.
DR. MC DONALD: Then I think some of the things we discussed this morning to fill it, we should try to fill it.
DR. SUJANSKY: We'll circle back around and do it next time. I think we would all agree, it is more important to discuss the really substantive recommendation issues, and then if there is time, we will come back. If it is not, we can try and do it over e-mail and so forth. I think we will be able to figure something out for that.
DR. COHN: I think we need to make a distinction between wordsmithing and substance.
DR. SUJANSKY: Right.
DR. COHN: A question or comment?
DR. GRIFFITH: A quick question. In the previous paragraph, first sentence, is that entirely true? The testimony yesterday, I thought was==to clean up NDC codes, which is pretty important, would require regulation, I thought.
DR. SUJANSKY: Yes, that is true. From what we heard from the FDA, it would require regulation, but my recollection is, we agreed in the recommendations not to mention NDC or any changes required to NDC codes specifically.
MR. BLAIR: Just for a point of clarification, NDC codes were recognized by the NCVHS as a one of the code sets that is part of the financial and administrative HIPAA transaction sets. So they already are designated as a HIPAA code set. If we wind up making any other recommendations on that, there would be a lot of other interested parties that would need to testify, and we are really trying to stay away from expanding the scope of our deliberations, to not cover NDC specifically.
So NDC was presented to us by Randy Levin yesterday morning as background information, but not for us to recommend changes.
DR. SUJANSKY: I'm sure we will discuss that topic a little bit more as we move on.
NCVHS recommendations for PMRI terminology standards. First paragraph here, a "core set" of PMRI terminology standards. NCVHS recommends that the federal government recognize a "core set" of PMRI terminology standards as a national standard. This core set compromises the minimal set of terminologies that one, are required to adequately cover the domain of patient medical record information and two, meet essential technical criteria to serve as reference terminologies.
Furthermore, the NCVHS recommends that these terminologies be mutually crossed referenced to create a cohesive, internally consistent and minimally redundant terminology resource. Also, the NCVHS recommends that the Secretary designate the National Library of Medicine as the central coordinating body to manage this terminology resource and coordinate its ongoing maintenance and distribution.
The core set terminologies recommended for the PMRI standards are SNOMED CT as licensed by the National Library of Medicine, logical observation identifiers, names and codes, laboratory subset, and RX Norm.
DR. YASNOFF: The Secretary is well versed in these issues, but when you are making a recommendation that these should serve as reference terminologies, couldn't we at least put in a footnote saying what we mean by that?
DR. SUJANSKY: He hasn't read the seminal papers?
DR. YASNOFF: Oh, he has, he has, but just not recently.
DR. COHN: Other comments?
DR. MC DONALD: I would like to suggest we at least think about putting in, recognizing an initial core set, to be consistent with what our discussion is.
DR. COHN: I think what we are doing is recommending that there be a core set, and then in the second paragraph, basically we are talking about these as the initial elements of that core set. Does that make sense?
DR. SUJANSKY: Shall we put it here? The initial core set of terminologies recommended for the PMRI standards are.
DR. COHN: Does that make sense, Clem? I think what we are trying to get at is that there needs to be a core set.
DR. HUMPHREYS: I would think that if you did that down in the initial core set and then in the second sentence of the first paragraph, you say this core set should comprise, because you haven't finished it yet.
DR. SUJANSKY: Okay.
DR. COHN: Bill.
DR. YASNOFF: I think it would be useful in connecting the dots if at the end of this paragraph there was a sentence added that said, we take note of the fact that LOINC has already been adopted through the CHI process for government-wide use.
In essence, we are expressing our agreement with what you have already done, but I think it would be good to explicitly indicate that we know about that and we are supportive in this recommendation, and that these things are connected.
DR. COHN: Bill, should we stick that down into the LOINC section rather than --
DR. YASNOFF: Sure.
DR. COHN: I agree with you. I think that is a good idea.
DR. YASNOFF: Where it goes I don't think is as important as making sure that in this letter, it is clear that we are cognizant of these other things that are going on, and in this case are expressing our agreement with something that really is already done.
DR. COHN: Okay. Walter, you still haven't fixed that first sentence there, in the second paragraph.
DR. SUJANSKY: The initial core set?
DR. FITZMAURICE: It doesn't need to be the first. I was wrong. I think what Betsy said was right.
DR. SUJANSKY: I thought we wanted to do both. Should comprise, and then also say, the initial core set.
DR. HUMPHREYS: I would say the initial terminologies recommended for the core set.
MR. BLAIR: I would agree with that, because I think it is very confusing to refer to something as an initial core set. There are going to be other core sets. So I think we have one core set. This may be the initial list of terminologies that are included in the core set, and that we could use the phrase that this initial list, we intend to augment this list in the future, but still refer to it as a core set. Do you agree?
DR. COHN: Walter, do you want to read what you have there?
DR. SUJANSKY: Let me read what I think it should say, not what it says right now. The initial terminologies recommended for the core set of PMRI terminology standards are.
MR. BLAIR: Yes, right.
DR. COHN: They are nodding their heads. Clem, you don't like it?
DR. MC DONALD: No, this is another subject. Another sentence.
DR. COHN: Oh, a new sentence in that area? Please, go.
DR. MC DONALD: The sentence that starts with, furthermore, the NCVHS recommends, I think the goal is a cohesive, internally consistent minimally redundant terminology resource --
DR. SUJANSKY: Clem, I missed some of your words. Could you get closer to the microphone?
DR. MC DONALD: I think I talk too fast. Let me just say what I think we should say. I think we should take out the mutually cross referenced to create, because there are more things that have to be done than mutually cross referencing, and you may not want to have that be the only tool.
What you want to do is to end up with those phrases at the end. I would suggest we say that these terminologies be organized into a cohesive, internally consistent and minimally redundant. Then we have the NLM management to figure out how to make that happen, which would include --
DR. SUJANSKY: Would you agree though that they should be mutually cross referenced?
DR. MC DONALD: No, not necessarily. You can just as well do it by deleting some of it. There are lots of tools for getting to this goal. In some cases you may get into copyright problems.
DR. SUJANSKY: What would be a copyright issue?
DR. MC DONALD: Let's say if RX Norm went through and linked to all of the codes, that would then be a derivative work. There are issues, so I don't think we should be that specific about how we get this done. The goal is to get it done, not how to do it. There is some of the cross referencing, there may be lots of tactics that are used.
DR. COHN: Bill or Betsy.
DR. HUMPHREYS: I would just comment that I tend to agree with Clem's point here. I think the goal is as stated, and the issue is exactly how to achieve this. In effect, all of the connections between the three and the labelling of them as being the PMRI set or whatever we are going to do, all the connections could be in the UMLS meta thesaurus and not necessarily have to be independently maintained over in LOINC and RX Norm and SNOMED. The cross referencing does imply that this is being handled on all ends of the problem.
Maybe that is the best way to do it, but I think we don't know that yet.
DR. SUJANSKY: I just want to address Betsy's point.
DR. YASNOFF: I think I can help with this. I agree with Clem's point as well. The goal is to organize them. Perhaps you would be willing to say at the end of that sentence, so that for example, they can be mutually cross referenceable.
In other words, I am trying to get to the point that, if you just say that these terminologies be organized, cohesive and so on, again, it is not clear what you mean. So by saying one of the tasks, if appropriate, would be to mutually cross reference them, then what you are saying is more understandable, even though you are not prescribing specific actions.
DR. SUJANSKY: We obviously have to remember, just to follow up on those comments briefly, that this is obviously going to be read by many others besides the Secretary and people within the government. This is something that is going to be reviewed by everyone in industry and the potential users of this standard.
Up to this point, they have already been using these terminologies and many terminologies, and the problem, they have discovered, is that there are many terminologies, and they don't fit together and they don't work together well, and it is confusing how to see a set of terminologies. So this is by way of reassurance that there is a plan to make these work together. It is clear and relatively specific. This is the answer to the question that arises in peoples' minds at this point, this is great, but how is it going to fit together. So that is the intent of mentioning that.
DR. MC NAMARA: I remain very concerned about RN Norm at this juncture for some of the reasons that we heard yesterday. We heard testimony that RX Norm has not yet been tested for the purpose about which we are including it as a core set. We know that RX Norm is new. Its relationship to NDC has not been thoroughly established. I think it is premature to include such a new untested vocabulary system for drugs in this as a recommendation to the Secretary.
DR. COHN: I think we still need to get this handled here, and then we can move on to that next issue.
DR. MC DONALD: Can I come back to the logic? Firstly, you could absolutely create a cohesive set of standards by just chopping and cutting. Secondly, you can't guarantee to cross reference some of these things, because they are not synonymous. So I think this is over prescribing to the Secretary's level.
The goal isn't to do a certain thing, it is to get to a certain point. I think when we start putting methods in, at this level, a very high level spec, it seems to be way too tight.
DR. SUJANSKY: Well, cross referencing isn't necessarily implying that all this synonymy is mapped. I think the cross referencing is fairly general. It is not specifying exhaustive cross referencing, it is not specifying whether synonyms are hierarchical relationships or anything else. It is just simply saying that there will be cross references among them, and the details are dee dee dee.
DR. MC DONALD: But you are not permitting any other ways to get to a cohesive, internally consistent, minimally redundant --
DR. SUJANSKY: I don't believe it is excluding anything, either.
DR. COHN: Walter, let's take some other comments here.
DR. STEINDEL: I think there is a general consensus coming around from the subcommittee that words should be organized, and that there is disagreement about the scope of mutually cross reference, that it may be too prescriptive, used in this general letter, and that the Library should decide the mechanisms that they need, which probably will include mutually cross referenced.
DR. SUJANSKY: There was discussion of putting cross reference -- one of the committee members suggested putting it at the end. Is there any consensus on doing or not doing that?
DR. STEINDEL: I would prefer that it was just eliminated.
DR. SUJANSKY: Eliminated, because I don't think at this point it is useful to list examples to the Secretary, and then the Secretary asking the Library, have you made progress in this area.
DR. COHN: I think I agree. The question is whether there is a better word than organize. I'd let Walter sleep on that. But I think we are all agreeing that mutually cross referenced are not the right words, organize may be the right word, or Betsy may sleep on it and come up with a better word. I think we are just wordsmithing that at this point.
DR. SUJANSKY: I hear you. How about for now, we change it to, NCVHS recommends that the terminologies be organized into a piece of internally consistent.
DR. HUFF: What about coherent in place of cohesive? Cohesive just says that they stick together; coherent says that they make sense together.
DR. SUJANSKY: Okay.
DR. HUMPHREYS: You want a lot, don't you?
DR. SUJANSKY: You want a lot; I was going to say the same thing.
DR. SPACKMAN: Eliminating the reference to cross referencing suggests that the language now says something that I don't know that we can do. That is, that the terminologies be organized into a single terminology resource. It depends on what we mean by that. Having the words cross referencing in there gives me confidence that we can do it. Leaving it this way makes me wonder what exactly is being entailed.
These are independent terminologies, and as far as I can tell, will remain independent. So the idea that the Secretary is somehow going to coerce them into a structure that doesn't fit is potentially implied by this.
So I strongly think that the notion of cross referencing is as strong a notion as you can reasonably expect.
DR. HUFF: I think that is exactly the point. In fact, I think it might take a different route than that. What you might say is that we are not going to use certain parts that you deprecate and cut out parts that are not coherent, so you actually are going to coerce it into a unified thing. I would grant that that is not known right now, and that is exactly the reason why you don't want to put in mutually cross referenced, because that is only one way that this could be accomplished. In fact, that might be desirable.
Cross referencing to me implies that there is redundancy, and you are compensating for the redundancy by making cross references between these things. A more desirable way is to chop out the redundancy, define clear boundaries between the parts that you want to be coherent, and just eliminate.
MR. BLAIR: Could I suggest that we are getting down to virtually an architecture discussion here, and I think that that is at a deeper level than the subcommittee really is prepared to recommend. So I think that the general language, and leaving it to the discretion of NLM at this point on how to facilitate the communications between the terminologies nd the mapping between the terminologies is more appropriate for the level of our recommendations.
DR. COHN: Steve, and then maybe we can move on from this.
DR. STEINDEL: I was going to move on to the next sentence.
DR. SUJANSKY: Actually, before doing that, there was another comment made that I think we need to address. The previous commenter voiced some concerns about --
(Simultaneous discussion.)
DR. STEINDEL: That is in the next paragraph. I want to focus on the next sentence.
DR. SUJANSKY: Okay.
DR. STEINDEL: I hope Betsy agrees with me. I don't think we should just designate the National Library of Medicine, but I think we should also recommend that the Secretary provide resources to do this.
DR. SUJANSKY: That is also coming up in other paragraphs, so we can address that.
DR. STEINDEL: Okay, if it exists elsewhere. I just want to make sure that Betsy is not being asked to produce this coherent whole and then loses 40 FTES or something.
DR. COHN: Marjorie.
DR. GREENBERG: I'm not quite sure if I understood Jeff's response to Dr. Spackman. But since he chairs one of these major terminologies that is being recommended here, I am somewhat -- I think we need to think about that, and whether the subcommittee is recommending to the committee and then the committee recommending to the Department that there be a single non-redundant resource as Stan said, or whether in some cases it might be cross mapping or mutually referencing or whatever.
MR. BLAIR: I didn't have my response as a response to Ken Spackman particularly. It is just that there are a number of discussions here as to what should be the mechanisms. They get into architecture discussions, and we virtually could have probably several days of hearings for us to get to the level at which we could wind up agreeing on what should be the proper architecture mechanisms. So I just felt like we probably -- I don't think we are prepared to get down to the level of describing the mechanisms.
DR. GREENBERG: I agree with that, but I think by changing the language the way you did, about being organized into a coherent, minimally redundant terminology resource, you may be coming out for one approach, as opposed to --
DR. MC DONALD: Marjorie, that has always been in there, the internally consistent minimally redundant is nothing new. The only question is, are we truncating the mechanisms by saying one mechanism.
The previous sentence says, this is the only thing you are going to do, and you are going to hope that is going to happen. I added the word hope. This is saying, this is what we want to get to, and we have confidence and expertise -- it is just like any other kind of regulatory thing; you don't tell them exactly how to dot the I's and cross the T's.
DR. SUJANSKY: Again, I know we are beating a dead horse, but I do think this is important, and I do think that the signal that this sends to the whole industry is important. I am concerned this is going to send the wrong signal.
The only thing mutually cross referenced is saying is that they are going to be somehow connected. What I hear Stan saying is, maybe they won't, maybe we will just make sure --
DR. HUFF: No, I --
DR. SUJANSKY: -- finding out parts of one that are in other, but we are not going to necessarily know. I think personally, I am concerned about sending that message.
DR. HUFF: No, I think my point would be that mutually cross referenced, you are taking it with a different meaning than -- if we are saying for instance that LOINC should have its primitive parts defined by reference to other UMLS concepts, that is absolutely perfect.
Mutually cross referenced doesn't necessarily mean that. Mutually cross referenced might mean that LOINC codes and SNOMED codes and other things point to each other and say this thing is the same as that thing, which to me wouldn't be a good outcome.
DR. SUJANSKY: It may or may not mean that. What I am saying is, it is unclear what mutually cross referenced means.
DR. HUFF: That is why I want to take it out.
DR. SUJANSKY: But it was taken out on the grounds that it was too specific. But now you are saying you want to take it out because it is not specific enough. I think we need to get clarity on that.
DR. HUFF: We can have more than one reason if we want to take it out.
DR. SUJANSKY: That's right. I'm saying we need to have parity on it.
DR. COHN: Betsy has a comment, I think Steve has a comment, and then we have a comment from the floor.
DR. HUMPHREYS: I am just trying to think of a way around this. First off, obviously these things exist, and there isn't anything in here that means that we are going to destroy parts of them, if that is not mutually agreeable to the people that developed them.
The way to do this however is not -- the goal is clear to me. If you mean cohesive or coherent, and if you prefer cohesive, that is fine with me, too. Internally redundant, minimally redundant; minimally redundant doesn't say no redundancy, it says minimally redundant.
I agree with Stan that one of the ways to achieve minimum redundancy is to say, we will stop the development of this over here, and we will only do it over there. I think that probably is one piece of how we will get to where we are going.
The other question is, as I say, I see it from another perspective, which is, does each of these have mutual cross referencing or does the cross referencing take place within the UMLS? It is very specific, what you have got here.
DR. SUJANSKY: But mutual cross referencing doesn't imply that where it is cross referenced and how it is cross referenced, whether it takes place in the NLM or --
DR. COHN: Walter.
DR. SUJANSKY: I just want to understand Betsy's point.
DR. HUMPHREYS: That is how I would read it. What we are understanding the words mutually cross referenced, it is describing a specific technique to achieve this goal. What we are saying is that there are multiple techniques to be used to achieve the goal of the consistent and minimally redundant terminology resource. The National Library of Medicine should have available to it the whole gamut of tools that may be available, and mutual cross reference is one of those tools.
DR. ASHMON: Diana Ashmon, SNOMED. By taking out the word mutually cross referenced, I think what I am sensing is that the word resource then becomes an ambiguous term. One alternative to retain what I think is the intent of the committee might be to substitute for the word redundant terminology resource, substitute the words redundant core terminology set. So it keeps the notion of the set alive as you are talking about achieving a coherent consistent terminology.
So again, I would recommend that perhaps the word redundant core terminology set be a substitute for redundant terminology resource.
DR. COHN: Bill, do you have a comment?
DR. YASNOFF: Yes, actually it was very close to the comment I was going to make. By saying mutually cross referenced, you have implied multiple terminologies, but by taking it out, now there is an implication that there is going to be a single unified terminology resource.
DR. HUMPHREYS: I think Diane's wording takes care of that.
DR. YASNOFF: The wording I was going to suggest is a little different, to say consistent and minimally redundant set of terminology resources.
DR. SUJANSKY: I agree that that clarifies it, but I am concerned that now the message is even more distressing. It is not even necessarily going to be a single resource. It could maybe just be three terminologies that are somehow --
DR. YASNOFF: We are not prepared to recommend.
DR. SUJANSKY: I hear that, I hear that.
DR. YASNOFF: There is clearly consensus in the group. We are not recommending that all terminologies be merged into a single one. So we don't want to recommend that, and so we don't want to imply that. At least, that is what I am sensing.
DR. HUMPHREYS: I think that Diane's wording helps here, so I would vote for internally consistent and minimally redundant core terminology set.
DR. SUJANSKY: Everybody okay with that?
DR. COHN: Yes, I'm okay with that. I have one additional suggestion.
DR. MC DONALD: Can we get that in?
DR. COHN: But I actually had a suggestion as we move on this that we actually add back mutually cross referenced. How about this? They be organized into a coherent, internally consistent, mutually cross referenced and minimally redundant core terminology set. I think we do want it to be cross referenced.
DR. SUJANSKY: But that is the same --
DR. COHN: No, that is one of the requirements. I think we do want them to be --
DR. HUMPHREYS: If you take out the mutually and say cross referenced, we'll be fine.
MR. BLAIR: Betsy just may have given the link. She just took out mutually --
DR. COHN: And just call it cross referenced. Is that okay?
DR. SUJANSKY: Okay, I'm fine with that. I think that takes us most of the way to what I was concerned about.
DR. COHN: Sometimes it looks different than what we expected.
DR. MC DONALD: Where do you want to put in cross referenced?
DR. GREENBERG: Before core, right?
DR. SUJANSKY: Simon was suggesting just adding it as an internally consistent, minimally redundant cross referenced core terminology set.
So let's see how that reads now, make sure it is still proper English.
DR. HUMPHREYS: No, it's not. We knew that already.
DR. SUJANSKY: Furthermore, the NCVHS recommends that these terminologies be organized into a coherent, internally consistent, minimally redundant and cross referenced core terminology set. One can just about parse that.
DR. COHN: Okay.
DR. SUJANSKY: Okay, we're all okay. Also, -- we read that, no comments on that part, right?
DR. MC DONALD: Do we need and? I don't think you do. It is a bright sunny day. You don't say a bright and sunny day. I think we don't need the word and in there. That is not hat was suggested.
DR. SUJANSKY: Okay, that's fine. Do you need the comma?
DR. COHN: Actually, it does need the comma. Otherwise, it becomes a minimally redundant cross reference.
DR. SUJANSKY: I call the question on it.
DR. COHN: You call the question? Are we done with this one? Knowing that Walter can play around with this, do we have consensus on the comma? Shall we move on?
DR. MC DONALD: The list of terminologies. We still have an open question on that.
DR. COHN: Why don't we handle that in the body?
DR. STEINDEL: Now, a question on style at this point. This has nothing to do with the list. You listed SNOMED CT as licensed by the National Library of Medicine, LOINC laboratory subset. I think the parenthetical expression should be consistent.
DR. SUJANSKY: I agree. I tried to think of something to put in there when I changed this, but I couldn't. RX Norm, the NLM --
DR. STEINDEL: The RX Norm would be developed and maintained by the National Library of Medicine or something like that.
DR. SUJANSKY: Or leave out the parenthetical.
DR. HUMPHREYS: You need the second one because it is only a piece.
DR. SUJANSKY: The second one is definite.
DR. GREENBERG: You need the first one, too.
DR. HUMPHREYS: Why don't we go back after it is discussed later on? It may become a non-problem.
DR. COHN: Let's keep going.
DR. SUJANSKY: I agree with you Steve, but I couldn't think of anything.
DR. HUMPHREYS: Maybe you should not make it parenthetical LOINC, but say period and then say laboratory subset. They are two different things.
DR. SUJANSKY: I think we are getting contorted because of this parenthetical issue, and maybe it is not worth it.
DR. COHN: Okay, let's move on.
DR. HUMPHREYS: On a real stylistic thing, under the core set, you said NCVHS recommends, and under SNOMED CT and the rest of the paragraphs, you refer to the NCVHS. So it is either the NCVHS when it is the subject of a sentence, or it is NCVHS, one or the other.
DR. SUJANSKY: What do you prefer? No the?
DR. COHN: No the.
DR. SUJANSKY: But if I spelled it out to what it actually refers to, we would have to have a the.
DR. COHN: Walter, I think we can take this particular piece under advisement.
DR. HUMPHREYS: I'm sorry I brought it up.
DR. GREENBERG: Our official writer, Susan Canann, has recommended that we not use the the.
DR. COHN: That's fine. Let's get back to the content here, please.
DR. SUJANSKY: Okay. Where were we? NCVHS through the terminology evaluation process as described above has identified SNOMED CT as the general terminology for the core set of PMRI terminology, breadth of content, sound terminology model and widely recognized value of SNOMED CT qualify it as a general purpose terminology for the exchange, aggregation and analysis of patient medical information.
The broad scope of SNOMED CT itself and the inclusion within it of content from other important health care terminologies such as the nursing specific terminologies, NANDA, NIC and Omaha, allow SNOMED CT to accomplish much of the patient medical record domain. A license agreement between the College of American Pathologists and the National Library of Medicine make the use of SNOMED CT in the United States available free of charge.
DR. COHN: Bill and Susie, then we have a question from the floor.
DR. YASNOFF: For consistency, where you say patient medical information, shouldn't we say patient medical record information?
DR. SUJANSKY: Yes.
DR. BEBEE: I have laryngitis, so I apologize first. I suggest that we list the nursing specific terminologies, all of them that are there, rather than such as, because there are other ones of the 12. HHCC is one that has been wrapped, I believe.
DR. SUJANSKY: Is that true, Virginia?
DR. SABA: There are 13. I don't know how many are in SNOMED, so you want to list those that are in SNOMED.
DR. SUJANSKY: But is HHCC among those?
DR. SABA: Yes, it is. That is the only one that is public domain. All the others are copyrighted. So it is misleading. If they are going to use the SNOMED version, that is one thing, but if they want to use the specific copyrighted version, then they have to file another separate license.
So either you list them all and describe how they are used, or not list any of them.
MR. BLAIR: If we change the such as to including, and then you wind up -- and which ones have we left out that are in SNOMED?
DR. HUMPHREYS: I think the perioperative nursing --
DR. SUJANSKY: I have the list. I am happy to do that.
DR. SABA: But you still have a problem between those that are public domain and those that are copyrighted, because they change in SNOMED when they are copyrighted.
DR. SUJANSKY: What we are saying here is content from other important health terminologies. So we are not implying that entire copyrighted terminologies are part of SNOMED. I don't believe we are implying it.
DR. HUMPHREYS: I was thinking that if you changed it to concepts, instead of saying content from, if you said concepts from, that might address Virginia's issue.
DR. SABA: That would be better.
DR. SUJANSKY: I'm fine with that.
DR. SABA: I'm fine with that.
DR. SUJANSKY: Everyone okay?
DR. SABA: And I would list them all, the ones that are in SNOMED, to be fair.
DR. COHN: Marjorie?
DR. GREENBERG: Y0u might just want to say, such as several nursing specific --
MR. BLAIR: No, including.
DR. GREENBERG: Or including several nursing specific, not mention the actual terminology.
MR. BLAIR: No, we want to specifically identify them.
DR. SUJANSKY: Yes, I think that is the point, and it is a good one. We shouldn't just mention some and not all that are included in one way or another in there.
DR. COHN: Carol.
DR. BICKFORD: Carol Bickford, American Nurses Association. Although these terminologies were created by nursing for nursing, they have application to other health care providers and other entities. So it may be that we would like to include some thinking that it is languages that support nursing practice, but it also allows other entities to appreciate it.
When it is nursing specific, it is still constraining from the -- we have done the work, and it is specific to nursing, but there is significant value in all of our products to other providers.
DR. SUJANSKY: Carol, wait, before you sit down, do you have a recommendation as to how we can replace the specific phrase nursing specific terminology?
MR. BLAIR: Yes, she said languages that support nursing practices.
DR. SUJANSKY: Is that correct?
DR. BICKFORD: You can list the terminologies that support nursing practice. I'm just bringing to the attention that these are very useful for other professionals besides nursing, although we are the majority users, and we have done the work to make it.
DR. SUJANSKY: Such as terminologies that support nursing practice, i.e., -- I'm sorry, Carol?
DR. BICKFORD: I'll give you specific language.
DR. SUJANSKY: The subcommittee needs to be able to test the language.
DR. HUMPHREYS: Maybe it may be a way to refer to them in terms of credit when it is due, and Carol's good point is to say terminologies developed to support nursing practice.
DR. SUJANSKY: I think that's good. How about if we say, other important health care terminologies, including terminologies developed to support nursing practice, such as blah, blah, blah.
MR. BLAIR: Including.
DR. SUJANSKY: Including, but then instead of i.e., say such as, but then list them all. Or we could say i.e. and include them all. I don't like to use i.e. That's fine, either way is good.
DR. COHN: Other comments on this?
DR. HUMPHREYS: The last sentence, I think it is a little strange to say make the use of something available free of charge. Use free of charge, maybe?
DR. SUJANSKY: Yes, it is a little awkward.
DR. HUFF: I had a question. I'm not sure what it means to be the general terminology. I would be more comfortable if we said SNOMED CT as the source for clinical diagnosis, diseases, signs and symptoms, body location, give some idea of -- I guess it is a subtle difference between -- I don't know what general means. Does that mean we use it anywhere we don't have something else?
DR. SUJANSKY: I would just say, maybe the very next sentence, the broad scope of SNOMED CT itself and the inclusion of, et cetera, allow SNOMED CT to encompass much of the patient medical record information domain. That speaks to what is meant by general, much, but obviously implying not all.
DR. MC DONALD: Yesterday we heard testimony about devices, and that could be a much larger area that might be another domain.
DR. HUMPHREYS: Yes, but I don't think there is any implication here that it covers all, just general broad coverage, which it does.
DR. SUJANSKY: I think we would all agree that it has more general coverage than any of the other terminologies that are under consideration.
DR. COHN: I actually agree that we should leave it a little vague. In this, we think we can sit down and precisely identify every domain that SNOMED CT -- which seems to me to be way beyond the scope of this letter.
DR. SUJANSKY: Because again, we don't want to dictate to the NLM and so forth which areas should be carved out of SNOMED and designated for another terminology and so forth. So if we list each individual domain area, we would be at risk of doing that.
DR. MC DONALD: But I would still be more comfortable if we said something like SNOMED as the source for many domains of clinical, or something like that. It is a subtle difference, saying yes, we want to use all of the things that are there, versus saying everything that we don't state somewhere else is part of this general --
DR. SUJANSKY: Again, it is not implying we are going to use everything that is there by saying it is a general terminology. I personally don't see that implication. If there is agreement that there is that implication and there is concern about that implication, then I'd be happy to change it.
I think obviously it is language, and we are all from different backgrounds and with different perspectives are reading different things into the meaning. But obviously not everyone who reads it will --
DR. MC DONALD: I guess it is probably the "the" that concerns me more than the general. If you have the general terminology, you can only have one of them that is the general terminology.
DR. SUJANSKY: That is the intended meaning. Remember, going back to our general classification of the terminologies in the analysis and so forth, we were of the mind that we needed to pick one that was in that general category. We did have a general category.
That is the other point to make. This is going to be supported by documents that we have prepared. Obviously not everyone will read those documents, but certainly some people will, and this will link to that.
But again, if we want to change it to a general or something like that, I'm fine with that.
DR. COHN: I think at the end of the day, we are talking about that core set of minimally redundant da da da da da. Then you can talk about age-old terminology, duplication, there could be multiple general terminologies. I don't think this gets us where we want to go. It provides a conflicting message.
DR. MC DONALD: Could I just make a suggestion? Maybe the place to do -- there is no specification at all to a naive user, what this is.
MR. BLAIR: I'm missing your words.
DR. COHN: You need to get closer to the microphone.
DR. MC DONALD: Maybe if you put something after breadth of content, including diagnosis, symptoms, findings, anatomic sites, kind of like the nursing thing; you give people a sense of what it does contain, and maybe a little sense of what it doesn't contain as a result. It's a lot better if people understand it better.
DR. SUJANSKY: Are we in agreement that that is a good idea? If we are going to do that, actually I would suggest putting it in the next sentence, just so it will read a little better. Say, the broad scope of SNOMED CT, including X, Y and Z.
DR. MC DONALD: Let's do some of those X, Ys and Zs.
DR. SUJANSKY: But let's make sure we are in agreement that that is a good idea.
DR. MC DONALD: That sentence is a good place for it.
DR. COHN: I guess, Clem or Stan, since you are both having issues with this, what exactly is the issue here, so that we can make sure that we address it? There are many places in the letter that we leave things as being -- what we are doing is trying to provide directional guidance. We have asked the NLM to provide on the ground coordination, which I understand that you are actually part of, Stan. I don't have concern based on what we are seeing here, that we are somehow providing the wrong communication, but obviously you --
DR. HUFF: The previous wording of this said SNOMED CT as the core. What I am worried about is establishing some hierarchy within the core that makes SNOMED more pre-eminent than other parts of the core. That is what I am worried about.
DR. COHN: Ah, okay. That is a slightly different issue, which is that what we describe is more equal than others.
DR. HUFF: Clem, is that your issue, too?
DR. MC DONALD: No, my issue is that I think no one knows what this is in the whole world. That is the truth. They think, we are done with terminology. There are probably another two million terms that we have floating out there, everything including the very detailed devices through the medication stuff. So it is helpful to give specifics for the world to understand, that's all.
DR. COHN: Steve.
DR. STEINDEL: Going back to our reference document, which is the scope and criteria document, we don't really refer to a general set for the terminology in there. We refer to a Section 4.2, domain coverage of the core terminologies. We list in there, and we note this is not an exhaustive list, clinical disorders, diagnoses, subjective symptoms, observed findings, procedures performed by clinicians, laboratory tests and other tests, blah, radiology tests and test results, findings, anatomical structures, medications, clinical substances other than medications, medical devices and supplies, and I am abbreviating, social and care management concepts, standardized assessment tools, and then we add, and the list is not exhaustive.
So we really do prescribe a list in our reference document, and when we go back and discuss it here, we are using the word general, but not referring it back to anything in the reference document.
I see what Clem's point is. We probably should be a little bit --
DR. SUJANSKY: No, it is not mentioned in that document, but it is mentioned in the analysis document, for example, here.
MR. BLAIR: Could we do it in a footnote? We are trying to keep this so it doesn't drown people when they try to read it. So if we put a footnote by general, indicating the things that were included, that way it doesn't disrupt somebody when they are reading it.
DR. GREENBERG: Doesn't that other document you just had up define general?
DR. COHN: Well, actually that didn't define general, I don't think.
DR. GREENBERG: I don't think that -- this is actually a process that the CHI is going through, but I don't think the subcommittee has really gone through a process of looking at SNOMED for each one of these domains.
I think we have gotten testimony, et cetera, that gives confidence that there is good coverage of a lot of these domains, as he puts it, much of the patient medical record domain. When you start mentioning individual domains, that could be a little pre-emptive, or imply work that hasn't really been done.
DR. COHN: Yes. As well, it may provide some confusion, just because there are some pieces where part of a particular domain may actually be covered by SNOMED, but you need the increased granularity of additional terminologies. So I don't think we necessarily want to get into that discussion in this letter.
DR. HUFF: Let me suggest some specific wording. What I would say is, the first part of the sentence would say the NCVHS through the terminology evaluation process described above has identified SNOMED CT -- and here is where the change comes in -- SNOMED CT has the best source of a broad range of general medical terms.
DR. HUMPHREYS: That is not good. They have got a lot of specific terms, too. That sounds weird.
DR. HUFF: There is a broad range of whatever you want there. We can take out the general. But as a best source for --
DR. GREENBERG: A broad range of terms needed for patient medical record information, something like that.
DR. SUJANSKY: To address your specific concern, which is the potential misconception that somehow SNOMED is more equal than he other terminologies, perhaps in the next paragraph we can say, recognizing that -- in the paragraph where we talk about LOINC, we can say, recognizing that SNOMED CT does not contain all the content required for patient medical record information terminology standards, and then we talk about --
DR. HUMPHREYS: No, I think that is worse, not better.
DR. SUJANSKY: But would it not address -- Stan, would that address your concern?
DR. HUFF: That's okay. If the rest of the committee doesn't think this is good, then just leave it the way it is, that's fine.
DR. MC DONALD: Let me try another tack. I think SNOMED dominates in atomic concepts. That may not fit at all, but there is this subtle but important difference when one gets into other kinds of things.
MR. BLAIR: Would you get to the microphone, please?
DR. MC DONALD: I'm close, I think. I can't chew it.
DR. COHN: Clem, the problem is, you don't actually talk into it because you go like this.
DR. MC DONALD: I'm sorry.
DR. COHN: You could put it to your left and that would work.
DR. MC DONALD: Like this?
DR. COHN: No, move it to your left because that is how you talk.
DR. MC DONALD: I talk to the left?
DR. SUJANSKY: It sounds like maybe it is okay the way it is?
MR. BLAIR: Yes, let's leave it there. There are so many other issues that have substance, that we really need to --
DR. SUJANSKY: So we have worked through that whole paragraph then. Let's go on to the others.
DR. MC DONALD: One simple sentence at the end, unless you did this already, about the license made use of, et cetera?
DR. SUJANSKY: Yes.
DR. MC DONALD: Did that get fixed?
DR. GREENBERG: Yes.
DR. SUJANSKY: We took out available. Let's just make sure that reads okay. License agreement, should read, makes the use of SNOMED CT in the United States free of charge.
DR. GREENBERG: Or maybe free of charge in the United States.
DR. MC DONALD: That is what I was going to suggest.
DR. HUMPHREYS: The license between the CAP and NLM doesn't address how it is used outside the U.S.
DR. COHN: Now, do we want to try to deal with LOINC before we take a break?
DR. HUMPHREYS: Be very quick here, right? I believe that since neither Clem nor Stan can help us here except to nod their heads yes or no, I believe that it is correct to say it is a terminology for the representation of laboratory test results, is that right?
DR. SUJANSKY: Let the record show Clem is nodding his head.
DR. HUMPHREYS: All I know is, Stan tells me there are clinical attributes, which means these are not tests.
DR. SUJANSKY: The next sentence says lab LOINC is already used in many systems for ordering laboratory tests and reporting laboratory test results.
DR. COHN: Susie and then Steve.
DR. BEBEE: I think this is where we wanted to add that we take note that LOINC has already been adopted by HHS.
DR. SUJANSKY: Right, thank you.
DR. STEINDEL: And also, you want to remove "the" NCVHS.
DR. SUJANSKY: Very good.
DR. STEINDEL: My other comment was the same as Susie's.
DR. LINCOLN: The VA would like to comment on LOINC, too. We were using it for our lab results now, and we are using the clinical document architecture for our document titles and subsections of titles.
DR. COHN: Are we okay with this paragraph?
DR. STEINDEL: Simon, just as a style comment, when he is expanding that parenthetical remark that he just put in there, should we refer to the July letter that we sent to the Secretary?
DR. COHN: I don't -- no.
DR. STEINDEL: I was just asking the question.
DR. COHN: Everybody okay with this? I'm going to suggest that we take a ten-minute break. Actually, this is turning out to be hard going, but I don't think it is going to get any easier. So let's take a break and take a breath, and come back in ten minutes.
(Brief recess.)
DR. COHN: So we have made it through two of the three elements of our core terminologies. I think we need to obviously consider the question about RX Norm. We have heard one comment of concern about it not being ready for being so named.
Is Betsy here?
PARTICIPANT: She will be back in a minute.
MR. BLAIR: How about Stuart Nelson? Is he here?
DR. COHN: I don't think Stuart is in the audience. I think we need to look at the wording, and we also need to reflect on its readiness as a part of the core terminologies.
MR. BLAIR: Should we go on to the next item and then come back to that as soon as Betsy gets back? Or I don't know if we can even go on to something else.
DR. SUJANSKY: The next area I think talks about the NLM's role, so we won't get far. The next area actually talks about the important related terminologies. We could go to that. Is everyone here except Betsy?
DR. COHN: I think Bill has actually left.
DR. SUJANSKY: I'm happy to start, and we'll come back to that section.
DR. COHN: Okay.
DR. SUJANSKY: Steve is out of the room, too. We skipped over the next section which is about drug terminology, and then the next section has the heading, mappings among the core set of PMRI standard terminology. To form an cohesive, internally consistent terminology resource, a core set of PMRI terminologies must be integrated.
This integration should consist of mappings between equivalent or hierarchically -- is this in the letter that was printed out? Okay, I'll go on then, sorry. This integration should consist of mappings between equivalent or hierarchically related concepts across the core terminology, such that these relationships are explicitly represented. The mapping relationships should be part of the content of the PMRI terminology standards maintained in concert with changes to the constituent terminologies.
DR. FITZMAURICE: Now, Walter, this takes care of the mutually cross referenced issue above, right?
DR. SUJANSKY: It expands on it, yes.
MR. BLAIR: Please understand, this is two different types of mappings that we are discussing.
DR. FITZMAURICE: What are those two? I'm not sure I'm with you on that.
MR. BLAIR: This section is referring to the important related terminologies.
DR. SUJANSKY: I'm sorry, Jeff, it's not yet. That section is next. This is referring to mappings among the core set.
MR. BLAIR: Okay.
DR. COHN: I'm actually sorry that Betsy or Clem are not here. I am actually wondering whether we even need that whole section with the other parts we have put in.
Actually we have got Betsy here. Maybe we should go back to the RX Norm discussion and then move into this one, because these are all things that relate to the work of the NLM.
DR. SUJANSKY: Okay.
DR. COHN: Does that make sense?
DR. HUMPHREYS: Sorry.
DR. COHN: Betsy, we were holding up discussion of the RX Norm until you arrived.
DR. HUMPHREYS: Sorry.
DR. COHN: Then we realized that the next section included work that we were also expecting of the National Library of Medicine.
DR. SUJANSKY: This language of course is all new since yesterday, so it is not in anything that was sent out, or printed copies of anything. Again, this is the initial cut at this language.
RX Norm. RX Norm is a medication coding system that is currently under developing within the unified medical language system of the National Library of Medicine. The goal of RX Norm is to provide non-proprietary and clinically relevant coding standards for medications. Creation and maintenance of RX Norm are facilitated by the availability of drug product information in electronic form from the FDA.
Principal deficiencies of RX Norm in its current form are one, the population of drugs within its current model is not yet complete, and two, the current model lacks allergen classes, therapeutic classification system and comprehensive mappings to NDC codes. These components are judged important for RX Norm to be accepted and adopted as a national standard for coding medication information.
Further recommendation. The NCVHS recommends that the federal government increase funding to the RX Norm project to accelerate the population of the current terminology model and to supplement RX Norm with applicable allergen classes, therapeutic classification system, and comprehensive mappings to NDC codes. NCVHS also recommends that the Secretary support the FDA's efforts to improve the timeliness and efficiency with which it collects drug product information. The success of these efforts will streamline the maintenance of RX Norm and enhance its value as a national standard for coding medications and facilitating patient safety activities.
All right, go at it.
DR. STEINDEL: Excellent. Let's move on.
DR. HUFF: Yes, that was well done.
DR. FITZMAURICE: What terminologies do doctors now use when ordering drugs for patient care?
DR. COHN: Sometimes we use J codes.
DR. FITZMAURICE: That is for billing. They don't order with J code, right?
DR. COHN: I don't know. My understanding is an NCPDP script text is the way you order a drug at this point.
DR. FITZMAURICE: So they negotiate with the pharmacy. What they call a drug -- if I call it this, you will fill this?
DR. COHN: I think they are so happy to be able to read it, if it is legible I don't think they complain too much.
DR. HUFF: With Intermountain Health Care, we use codes that are derived from First Databank.
DR. COHN: Steve.
DR. STEINDEL: My sense on mulling over the discussions yesterday and actually seeing the language which except for some crafting is pretty much what I envision the language to be, what we are putting in here is a very conditional recommendation. Probably what I think is a good approach is to actually make the statement that we are not prepared to make a recommendation at this time, that it is conditional. The drug terminologies that fit in this mold are still in a state of flux, and that we are going to be revisiting this and making a recommendation to the Secretary. Then I think we should go on and note that we are considering RX Norm and add the supportive type language, and add language that RX Norm cannot stand without the standards development efforts of the FDA that is needed to support it.
Then when those efforts reach fruition, we can return to the subject and make a firm recommendation to the Secretary, something along that idea.
DR. COHN: Betsy, one of the reasons we wanted to make sure you were here was, we want you to reflect on what you think the current maturity or completeness of RX Norm is, at least in terms of what it is trying to do.
DR. HUMPHREYS: I think that if the committee agrees generally with this language, then I agree with Steve that it sounds too provisional for a recommendation.
I do feel that it to me is an open question, whether the drug terminology portion of a core terminology set is -- RX Norm as defined doesn't have some of these things. The therapeutic classification system, for example, doesn't have that. If we were to create an RX Norm, it would become the third one that we have. SNOMED has one, the medical subject headings have one, et cetera.
So the issue again is, what can we make available in the core set of terminologies linked in an effective way, and if we can do that effectively without creating yet another therapeutic classification from ingredients in drugs, we might do it that way.
So that is an issue I have about a specific recommendation to make these additions to RX Norm, versus the fact that obviously they should -- perhaps everyone agrees that they should be available in the core set.
DR. SUJANSKY: My intent in the language of the further recommendations was to not state that these other component areas should necessarily be added to RX Norm per se. I was struggling for the right word, the word I selected was supplemented, supplement RX Norm with applicable allergen classes and so forth.
DR. HUMPHREYS: Yes. I guess the thing is, up above on that same point, it might be phrased differently. The last sentence seems to imply --
DR. SUJANSKY: That reads, these components are judged important for RX Norm to be accepted and adopted?
DR. HUMPHREYS: As opposed to saying these components are judged important for a core set terminology set or something.
DR. SUJANSKY: Fair enough.
DR. GREENBERG: Also, one solution may be to not add these things to RX Norm. Then I don't think you want to refer to them as deficiencies. You might want to say gaps or something.
DR. SUJANSKY: Right.
DR. HUMPHREYS: As far as NLM is concerned, we could deal with this recommendation in either direction.
DR. HUFF: I guess one of the question is, what is the material difference between saying what Steve said, versus approving this as a core terminology.
I agree with these things that need to be done. I don't see any better solution or one on the horizon that would provide this non-proprietary coding system for drugs in this area. I hope we are not implying by this statement noting the deficiencies that we think any of the other ones above are perfect, because in fact, there are big areas where they haven't been used for the purposes that we were conjecturing, either.
So do we just think there is a dramatic difference in the maturity here? We think it is big enough that we need to note it.
DR. SUJANSKY: People can use LOINC, for example, even though it doesn't have every single lab test.
DR. COHN: Steve.
DR. HUFF: You can use drug ones without having every single drug in them, too, though.
DR. STEINDEL: Stan, the subcommittee hasn't seen this yet. Just from a style point of view, CHI has run into this type of situation in several of the terminology domains. We have noted that there is a rising terminology, one that appears to be the one that we will recommend, but we are not in a position to recommend it, and have suggested that we re-look at the situation in six, 12 or whatever months. So that is the concept that I would like to get across in the letter.
DR. HUFF: I guess I come back to my first question. What is it that we want to have happen here? I wouldn't like anything we do here to either impede the progress in creating RX Norm or to stop people from using RX Norm, because we didn't approve it.
DR. SUJANSKY: The flip side of that is, without including sufficient provisos, there may not be awareness on the part of the Secretary and the government that more work and perhaps more resources need to be put into RX Norm before it can meet the needs of the standard. So there are two sides of that coin.
DR. COHN: I think we have some people to make comments, please.
DR. LESH: Kathy Lesh, the Kevver company. I just want for my own edification, is RX Norm a medication coding system? I don't believe that that is what it is trying to be, is a coding system.
DR. SUJANSKY: Well, it is a system for assigning codes to medication concepts.
DR. HUMPHREYS: Not exactly.
DR. SUJANSKY: How would you describe it?
DR. HUMPHREYS: I would say that basically what we are trying to d is have a controlled terminology for clinical drugs and for selected related drug terminology like dosage forms and components and so forth. So it is a controlled terminology for the clinical drug, is what it is.
DR. SUJANSKY: Okay, so before we leave that point, do we want to change -- or I should say, how do we want to change this language? Relevant terminology for medication, instead of coding standard?
DR. LEVIN: Randy Levin for FDA. As Betsy was just saying, AHRQ's Norm is for a specific concept in all the medication and drug terminology. There are many other concepts. So if you are not capturing those, you are going to be missing a lot of things.
From ingredients is also another concept. The product itself is a concept. That is not captured by RX Norm. That includes the inactive ingredients, the appearance, the package product, the dosage form. Those are all concepts that are not captured by RX Norm. If you want the whole armamentarium of drug terminology, then you need all those things. RX Norm is one piece of those. So you are not clumping everything into RX Norm. RX Norm is one concept.
Betsy, is that --
DR. HUMPHREYS: Yes, I think that it seems to me if you are going to discuss the drug terminology in this recommendation, then there is the issue that there are many different levels at which there need to be standard terminology for drugs and medications. What we think is, if you think it, this one is either provisionally there for these pieces of what we need, and we can identify which pieces they are, and we can recommend it for that. But I do think it is kind of important to indicate that there is such a thing as the NDC level. Otherwise, as all of these charts they were showing yesterday show, there are all these different levels at which you can control drug and medication terminology.
DR. LEVIN: So when someone wants to talk about a product that is out there, they are not going to be able to express that using RX Norm. They are going to have to use the product code to describe the product.
DR. SUJANSKY: I agree. But again, we need to remember and revisit that we are talking about patient medical record information standards.
DR. HUMPHREYS: Yes, but all of them are relevant.
DR. SUJANSKY: Yesterday we tried to draw a distinction between knowledge and coding and so forth, so I think we --
DR. LEVIN: This is a code on a product. If you want to describe a product in a patient medical record, you are not going to do it using RX Norm. You are going to need a product code to do that. That is not a knowledge base; you are just describing a product.
If you are going to describe ingredients, you want an ingredient code. So those things are very relevant for safety issues and other things. All the inactive ingredients are not being covered by the concept of RX Norm. That is just one piece of the whole armamentarium that you want for product codes.
DR. SUJANSKY: I think the issue we need to consider, and I'm not sure what the answer is, is what is needed for patient medical record information, for electronic medical records, what is needed. Are drug ingredients in the medical records?
DR. HUMPHREYS: Yes, because generally that is the level at which you are going to show an allergy.
DR. SUJANSKY: But the allergen classes address that, which is one of the gaps --
DR. HUFF: They overlap with the ingredients. They are essentially the same thing.
DR. SUJANSKY: They do, but if you need them to designate allergies, why do you need them for other things?
DR. HUFF: We put them in the medical records. It is not a theoretical question. We put the ingredient names in our medical record today.
DR. LEVIN: Also, you want to put the product in there, too. That is how we use surveillance on adverse events.
DR. HUFF: What we actually do is, we take the product, we know from the knowledge base the ingredients that are in the product, we store the ingredients in the record, because that is a much more efficient way of checking for the allergies.
So he is exactly right; you need the ingredient codes, or at least we need the ingredient codes. Maybe that is not universally done, but we store the ingredient codes today in the electronic medical record. Not only that, those ingredients are the things that you want to tie to the measurements that are made in the laboratory, because if the ingredient that you measure in a toxicologic study, when you are measuring drug levels in the laboratory, it is the ingredient --
DR. LEVIN: That can be derived from the knowledge of the medication.
DR. HUFF: You need an identifier for the ingredient, though.
DR. SUJANSKY: It is an important issue, because we are talking about scope now, and we are basically addressing, is RX Norm sufficient or not. So this is the point that we need to address. We are hearing that it may not be.
MR. BLAIR: When I made the proposal yesterday, my thought was that RX Norm could serve as a reference terminology for the clinical concepts, and that the drug knowledge bases and NDFRT could all be mapped to it. I thought that that was a compromise that would allow us to move forward in a way where we are not selecting one of the drug knowledge bases over another, because they all serve slightly different markets and slightly different set of users.
If we start to expand our discussion into the realm of the drug knowledge bases, then I think we are going backwards. Yes, they are all needed, but within our core set of PMRI terminologies, what we are looking for more is the terminologies that serve as reference terminologies, which I thought RX Norm was more appropriate for.
DR. HUMPHREYS: Just let me clarify one thing. If you have your presentation that Stuart gave yesterday, what is included as part of RX Norm is active ingredient, drug component, clinical drug and dosage form, and the relationships among all of those. In cases where we know a brand-new product, the brand is connected to the clinical drug. NDC codes when known are attributes, but they are not separate concepts.
So the basic core of this is active ingredient, drug component, clinical drug and dosage form. As Randy pointed out, when the FDA starts distributing its entire ingredient field, we would include their identifiers and so forth of the active ingredients, and then we would have a source of the inactive ingredients which are not part of the RX Norm model at the moment.
So those are the four things, the active ingredient, the drug component, the clinical drug and the dosage form.
DR. SUJANSKY: So are you suggesting we should add inactive ingredients to the things that need to be added to RX Norm?
DR. HUMPHREYS: No.
DR. SUJANSKY: No? We are hearing inactive ingredients need to be there, right?
DR. HUMPHREYS: Inactive ingredients?
DR. MC DONALD: I think the challenge here is, if we listen to the challenges in health care, medication errors -- I guess I am talking about of both sides of the street on this one -- that is the big force. So the core of RX Norm is the glue that is going to hold together lots of things. I think the challenge is, we can never stop extending what else could be useful. The instrument and devices discussion is exactly the same.
So I would like to think we want to narrow this down as something we can support strongly and say this is good, and get it done. We have been talking about standard drug code with some of the people in the room for six years. So however we can make this happen fast, I think is the right way to do it. I don't think we want to expand it to everything it could be, because that will just slow it down. There are some good products that do lots of things; we don't have to replace.
DR. COHN: I second what you are saying. I am trying to think of whether -- I think we are dealing with first of all getting a real handle on what this is. Obviously, you are speaking of the fact that a lot of deficiencies get solved with integration with other terminologies, which of course is one of the things we are asking the NLM to do.
The question is, do we support this as a core set, or do we take the other side, which is review in six months, or do we create a new concept, a provisional part of the core set terminologies, needs a little more work, we think it is very likely going forward, highlight it in that sort of a fashion, with the idea that we will be updating the Secretary more over the next couple of months after it is more thoroughly tested.
DR. MC DONALD: Because it is such an important part of managing health care, --
DR. COHN: Absolutely.
DR. MC DONALD: -- and the communication of medical information is so hard, because we don't have a single code that we can attach to, I would support making it a core set, but making the scope narrowly than what is in the presentation, at a level that would be acceptable to the community at large. I think we just really need to get going with this.
DR. COHN: Steve, did you have a comment?
DR. STEINDEL: I'm mulling a little bit over what Clem said, but in terms of what Simon said, I think either way we do a recommendation for our RX Norm is going to come across as a provisional recommendation. If we say we recommend it and we put this type of language in, it is going to come across as a provisional recommendation. I think if we are going to do that, then we ought to state right up front that we provisionally recommend, and that the following things need to happen before it becomes a full --
DR. MC DONALD: I would take out much of this. It is damning by faint praise. You could take any codes that we have talked about or accepted, and you could pick out a part and say it looks like a pile of dung.
DR. STEINDEL: Clem, there are some things that are in the works that need to be finalized for the RX Norm system to be fully mature and be able to work. Some of it has to do with the FDA putting certain processes in place, one of which is an organized feed to the NLM. Another is a better organization of the NDC codes, so there is a better reporting of what drugs are out there.
DR. MC DONALD: But that would be gravy. I don't look at that as being the core thing.
DR. HUMPHREYS: I think that the issue is, we can be talking some time before the FDA will be able to feed data to us automatically. I hope that we are not talking a long time. I hope that we are talking within our lifetimes, this year, whatever.
But it seems to me that the issue is, in the view of the committee or anything else, what is the missing piece in our RX Norm, what is missing, and what needs to be fixed in order for it to be accepted. Is what needs to be fixed that it has to have this update frequency, that we have to know that it has what percent of approved drugs at the clinical drug level, what needs to be fixed.
I think if that can be identified, then it is provisional based on that being fixed. I'm not quite sure what finished is, in the mind of the committee. I'm not trying to be difficult; I just don't know where the bar is. Maybe CHI can define it, I don't know.
DR. COHN: Marjorie.
DR. GREENBERG: One possibility would be to do a separate letter on this. I really agree with Steve, that this is a different level of recommendation, or needs to be a different level, than the other two.
For example, there has been no testimony from users of RX Norm, because I don't think it is ready to be used.
DR. SUJANSKY: You're right, Marjorie, not RX Norm. There was one user of NDFRT, but not RX Norm.
DR. GREENBERG: I understand the interest in getting something out there. I think the committee should be very supportive and urge resources and if not the completion of this work, getting this work to the point where it really can be used.
DR. HUMPHREYS: I think the interesting part about all of this is that actually, it gets back to Stan, and I think Clem may have made this point; it is really quite usable the way it is.
DR. GREENBERG: Then maybe it should be recommended for what it is now used for.
DR. HUMPHREYS: It is quite useful and usable the way it is, but I will absolutely not tell you that it has had a lot of use.
DR. GREENBERG: It hasn't.
DR. HUMPHREYS: So in some sense, I am not disagreeing with your point of view.
MR. BLAIR: What if we take a language that Walter prepared for us on RX Norm, and instead of placing that among the SNOMED, the LOINC and RX Norm, it would be two there that we are recommending be a part of the core set of terminologies, and we move that language down to the section in the latter which refers to the ongoing or proposed additional research area? That wording I think is still appropriate. We could use either the word conditional or provisional. I feel more comfortable with conditional. That way, the issues are there.
We want the issues to be there so they will be addressed. We want to encourage expeditious growth and enhancement of RX Norm and the mappings. So I think that language helps that. Maybe you want to work on the language if we moved it down.
DR. COHN: Stan has a comment.
DR. HUFF: I think if we think about what we want to have happen, we want RX Norm to be supported, and we want people to use it. It is probably only going to get better by use.
So I would rather take the tack of saying we approve it. Six months from now, we are not going to choose something different. We are going to choose RX Norm. So I would rather say, let's choose RX Norm. Let's tell people up front that we think there is additional work that needs to be done, but let's choose it, because it is the basis for taking the next step forward. It is essential for this to exist for us to be able to do a lot of the quality assurance and improvement in medication ordering and correlating that to cost and other things.
I would be much happier to approve it and move forward, recognizing that it needs more work. But the difference to me is that if we don't approve it now, then people would not do the kinds of experimentation that you would like. You would like for people to say, oh, the NCVHS has supported this, it has been approved, let's start trying to use that. So the application developers and other people would actually try to use this, and then it would be enhanced appropriately for their needs. If we don't approve it as a core, then I think that gives people a reason to say we are just going to keep doing what we are doing, instead of actually moving to a standard.
So I think approving it as a standard creates the momentum that we want, but you don't want people to go into that thinking that this is shrink-wrapped. In fact, we don't want them to think that about any of the other terminologies, that these things are shrink-wrapped, ready to use, all you have got to do is drop them in, because there is a lot of work between this approval and when these things all work in peoples' applications.
So again, I would be more in favor of approving that, and just being up front with people about the fact that this may take some more work, just warning people that it is going to take work to get from where it is at now to a shrink-wrapped, easy to use thing.
DR. SUJANSKY: Just a one-second addition to Stan. Personally, I happen to agree with him. But factually, I would just remind everyone that in the very first draft of the letter, we listed RX Norm among the three and then in parentheses we put something like, with enhancements, with additions.
The recommendation was a little bit stronger than saying we were going to do additional research, maybe a fair bit stronger, but still provisional. That is how we handled it before.
DR. LINCOLN: I always have to agree with Stan, because otherwise he assigns me more lecture hours every semester. I have a little bit of conflict of interest there.
I just had two comments. One comment is, people have commented about the relative lack of testing or experience with RX Norm. It is important to remember that RX Norm and NDFRT are based on the VA national drug file which has been implemented for over 15 years and is maintained, and last year resulted in the writing of 75 million prescriptions, most of which, about three-quarters of which, were transmitted electronically to central mail-out pharmacies. So the seed set of terms if you will for RX Norm in large part comes from NDF, and NDF has a substantial and pretty time-honored track record.
The other point I was going to make with regards to ingredients is that the inclusion of uni codes in RX Norm would allow a link to ingredients, which might be useful to people.
DR. COHN: Thanks. Randy, did you have a comment? I understand you have to leave. Did you want to comment or not?
DR. LEVIN: I was just talking with Steve. The concept of an inactive ingredient or things that related to the product itself are different than RX Norm. I would just express that. I look at it as a whole set of codes, as a whole set of concepts. RX Norm is one piece of that.
As far as standards, people have been using product codes and package codes for a long time. Those things are in existence now, they use it, and those things are something that needed to be improved as far as how they are maintained and up to date. We discussed that already, but the product and package codes are available and used.
Some extra things that people find useful are finished dosage forms, codes for finished dosage forms and codes for ingredients.
Now, all these things will be listed. For each RX Norm, there is going to be many products listed under that. For each ingredient there will be a number of different RX Norm. They are all a group. I don't see separating them out. I see that you might bring them up as a package, that you want all those codes.
DR. STEINDEL: Simon, I can comment. As I was talking with Randy and thinking about this more, I am kind of inclined to agree with what Marjorie has said a few minutes ago. This is a much more complex issue that we can't really handle in this letter. It might be the best approach to take is to handle it in this letter with just a simple statement that we have started to look at the issue of drug coding, we find it is more complex than can be discussed in this letter, and that we are holding further investigations.
In part, I say that because some of what Randy is referring to occurs in the CHI documentation that we will be bringing in front of the subcommittee probably in October, in their medication quote-unquote report, which actually consists of six separate reports.
That goes into what Randy is talking about. As they investigated that domain, they found that there was a lot of interlinked needs for terminology in that area. I think this needs to be exposed to the subcommittee before the subcommittee can really make an intelligent recommendation to the Secretary.
DR. COHN: Michael.
DR. FITZMAURICE: At the risk of disagreeing with Steve, Marjorie, I tend to side along with Stan that it is time to do something, but I am bothered by the fact that RX Norm doesn't do as much as I would like it to do, and more is needed.
I think for allergy checking and patient safety and other clinical care needs, we need those things, like identifying the active and inactive ingredients, the manufacturer, perhaps the box. That can all be part of clinical care. For ordering, you might be able to get by with RX Norm, but you need the ingredient codes, the product codes. All of this is maintained by the FDA.
So I would feel more comfortable, if we did wind up adopting RX Norm, by saying, also as part of this we need to support not only NLM and RX Norm, but we need to support FDA in supplying all this other information and maintaining the codes for these other particular categories. We can wait and look at what CHI has been presenting to us, but there is a time for action.
DR. SUJANSKY: And that is actually the essence of the CHI recommendation. So if we want to summarize it today in that form, I'm fine with that as well.
DR. COHN: I almost understood what just happened.
DR. SUJANSKY: If I can summarize, Simon, what Mike was saying, where we add the terminologies that are going to be supported by the FDA that are needed to support the medication terminologies and impart RX Norm as well, that is part of the six or so, whatever number it is, recommendations from the CHI that would be discussed in more detail at the subcommittee in the October time frame.
What Mike is proposing is that we just accept what the FDA has told us, which they told us yesterday, about the need for this information as part of the drug information, and move it forward with language around it that this effort be enhanced and supported as needed, both within the National Library of Medicine and within the FDA.
I am comfortable with that approach, too. It is just that really, the subcommittee hasn't heard the details of that. So it is up to the subcommittee if they --
DR. COHN: Did you want to comment?
DR. ECKER: Now my comment makes sense again.
DR. COHN: Then let me try to put things together, because otherwise we're going to spend the next four hours on this.
DR. ECKER: Carol Ecker from Medi-Span. We talked yesterday about the RX Norm and also about the -- I believe the acronym was UIIN, the unique number from the FDA for ingredients. Making those both available under this recommendation in this letter, vocabularies to be used in identifying drugs, is something that Medi-Span would support and would recognize.
Going to the steps further of where the FDA is eventually going with the other levels, I can see that as a future, but trying to deal with what we have available today, I think we would need to limit it to what is available today, whereas those are more coming in the future, and then we would need to account for them at a later time when the delivery and the updating and the availability of them is more ironed out.
DR. COHN: Can you help put this together?
DR. MC DONALD: That is a tall order. But the segue really is the same thing; I think we ought to focus on -- I want to figure out a way for everybody to be happy.
DR. COHN: Me, too.
DR. MC DONALD: I don't think it is fair to say that you can't accept RX Norm until it has everything mapped, NDC code. Nothing, and I think this is true of very, very good, very, very rich software systems, have every NDC code mapped. So I think that is a bar that is silly, it is not reachable.
Plus, I am torn with what Randy wants, but we don't get anything if we keep making the ball too big. So I think we should propose the RX Norm as the identifying codes for the drugs, and talk about supporting the FDA's direction, rather than say that RX Norm isn't done until it has done everything that one can think of, because it just gets too big.
DR. COHN: I don't think I have the answer. I am hearing a wide diversity within the committee and also around the room. I think I am hearing A, that we all tend to think that RX Norm is a critical piece of the solution, though some of us may or may not think it is all completely tested yet, but it certainly appears to be a critical piece of the solution.
I think what we are also hearing is that it alone may not be completely sufficient. I think that is the discussion. I think I hear from Clem and others that this is like a really good piece for what it is, but it doesn't solve everything. Is that sort of the right word? Or that there are additional gaps that need to be filled.
We also need to admit that this is a very new system, that people really have not had a chance to thoroughly evaluate. Maybe we really are somehow -- on one extreme we have it as a complete recommendation. There is also a piece called further research, or even worse, it is called a whole completely separate letter.
Maybe there really is that middle ground or middle concept, which is that provisional recommendation, recognizing that RX Norm appears to be very good for its purpose. We recognize that there are deficiencies and holes. We provisionally believe that it should be part of the core terminology, but we will be following it and revisiting it in six months with additional communications, and believe that there need to be other parts of the puzzle that need to be put together.
Now, of course I didn't write that down, so I can't replicate it, but is that sort of where we are coming to?
DR. HUFF: Let me ask one question first. In the work that CHI is doing, is there a sense, Steve, that RX Norm would not be approved? That there is an alternative to RX Norm that is viable?
DR. STEINDEL: I'll let Randy address that. He was part of the CHI work group on that.
DR. LEVIN: RX Norm was one of the recommendations, along with the other terminologies, including the unique ingredient identifier and the other pieces.
DR. HUFF: So given that response, again, I am back to the point where not approving it now just delays support and/or implementations that would use it. If the consensus is that RX Norm is important, that it is going to be approved, that there is not a viable alternative, let's say now that we believe that, and let's acknowledge where it has gaps and where maybe it needs enhancements, but otherwise, it puts us past the September time frame, so the full committee wouldn't be able to examine this until when again?
DR. COHN: November.
DR. HUFF: November. It goes on. I would like to send an early signal that this is the direction we think we want to go, and let's encourage people to experiment and begin making applications and implementations of it, but recognize that it has gaps and things that we want to enhance.
DR. COHN: Jeff.
MR. BLAIR: I share a lot of Stan's feeling. Now, we don't approve, we recommend. I think that if we could find a middle ground, a common ground, what I would like to see is that we are recommending in the first section, but list it as either provisional or conditional, and say the reasons why it is provisional or conditional, for the purpose of encouraging that those deficiencies be addressed.
Now, I say that with -- that is what I would like to see. I guess at the same time, I am especially interested in, Steve, your thoughts, whether you feel that that would create any problems if we did it that way.
DR. MC DONALD: It is all in defining what the It is. You can say this it, which is part of the bull's-eye, is deficient, or you make the bull's-eye smaller and you make it perfect.
What I am really trying to say is that instead of saying RX Norm is deficient, we say it is not sufficient for the whole problem space. Then the whole problem space needs this additional contribution that can come from industry and the FDA.
MR. BLAIR: What Walter read to us, I thought, that paragraph, I thought covered that.
DR. MC DONALD: But I wouldn't say the deficiencies are RX Norm's. I would say additional requirements of the problem space.
DR. HUFF: This makes some things feel a lot better. You can see that the motivation up there is to make resources down here.
DR. MC DONALD: I would still couch it in terms of, the problem space has some expanded requirements.
DR. HUFF: I think that is a better way to say it, because you don't get into the damning by faint praise.
DR. SUJANSKY: I guess we are circling around a -- maybe I can distill the key issue, that is, whether it does more good than harm or more harm than good to recommend RX Norm, clearly acknowledging gaps and additions that need to be made, deficiencies, whatever we want to call it, pointing those out.
DR. HUMPHREYS: While clearly acknowledging that it does not solve the entire medication and drug terminology problem, which it never will. So that needs to be acknowledged right up front.
DR. SUJANSKY: Right, for patient medical record information.
MR. BLAIR: And we call it a conditional or provisional recommendation.
DR. SUJANSKY: Yes, we communicate that in some way. Does that do more good or more harm in terms of furthering patient medical record information terminology standards in the U.S.?
DR. LINCOLN: I just have a comment about completeness. At the OI field office in Salt Lake, we are putting together our health data repository. We have been looking at RX Norm and SNOMED and a bunch of other terminologies for completeness.
Ain't nothing out there complete, as Clem and Stan have pointed out. We have been trying to throw SNOMED at our health factors, which we use in our health maintenance checking and against our unresolved narratives from our problem list application, and a lot less sticks than you might think.
It is the same for RX Norm, it is the same for any controlled terminology. They all have to be improved. I think it would encourage work on RX Norm if there was a strong signal sent that RX Norm was recommended as a standard for clinical drugs, and improvements were recommended for other axes or components which are not currently present or which need to be enhanced.
DR. COHN: Steve, do you have a comment? Then maybe we will try to come to some sense of --
DR. STEINDEL: I was looking at the actual presentation to CHI of the medication recommendations while we were talking. It follows the thrust of what we are talking about, about recommending the FDA standards that are developed as part of the medication terminology. Whether we do it as a codicil in the paragraph that we talk about RX Norm or an explicit statement up front I think is just a matter of style, as long as it is clearly stated that this is needed to augment the RX Norm.
RX Norm was provisionally recommended for semantic clinical drugs in CHI. The provisional recommendation was not because it was incomplete, because all these terminologies as Mike just pointed out are incomplete, and they are always going to be incomplete, but just that it is in development.
I think there is a subtle difference between something that is incomplete and something that is in development. There are still some edges that need to be tweaked out with respect to RX Norm before Stuart and the National Library are to say that this is a production system.
Whether that is going to happen tomorrow or whether it is going to happen six months from now, that is something that the Library can answer and Stuart can answer. But I think that is what concerns CHI and the medication group that was looking at this, was that the words they were hearing was development, not completeness. If we can be assured on the subcommittee that we can say something about the development phase, whether they are out of it today, in which case it would not be a provisional recommendation, or they will be out of it in a few months, in which case we can say that this is provisional and we are assuming that the National Library has the resources, et cetera, to move into the production phase by the end of the year or whatever the case may be, that is the type of wording that I think we should work for.
Betsy, do you have any comment on that?
DR. HUMPHREYS: Yes. If you want to deal with it on that basis, then I would come back to you with some assessment, date estimate, of what I would give you. I can do that quickly, but I don't want to make it up right now.
DR. STEINDEL: Yes, that was the reason for CHI's provision, was the word development.
DR. COHN: Marjorie.
DR. GREENBERG: One option is to include it among the initial set, if the committee feels strongly that this is definitely going to be what you want. But instead of putting conditional or provisional, put in development, so it is clear that it isn't ready yet.
DR. STEINDEL: It is our recommendation that is conditional or provisional.
DR. GREENBERG: I know, but the thing is, if the recommendation is conditional or provisional, you could change it. This is what we think, but maybe not. If you feel more strongly that yes, this is going to be it, but all this work needs to be completed, then what you really need to acknowledge is that it is still in development. Then in the text, explain that we have these needs. Even so, it would meet these needs, it wouldn't be sufficient for all the needs in medication.
I think that is really important, because there might be people who would think, oh boy, I can go get this and it solves all my needs in medication.
But this is that the committee is really supportive that this should be ready for prime time. That has a different feeling, and conditional or provisional might make it sound like you could change your mind. I realize maybe you think people wouldn't want to touch it because it is still in development.
DR. HUFF: I like the spirit of that, but I don't want to say it is still in development, because I don't think we have any evidence that it is any worse for the purpose we are conjecturing than LOINC is or that SNOMED is. It is just earlier in its life cycle.
DR. GREENBERG: But it is. You don't have all sorts of people using it, you don't have --
DR. HUFF: But SNOMED is not in production systems for the purposes for which we have adopted it, either.
DR. GREENBERG: That is true.
DR. HUFF: If the bar is that it is already in production, the only one we could probably approve is LOINC. So in the spirit of what Clem was saying, I think if we said something --
DR. GREENBERG: We did hear at the last hearing about people who were using SNOMED.
DR. HUFF: Right, it is used.
DR. COHN: Or used more than you think.
DR. GREENBERG: But not in the full sense.
DR. ASHMON: I just want to go quickly on the record to say that there are in fact a wide number of systems who are using SNOMED in production systems, even for things like clinical decision support for drug therapy.
So I think it would be a real injustice to say that there aren't production systems using SNOMED today, and that should be reflected in the record.
DR. COHN: Some organizations including my own do use SNOMED.
DR. HUFF: But anyway, I think the important points are if we said this, if we said we approve RX Norm as a non-proprietary terminology for clinical drugs. Then we say, we recognize that we need other medication terminologies in the future.
I think that clearly states what our intent is. We say, we approve this not as a medication solution, but as the set that we approve for non-proprietary terms for clinical drugs. That limits its scope to say what purpose we intend it to be used for, and not describe it as developmental or provisional or preliminary or those other things.
DR. COHN: Steve, do you have a comment?
DR. STEINDEL: I think one comment is that as I understand it, while it is very close to being in production, it really is not in production. So while somebody can probably extract RX Norm from the UNLS right now, it is something that the Library may not endorse for use as a reliable source.
DR. HUMPHREYS: We think that what is there is reliable information. I think that we feel -- and something we are addressing right now is, we don't feel that we have the level of documentation that people would want for production use. That is a deficiency that I expect to be fixed in a few months.
DR. STEINDEL: I think that is what I was referring to. It is not the quality of the material that is in the UMLS at this time under RX Norm, but the system that is represented.
DR. SUJANSKY: But for the record, Betsy, it may be useful to express if you can what percentage of the clinical drugs using the RX Norm definition that exist, that you feel exist, even just in the prescription realm? Do you have a feeling for that?
DR. HUMPHREYS: As Stuart says, the problem is that we can't go and get the universe of those approved drugs. It is not easy to find that. But we believe in terms of prescription drugs that we are very close to general coverage of all the other drugs for prescriptions, not for over the counter.
DR. SUJANSKY: So over 90 percent, let's say.
DR. HUMPHREYS: We think so.
DR. SUJANSKY: I think we heard that from Stuart yesterday, but I think the real answer to your question is that it was incomplete and will always be incomplete, is true with every terminology.
DR. HUFF: But if it is 50 percent, that is a lot different than 95.
DR. SUJANSKY: But if that 50 percent could be solved in a month, it is irrelevant. But we heard yesterday it was closer to 95.
DR. COHN: I think we can spend the next four hours discussing this. Obviously we are going to be talking about it in October. I am happy to do whatever the will of the committee is, but I think we are grappling with wishes, hopes, desires.
DR. MC DONALD: Could I make a motion?
DR. COHN: Let me make a comment before you make a motion. I am hearing one, that the system is despite everyone's best wishes, not in production yet. So I think Marjorie is actually technically right, the system is still in development. Now, it is close, but I guess the question that the committee has to wrestle with is, are we recommending things as national standards without condition in development? I still come back to that provisional, that seems like the best thing, but once it gets in development, we will review it and present final recommendations.
Now, Clem, do you have a comment about that?
DR. MC DONALD: I was going to try to make a motion, unless you think we can come to --
DR. COHN: That's fine. I think if we can get the sense of where the committee wants to go, we can do something. We can actually defer this whole area until we have heard from the CHI, if we want that. That is not really my desire, but we certainly don't seem to be coming up with an easy solution.
DR. FITZMAURICE: I guess I am hearing somewhat of a comfort level, although maybe not the same comfort levels as for other issues, that making a direct recommendation of RX Norm as one of the core terminologies is acceptable, and accompanied by a strong support of resources and timeliness for the National Library of Medicine or the RX Norm project and with Food and Drug Administration, or the medical information system development.
I tend to think of the medical information system terminology as RX Norm, it is NDC, it is a lot of things coming down in columns that identify a particular drug, depending upon whether it is for patient safety purposes or ordering the drug or doing allergy checking. A lot of things are needed, but certainly RX Norm is needed.
I would be comfortable with making a direct recommendation for RX Norm, when accompanied by resources and timeliness devoted to these other things that need to be added for a medication information system.
DR. HUFF: Second. Is that a motion?
DR. GREENBERG: Mike can't make a motion.
DR. HUFF: You can make a motion to move Mike's --
DR. MC DONALD: I'd move your motion, Mike's wording.
DR. HUFF: I would second that motion.
DR. COHN: So let me just clarify, so this is without reference, the fact that it is in development.
DR. MC DONALD: Yes.
MR. BLAIR: Let me make a comment if I can.
DR. COHN: Sure.
MR. BLAIR: I am torn, because I really, really had hoped that we would be able to recommend this additional piece of the core set of terminologies. Yet, I have real respect for the concerns that have been voiced. Therefore, I don't feel like I could vote for it if it is included in the top level, but I don't want it to be out of the letter.
I could vote for it if it is in that area of additional study, saying that we feel like RX Norm plays an important role, there is no other terminology that is being developed that will play that role. Therefore, we support the additional resources to make that available as soon as possible so that we can give it our full endorsement, or something like that.
DR. MC DONALD: That is not the motion. The question is, would you accept -- maybe we could come up with an amendment that wold say we support it for use as a deliverable package which is expected in the next -- probably by the time we get it approved, but when NLM has it ready for production delivery. I think the production part is how do you ship it out and all that sort of stuff.
DR. COHN: I am going to even give another view, because I can't support your recommendation without an amendment. I can go a little further than Jeff can, which is to describe it as a provisional recommendation, recognizing that it is in development, that it appears to solve X problem. We recommend additional resources, and we will review it once it is in production, and hopefully at that point be able to give it final approval. That I would support.
MR. BLAIR: If that is a proposal, I can second that.
DR. COHN: It may be a friendly amendment, is what it may be, if you are willing.
DR. MC DONALD: Maybe we should get some words on the board. I don't think saying under development; I'd say rather put it as conditional on when it goes into production.
DR. COHN: I was just speaking of the differentiation between -- saying that the reason we can't fully recommend it now is that it is still in development, when it goes into production we will review it, and expect to be able to approve it at that point. The idea is that we will be reviewing this area, which it sounds like we will with CHI in October. This is not an area that we are going to forget and go away from.
DR. SUJANSKY: Let me try to summarize a little bit. Here what we are talking about is, Simon and the others, maybe besides Jeff, may be comfortable actually including RX Norm among the terminologies that are recommended, but clearly indicating that it is in development, and that we are not recommending it in its current state, but in its complete state.
Specifically -- and this is the part I am adding -- would it be helpful to maybe not right now, but before the recommendation is complete, to specifically identify those areas in which it needs to be augmented, as we tried to do here? That might satisfy Jeff.
DR. MC DONALD: No, but it is not the completeness, it is the scope. The scope of RX Norm is pretty darn complete right now.
DR. SUJANSKY: That is not what we heard from Randy Levin.
DR. MC DONALD: The scope of the problem space isn't complete. That is the part we keep jumping back. We are also going to support the FDA's efforts to allow the whole problem space. But the reason why some people are nervous about RX Norm is only because they haven't sent out disks yet. They don't have what it is defined as, and they are not shipping them out.
We will never get there. If we make everything has to be done at once, we will never get anything done. I am totally against trying to tie it to all the things we need in medication terminology. I think we should say we want these other things in the medication terminology, but as a parallel statement.
DR. SUJANSKY: How does Jeff feel about that?
MR. BLAIR: I am so torn. If we wind up putting it in the ones where we recommend, then I think we have to have more caveats there. If we put it in the to be studied, then I think we can have a simple sentence that just basically indicates the value of it, and that as soon as it is -- the development is completed, that we expect to be able to approve it. So it depends on where you put it.
DR. COHN: And actually, what I was recommending is that it have its own section called provisional recommendation. I was not including this as part of the core set of terminologies recommended. I thought there was another heading here that was before additional research which said provisional recommendation, RX Norm in development, necessary element, but not sufficient for the entire drug problem or whatever, and once it goes into production, we will revisit. The appears that the best solution for this will hopefully -- we will be provide to provide final recommendations within six months or whatever. That is what I was recommending.
DR. MC DONALD: I am hearing what you are saying, and I am guessing that the only thing that will get a consensus is to push in that direction.
I want to say that I am torn. I will draft up a thing that we can put on the board after lunch. The reason I am torn is, I am supposed to also check on something in my office, but I'll --
DR. COHN: Well, we are obviously moving ourselves well into lunch here. We actually need to look at the agenda in the afternoon. I guess I need to check with everybody. Given how easily this has gone so far, --
DR. SUJANSKY: We're about halfway through this letter.
DR. GREENBERG: I can tell you that the speakers for the afternoon sessions have all arrived.
DR. COHN: They have not?
DR. GREENBERG: They have all arrived.
DR. COHN: They have all arrived. I guess the question would be, after we discuss ICD-10, can we come back to this?
MR. BLAIR: Will we have a quorum? Will Betsy be here?
DR. COHN: Betsy is not part of a quorum.
MR. BLAIR: I understand, but Betsy will be here?
DR. COHN: Betsy has already left, so it doesn't matter.
DR. GREENBERG: She has left already.
DR. COHN: Are the committee members here the remainder of the day?
MR. BLAIR: Yes.
DR. COHN: Clem, you are here for the remainder of the day?
DR. MC DONALD: Yes.
DR. COHN: Okay, so I think we have a quorum, so that gives a little bit of time to work on this also.
We have worked our way well into lunch here. I know Betsy doesn't like the fact that we only take 40-minute lunch breaks, but --
DR. GREENBERG: She has left anyway.
DR. COHN: -- she has left anyway. So we will adjourn until 1 o'clock.
(The meeting recessed for lunch at 12:23 p.m., to reconvene at 1:00 p.m.)
DR. COHN: Everyone, we are sorry to be running a little late here. I'll call the meeting back to order.
Agenda Item: ICD-10 Impact Analysis
Status Report
The discussion obviously this afternoon is on ICD-10. We are delighted to have Martin Libicki coming back to us from the RAND Corporation to talk a little bit about the impact analysis, not so much about what all is in there, but more the process and observations at this point, to help guide the subcommittee.
So I want to thank you for coming. We will be having an open mike later on in the afternoon, as well as having discussions bout the impact analysis. We will also be having a report on the pilot report from AHA and AHIMA. That is this afternoon.
DR. LIBICKI: Good afternoon, Mr. Chairman and members of the subcommittee. My name is Martin Libicki. I am a senior policy researcher at the RAND Corporation, a nonprofit institution that seeks to improve public policy through research and analysis.
My presentation today will cover the work that the Science and Technology Institute of RAND did on the costs and benefits from switching from ICD-9-CM to ICD-10-CM and ICD-10-PCS. This work, it should be noted, has not been completely through the RAND quality assurance process.
RAND was asked three questions. First, what are the costs and the benefits from switching from ICD-9-CM's diagnostic codes to ICD-10-CM? Second, what are the costs and the benefits of switching from ICD-9-CM's procedure codes to ICD-10-PCS. Third, if it is advisable to switch to both ICD-10-CM and ICD-10-PCS, should they be done sequentially or simultaneously.
In answering these questions, we drew on three sources: Prior testimony before the subcommittee, a literature review, and interviews by phone, mostly by experts and stakeholders throughout the health care sector. From the middle of April until the beginning of August, we conducted 87 interviews in which we talked to 102 people.
Our presentation today will focus first on costs and then on benefits. We will not repeat the introductory and methodological material from last May, nor go into detail about the codes themselves, except to note briefly that their most salient feature from our perspective remains the greater specificity they offer, a mostly twofold increase in codes from ICD-10-CM, and a 50-fold increase from ICD-10-PCS.
Although neither ICD-10-CM nor ICD-10-PCS were designed to restrict one to many expansions of their predecessors, ICD-10-PCS by dint of its 50-fold increase is de facto a one to many expansion, while ICD-10-CM is only a one to many expansion for about two-thirds of its codes.
Now for costs. We estimate the costs of the transition to be $600 to $750 million for the first year, and perhaps as much as $20 million a year thereafter. These figures though do not include the cost to CMS of revamping their information systems. CMS has only begun to estimate such costs department by department. There are also some types of individually smaller costs, not separately estimated.
This number also assumes that the switch to ICD-10-CM and ICD-10-PCS occurs simultaneously. If considered separately, one can add up to another $200 million to that number, with costs more or less evenly split between ICD-10-CM and ICD-10-PCS. Almost all of the potential loss in productivity however is associated with ICD-10-PCS.
The cost of training people to use ICD-10-CM and ICD-10-PCS can be subdivided into the cost of training hospital coders, non-hospital coders, code users and others, notably physicians. The amount of training required for coders has been studied, and results from other countries serve as benchmarks.
Start with the roughly 50,000 hospital coders. The time it takes to get them to proficiency in new codes, mostly ICD-10-PCS, has been estimated at several days to a week by AHIMA among others. If such retraining costs $2,000 split between expenses and lost work time, then the training bill in this category is roughly $100 million.
One caveat though merits note. Gail Grant from the Veterans Administration informally tested ICD-10-PCS on coders, and concluded that while good coders essentially had no problems with the switch, others might have, because their knowledge of anatomy and physiology was incomplete. If the latter have to bolster their knowledge of such fundamental subjects, this would be an endeavor of weeks to months, and using a cost of $10,000 each and multiplying by the least proficient 20 percent of coders would add another $100 million.
However, that said, there is no evidence that such broad training was required in Canada, where an analogous transition was made in coding procedures, except for the code expansion and procedures being a factor of six rather than 50.
In Canada, all told, it took two days of intensive training plus additional follow-on sessions of a few hours each to get their coders up to speed. Initial losses in productivity disappeared largely or totally within six months. The training experience for Australia was similar.
What about teaching the code to new coders? In other words, is ICD-10-CM and PCS harder or easier to learn? Advocates of ICD-10-PCS say that its logical structure does make it easier to learn. Conversely, there are more individual codes from which to choose in both ICD-10-CM and ICD-10-PCS. But in either case, learning codes actually constitutes a small slice of what coders are in fact taught, and turnover in the profession is low. It has been estimated at under ten percent a year. Thus, we calculated that ICD-10-PCS and CM were not necessarily harder to learn than ICD-9.
What about those who code only ICD-10-CM, that is, those who deal largely with outpatient records? With 200,000 doctors' offices, one can think in terms of 200,000 coders. Yet, almost all the diagnoses in any one office usually involves only a small percentage of the total world of diagnostic codes. Some offices preprint the codes, either in hard copy or on disk.
Many physicians' offices do not record diagnoses with the highest degree of specificity in any case. It is not unknown for physicians, for instance, presented with a choice of 40 different diabetic codes in ICD-9-CM, to use the most general code in billing. Thus, little or no training will be asked for in many offices. Some outpatient coders will simply train themselves as they do, and there are potentially efficiencies in training that may be available by using the Web as an instruction vehicle.
More broadly, based on information from AHIMA, we used a figure of eight hours for the cost of training people who only used ICD-10-CM, and for whom coding is only a small percentage of what they do, yielding a cost figure of $100 million to train coders outside of hospitals.
As to the retraining of physicians, one person we talked to indicated the requirement for training would depend on how different ICD-10 was from ICD-9. At the high end, another estimated that each physician would need four hours of education in the new diagnostic codes, plus two further two-hour sessions in the first year for a total of eight hours. If multiplied by a half a million physicians in clinical practice, this sums to roughly 2,000 man years training, at a cost of several hundred million dollars. Nevertheless, no physician that we talked to said that they themselves would have to be retrained. Professional coders for their part are extremely wary of having doctors do their own coding.
Finally, as noted, many physicians' offices for better or worse rely on super-bills to capture their diagnosis for the record. If ten percent of all physicians take ten hours of training, the overall cost in this subcategory would be roughly $50 million.
Code users may also need some time to become familiar with the new codes. Perhaps half of the quarter million employees at peer organizations work directly with codes. It is unclear how much training they will get. Formal training in ICD-9 remains the exception today. To some extent, employers may count on people seeing codes and descriptions on claim forms often enough to link them together, and no payor has mentioned any formal training program to us, although few have yet to focus on the issue.
If the average employee needs the aforementioned eight hours of training, then roughly $50 million worth of training man years would be required.
There may also be some loss in productivity, that is, some extra work that people have to do. CMS formally tested ICD-10-PCS in 1998 by using two clinical data abstraction centers. Diane Kapro's report indicated, quote, no significant time difference in coding was found between ICD-9-CM and ICD-10-PCS. FMAS however found that ICD-10-PCS generated more codes than did ICD-9-CM, and that each record on average took 1.7 million minutes longer to code, or when multiplied by 1.8 million discharges a month with procedures associated with them, we are talking about 50,000 extra coding hours, or roughly three million dollars.
However, that is only initially. If one assumes a six-months learning curve, at the end of which two-thirds of the difference disappears and one-third of the difference persists, in the long run, productivity difference is roughly $12 million a year. But these data must be qualified by the observation that 200 records is a very small sample on which to base such conclusions. The true estimate may also be lower, to the extent that most hospitals use automatic encoders these days.
Another possible source of productivity loss can stem from the fact that these codes, requiring as they do more of the patient record, will require in some cases that more information be added to the patient record to fill out the codes. If one focuses on ICD-10-PCS and assumes one, that 80 percent of the patient records that either prescribe routine procedures or are already well documented, two, coders get back to provides half the time they need information and three, physicians need to spend five minutes for each such patient, then again, doing the multiplication, we are talking about 15,000 more physician work hours, or about two million dollars worth in the first month, and again, this figure would decline as physicians get used to what they have to note in the record whenever certain types of procedures are done. So we figure about eight million dollars a year thereafter.
Systems upgrades account for the other half of the costs of transition. Providers will have to adjust their billing and administrative systems to handle new codes, which in ICD-10 and PCS are longer and contain alphanumeric characters rather than just numbers, as in most of the case of ICD-9-CM. Most providers however buy their systems from vendors, and most of the provider- vendor relationships have a software maintenance component which pays for periodic upgrades, especially in response to government regulation.
Nevertheless, as Bob Hardesty from Cerner testified before the subcommittee, quote, it would be expected that there would be professional service fees for planning, testing, training and version implementation. Providers themselves may have to do some of their own planning, engineering and using end user services, and testing and insuring backwards compatibility with earlier data and systems integration.
Overall, the level of concern we found with such cost was modest, with neither AHA nor FAH raising such issues before the subcommittee, both of them being more worried about covering training costs. Major hospital chains that we talked to similarly were not terribly concerned about the change, believing that they themselves would need little additional consulting, and for them, the major cost would be testing and integration.
One hospital chain representing .5 percent of the total U.S. inpatient market estimated that the cost of testing and integration would be a maximum of a man year. Another said it would be close to a trivial event.
The Veterans Administration, which accounts for roughly three percent of the total patient population, does not plan to spend more than a million dollars. But there were also outliers, quote, easily a million dollars of work among hospitals. The outliers may account for a disproportionate share of costs. This makes estimates difficult to generate on the basis of small sample sizes; it is a statistical phenomenon. Thus, we used a $100 million transition cost as a plausible estimate in this case.
As for software vendors, at the high end, two of them we spoke to expressed the need for as much as three years to make the change. The man year estimates from two of them were roughly one man year for every ten employees on the payroll right now. A third major vendor cited an estimate of 50 to 100 man years for total adaptation to ICD-10-PCS and CM. Others such as Quadra-Med, Cerner, Soluthianth, Core Solutions, Symmetry and Engenics talked in terms of a few man years of effort. So we used the figure of $100 million for the sector, but it could be less than that.
Fixing payor systems, actually adjusting payor systems, is also a large cost element. Payors vary widely. Some, generally the larger one, develop and maintain their own systems, and their costs could be large, and they seem to fall either into a lower rate of 40 cents per member, or a higher rate of a dollar per member. Others, perhaps buffeted by earlier changes required by Y2K or HIPAA, have come to rely on vendors rather than run their own systems, and face fewer direct costs and perhaps fewer costs overall.
Based on those who own and run their own systems, we calculated the cost of adopting software to accommodate ICD-10 would run between $100 and $250 million if extrapolated to the entire payor base. We note however that at least one section of the payor business, fiscal intermediaries that handle claims for CMS, would in any way have to expand their code lengths that they accommodated because of HIPAA regulations that come into effect on 1 October.
Last is a word about how costs are distributed. As a rough estimate, hospitals are likely to bear 40 percent of the cost, outpatient facilities 25 percent of the costs, and payors, 35 percent of the cost. As noted, this does not include CMS. Within these categories, we found that as a rough rule of thumb, the likely cost that would accommodate a code change are proportional to the number of patients providers see or the number of members that payors have. In other words, geographically and through most other ways, they fall relatively evenly.
We also were not able to detect or predict any systematic shift in payments as a result of the change in coding between ICD-9-CM and ICD-10-CM and PCS. The current plans are that the DRGS will not be redefined until sometime after ICD-10-CM and PCS goes into effect. Insofar as CMS is a very large fraction in the total inpatient and outpatient sector, that probably has a knock-on effect for the rest of the business.
Now for the benefit side of the ledger, where we note that the discussion will be more qualitative in nature, and designed to note the areas where benefits may be expected.
We are currently engaged in developing scenarios to generate benefit numbers for each benefit category, but at this time, the costs and the benefits we talked about are not directly comparable.
Possible benefits could accrue in six categories. One, more accurate payments. Two, fewer erroneous, rejected or fraudulent reimbursement claims. Three, better understanding of health care. Four, improved disease management. Five, enhanced ability to evaluate provides. Six, more timely intervention for emerging diseases.
Category one, more accurate payments. As one researcher has argued, the predictive power and clinical utility of a DRG classification system is limited or empowered by the precision of the available diagnosis and procedure codes. This dictum applies with particular force to ICD-10-PCS. With more codes available, it would be possible to differentiate between simple and complex inpatient procedures that are currently lumped together. Different procedures therefore can be reimbursed differently.
This in theory is a net plus, if procedures had been optimal for patient care and were not performed because providers feared they would lose money in doing so. However, in practice this possibility was not raised by any provider and certainly not by any payor that we talked to.
In recent years, CMS has started a program to reimburse selected new procedures at rates higher than what is indicated by the DRGs that such procedures are associated with. The limitations on expanding ICD-9-CM, Volume 3, are -- not the only one, but a very major reason that such procedures have been approved. With ICD-10-PCS, the approval rate for additional payments for new procedures could rise to several dozen a year.
Category two, fewer erroneous, rejected and fraudulent reimbursement claims. ICD-10-PCS is more logical than its predecessor. Its hierarchical structure is more transparent, and ICD-10-CM is also clear. All these give grounds to believe that in the long run, there will be fewer erroneous, rejected and fraudulent reimbursement claims.
On the other hand, in the short run it will be a new code, and it does present far more codes from which to choose. The GAO report argued that by making more categories which for the sake of argument assume is done by adding more lower-order bits to the code, therefore, more errors will result. If these errors result from mis-coding only ont the lower-order bits that are added, one still gets more information from having such bits, if those bits are usually correct, as long as errors in those bits do not come as a surprise. Then one would have, if one had no lower-order bits at all.
The best guess on errors at this point is that error rates are likely to rise initially until people are familiar with the codes. Whether such rates eventually decline to below what they were before the switch is hard to predict at this stage.
As the work group for electronic data interchanged observed, ICD-10-CM and PCS incorporate much greater specificity in clinical detail, which may help reduce the number of cases where copies of the medical record need to be submitted for clarification for claims adjudication. By one estimate -- and this by the way, this send-back thing, is not a trivial issue. By one estimate, only one in five hospital claims make it through with complete payment and no additional questions asked.
Nevertheless, almost everyone who commented felt rejection rates would rise initially as people unfamiliar with the codes put down the wrong code, or as payors who have to learn new codes err and reject correct codes. At some point, the reduced rejection rate for more information is likely to exceed the increased rejection rate from residual errors.
As for fraud, the WEDI report goes on to argue, if a new coding scheme were implemented, there would be increased opportunity for fraud in the beginning, when people were less familiar with the new codes. It might be more difficult to detect potential duplicates, unbundled services or up-coding of procedures during the transition when two versions of code sets would be in effect.
In the longer term, it is possible that fraud could be reduced, since ICD-10-CM and ICD-10-PCS are more specific, and there are fewer gray areas for coding.
The bottom line is that although these reductions in error rates in rejected claims paperwork and fraudulent claims are all reasonable expectations, the reverse is likely to be initially true. People are apt to be somewhat uncomfortable with the new codes, particularly in the face of laws ranging from those dealing with false claims to those dealing with prompt payment, but such discomfort will pass, and after some crossover point, a net reduction in returned claims, in fraud and even errors are possible.
Category three, better understanding of health care. The tens of millions of annual hospital discharge records should hold enormous potential for assessing the relationship between health care and outcomes.
One prominent researcher observed that ICD-10-CM had the potential to say a good deal more about the quality of care in its codes, so that data could get used in a more meaningful way, enabling better understanding of complications, better designing of clinically robust algorithms, and better tracking of the outcome of care.
We observe that the ability to know, for instance, whether or when laparoscopic compared to open surgery improves health outcomes could affect thousands of lives and billions of dollars. More generally, one should never discount the possibility of a breakthrough in research.
Nevertheless, several factors could mitigate its potential benefit. For instance, studies that infer conclusions from hospital discharge statistics are limited inherently in their ability to control population samples or data quality. Furthermore, some of the gains from fine distinctions are vitiated by the fact that when you make finer distinctions, you get smaller population sizes within each category, and this in turn erodes the statistical significance.
Indeed, many researchers go the other way. They agglomerate information to super categories, not only to create statistically significant sample sizes, but to avoid the misleading effect of minor coding errors or coders unwilling or unable to make the fine distinctions that ICD-9-CM already allows.
Many software companies specialize in group software precisely to support such analysis. True, a finer, more scientifically precise set of distinctions can help form more precise groups, but only somewhat. The prevalence of coding errors and the likelihood that small errors will increase initially when ICD-10-CM and PCS is introduced also dilutes the value of such data research, at least the initial research. Errors, as long as they are randomly distributed, are unlikely to create spurious findings, but they are more likely to obscure findings that would otherwise be extracted.
Also, long-range studies will require some reliable means of translating between existing records in ICD-9 and future records in ICD-9 and 10-CM and PCS, a point of concern made by several researchers.
Last in this category, hospital discharge data sets alone do not say what happens to patients after they are discharged, how long they lived, how they fared, whether they re-entered the hospital, even the same hospital.
There are however two ways around the problem. One is that AHRQ maintains a database with ID numbers, and if one can get permission to access such data, which a few -- perhaps not enough -- researchers have, more followup information would be available. Payors and data consolidators that work with payors do have access not only to names, but inpatient records as well. Mining such data has already generated useful conclusions about procedures such as coronary artery bypass grafts. It is also possible to amalgamate the records of payors if payors believe that that is a good thing, and there are organizations working to persuade payors thusly.
On net, greater specificity would seem to be beneficial. Exactly what we will learn from greater specificity is hard to forecast with any confidence, but such uncertainty typifies all serious research.
Category four, disease management, a process by which care management organizations work with patients, especially those with chronic conditions such as diabetes, asthma and heart disease, to insure that they are receiving the correct quality of care, are given reminders of treatment and referrals to classes and other soft management techniques.
The point here is to postpone or eliminate the onset of the more serious conditions that send people to hospitals. True, some of these services could be performed by primary care physicians, and primary care physicians would rely on personal knowledge of or detailed notes on patients rather than on discharge codes. Yet, many patients lack primary care physicians or have no such single physician who can manage the patient.
Since much of the work of disease management -- finding eligible patients, sending them reminders, giving them information -- can be automated, organizations can run through patient records en masse and pick out patients who need disease management and deliver such management aromatically.
The potential benefits for disease management are large. Diabetes for instance kills 20,000 people a year. If thanks to better codes diabetics were better identified and their conditions more precisely managed, and if as a result the death rate from diabetes dropped as little as one percent, the benefits from that alone would more than pay for all the costs noted above.
Again, however, a tempered optimism is called for. As it is, many primary care physicians do not use the specificity already present in ICD-9-CM, and it is unclear whether they will react to more specific codes by using distinctions that currently lay unused.
Others produce diagnoses that indicate not the underlying cause of the encounter such as diabetes, but its proximate cause, such as kidney dysfunction. One respondent has suggested that some of the differentiation in diabetes, e.g., neuropathy, exceeds what are typically recorded in doctors' notes.
As a result, many care management organizations use other indicators to differentiate diabetics from one another. Some use hemoglobin-A1B tests and diagnostic codes for comorbidities. Perhaps half of all payors classify diagnoses by broad ETG, episode treatment groups, of which the most popular set has 500 to 1,000, and thus are more aggregated than ICD-9, let alone ICD-10.
The more sophisticated algorithms use a mix of diagnoses, procedures, lab tests and medications prescribed to determine what the patient may be suffering from, and how far the patient has progressed in the course or his or her disease, and thus whether disease management is warranted.
The finer details from ICD-10-CM and PCS cannot hurt the implementation of this algorithm, but responses to date do not suggest they will contribute that much more accuracy to ETG determination, either. One provider is moving away from codes altogether and going over to SNOMED.
Category five. By analyzing its own cases, a provider could determine what procedures it does well relative to its peers and what procedures it should find others to do. It can use such information to reallocate its resources and promote itself to referring physicians. Better codes for procedures particularly aid this process. Providers may want to also go back to discover using the results of a particular technique of device, and codes that include such detail make it easier to identify which records to pull in finding out this information.
Payors can also use discharge data to rate providers. One payor indicated it would make such ratings public via the Web, and it would begin to use such ratings to designate top-tier providers to whom patients would be steered. Again, however, caveats are in order.
Since many procedures are billed in both ICD and CPT, payors at least have access to the greater specificity already provided by CPT, which although not a perfect substitute for ICD-10-PCS, does fill in some of the gaps that currently exist in ICD-9-CM, Volume 3.
The fact also remains that providers are the ones who will supply diagnoses and procedures. In other words, we are asking them to write themselves. This has presented a lot of issues in the past. DRGs for instance have crept upwards from less serious and resource intensive categories to more serious and resource intensive ones. Both New York State and California have asked providers to indicate whether a particular patient condition, that is, a particular diagnosis, was contracted in the hospital or was present on admission. They have received inconsistently reported results.
In a study of diskectomy patients, Drs. Romano et al. concluded I think ICD-9-CM complications were under reported, especially at hospitals with low risk adjusted complication rates. The aforementioned coding errors don't help, either. Nevertheless, increases in specificity are still likely to be helpful in evaluating providers, especially in their ability to carry out new procedures, the very ones that currently get short shrift in ICD-9-CM, Volume 3's code book.
Category six. It is also theoretically possible that the more precise categorization of diseases could help the public health community detect emerging diseases, including bioterrorism, faster. If better diagnostic codes could improve the odds, even by a percentage point, of heading off the next SARS-like disease, the switch could pay for itself many times over through that benefit alone, inasmuch as SARS was said to have cost the Asian economy over $11 billion.
Similar benefits may come from being able to spot clusters of diseases that might be traced to environmental or occupational conditions. That noted however, ICD-10-CM was not designed for syndromic surveillance, and access to hospital discharge records would have to be accelerated in order to provide for timely intervention in the case of a fast-moving epidemic.
The last question RAND was asked: If switching to both ICD-10-CM and ICD-10-PCS is warranted, is it better to do both at once or one after the other? Many witnesses to the NCVHS and those we interviewed shuddered at the thought of trying to implement new codes, that is, ICD-10, before they had finished complying with HIPAA requirements, but no one who testified or who we interviewed suggested that it was more cost effective to first implement one code and then implement the other. Canada, which also implemented comparable changes in diagnoses and procedures, did them together. There are several reasons switching both at once makes sense. It is easier to get people, notably coders, mentally prepared for a mental change once than getting them prepared twice. It is also more efficient. Travel for training only has to be done once. It is easier to review a complex billing software system to look for where ICD-9-CM is used and has to be replaced once than to do it twice, first for Volumes 1 and 2, and then for Volume 3.
Simultaneous change also means testing once. Dealing with transition issues such as getting through errors of mis-coding or misinterpretation is best done all at once, rather than two periods separated by months or years. People can focus attention and resources on the process of making a smooth changeover and move on.
Finally, the long-awaited ERGs, dependent as they are on both diagnosis and procedures, would have to be further postponed if they had to first wait for a diagnosis to be changed and then wait for procedures to be changed.
The chart I have at the left has a number of cost estimates on them. The ones with asterisks are in our opinion the ones which would be increased in cost if we went to a sequential rather than a simultaneous change.
In conclusion, we have endeavored to estimate the cost and characterize the benefits of switching from ICD-9-CM to ICD-10-CM and PCS. But in concluding, we thought it would be helpful to introduce a broader perspective in the discussion, and the broader perspective is this.
At almost a trillion and a half dollars a year, the U.S. health care industry is enormous. It would not take a very large percentage shift in its efficacy or efficiency to make the kind of economic difference that could easily swamp the kind of cost estimates that we are talking about.
But if you look at the history of how people improve themselves, if you look at the history of continuous process improvement, every other industry has found that if it wants to get continuous process of improvement, if it wants to improve how it goes about doing its business, the first thing it has to do is, it has to measure itself better. It has to measure its inputs and it has to measure its outputs. Only after such measurements are made and validated can alternatives by systematically compared.
As an industry, the health care sector has been a laggard in the approach of measurement, leaving great room for improvements. Codes are not the only form of measurement, but codes are in integral form of measurement. As such, the use of systems, or at least the use of information compared in these codes, is central to the measurement process, and thus central to any program of continuous process improvement. Anything that improves such codes provided that they actually contribute to measuring what you want to measure, and they contribute to measuring it better, cannot help but have positive long term ramifications.
But all this is predicated on the idea that we know what we want to measure to improve outcomes, and how we want to measure them. Many people we had talked to had done a great deal of thinking on this score.
Patient safety advocates for instance really like the present on arrival indicated that we talked about in California and New York. People who measure patient quality of care are interested in whether providers comply with HEDIS standards. For instance, are ACE inhibitors or beta blockers given to heart attack patients within the first 30 minutes in an emergency room.
Others have noted that they would find indicators that differentiated diagnoses by severity to be quite useful. One researcher found that in tracking septicemia using ICD-9, there was only a code for urinary infections, NEC, which mixes the kidney, the bladder and the urethra, conflating a condition that is potentially fatal such as urosepsis and a condition that merely indicates the presence of a pathogen somewhere.
Also, there are many cases where severity can be indicated through laboratory test results, such as hemoglobin A1C for diabetes. Overall, there are good grounds for believing there will be benefits, and that these benefits are likely to exceed the initial break-in cost within a few years of when the cost is adopted.
What ICD-10-CM and PCS permits is a considerable expansion in what we can know about the health of the population and the care it receives. The extent to which this information can be properly and intelligently exploited is hard to gauge at this point, but the potential gain is large if the opportunities are taken.
DR. COHN: Thank you very much. Questions?
DR. FITZMAURICE: Does the potential gain exceed the potential cost?
DR. LIBICKI: I believe so. But I am going to say honestly that there is large room for error in the costs. I once told somebody that it was a factor of two, and they were surprised that I was even that confident. There is an even larger chance for error in the assumption of benefits.
When I said back in May that I would love to come up with a number of 1.47 or .78, but I wasn't going to do it, that's why. It is very difficult in general -- if I may start to talk a little bit off the topic on this thing, but I think it is relevant. RAND is an organization -- about 60 percent of its work is with the military, and RAND is an organization that believes in quantification.
A number of my colleagues have been spending a lot of time trying to figure out whether the transformation that Rumsfeld talks about, what kind of benefits you get. When you talk about military modelling, you have got a very defined set of objectives. You have got a very well-trained hierarchy that actually carries out what they are told to do.
You have a long history of cost-benefit analysis. You have accepted ability to do models and scenarios that are highly refined. And after all this, and these are very smart people, you have a very difficult time quantifying what transformation is going to do to them, except in some specific cases.
The health care sector does not have that kind of history of quantification of benefits. To ask the question, I am going to give you better information, what is the benefit of that, is extremely difficult, because you have to trace it through the decisions that people make. That is why we are going to use scenarios.
Now, these scenarios are only as good as the assumptions built into them, which will give you an order of magnitude. It will give any person who wants to further research a topic an initial sense of where to start looking. But I do not wish to pretend to a precision which I cannot confidently feel.
DR. FITZMAURICE: On achieving the benefits and the costs, what I gather is that the current health system isn't doing as well as it could with the coding system that it has, and with a new coding system it could do much better. But it sounds like there are incentives that are gong to be needed as well, to use these coding systems for this purposes for which you --
DR. LIBICKI: I believe that is a fair statement, yes.
DR. COHN: Clem.
DR. MC DONALD: Maybe other folks in the audience may know better. You mentioned ICD-10-CM. Is that a done deal? That is one question.
The second question is, ICD-10, I realize just recently, is not open to public use like ICD-9-CM is. I understand ICD-10-CM will be. ICD-10-CM is proprietary, class three in the UMLS vocabularies. So I would just like clarification about, we are losing the free use if we go to ICD-10 which we have in ICD-9. That is the first question.
The second thing was about the Canadians and the Australians. You said they had a new coding system, but they only increased their codes by sixfold, which seems digestible compared to 50-fold. I wondered, do we have that choice, to get something less rich.
When you talked about the fraud and abuse, I have a sibling and a lot of friends in medicine and private practice, and they feel the more complicated the game is, the better shot -- they think it is a game. This fraud thing is just a way to get the dollars down, manage the CMS levels, and this would just give them more opportunities to find traps, because it is a more complicated a labyrinth they are going to have to run through. So there will be obviously more fraud.
So I guess those are three separate ones. I can hold the other ones.
DR. LIBICKI: I think on the first question, I am going to have to defer.
DR. GREENBERG: ICD-10-CM will be in the public domain.
DR. MC DONALD: That is what we would use, not ICD-10?
DR. GREENBERG: Right, we would use ICD-10-CM. Actually, there is a copyright on ICD-10, if that is what you are referring to, although in this country, ICD-10 is only used for mortality, and it is really available to everybody who needs to use it for mortality.
DR. MC DONALD: Not free. I tried to get it, and I can't get it, ICD-10. It is copyrighted and you have to get a license.
DR. GREENBERG: I was referring to the people who need to use it to code all mortality data. You are talking about getting the system.
DR. MC DONALD: Yes. So the question is, are we going to lose -- now, CM you can pull off the web, so that is definitely there and it is free.
DR. GREENBERG: ICD-10-CM will be in the public domain in the same way that ICD-9-CM is.
DR. LIBICKI: The second question. We did not evaluate whether 18,000 is better than 200,000 codes. Having said that, I really doubt that all 20,000 codes are going to be used. There are just some categories for which you are just never going to get any patients. Only after we can use the code will we figure out how many will get used on an annual basis.
The third question had to do I believe with fraud and gaming the system and CMS. A lot of it is controlled by the fact that everything gets funneled through a DRG anyhow.
DR. MC DONALD: That is not true.
DR. LIBICKI: For inpatients. For outpatients I think it is different.
DR. MC DONALD: Well, there are a lot of complexities. You medical necessity ties to the diagnosis. There are just all kinds of interactions. What the physicians think about it is, the government is gaming them, not that they are gaming -- I can tell you stories, where these are completely things. USCD when they came in and did all the measurements, they found that the government -- they had undercharged Medicare by three million and had overcharged by two million, which suggests to me honest error, and they got three million plus fines, rather than -- not the difference. They should have got money back.
DR. LIBICKI: That is all a good point.
DR. MC DONALD: There was one other piece regarding the hemoglobin A1C in diabetes. Have you really looked at those codes?
DR. LIBICKI: Have I looked at the diabetic codes?
DR. MC DONALD: They are the ugliest on earth. There are a couple of other ones that are almost as ugly, in terms of rational sets of choices that humans should be forced to make in making the coding. So you have diabetes with complications, without compilations, with this complication, without the other complication. You have got to have another kind of code with it if you put in neuropathy with diabetes, or it doesn't even count.
It is really not for human use. Computers can do it. But the problem is, we want physicians to be online doing all this kind of stuff. It is a fraud not to go to the more specific code. They do for practical reasons go to the general one, because they know what that is saying. It is insulin dependent or not dependent. Those we can get to, those make sense.
Beyond that, the stuff looks nutty to a physician, some of us anyway. Is there any chance to get any of that fixed before we have to go to more specific nutty ones?
DR. PICKETT: With all due respect, while you seem to have termed it as nutty, there are a number of clinicians that we work with in the development of those codes who do not feel so.
DR. MC DONALD: We have nutty physicians, too.
DR. LIBICKI: I'm not a physician, but just in terms of the structure, it turns out that ICD-9-CM in the diabetes area cross-walks into ICD-10-CM very neatly. In fact, I was able to get it down to a half a page of narrative.
So at least not from a physician point of view, but simply from a categorization point of view, --
DR. MC DONALD: Let me make a plea. Why don't we say they have got severe diabetes and put the hemoglobin A1C, and say what the complications are, separate codes? Just get rid of these combination codes. They are nightmarish. They don't show up uniformly. I have asked this before. I don't expect you to change.
DR. COHN: Actually, I have a comment and a question. I had an opportunity recently to look at frequencies of the five percent sample from Medicare relatively recently. I was just astonished at the number of NRS or non-specific codes that were being used in diagnosis, as opposed to what we know are a large number of relatively specific codes.
I don't know if it necessarily confirms some of the comments you were making, but certainly without incentives, people tend to seek those common denominators, which I think is really what you were alluding to.
Now, the question I had had as I listened to what you were talking about, one of the key areas I am concerned about is the transition, transition costs, transition vulnerabilities. To my view, one of the issues that I have a good sense about is this issue of data in the transition, and I mean by that, mapping.
I have heard comments made by NCHS and others about mapping. I have to say, I have never seen it, and I have never seen any independent evaluation of the quality of the mappings between ICD-9 and ICD-10. I'm just curious; in your evaluation, did you have a chance to -- it looks like you have looked at the diabetic hierarchies. Did you have a chance to look at anything else? Were there any other reports or otherwise that allowed you to develop an opinion on this?
DR. LIBICKI: In ICD-10-PCS, the mapping is fairly clean because there is such an expansion of codes. In only 50 cases that I could find them cross-walked was there as many as two old codes to one new codes.
The diagnostic side of the mapping is more complicated, but part of the complication arises because we are losing information that we now believe is not very important. A good case is tuberculosis. If you look at the tuberculosis codes, you actually look like you are losing detail. You go from -- these are rough numbers -- 350 tuberculosis codes to 50 tuberculosis codes. But if you take a closer look at what is happening in ICD-9-CM, for every one ICD-10-CM code, there are seven ICD-9-CM codes which indicate how we found the tuberculosis. Right now, there is a standard way of finding tuberculosis, so we don't worry about that kind of measurement anymore.
So in essence, what looks like a one to many cross-hatch is not that at all. You have to look closely at the codes.
More generally, if you look at the codes and you say, we used to take this information, and now we don't think it is clinically meaningful anymore, then technically you have a cross-walk problem. But in practice, what you were doing was encoding irrelevant differences in the past.
Now, having said as much, it is not a clean cross-walk. But how dirty it is, whether you want to aggregate up, what information you are losing, that is a tough question. I hate to make broad, across-the-board judgments. I think it is tractable, because I think in the end when you do analysis, a lot of people group upwards anyhow.
DR. COHN: I guess I should ask the same question for Donna. I apologize, because we have had discussions and presentations about this over the years, but have there been other published evaluations or third-party evaluations of the mappings or cross-walks that exist?
DR. PICKETT: I'm not sure what you mean by a published evaluation. We had an earlier crosswalk that was posted during the open public comment period in '97-98. We then took that crosswalk down because of the amount of work that went into providing additional modifications to ICD-10-CM based on those public comments, and then subsequent work done by staff to incorporate some additional changes over time.
So we have not finalized the current crosswalk, but now that AHA and AHIMA have done a pilot and we are able to see what those results are and see how they will impact the classification, we then will be finalizing the crosswalk.
DR. COHN: So I guess the answer to my question is that the crosswalk hasn't been finalized, therefore it really isn't ready for independent review yet.
DR. PICKETT: Would you say that again? I can't hear.
DR. COHN: It sounds like the crosswalk hasn't been finalized. Therefore, it isn't really ready for independent review. Is that correct?
DR. PICKETT: I'm not sure what you mean by independent review.
DR. COHN: As a health care delivery system that has a large database, if I am going to take a data mapping and apply it to my data during transition, I would obviously want to have independent assurances that the mapping is of high quality and good fidelity, as well as describe where it is clean and where it isn't.
I recognize that I consider that to be a central part of the transition strategy. I was just trying to get a sense of where we were with that.
DR. PICKETT: No, in terms of that type of review, no, that has not been done yet. But we are also aware that having completed a crosswalk and making it available, those who were using it for purposes such as reimbursement would look at it very differently than a pure crosswalk.
By example, a new concept in ICD-10-CM that was not present in ICD-9-CM at all, we would not crosswalk that back to an ICD-9-CM code, because the concept did not exist. However, I have been told that for reimbursement purposes, where you are trying to do mappings for DRG purposes -- and I think there is another expert in the room that could probably speak to it better than I could -- you would have to have a mapping, because everything that is in the new code set has to go back to an old code in order for it to be applied for whatever other purpose. But for the purposes of a pure crosswalk, the tool that we would make available would not force a backward mapping if the concept did not exist in ICD-9-CM.
So there would be some differences in terms of how you would look at it. But if you were just looking at the diabetes codes in ICD-10-CM going back to diabetes codes in ICD-9-CM, that would be a little bit different than new concepts that were never included.
DR. COHN: Other questions or comments? I got off into a slightly technical area, but one that bears a lot on costs and risks of the transition. Questions around the table? Do we have any questions from the floor?
DR. MC DONALD: I was just going to say, I thought it was a thoughtful commentary. I appreciated the analysis.
DR. LIBICKI: Thank you.
DR. COHN: Yes. Let me ask a question about the analysis then. These numbers look a little small in some cases. You indicated that a few things could be off by a factor of two relatively easily. Are you going to be further refining these numbers, or are these the final numbers that you are standing behind at this point?
DR. LIBICKI: Let me just pull this back.
DR. COHN: I guess it is the old story that a million dollars doesn't buy what it used to.
DR. LIBICKI: No. The only place where I think the numbers are going to be low is in the cats and dogs, a lot of the miscellaneous stuff. But it is precisely because it is cats and dogs.
If I had to make an informal review, which of these numbers are high and which are low -- you may wish at some point to chime in -- I'm pretty happy with the inpatient numbers, because we are dealing with a population, all of whom are going to get trained, or pretty much all of whom are going to get trained.
The outpatient numbers, I assume 100 percent training, and it is not going to be 100 percent training. There are a lot of people who are basically clerks, and fill out these sheets and whatever.
In terms of the training time, it might be higher, it might not be higher. So that is my sense on that number.
On physicians, there is a huge variability. An issue has been raised, physicians have to do continuing medical education, do so many hours, and would it count if this was part of their hours or not.
In terms of users, I think the number is high, because most of these people don't get trained anyhow in ICD-9-CM, so I don't think they are going to get trained in ICD-10-CM, but we assume that they would.
When you get down to providers, that $100 million figure, the cost of fixing hospital systems, is very soft. It is very soft for the reasons I said, because if it turns out that most of the costs are borne by a small percentage of providers, it is very difficult to get accurate statistics by polling, because you may hit three -- statistically speaking, that becomes very difficult.
A lot of people -- you call a lot of people and ask them what they think, and some people are a lot worried and some people are not worried. You almost need to take a Rohrschach test before you ask them, because you end up, I found an optimist today, I found a pessimist today. That is the sort of thing here.
In terms of vendors, I think the number is a big high. Vendors as a whole thought they could do it in a fairly clean way. That is, the people who sell software.
In terms of payors, I am fairly comfortable with the payors, and here I have a range. I realize that somebody testified -- the highest number I got was from Jim Daley from Blue Cross of South Carolina. But I have reasons to believe that that estimate may be a special case for a lot of the things having to do with Blue Cross of South Carolina. Then I have gotten some very low numbers as well. For CMS I have no data whatsoever.
Then there are the odds and the ends and what you should include. There is one category of cost that I didn't call a cost, but a negative benefit. That is a lot of the break-in stuff on confusion. Taking a look over a ten-year period, I am not saying how much more fraud you are going to get the first year, but how much more fraud you are going to get over a ten-year period, and there are reasons to believe it will go up and then down.
Similarly for send-backs. In the beginning there is going to be confusion. In the beginning, more stuff is going to be sent back. Then that goes up and then it goes down.
So I take a look at the total. So some of what you may be thinking of as costs are in fact subtractions from benefits.
DR. COHN: Help me with this one a little bit. These are all a net average for a little over ten years.
DR. LIBICKI: Most of these costs are transition costs. That is, they take place within the first year.
DR. COHN: That is how I had taken your discussion of the ten-year --
DR. LIBICKI: And then there is a loss in productivity, which I estimate again very roughly at $20 million a year.
DR. COHN: You had commented, and perhaps Karen can also comment, on fiscal intermediaries.
DR. LIBICKI: Right.
DR. COHN: You had mentioned it as part of the HIPAA implementation in the middle of October, that financial intermediaries were being required to modify their systems to accommodate --
DR. LIBICKI: No, they will be required to modify their systems to accommodate a seven-character alpha numeric code.
DR. COHN: For diagnoses?
DR. LIBICKI: For diagnoses and procedures, irrespective of whether ICD-10 comes out, because this reg was made in anticipation that ICD-10 would. I think I'm right on that.
DR. COHN: I was actually looking to Karen.
DR. LIBICKI: Does that sound roughly right?
MS. TRUDEL: The only thing I can think of is that you are referring to the specifications that are already in the implementation guides for the transactions and code sets, which doesn't necessarily mean that any internal systems are being touched.
DR. LIBICKI: Okay, so we are talking about --
MS. TRUDEL: It is a more complicated issue than that there is going to be a complete change.
DR. LIBICKI: Okay, but that particular statistic did not go into these numbers. In other words, that particular fact did not go into these numbers. It was what we call a consideration.
DR. COHN: So I should then take the payor number as including the intermediary transition, or not?
DR. LIBICKI: We didn't separately estimate the FI.
DR. COHN: Is that under CMS EVD?
DR. LIBICKI: No, no, no, it is under payors.
DR. COHN: Oh, I see. So this was more of a, yes, they may have to do some of this stuff or include that in the cost.
DR. LIBICKI: That's right.
DR. COHN: Thank you.
DR. LIBICKI: In other words, it was sort of a bias factor. It might be lower than we think.
DR. MC DONALD: Did you entertain the costs involved in some of the systems that used ICD to compute or do other things? One of them is DRGs, which I'm sure you must have gotten into that. Someone is going to have to rebuild all those tables and redo the statistics and all or whatever.
But the medical necessity which we have been recently looking at very closely, we have got -- I think the file that is provided by our intermediary has got like 200,000 pairs of diagnoses and procedures in them, that have to have medical necessity validation, which really means in our computer system that when a physician puts in a general diagnosis, they have to open up and answer a question about three or four more specific ones, because they have to answer at a specific level, which really makes it more expensive for each test you order than it would have been without.
Now, if we multiply the procedures by tenfold or something like that, 50-fold, and we multiply the diagnoses made by double, we may have like a 20-million file record and we are going to have to look at 20 or 30 different diagnoses to do the thing. When you have ten tests, it could make it much more difficult.
DR. LIBICKI: It could be, but that is where a clean crosswalk comes in. That is to say, if you have the logic of the old crosswalk, and you can translate all the new codes into the old codes, then in protocol terms, you take the new stuff, translate it into the old stuff, process it as the old stuff and work the number out the back end.
As the DRGs change, however, then you are going to have to also fix the logic on that. But at least for the initial transition, that is where crosswalk comes in to reduce cost.
DR. MC DONALD: The thing that snags the process is these other side rules. This I think applies to physicians as well. If you are supposed to code the most specific term period. That is tied into the regs via something else. I don't know how often it is enforced, but that is the part you would like to decouple. Then you could find the level that is affordable and workable.
But if you do that, and then you are not in the most specific ICD-10, you are still screwed. It might be a pretty specific ICD-9-CM.
DR. PICKETT: Question. Dr. McDonald, when you are talking about the pairs of codes, the diagnosis and the procedure codes, are you talking about CPT codes, or are you talking about Volume 3, ICD-9-CM inpatient procedure codes? Because there would be a difference.
DR. MC DONALD: This is outpatient, so it is CPT.
MS. LEON-CHISEN: I believe what you are talking about are the local medical review policies that are developed by the FI, so they may be to different from state to state. Those have ICD-9-CM diagnosis codes, and then for the procedure, those would be CPT. DR. MC DONALD: That's correct.
MS. LEON-CHISEN: So it would not be ICD-9, Volume 3 codes. So it would not be affected by ICD-10-PCS. Only the diagnosis portion would be affected.
DR. MC DONALD: Okay. You listed outpatient here, and it had me a little bit off balance.
DR. COHN: Let me just follow up with a question.
DR. LIBICKI: Let me just add, LMRP is included --
DR. MC DONALD: LMRP?
DR. LIBICKI: Local medical review policies.
DR. COHN: I think you just made a reference -- I think going along with the mapping question that I had had; you were describing that as part of the transition strategy. I am presuming however that your costs include at least in system costs, actually going in and modifying the systems to enable them to fully take the full ICD-10, the PCS or CM code. As you were just describing, this is a technique to get in easily, that is correct?
DR. LIBICKI: Pretty much.
DR. COHN: You're not talking about enveloping an old claims systems using ICD-9 and just costing out the translation wrapper around it?
DR. LIBICKI: No, it is not going to be as simple as a translation wrapper, because if it were, these costs would be actually somewhat lower. But I would imagine that a lot of these people will at least try to use translation wrappers as much as they can early on, and then revisit the policy when the DRGs change.
DR. COHN: Yes, but what I am wondering is, do these costs include that work being done?
DR. LIBICKI: Roughly, yes, because those are the two specific questions we asked to payors. There are two things we are interested in. We are interested in the change in the logic and we are interested in also in the change in being able to handle a larger character set.
DR. COHN: Okay.
DR. LIBICKI: For at least South Carolina, it looked like the ratio was one logic to two. That is, two-thirds of the costs was on the character sets and one third was on the logic, which strikes me as high.
DR. COHN: Other questions from the subcommittee on this? We will obviously be hearing more as we move on. I know in September there will be an occasion for us to talk more.
DR. FITZMAURICE: When do we expect the final report to be in our hands?
DR. GREENBERG: Hopefully if we move on the schedule we are on now, everyone would have the final, final report a week before the full committee meeting. But the subcommittee members and staff would see it earlier than that obviously, so we have a lot of work to do to get to that. That is our target.
DR. COHN: I think we will hopefully try to reflect towards the end of the day on the process of getting from here to there after we have had a chance to listen and hear more and talk a little more.
What we are going to do now is take a break for 15 minutes. When we come back, we'll open up a period if people have comments or questions from the floor, an open mike period, and then we are going to have Nelly and Sue to talk about the AHA and the AHIMA pilot test, and then after that we will talk a little more about what we see as the next steps to get us from here to there.
Now, we will probably have a chance to reflect either today or tomorrow on our discussions this morning on PMRI. So let's take a 15-minute break and come back.
(Brief recess.)
Agenda Item: ICD-10 Impact Analysis
Open Microphones
DR. COHN: I apologize, we are running a little bit late. We had Martin talk some about the impact analysis. There going to be multiple other times over the next couple of months for the public to make comments about the analysis the NCVHS will be recommending to the Secretary, but we wanted to at least allow some time now, basically an open microphone session for anyone in the public who wants to make some comments, either related to the impact analysis or things related to ICD-10 that they feel the subcommittee should know.
So with that, we'll open the microphone if anybody wants to make any comment. Make sure you introduce yourself for the transcript.
DR. FRASER: Hi. My name is Chris Fraser, and I represent the Academy of Professional Coders. I was really torn if I should talk today or in concluding remarks tomorrow, but I did paraphrase Dr. Libicki in my comments, so I thought I would make my statement today.
It is a four-page statement, I've condensed it, so it will only run about three minutes.
Dear Chairman, members of the National Committee on Vital and Health Statistics, Subcommittee on Standards and Security, and audience. I appreciate the opportunity to discuss the transition to ICD-10-CM in relation to its impact on the coding community. As a representative of the American Academy of Professional Coders, I can speak to our support of an expanded and enhanced code set for reporting morbidity statistics, and I also can speak to the desire of the AAPC to be involved in the process of education and training, if such a transition takes place.
Before I continue on these two points, however, I would like to preface my comments with an invitation to the Center for Medicaid and Medicare Services extended to the AAPC last year at the time of the AAPC national convention. In part, CMS acknowledged the contribution of coders to health care, and encouraged the AAPC participation in issues common to both entities. This statement today represents our willingness to be a partner in the transition, and we graciously offer our assistance in the development and delivery of educational materials to the enhancement of the coding community.
The AAPC believes that a partnership can result in developing an approach that will help to ease any uphill battles anticipated in terms of convincing the health care community, specifically coders, of its importance.
For this statement, I used the white paper entitled, The Role of Technical Standards in Today's Society in the Future, written by Dr. Libicki, and taken from the book he co-authored, Scaffolding the New Web: Standards and Standards Policy for the Digital Economy. Although the paper dates from September 2000, his points about standards in the digital economy lend themselves to the NCVHS subcommittee discussions.
My statement is in two sections, why the AAPC supports enhanced standards for coding health care information, and why it takes partnering for standards to succeed in the transition. For the sake of time, I am jumping to the second point which, if you have the paper, begins on page three, and that also I will summarize.
Dr. Libicki writes, if and as standards take hold, they can be expanded outward. The prospects of success may be gauged by the record of successful standards that started small and grew, rather than those that were born complex. Yet, as a practical matter, the ability of industry to develop coherent semantic standards depends on the health of the standards process.
A healthy process depends on the following path in relation to the success of the digital standards which I have applied to the code set standards.
First, let the market decide. Dr. Libicki states that a standard may also be judged by the fairness of the process in which it was developed, and the process of which it is disseminated to the users. In time, momentum develops for a preferred tag set. Once this momentum is recognized, competing alternatives are discarded and consolidation proceeds rapidly. Everyone ends up speaking the same language.
The AAPC believes this can be applied to momentum in diverse methods in delivering the information education to coders. Consolidation proceeds rapidly, as long as diverse members of the delivery system work together toward the common goal to speak the same language.
For example, the AAPC, which represents 30,000 coders, claims expertise in physician coding and could act as a vital link between the delivery system model and the physician coder.
Two. Have different communities each decide. Dr. Libicki writes that the boundary between sectors has never been easy to delineate, and is not always meaningful. Because every standard rests on its own business model, multiple business models makes business process systems unwieldy to create, operate and maintain. The AAPC believes if the spirit of coding is universality, the education of the delivery system must be broad in its spectrum and knowledgeable about its audience in order to reach the wide span of its practitioners.
Benefits of partnership include these four points. Partners interested in achieving the same goal, cut overall coordination costs by letting everyone know where they sit in the rapidly evolving universe. Partnership provides a neutral meeting ground for various efforts, and could also document the current status and plans of the various groups to look for possible gaps, overlaps and contradictions.
Three. Partnership presumes a public interest and timely standards that neither split the user community into disparate camps nor biases the marketplace in anyone's favor.
Four. Partners can become an important part of the voluntary regulatory process, since they can work together in delivering the same message to a diverse audience.
In summary, Dr. Libicki asks if the ecology of standardization is as healthy as it should be. Can standard guides still rise above competition without ascending the sterile heights of irrelevant perfectionism? To what extent can open transparency in the creation of a standard substitute for more legal definitions of fair and democratic?
The AAPC believes standards that prevail in the market necessarily reflect market forces, the desires of consumers, the strategies of players and yes, the role of public policy. Second, as Dr. Libicki writes, it helps to understand how standards work if deleterious effects of such policies are to be minimized.
The AAPC agrees. We know how the standards of coding work for coders, and we know how important it is for standards to reflect market forces and the desires of consumers in order to prosper. The AAPC is an organization that represents the physician coder, and through our 16 years in the business, understands the needs of the coding community.
In conclusion, the AAPC offers its help as partners in enhancing the ability to standardize and advance information technology among the coding community.
Thank you for your time.
DR. COHN: Are there comments? Yes, Michael.
DR. FITZMAURICE: Could I just ask a question? Are you for moving from ICD-9 to ICD-10 for diagnosis and for moving from ICD-9-CM to ICD-10-PCS for in-hospital procedure coding?
DR. FRASER: We support what you decide.
DR. FITZMAURICE: Thank you. Are you married?
DR. FRASER: We are here as representatives. We want to get involved. We want to be a part of the process. Any other questions?
DR. COHN: Any other questions?
DR. FRASER: Thank you.
DR. COHN: I think Michael may have asked his last question for the day here.
DR. TENET: Robert Tenet with the Medical Group Management Association. I just want to start off by saying we share Dr. Cohn's concern that these numbers may be looking a little bit low.
But my question is one about cost shifting. When you look at these numbers that are in categories, and it looks wonderful, and it looks like everybody is going to be pitching in to pay for all of this, but in reality, when you look at for example the vendors that are going to be spending all this money, that money is going to be cost shifted to the purchasers, meaning the physician practices that need to purchase the software to upgrade their systems.
Similarly with the payors, you spoke to apparently a lot of payors around the country. Maybe I can ask the question, how many said that they were prepared to absorb all of the costs and not pass those costs on through increased prices of premiums.
So I think it is important to realize that though we can categorize the costs, eventually the costs come down to a point where they can't go any further, and that is usually the physician practices.
So I ask two things. One is, if you are admitting that these cost estimates are variable up to two or four times, it might be valuable to include those ranges in the report to give a more accurate presentation of the potential costs. Then also talk about two other things, one, cost shifting but two, I think Dr. Fitzmaurice asked the tough question -- the first one, not the second one -- which was, do the benefits outweigh the costs.
Then you've got to ask yourself, by category, if these cost shifts occur to the providers, what is the incentive for them to move forward with these coding changes, what are the benefits to the providers.
We all know about the many benefits for health care in general. If the benefits are not apparent to providers, is there some method that we can use to incentivize that community to move forward with any changes.
DR. LIBICKI: Well, let me see if I can respond a bit. The notion of putting down in the conclusion some general confidence in the quality of the estimate is not a bad idea, one. Two, the report makes the assumption that the costs are going to be shifted away from vendors. In fact, in the end, consumers are going to pay all the costs one way or the other.
But the reason it is an assumption and not a finding is because cost shifting, particularly when you are talking about the shifting of a small cost in a large sector, is one of the slipperiest fields in economics. It is just to make a blanket assumption that ultimately, the consumer pays for all this sort of stuff, and leave it at that.
In terms of incentivizing the community, there are two incentives that we are talking about. First is the incentive to actually use the codes. I am less worried about that, because you have the force of law behind you. When you talk about the incentive to use the codes well, however, that is a different issue altogether. But that is actually outside the cost.
If you are a payor and a provider, you have to learn the new codes because of the force of law. You have to change your computer systems because of the force of law. But you don't have nearly as great an obligation to fill out the codes with the care and the thoughtfulness that they may deserve. For that we may have to think in terms of incentives, but I didn't go there in the report.
MS. LEON-CHISEN: Number one, I'd like to commend Dr. Libicki on doing a very tough job. I know that we have been asking ourselves and a number of other organizations through this committee how much will it cost, and it has been very, very hard to quantify those numbers.
Having said that, I'd like to see some sort of calculation of the cost, meaning if we consider how long we have been working with ICD-9-CM, 20-plus years, when I was a kid in school we thought ICD-9-CM was going to be around for nine years, ten years, and we were going to have another version.
If you would take a look at how much would it cost to implement and then spread it out over, I would say a minimum ten years, 20 years, maybe Donna can say when we are going to have ICD-11, perhaps it gives us a better idea in terms of what is this investment really mean, and what does it mean to invest it now and to get benefits.
I know the benefits part was also very hard to quantify, but some of those things will be around for at least ten years. Perhaps the negative side, what would be the cost of not moving to ICD-10? What happens if we stay with an outmoded coding system that is not able to represent the conditions and services that are being provided today?
DR. LIBICKI: Let me tell you how I'm handling it. First of all, I'm using a ten-year base, for precisely the reason you talk about; there is an average of ten years between codes.
All the benefits are calculated relative to the baseline, which is that we don't make the implementation. That is a technique that we use. In other words, when we for instance talk about the fact -- and I mentioned it in testimony -- that ICD-10-PCS allows you to put in more special procedures, the assumption was that the current system doesn't allow you to do it because it runs out of codes.
So yes, good suggestions.
DR. MC DONALD: My last time, because I think I am swimming way upstream. I don't think that there is huge value to the whole system in doing these changes. I just want to get that on the record. I think most of the discussion has to do with little nitsy-gritsy things, has to do with billing and other sorts. If you really want to get detailed on procedures and patients, you look at CPT codes, which go to HCFA all the same anyway. There is the detail. That is number one.
Number two is, I think we should be looking at how this should never happen again. Why are we making a code system that has to be changed every ten years, ever, anyway? Why aren't we using SNOMED for coding the diagnoses, and be done with it? Because then we don't have to keep changing in cycles every ten years.
I think we are doing some dumb things here, because of pressures of different kinds, different forces on different parts of the medical industrial community, and it is going to come out this way. But I think we shouldn't be having to do this. It has to do with the fact that these code systems don't satisfy the things we are requiring when we look at other code systems, in terms of not having had the structure built into the codes. They run out of space, and all these other silly things.
I think we are still doing it. Even ICD-10-PCS, with all of its big numbers, it still has limited ranges for some of the spaces. So I'd just go on record. I think it is not a great idea. I think we should do a better job, and if we ever do it again, by God, let's not do it the same way. Let's make it once, so we are out of that game of those funny structures.
DR. COHN: Martin, would you like to comment?
DR. LIBICKI: Actually, I'm going to pass on that one.
DR. COHN: Okay. Marjorie?
DR. GREENBERG: I just would like to mention, and reflecting on yesterday's discussion, there was a fair amount of recognition -- and this would just apply to ICD-10-CM, which is an international classification -- but that FDA and other parts of the government as well as other countries have certain obligations, whether they be formal treaties or other agreements, to use certain international classifications. That does influence decisions that have to be made about what classifications are going to be used.
We have a similar agreement -- this is FDA, has agreements related to the global medical device nomenclature, and some related to MEDRA, which weren't actually on the table yesterday, but we do have similar agreements, the U.S., as well as other countries, to use the WHO -- the latest version of the international classification of diseases.
I just think that should be recognized, since it was recognized yesterday about other classifications.
DR. COHN: Michael and then Karen.
DR. FITZMAURICE: I think there is a lot of wisdom to some of what Clem said, which is assumed. But I want to ask Martin a question, sparked by what Clem said. He said, why do we change these systems every ten years, why have this discontinuity? If we know things need to be put into it, changes need to be made, why can't we make change continuously over time, and not every ten years.
Martin, is there some body of thinking that says you want to make continuous changes in a coding system because it is cheaper in the long run to make those changes, periodically updating, as opposed to, save them all up and make them all at one point in time?
DR. LIBICKI: In fact, we do both.
DR. FITZMAURICE: I know we do both.
DR. LIBICKI: That is, we are updating 9-CM while we are doing 10.
I almost hesitate to get -- in fact, I will hesitate, and just say that is a good comment. But it opens up a lot of interesting issues having to do with how you encode information and what information you want to include, and what information is built into the codes themselves.
We don't update the words we use every ten years. What was hypoglycemia ten years ago is going to be hypoglycemia now. We may add a couple more adjectives to it, but we don't update it in the same way.
Let me just stop there, before we get into some serious trouble.
DR. COHN: Karen?
DR. GREENBERG: Against my better judgment, I do feel obligated to respond somewhat to what Clem said.
First of all, 10 was developed with the idea that the old system of strictly a numeric code was limited, from the point of view of expansion. This is a different issue of meaningless codes versus codes with meaning in them. But in any event, 10 is an alphanumeric code in order to provide more opportunity for expansion, and incremental expansion as opposed to wholesale replacement.
DR. MC DONALD: 10-CM?
DR. GREENBERG: No, 10, which 10-CM is based on. So that is why World Health Organization and the member states all adopted 10. One of the reasons was of course that it was more up to date than 9, and they had a long history of these ten-year or in this case, it was a 15-year revision. But it was designed in order to have a longer shelf life, by going to alphanumeric.
Secondly, they adopted a process by which 10 could be modified incrementally, just as we had been modifying 9-CM incrementally. So whereas in the past, we were never able to make incremental changes to ICD-7 or 8 or 9 within that ten or 15-year period, now that ability does exist. But it only exists in 10, it does not exist in 9. It did exist in 9-CM, but the international standard, it only exists in 10.
We have begun to use that process. Last week, Donna was reviewing something like 250 or 300 possible recommendations at an international level. So that is one point that I want to make. I think that going from 9 to 10 was a recognition of having a more robust classification, and that would have a longer shelf life.
The other thing is that I really do feel strongly that it is not a question of terminology versus classification or clinical terminology versus classification.
I think this subcommittee and the full committee have gone on record that there is a continuum, and that in the code terminology continuum -- that is the terminology that was used; I know we had some discussion about that at the last meeting, as to whether the word terminology applies to the whole continuum.
But that is what it says in the PMRI report, and also is the way it is viewed by the World Health Organization and the collaborating centers. You have very detailed clinical terminology, and then you have more aggregated ones, which are classifications. There are different applications for each.
The goal is that if you can collect in systems that collect information at that more granular level, such as being recommended for electronic health records and in some cases being implemented, you would then be able to use that data to map to the classifications that you use for the purpose the classifications are needed, which is statistical applications and reimbursement applications and reporting to state health departments and all of that, hospital discharge data, what have you.
Then if that is the case, even if new classifications come in in the future, as long as you have a good rule-based map between the detailed clinical terminology and the new classification, it is almost invisible to you. So one way for you to implement 10-CM when we do that kind of map, is to implement SNOMED. It may not be completely that easy, but clearly that is our goal.
But I think to not recognize that there are purposes for which classifications serve is short-sighted, and to want to stay with a classification, given that there are purposes of classifications that serve, do you want to stay with one that is 25 years old, has some limitations?
I think if you want to implement 10-CM or 10-PCS, for that matter, through an agreed-upon, certified mapping that hopefully we will come up with from SNOMED to those classifications, fine. That is one way to do it. Then if there is another version that is adopted, you could do that again.
DR. COHN: Let me just go on the record. Some of us may resemble that remark.
DR. GREENBERG: He gets the last word.
DR. MC DONALD: I don't disagree with classification being important, that is for sure. My complaint really is the code space that is making us do this. I think PCS definitely is going to run out of code space even as long as it is, because it has got two characters for things that are going to get bigger than that.
So we are already putting ourselves in a box, and I think we should recognize that contradiction.
The second thing is that you really can't map these different kinds of hierarchies because it gets very, very complicated. You look at these diabetes rules that aren't just climbing a hierarchy; you've got to look right and left and make other kinds of inferences. So it is hard. With PCS, it is more granular than any coding system we have now.
DR. GREENBERG: I thought the vision of this committee was to collect information once, and then you use it for multiple purposes.
DR. MC DONALD: Yes.
DR. GREENBERG: There is still going to be a need for classification, so you are going to have to figure out a way to do it.
DR. LESH: Hi, Kathy Lesh, the Kevver Company, although I am speaking for myself, not for the Kevver Company, on this.
I agree with Dr. McDonald, in where we are going with this. It is my understanding that the ICD was developed and is being used internationally for reporting of morbidity and mortality statistics, not for billing purposes. I think that is where we run into a lot of problems using ICD in its billing purposes. I know that is where this committee has said that for HIPAA, that is where we are. I also think that it is for lack of anything better.
Now, I haven't attempted to use SNOMED-CT as a billing -- nor was it developed for that. But if we are going to be using the clinical data, there would be a reasonable crosswalk, if you want to call it that, into the billing data, without having to dip into ICD-9 or 10, as the case may be. But we could still use a cross-map between SNOMED and ICD for doing our international duty of reporting the morbidity and mortality statistics.
DR. COHN: Other comments? Why don't we move now to a discussion of the pilot study? Nelly Leon-Chisen and Sue Prophet-Bowman are going to co-present. Then we will talk a little bit about the calendar for moving forward on this and other committee issues.
Agenda Item: ICD-10-CM Pilot Study
DR. PROPHET-BOWMAN: Good afternoon. I am Sue Prophet-Bowman from the American Health Information and Management Association. I have with me Nelly Leon-Chisen from the American Hospital Association.
Over the summer of 2003, over the past month or so, the American Health Information and Management Association or AHIMA and the American Hospital Association, AHA, conducted field testing of ICD-10-CM with the primary purpose of this testing being to assess the functionality and utility of applying ICD-10-CM codes to actual medical records in a variety of health care settings, and to assess the level of education and training required by professional credentialed coders to implement ICD-10-CM.
There was no external source of funding for this project. It was conducted using the existing resources of both associations with the generous support of Ohio State University Health Informatics and Information Systems staff.
Participants in this project received no remuneration for their participation. It is important to note that this project focused exclusively on ICD-10-CM, not on ICD-10-PCS at all.
AHA and AHIMA solicitated health information management professionals from across the country to serve as volunteers via announcements in our various publications and e-mail communication. A total of 187 individuals actively participated in the testing. These are individual health information management professionals, as opposed to organization. Some health care organizations had more than one individual participating in the project.
These participants represented a cross-section of all geographic regions of the country, with almost every state, including Alaska and Hawaii having representatives in the sample.
Some volunteers were able to participate during their work hours, whereas others actually took the time to participate in this project on their own time, but with using their organization's consent to use their facility's medical records.
Ongoing communication with the participants and access to the resources necessary for participation in the project were provided through a virtual, what we call community of practice accessible via the AHIMA website. The resources provided included training materials, coding guidelines, and a link for submitting the survey form, which was a web-based process that I will talk about in a little bit.
In addition, the community of practice or COP, as we call it within AHIMA, allowed communication between the participants as well as between the participants and the project coordinators. Participants could pose questions pertaining to the training materials, coding guidelines, data submission process, or other aspects of the project on the community of practice. Project coordinators posted responses to those questions about the project within 24 hours. Only participants in the project and the project coordinators had access to the community of practice.
Staff of the National Center for Health Statistics provided official answers to questions related to the training materials or coding guidelines. Any issues pertaining to the use of ICD-10-CM or the coding guidelines or areas where more training would be needed were specifically identified as part of the data collection.
All participants were required to listen to a two-hour archived audio seminar accessible via the Internet. This audio seminar included a slide presentation. Presenters were National Center for Health Statistics staff members. All participants were also provided a copy of the ICD-10-CM coding guidelines, which they were required to review prior to starting the project. It is important to note again that the only training these participants had were two hours of a non-interactive audio seminar training.
A panel of professors and researchers at the doctoral level from several academic institutions reviewed and advised on the research methodology, including the instructions on the record sample selection, process for coding records, and the process for submitting data. Participants were required to use the ICD-10-CM version, made available by NCHS in June, 2003.
Since user friendly electronic tools for using ICD-10-CM are not yet available unfortunately, printed copies of the index and the tabular parts of the coding system had to be provided to all participants for use in the project.
The primary limitation of the study had to do with the lack of availability of user friendly tools. The ICD-10-CM alphabetic index is the means by which diagnostic terms are located and the appropriate codes or code categories are identified. Locating a diagnostic term in the index is the first step in the process of determining the applicable code.
Unfortunately, the only tool available at this time for using the index was extremely unwieldy, cumbersome and difficult to use. Currently, a variety of hard copy and electronic tools are available to facilitate the process of looking up terms in the ICD-9-CM index. For example, the index in ICD-9-CM code books uses standard column formats and headings, font styles and indentations with standard tab settings to enable people to easily find a term. Electronic ICD-9-CM coding products use search engines for locating terms in the index. However, the only available tool for using the ICD-10-CM index for this project was a hard copy printout with confusing indentations, infrequent main headings, and lack of font style changes or other characteristics that would facilitate the ability to locate a term.
It is important to recognize that this problem was unrelated to the ICD-10-CM structure itself, but rather related to the available navigation tools and the format of the page layout, and this issue will be resolved when ICD-10-CM is implemented as a variety of user friendly, easily navigable tools for locating entries in the index will be available in both electronic and paper products at that time. However, we have to acknowledge that the lack of user friendly tools did have a detrimental effect on the project, particularly in terms of time comparisons, coding records in ICD-9-CM and ICD-10-CM, and the ability to easily find diagnostic terms in the index, less measurable but equally significant as the impact an inadequate index tool likely had on the participants' overall perception of ICD-10-CM.
Although we assured the participants on numerous occasions during the course of the project that much better tools for using the coding system would be available after implementation, their frustration with the available tools was expressed in a number of comments provided in response to a followup questionnaire after the project. It was difficult for the participants to separate problems with the tools for using the coding system from the coding system itself.
Participants were encouraged to code 50 medical records apiece if possible, and to randomly select their record sample from the discharges or visits of any month or months from 2003. The sample was to be representative of the variety of diagnoses treated by the facility whose records were being coded.
Records were to be selected without regard to payor. For instance, if the organization provided both inpatient and outpatient services, then both inpatient and outpatient records were included in the sample. Multiple project participants from the same organization were allowed to divide the responsibility for coding the different types of records among their participants.
If the same individual was coding both inpatient and outpatient records as part of the project, half of them were randomly selected from a specific month's inpatient discharges and half from the same month's outpatient encounters.
Participants whose facilities have a low volume of discharges or visits per month were allowed to select records for more than one month. Participants were also instructed to coordinate the record selection process with other participants from the same facility to insure that more than one participant did not accidentally select the same record.
They were instructed not to cherry pick easy or difficult records or to select a particular diagnostic category. Only if the facility treated a limited number of diagnoses, say a psychiatric facility, was it acceptable for the record sample to be limited to those types of cases.
The data collection period was from June 30, 2003 through August 5, 2003. Only records of discharged patients were eligible for inclusion in the study, meaning no patients who were currently hospitalized. All records had to be complete, which was defined as the availability of all ancillary test results and clinical documentation, including the discharge summary.
Both ICD-9-CM and ICD-10-CM diagnosis codes were assigned for each medical record. If the record being coded for the project had previously been coded in ICD-9-CM, the participant was required to disregard the original ICD-9-CM codes, preferably have no access to them at all, and re-code the record in ICD-9-CM.
ICD-9-CM codes were to be assigned in accordance with the ICD-9-CM official guidelines for coding and reporting, and ICD-10-CM codes were to be assigned in accordance with the official guidelines for their reporting. No procedure codes were assigned as part of this project.
The entire medical record was to be reviewed when assigning both ICD-9-CM and ICD-10-CM codes. Participants were instructed to assign codes as completely and accurately as possible according to the existing medical record documentation. They were instructed not to query physicians for clarification of documentation regarding either ICD-9-CM or ICD-10-CM codes. The rationale for this was to limit the variables in the coding process as much as possible in order to maximize the comparability of the data.
The data elements to be collected were determined by AHIMA and AHA staff in conjunction with consultation with researchers at several academic institutions. Data was submitted via a web-based survey tool developed by the Ohio State University School of Medicine. The survey tool was housed on the Ohio State University OSU server.
Participants were required to complete three separate surveys. The demographic survey was completed once by each participant and collected data about the participants' background and the type of organization where they were employed. The record survey was completed once for each record coded, and collected data on the process of coding the record in ICD-9-CM and ICD-10-CM. The followup survey was completed once by each participant at the conclusion of the project, and asked for participants' general impressions of ICD-10-CM and opinions related to implementation.
To validate a sample of the accuracy of the ICD-10-CM coding, a sample of participants, which was every other participant in our participant database, was selected to submit additional data on every fifth coded record. The additional data included the specific diagnoses documented in the medical record and the ICD-10-CM codes that the participant assigned. A total of 361 validation forms were submitted, which represents 5.8 percent of the total number of records coded for the project.
AHA and AHIMA professional coding staff recorded the validation forms in ICD-10-CM without approach of the codes assigned by the original participants. The codes assigned by the participant and the validator will be compared and the discrepancies will be reconciled, but this review has not been completed yet.
Unfortunately as we indicated earlier, we need to know that the unwieldy tool for using the index did have a negative impact somewhat on the result of the project, particularly in the area of the time required to code a record and ability to locate a term in the index.
Regarding how the results were tabulated, OSU health informatics and statistical staff cleaned the data, tabulated the results, and reported the results to AHA and AHIMA.
Regarding the demographic profile of the participants, all of the participants had a general health information management or coding credential, and most, or 85 percent, had one of the two general health information management credentials of a registered health information administrator and registered health information technician, one of those two. Fifteen percent had a specialized coding credential only, and did not have one of the general HIM credentials.
Over half, 57.8 percent, of the participants worked in a short-term acute care hospital. It doesn't mean they necessarily only coded records from a short-term acute care hospital. We also collected data on the specific type of records they coded, which we will go over in a few minutes.
The next-largest employment setting was integrated delivery system at 17.6 percent. Other settings represented by our participant sample included consultants or vendors, 5.4 percent, physician office practices, 3.2 percent, nursing homes, 2.7 percent, educational institutions, 2.4 percent, government agencies or contractors, 1.95 percent, other ambulatory types of settings, meaning other than physician office practice, 1.6 percent, rehabilitation facilities, 1.6 percent, long term care hospitals, .5 percent, home health or hospice, .5 percent, behavioral health facilities, .5 percent, and then other at 2.9 percent.
The average bed size for the short-term acute care hospitals employing our participants was 365 beds, but the bed size ranged from less than 50 beds at 13 percent to greater than 400 beds at 42.6 percent.
Now I will let Nelly address the results from coding the records.
MS. LEON-CHISEN: Due to resource constraints, we were not able to collect the specific code numbers assigned on each record. The collection of actual code numbers would have necessitated a more complex survey design and data analysis process, and thus would have been more resource intensive and expensive than the methodology we adopted.
Instead, we chose to collect data on the number of codes assigned. This allowed us to evaluate whether ICD-10-CM would result in fewer or greater codes assigned per record, as well as how well each of the ICD-10-CM captures was represented in the record sample.
A total of 6,191 medical records were coded as part of this project. On average, 33 records were coded per participant. Almost half, 42.3 percent, of the records represented short-term acute care hospital inpatient records. Another third, 38.8 percent, represented short-term acute care hospital outpatient records. Almost ten percent, or 7.85 percent, were records from post-acute care settings, which encompassed home health or hospice, nursing homes, long term care hospitals, and rehab units or facilities.
Interestingly, six percent of the coded records were from physician practices. A small percentage of records from clinics, community health centers, freestanding ambulatory surgery centers and freestanding diagnostic facilities were also included in the records sample, but it is still significant, as it constituted a 177 records.
Ninety-six records representing both inpatient and outpatient encounters in a behavior health facility were coded. Only 34 records were reported in the other setting category.
Participants had experienced coding medical records in a variety of settings with 32 percent having experience in hospital inpatient coding, 30.6 percent having experience in coding hospital outpatient records, and 11 percent having experience in physician practice coding.
The number of ICD-9-CM codes assigned across the entire records sample totaled 19,009. This number may have actually been higher, as the multiple choice responses to the survey question were limited to five codes. For example, if a record had more than five codes, the respondent was forced to select the choice of, quote-unquote, five or more. A total of 18,794 ICD-10-CM codes were reported as being assigned across the entire record sample.
The responses to this question were also limited to five or more, so the actual total may have been higher. However, the responses to the question regarding the total number of ICD-10-CM codes assigned across all system chapters only indicated a total number of 17,450 codes.
It is not clear why there was a discrepancy in the responses to these two survey questions. Some participants experienced difficulty in using the drop-down boxes for reporting the number of ICD-10-CM codes per chapter, so it is possible that some codes were inadvertently omitted in the response to the survey question.
Chapters of ICD-10-CM are primarily organized by body system, with other chapters devoted to infectious parasitic diseases, symptoms, signs and abnormal clinical findings, injury, wasting and other consequences, external causes of morbidity and factors influencing health services.
According to the reported number of ICD-10-CM codes assigned per chapter, all chapters were represented in the project. The highest number of reported codes represented diseases of the circulatory system, followed by factors influencing health services, and then endocrine, nutritional and metabolic diseases. It is important to note that the reported number of codes did not necessarily represent unique codes, but rather, the number of times respondents reported a code in one of these chapters. So evaluation, if multiple participants assigned the same code or a single participant assigned the same code and multiple records, the code would show up in the respective chapter multiple times.
The length of time it took to code each record in ICD-9-CM and ICD-10-CM was collected. This time included review of the record to identify diagnoses to be coded, as well as the process of locating the diagnosis in the index and verifying the codes in the tabular list.
Time as measured by noting the time when review of the record began and the time when all relevant diagnosis codes had been assigned. Record review time was included in both ICD-9-CM and ICD-10-CM coding times. Data submission time was not included.
The reasons for the length of time were also captured, but those records where there was more than a five-minute difference between the ICD-9-CM and ICD-10-CM coding times. However, further data analysis is needed in order to account for differences in types of records and the impact of the cumbersome tools for using the coding system.
Data on the number of errors or conflicts in instructions, index entries and tabular notes were collected, as well as the number of instances when an appropriate ICD-10-CM code could not be assigned.
For each record, conflicting instruction entry or diagnosis that could not be coded, participants were required to submit a problem identification form. These forms identified the specific ICD-9-CM code or codes assigned, and if possible the ICD-10-CM codes assigned, as well as a narrative explanation of the problem encountered. Possible problems necessitating the completion of a problem identification form included the inability to find an appropriate code, difficulty locating a code in the index, or insufficient medical record documentation to assign a more specific code. The results from the problem identification forms required manual tabulation in web-based survey forms.
Although this data is still in the process of being analyzed, a cursory review of the problem identification forms indicated the difficulty in locating a diagnostic term in the index appeared to be the most common problem. In-depth review needs to be done in order to determine how many of these problems were due to the tool use or the index, as opposed to actual problems with entries in the index that need to be resolved prior to implementation.
Data were also collected on the number of non-specific ICD-9-CM and ICD-10-CM codes assigned. For each non-specific ICD-10-CM code reported, participants were required to submit a problem identification form describing the ICD-9-CM and ICD-10-CM codes assigned and the reasons that a more specific code could not be assigned. Data regarding errors or conflicts in the ICD-10-CM coding system and the use of non-specific codes are still in the process of being analyzed.
With regards to the followup survey, at the conclusion of the coding phase of the project, participants were asked to complete a followup survey questionnaire; 158 participants completed this survey.
Interestingly, 60.8 percent of the respondents indicated being at least somewhat familiar with ICD-10-CM prior to the start of the project. Participants may have gained some prior familiarity through exposure to the differences between ICD-9-CM and ICD-10-CM provided by seminars and other products, as well as the review of a previous draft of ICD-10-CM that was available on the National Center for Health Statistics website.
Regarding the amount of ICD-10-CM training they thought an experienced coder would need, 75.6 percent suggested three days or less would be adequate. The options they were given were very little, half or less, at least a day, two days, three days, and four days or more. Responses to these options varied from 4.4 percent for very little, or 31 percent for at least a day. Keep in mind that these participants had only had a two-hour training via the web.
The clinical descriptions of the ICD-10-CM codes were thought to be better than ICD-9-CM by 70.9 percent of the respondents. ICD-9-CM code descriptions were thought to be better by 12 percent of the respondents, however, 11.4 percent were unsure.
In spite of the problematic index tools, 63.9 percent of the respondents still thought the notes, instructions and guidelines in ICD-10-CM were clear and comprehensive, whereas 22.8 percent thought they weren't, and 13.3 percent were unsure.
Data analysis of the submitted problem identification forms is expected to provide insights into the notes, instructions or guidelines that participants did not think were clear and comprehensive, and may need to be improved prior to system implementation. This information will be provided to the National Center for Health Statistics in order that enhancements may be made to ICD-10-CM prior to implementation.
Of note is the fact that 76.6 percent of the participants felt ICD-10-CM was an improvement over ICD-9-CM. And 82.9 percent of the respondents supported migration to ICD-10-CM after gaining experience with this system during the course of this project.
Respondents were asked for detailed comments, but they indicated they did not support migration to ICD-10-CM or were unsure. The most significant comments were not supporting migration to ICD-10-CM, but that the tool for use in the index was too difficult and that it needed an overhaul, a concern over the cost of change and a concern over the availability of resources to assist coders. The most comment related to the need for a more user friendly tool, being an index that would be more user friendly.
Although the problematic tool for using the index won't be a factor once ICD-10-CM is implemented, and we asked participants not to consider this issue when selecting their survey responses, it still did have an unfortunate impact on the testing results, and some participants' perceptions of the system.
Not all comments supported a no response. For example, one person who responded that he or she did not support migration to ICD-10-CM commented only that he/she liked the alphanumeric format for body systems.
The most significant comments for being unsure about supporting migration to ICD-10-CM were problems with the tool for using the index and poor physician documentation that would prevent reaping the benefits from the greater specificity of ICD-10-CM.
Other concerns were the cost of implementation, shortage of coders, and that systems will need to change.
Those participants who indicated support for migration to ICD-10-CM were asked how soon implementation should occur. They were provided choices of one year, two years or three years or other. ICD-10-CM should be implemented in three years or less, according to 78.1 percent of the respondents. The 21 percent of the respondents who selected other were asked to provide a narrative explanation.
It was apparent that some respondents felt the need to implement, quote-unquote, soon, but were unsure as to the definitive time frame. The most significant related to the need to first improve the tools for using the coding system, and then implementation should take place as soon as vendors and payors can accommodate the change. Several individuals noted ASAP, with an exclamation point. Another individual indicated, could not answer yes, but feel it is in the best interest of our profession to get on with this as soon as possible.
Our next steps include reviewing the validation sample and identifying mismatches and code assignments between the project participant and the validator. Discrepancies will be reconciled and through review of the affected code or codes and relevant index entries, instructions and guidelines. Reasons for the discrepancies will be determined whenever possible.
Data from the problem identification forms, including those that indicate insufficient medical record documentation to support a more specific code, will be compiled by type of issue and relevant ICD-10-CM codes and analyzed. Identified problems will be verified in order to insure that they are not just a reflection of the difficulties encountered with the tool used to locate index entry. A compilation of the identified problems will be submitted to the National Center for Health Statistics so that they can use this information to make any appropriate modifications to ICD-10-CM prior to implementation. Once the remainder of the data analysis has been completed, a formal report will be issued. I know you have all been waiting for this; you would like to know what our conclusions are.
In conclusion, coding professionals who have participated in the ICD-10-CM field testing generally supported adoption, and felt that the system is an improvement over ICD-9-CM. This coding system can be applied to medical records in a variety of health care settings, as participants did not report more difficulty using ICD-10-CM to code records in one setting versus another setting. In fact, some of the individual comments in non-hospital settings indicated that they thought it was much more applicable to these settings than ICD-9-CM. Although they received minimal training, and most felt that additional training would be advantageous prior to its implementation, preliminary results tentatively indicated participants were able to appropriately assign ICD-10-CM codes in most cases. Further review of the validation sample and problem identification forms is needed to confirm whether this is true.
Many of the negative comments by those participants who indicated they did not support migration to ICD-10-CM or were unsure were related to their frustration with the tool they had to use to locate entries in the index. We recognize that this tool was extremely difficult to use, and unquestionably affected the results of this project, but fortunately this problem is easily rectified.
Any questions?
DR. COHN: Thank you both very much for a very interesting report. Thank you for coming and presenting it to us. I do want to congratulate you. It looks as I read this, it looked like the study itself was finished a little over two weeks ago?
MS. LEON-CHISEN: Do you really want to know how many hours Sue and I have spent together?
DR. COHN: I can imagine. Sometimes deadlines like this do make things happen, so if we played any role in helping getting this thing done, I don't know if we take credit for it, but we do want to thank you.
Actually, I have a question that others may have also. As I was looking through this -- and it may just be that I didn't understand the slide about training needed for experienced coders. I understood this was a self-selected group, and therefore one would expect that these are among the best and brightest, not exactly as Marty would describe, the bottom 20 percent or anything like this. Yet, of that group, it seemed to me that a substantial group of them felt that they needed four days or more for training just on ICD-10-CM.
Is that a correct reading of that slide? You gave averages and all of this stuff, but it looked to me like 22 or 23 percent of that group felt that they needed relatively extensive training.
DR. PROPHET-BOWMAN: Yes, their answers to the individual responses were all over the map, which is why we compiled it into how many of them felt that it was within the three days or less response, which is the number that we provided. Clearly, the majority had at least answered one day, two days or three days.
Now, which one of those choices they answered was somewhat variable.
DR. COHN: I guess I'm looking at the wrong slide. I'm looking at the slide called training needed for an experienced coder. Am I looking at the wrong slide?
DR. MC DONALD: I think something like 80 percent of them said more than a day.
DR. COHN: I'm not surprised at more than a day. What I was surprised at was that the best and the brightest -- and I remember, and Martin, you're going to have to help me with this one, in your study what you projected out as the amount of training the coders would need generally for the ICD-9-CM.
DR. LIBICKI: (Comments off mike.)
DR. COHN: I guess what I am wondering is, if you are talking about one day of training, you've got 23 percent of your best coders asking for more than four days of training. They were just talking about CM rather than talking about PCS. There is obviously some different expectations here about how quickly people will get up to high proficiency.
I thought this was very interesting. I'm not misreading that, am I?
DR. PROPHET-BOWMAN: No. It is true that 23 percent selected four days or more. However, the same exact -- actually, 23.4 percent, however 23.4 percent also selected two days, and 31 percent selected a day and 17.7 percent selected three days. So some of it may have been -- this is obviously based on their opinion and some of it of course may be a reflection that this group had extremely minimal training as part of this project, two hours, so where between two hours and four days they thought they could be.
DR. COHN: I'm going to say that I don't even consider a week of training to be too little for something if it is going to be your main tool on which you derive your livelihood from. So I'm not sure that it is the numbers we are talking about. I was just surprised that -- once again, these are the brightest coders I would expect that you would have.
DR. PROPHET-BOWMAN: And the question was specific to, what do you think an experienced coder would need, not a new coder to the profession would need. So it was specific to what they thought --
DR. COHN: An experienced coder would need, okay. So I think that is an important piece of information from your study that I saw anyway, so I want to thank you.
Other questions?
DR. MC DONALD: I would just suggest that this is firmer data than you had to make your estimate. It looks to me like you should double that if you are going to be doing ICD-10-PCS as well. At least you are talking a week or so for all coders. That tells me that would be the safer projection.
DR. COHN: Yes.
DR. LIBICKI: I did use a week for the ICD-10-CM for the inpatient disorders. Let me ask you a question. We are talking about people who pretty much do coding full time, correct?
DR. PROPHET-BOWMAN: We're not sure.
MS. LEON-CHISEN: We're not sure, because we didn't collect job titles.
DR. LIBICKI: How would that compare to somebody who was a coder in say a two-person doctor's office?
DR. PROPHET-BOWMAN: While clearly these were all credentialed health information management professionals, these were people who all had formal ICD-9-CM coding training. Now, some of the people we know for sure were educators, where they teach coding, so they don't actually assign codes in a practice every day, but they teach coding or they could be a coding supervisor, where they normally review coding as opposed to actually have to do the coding. But clearly, these were not the receptionist in a doctor's office that learned coding on the job or anything like that. They were educated people who understood ICD-9-CM.
DR. GREENBERG: I'm a little confused here. For the hospital inpatient coders, how much training did you estimate for them?
DR. LIBICKI: 50,000 people, and I used one week.
DR. GREENBERG: You did use one week, not one day?
DR. LIBICKI: That's right. For the people who don't do coding full time, I used one day. For the people who work in offices that only deal with a small -- if I work for an ophthalmologist I'm going to be coding ophthalmology diagnoses, I'm not going to be coding diabetic diagnostic codes.
DR. COHN: Michael and then Jeff.
DR. FITZMAURICE: I applaud the speed with which you did it and the participation you got from your members. That is really quite amazing. I wonder if there is more to the study that might come up later. Did they use stopwatches or did they time themselves using code one way versus another way? Is there something more that might be gained in the future from this study?
DR. PROPHET-BOWMAN: We will be doing a full report, which will have more information related to the time. They were told to time it with a watch and to include the review of the record time in both figures. In other words, the time it takes to determine what the diagnoses you are going to be coding are as an element of the coding process, whatever time that took. Obviously they would only have to do it once for coding in both systems, but they should add that time to both figures, is what they were told. But we are not finished analyzing the time results yet, but that will be in the final report.
Another area that we looked at that I know actually came up earlier today was the issue -- one of the concerns that we have been asked to look at was the issue of, does the documentation support specificity in the coding system. One thing we asked them to do, which they should really be commended for this, because it was a huge time commitment on their part, but every time they assigned a code that was assigned in ICD-10-CM as a non-specific code, meaning it had unspecified or not otherwise specified in the code title, they were asked to submit detailed information about that, and why they could not assign a more specific code.
Well, of course, that happens a lot. And they were asked to report what kind of record it was. Our hypothesis was that you are going to find more detailed medical record documentation in the hospital inpatient record than perhaps a physician's inpatient office record.
As you can imagine, that is thousands of pieces of information which we have not been able to analyze yet, since this study just finished August 5. But in those cases, they were asked to report both the ICD-9-CM we assigned and the ICD-10-CM code we assigned, so we could see if there is a pattern of certain code categories, where the specificity just isn't there.
DR. FITZMAURICE: Given the speed with which you did this study, some people might expect that since you are up tonight, that you might report those results tomorrow.
DR. COHN: I think Jeff and then Clem.
MR. BLAIR: This is an area that I don't have a lot of familiarity with. You pointed out the fact that this study was compromised, possibly severely compromised, by the fact that the tool wasn't available for a fair comparison.
I don't know the order of the magnitude of that impact. I don't know whether, if the tool was available, does that mean that the coders would have been able to do this 50 percent faster, twice as fast, four times as fast, ten times as fast? Because tools can make a huge difference.
Could you at least give me some sense in your minds -- and I know that this is not quantitative and deliberate -- but just to help me get a sense of the order of magnitude, if there was a tool available, what kind of impact in your judgment do you think it would have?
DR. PROPHET-BOWMAN: The Ohio State University people are actually looking at that to try to give us some kind of quantified estimate of what the impact had. Clearly, the issue of time is the question that was most severely impacted by the tool that they had to use for locating a term in the index.
I would guess, and Nelly can correct me if it is wrong, but I would guess that they could have coded the records at least twice as fast if they had had the current level of either manual code book tools or electronic tools that they currently use for coding with ICD-9-CM.
MS. LEON-CHISEN: I would state that even twice as fast would probably be a conservative -- if anything, because it was really the number-one complaint that we got. A lot of them realized that they spent a lot of time trying to locate terms, and they blamed that on the index and the formatting. It is not like anything they were used to seeing in a book. They realized that it was a draft form, that once you have books published -- just like we do today, publishers take care of making it nice and aligned properly and your eyes get used to looking for gold lettering and all of that.
So it is hard to judge, but I would say twice as fast.
MR. BLAIR: I have a lot of sympathy, because I used to see 20/20 before I was a coder.
DR. COHN: Clem.
DR. MC DONALD: I want to commend you. This is really a very nice study. In contrast to the previous study, you've got this sample size. You've got 6,000 records, you've got about 200 coders in this study, and they are scattered all over the place, so you've got good sample size.
There are some other statistics you can get into when you have nesting, but I don't think that is so terribly important. But what would also be interesting is what code space is being used for most coding. My guess is it is the 80-20 rule or something like that, so you probably don't have that much difference in the coding systems in the big heavy part of the front of the tail, because you are probably going to dealing with the same number of codes in that space.
But then the difference is, how much more specificity can you get? If you needed more help analyzing this, we probably don't mind helping you, but --
MS. LEON-CHISEN: Only if you promise to be objective.
DR. MC DONALD: I am also guessing that I think you are kind of stuck with what the speed would be. I would guess realistically, probably the speeding time for these codes would be the same when you had comparable coding systems. I'm guessing at least twice as slow, because that is why you are thinking it would be twice as fast, because the problem space is about the same.
I think though that there may be more learning curve on this thing in terms of the time to get up to speed than a week or so, because people do memorize these things.
MS. LEON-CHISEN: Interestingly -- and these were comments that were being posted in the community of practice, so they were not part of the survey, and also in some of the informal conversations I had with some of the people validating or that participated in the project. Some of the coders who had not worked that much with ICD-10 before actually said they were kind of scared of it before, but now having seen it, worked with it, tried to apply it to records, they are really glad they did. They thanked us for the opportunity. They said that it wasn't as bad as they had originally thought.
Everybody is afraid of change, so -- actually, I have some of these quotes. Some of them were pretty funny, because I think a lot of -- it is natural; they were afraid of change. They had been coding ICD-9 for so many years, and they were pleasantly surprised that it wasn't that difficult.
DR. PROPHET-BOWMAN: And even though they indicated in some cases the level of training that they needed and some of the frustration finding a term because of this index tool, however, when we did this cursory review of some of the problem forms that were submitted and some of the issues that were provided when they were provided the code number, clearly they in most places got to the right place, found the right code, seemed to know what they were doing, and they had only had two hours of training.
So it wasn't like they were throwing up their hands and saying, oh gosh, I have no idea what I'm doing, I haven't been trained on this, I don't know how to find a term, I don't know how to code. That was not really the result that was coming through. I think it was more lack of confidence, because they didn't know if the code they were assigning was correct.
DR. MC DONALD: Did you use the same code in all the coders?
DR. PROPHET-BOWMAN: They used their own medical records available at their own facility.
DR. MC DONALD: Did they code both the same way? Because then you could check them with a crosswalk, too.
DR. PROPHET-BOWMAN: Well, because of the process we were using to facilitate the complexity of this and make it as simple as possible, both for collecting the data and analyzing the data, we did not collect specific code numbers for every case that was submitted. They just had to collect a drop-down box of number of codes in a particular chapter, so we could get a sense of what chapters were done, and only in those cases where they had problems, they couldn't find a term in the index or they couldn't figure out what code to use, or they had assigned a non-specific code, only in those cases did they report the actual ICD-10 code number.
We have a validation sample in order to assess whether these coders were even in the ball park, that we are currently checking to compare their code with our validator's code in that sample, which is a five percent sample.
DR. HUFF: Do you think they retained the codes?
DR. PROPHET-BOWMAN: We asked them to assign an internal tracking number to the records, which they had to submit then on the survey form, in case we ever had to go back to them and find out any information about those records. But whether they collected the code numbers on those log forms, or they only collected, this is Mary Smith's record and they would have to actually go pull some other form to see what they had submitted, that I'm not sure.
But we did have a mechanism in place that they could go back to the records that were included in the sample if they needed more information.
MS. LEON-CHISEN: It is possible that they did keep that information. I am a coder, so I can speak to that. We are very detail oriented people. We also like to keep a lot of information. So I know if it was me, I probably would have kept them for the next ten years.
DR. PROPHET-BOWMAN: These are HIM professionals who of course are used to record retention issues. So we got calls about, how long should I retain these records of these codes. We told them a few months.
DR. COHN: It sounds like you could get sidebar consultation from either of these members.
DR. MC DONALD: Do you have the energy to do this big study on the PCS? That is probably where you will see more differences, or the advantages or disadvantages would be larger.
DR. PROPHET-BOWMAN: Actually, one thing that amazed us was how committed this group was. For one thing, we actually had, because of the time constraints we were trying to work under, some other delays in the project. We ended up doing this in the summer, which wasn't the ideal time. So we lost a few volunteers who had originally volunteered just because July is the best time for them to be committing a lot of extra time.
But it was amazing, the number of people -- and we still had those other people who volunteered but who had to drop out for one reason or another, we had actually several hundred who originally volunteered, and the ones that were able to stick with the study, it was really amazing, their level of commitment. We were getting survey forms posted at two and three a.m., before the deadline.
MS. LEON-CHISEN: Some of them were complaining that the servers were down in the evening.
DR. PROPHET-BOWMAN: It was probably doing maintenance or something on their server, because they were complaining at three a.m., they couldn't get into the server. The week before, some of them said they worked all day Saturday and Sunday for the two weeks before the deadline, and this is in July. They were spending every waking minute. One person says she dreams in ICD-10 now.
MS. LEON-CHISEN: But it is not that hard to learn.
DR. COHN: Let me just ask, Dr. McDonald brought up a question, are you intending to do some more study looking at PCS?
MS. LEON-CHISEN: We hadn't talked about it. We hadn't planned on it. Actually, the reason why we did this study was that it came up through this subcommittee many, many times about a pilot testing.
DR. PROPHET-BOWMAN: Because none had been done.
MS. LEON-CHISEN: Because none had been done, and basically that is why we did it.
DR. COHN: It has obviously been awhile, but I remember that the pilot testing done with PCS, it was just in some of their centers.
DR. PROPHET-BOWMAN: Actually, there were two stages of testing done for PCS. One was an informal testing that included volunteers. They were recruited through AHIMA and AHA, and they got -- was it a two-hour training also, in Orlando as part of an annual convention. But I don't think that -- because that was a first level of testing, I don't think there was any formal data.
DR. MC DONALD: The sample size of the measure study stunk. Your sample size is pretty decent, pretty good numbers.
DR. FRASER: I don't mean to run with your project, but this is very exciting. I was wondering if, considering the percentage of people who were involved with the project, only six percent were physician practice coders. I was wondering if your opinion of doing an extended study, or AAPC helping you with an extended study of physician coders using ICD-10, physician practice.
DR. PROPHET-BOWMAN: Let me be honest with you. This would have been a very, very expensive project, as you may understand, some of you who are used to doing surveys of this magnitude understand. This was something that we did not have money in our budget for.
Obviously it is very important to us, so we wanted to work with our members. Now, if you want to do something with your own members, go ahead and do that. We could not commit to doing something with another group because this pretty much stretched our resources.
DR. PROPHET-BOWMAN: We are using our members' dues.
DR. FRASER: But we could model something after this. I could talk to you later about physician practice.
DR. PROPHET-BOWMAN: One of the things that actually even allowed us to facilitate and manage a project of this magnitude is this pretty exciting resource we have for our website called communities of practice, which is basically almost like you would be in a room with 200 people, and you are in connection with them all the time, and you can answer their questions and follow up on their issues and respond to them quickly, and put all the resources in one place, and not having to deal with 200 people sending you e-mail messages, basically. It located all of the communication with these 200 volunteers in one place. You could post an answer once, they all saw it there.
One of the reasons, besides the fact that we are using our members' dues to basically fund this project, is that we really needed to use this community of practice to manage the magnitude of this project. And right now, that feature is only available to AHIMA members.
DR. FRASER: Will you make your full report available to everyone? It would be just fascinating to read.
DR. PROPHET-BOWMAN: Yes. I imagine it will be posted on our website as well.
DR. FRASER: Okay, thank you.
DR. COHN: Karen, please.
MS. TRUDEL: I just wanted to ask you when you thought the final report would be ready.
MS. LEON-CHISEN: How many weekends do we have left?
DR. FITZMAURICE: What time tomorrow?
DR. PROPHET-BOWMAN: Someone had asked if it would be ready by mid-September. I think that might really be pushing it because of all of these problem forms and issues that we still have to go through and analyze.
MS. LEON-CHISEN: Manually.
DR. PROPHET-BOWMAN: Manually. So I would say end of September would probably be the very earliest date that we are able to pull all of it together.
DR. GREENBERG: The full written report.
DR. PROPHET-BOWMAN: The full written report.
DR. GREENBERG: I think the idea was that you would also present at the September meeting, give maybe a shorter report or something?
DR. PROPHET-BOWMAN: We can. That wasn't mentioned.
DR. GREENBERG: Maybe that was up in the air, that hadn't been decided.
MS. LEON-CHISEN: But we would not be able to have the full report by then yet.
DR. PROPHET-BOWMAN: We will perhaps have some additional pieces of data pulled together, but it probably wouldn't be the final, final of everything.
DR. GREENBERG: If you have a pre-final.
DR. PROPHET-BOWMAN: Preliminary-plus, almost done.
DR. GREENBERG: -- that would be helpful. By the way, NCHS really wants to thank AHIMA and AHA for this incredible effort, that as they said a voluntary effort. I want to thank the field for their interest in 10-CM and for carrying this out. This was not something that we were in a position to do, certainly not in the time frame that they did it. And we couldn't have done as good a job, either. But we really do appreciate it, and I wanted to publicly express our appreciation, and also our apologies for probably ruining your summer. But it was for a good cause.
Agenda Item: Subcommittee Discussion
DR. COHN: Any final questions? Thank you very much for a job well done.
Let me pull out my agenda here. I believe we move into more of a discussion period for the next hour. I think I want to take a minute and just reflect on a couple of items about where we are and how we are going to get to the end of a couple of issues, just in terms of timing and all of this, just to set expectations for the subcommittee and for the staff.
Earlier today we did talk about the PMRI letter. I think we have come to perhaps an informal conclusion that at least at this point, that letter is going to be held and going to be further discussed in October at our subcommittee meetings, with the intent to bring that forward in November to the full committee.
I guess I would have to ask Jeff about whether or not it would be of any value to share a draft version of that letter with the full committee in September or not, or whether we should just wait until November. I think my sense is that it may make sense to wait.
MR. BLAIR: Yes, I share the same sense. I do think that Walter probably could go ahead and include the things we agreed to this morning, so that at least when we do have our session in October, we are starting with a draft that includes all of the things we agreed to this morning.
DR. COHN: Okay.
DR. MC DONALD: On that, could we actually get the copy? There were some parts we left open, and we could go through those parts before October, either by sending mails if it is just all edits, and if he doesn't agree, we might have a small conference.
MR. BLAIR: I think a lot of the concerns that we had might fade away, in the sense that with NLM and FDA being able to give us the reassurance we want, and some of the caveats.
DR. MC DONALD: I'm not talking about that. There are some parts where we didn't really write it on the screen. We basically said we will fill that in. Those parts we could really lock down if we had a chance to say, that is not how we thought of it, and if he says yes, that is right, we can just fix it. If there is disagreement, then we would --
MR. BLAIR: Sure. And Clem, weren't you going to try to give some wording to Walter?
DR. MC DONALD: I did, but I think it is moot. I think we should wait. I was trying to make a proposal by this afternoon but I think we should wait to gather information and then do it from that.
DR. COHN: Okay. So we will move forward on that, and we will think about it. I am obviously hopeful that at some point, either in September or October, that we get a report from CHI on their medication domain recommendations for terminology, which might help us also.
DR. MC DONALD: Is that Walter's machine?
DR. COHN: I hope not. That is a good question. I hope not.
Now, let's talk for a minute about the process of moving forward with ICD-10-CM and PCS deliberations. Obviously today we have started with an update on where we are with the cost impact study, and we have also heard I think a report on a pilot study going on.
Now, my understanding is that hopefully over the next week, 14 days, next week to two weeks, we will have a version of a report on the cost impact that we would hope would be of a heft that we could show it with the full subcommittee.
Now, what we are expecting at that point is to actually ask the subcommittee to review it. I think we will need a conference call to go through it, to make sure it is adequate for basically forwarding to the full committee. At that point I think we would also be releasing it as a public document, at least as a draft public document, for the full committee's consideration, so that others would be able to react to it. I think it is very hard for anyone in the private sector to react to a page of numbers and a verbal statement.
Now, I think our intent is, assuming that that goes as planned, to have a session at the full committee meeting in September, which would involve a presentation by Martin, and then discussion and probably a reactor panel. I think the intent is to develop a reactor panel to react to the contents of it, as well as potentially begin to bring forward additional information on cost and impact.
I think also, time permitting, it would be very valuable, for any members both of the subcommittee and the full committee that may not have been through the last couple of years of discussions on ICD-10, that we need to come up with some way -- help me if I am missing anything there, but to bring them up to speed, not so much on the politics of all of this, but more on what this is, what has been evaluated about it, what we know about both CM and PCS.
Obviously, Donna, I am looking at you as I say all of this, but I think part of it is probably you, part of it is probably the appropriate person from CMS or their designee, to talk about PCS, but obviously in a time-constrained fashion with whatever advance information needs to be distributed to the full committee, so they can come prepared in September with some knowledge.
Now, as you remember, there are two pieces here. One is a cost impact study, and the other question is, what does the full committee recommend.
MR. BLAIR: You used the words cost impact.
DR. COHN: Yes.
MR. BLAIR: Cost benefit, right?
DR. COHN: I think the official title -- I'm sorry, impact analysis.
DR. GREENBERG: Which includes cost and benefits.
DR. COHN: And benefits. I'm sorry, I mis-stated the name for it before. I was trying to be technical and name the report appropriately.
MR. BLAIR: Okay.
DR. COHN: There is an analysis piece, where we want to understand the results of our report as well as other reports that may come in, but we also need to develop recommendations.
DR. GREENBERG: Right. Let me just clarify. We can certainly provide background information also prior to the September meeting. But were you suggesting that you would also want maybe a presentation on ICD-10-CM and ICD-10-PCS before the impact analysis is reported? I wasn't quite sure what you were asking for there.
DR. COHN: No. I am privy to the schedule of that meeting, but I don't know what -- now that we have changed things around a little bit, I don't know what the time frame looks like. I just thought it would be valuable for committee members who either hadn't been exposed to this in awhile, and certainly the new committee members, get an opportunity to hear about the coding systems.
So I thought that -- once again, it is a scheduling issue, but maybe before Martin presents or some other time that day, it would be useful for there to be at least some period for them to be informed about code sets.
DR. GREENBERG: Well, I think we should provide them something in writing also.
DR. COHN: Oh, yes, I think I asked for both. I wasn't sure that that was sufficient.
DR. GREENBERG: There is not a lot of time right now in the agenda for --
DR. COHN: I was just hoping that since we have taken the PMRI recommendations off the table for September, that might open up a little bit of time. That was the only reason I was suggesting that.
DR. GREENBERG: That will open up some time.
DR. COHN: That will open up some time.
DR. GREENBERG: Okay.
DR. COHN: Now, I think that as I said, September is transitional. We are beginning to get information. We are considering the impact analysis, but we need to begin to take that and begin to integrate it along with other information that we receive, and begin to try to move forward some recommendations.
So obviously, in September we will be talking about the analyses, but we need as I said to begin to think about that next step.
Now, we do have hearings in October for those in the audience. I think everyone around the table knows that we have actually just changed dates for those hearings. Michael, I think we told you that, didn't we?
DR. FITZMAURICE: Did you change it today?
DR. COHN: We changed it earlier today, yesterday, I'm sorry.
DR. FITZMAURICE: I did not hear that.
DR. COHN: I'm sorry. What we have done is, we have moved hearing dates from the 21st and 22nd, and they are likely also going to be the 23rd of October, to the next week. I think realistically, we are talking about another three-day hearing, which will be the 28th, 29th, and 30th of October. During that time we will hopefully deal with the PMRI letter and get final closure on that for the November meeting, which is the week after, and we will also at that point take additional comment, testimony of others, both on any outstanding cost-benefit issues, as well as comments from the community at large in relationship to what the committee ought to be recommending in relationship to ICD-9-CM and PCS.
The hope would be at that point that we would have time for deliberations also, and that we would be able to come forward in November with a letter for consideration by the full committee.
Now, this is relatively ambitious, but at least it appears that there is time for all of this. There is obviously no guarantees that this would work as smoothly as we hope, but does this make sense? Jeff, please.
MR. BLAIR: Questions?
DR. COHN: Please.
MR. BLAIR: Should I assume also, should we assume also that the joint meeting with the privacy subcommittee will also be that week?
DR. COHN: No, there is actually going to be a saturday session in December.
MR. BLAIR: Okay, so that will happen then. What about -- we were going to have I think either a day or an afternoon that was going to cover the attachments, the CDA and the CCR. Would that be happening that next week in October as well?
DR. COHN: Well, I didn't mention them, only because I think that that is something that you and I need to plan.
MR. BLAIR: Okay.
DR. COHN: I think that in my view, the priorities for October, recognizing that we have sort of fallen behind now a little bit, are obviously the PMRI letter and trying to make progress on ICD-10. The next priority is probably anything that needs to come forward that is time-sensitive related to CHI.
MR. BLAIR: Right.
DR. COHN: And then obviously all of this other stuff. Do people agree with that priority statement? Stan?
DR. HUFF: I had a small seizure. Could you repeat?
DR. COHN: That's okay. As long as you don't have a large seizure, we are able to manage it during our meeting.
I was just speaking in terms of October. In my view, the priorities were that we needed to complete the PMRI recommendations and letter and bring them forward to November. We also needed to make progress and hopefully complete a letter after November related to ICD-10. The third priority was dealing with time sensitive CHI issues, and then beyond that, we needed to begin to deal with issues related to other PMRI issues, probably CDA, claims attachment issues coming forward, or otherwise.
But that is what I was positing. Hopefully there will be time for all of that, but does that make sense?
DR. HUFF: (Comments off mike.)
DR. COHN: Okay.
DR. GREENBERG: You said the joint hearings with privacy on security --
DR. COHN: Let me be very precise. We tried to put together a joint session with privacy on security, and scheduling issues seemed to become more problematic than actually trying to do the hearing.
So I think what we did was, we pulled back and decided that what we would do is schedule a hearing by the subcommittee on the security final rule and progress being made and issues being identified. We are inviting members of the privacy subcommittee as they are able to join us in December, but we decided that was getting to be too much trouble, to try to figure out how to do this all together.
DR. GREENBERG: It wasn't going to work as a joint hearing.
DR. COHN: That's right.
DR. GREENBERG: I guess I hadn't heard that.
DR. COHN: Well, this was something that became obvious over the last couple of days. Certainly if the privacy subcommittee can schedule it, we would be happy to co-sponsor that session with them, but as I said, it just didn't seem to be working. Michael?
DR. FITZMAURICE: It just seems we need time.
DR. COHN: Yes, exactly. Anyway, this is the broad view. In November, the idea would be bringing forward the PMRI letter as well as hopefully the letter for discussion by the full committee related to ICD-10 recommendations.
In December, as you all remember, we actually do have hearings in December, and security will likely be the issue there. Depending on how far we are with CDA and CCR and others, that may be a topic. Obviously CHI is a topic, and there is actually a DSMO yearly update that probably needs to be considered at that point.
I have probably forgotten two or three other things, but this is actually pretty -- in case you are all bored at some point, I think we have enough issues to keep everyone busy.
Susie, what are we missing?
MS. BEBEE: Are we pushing electronic signatures to next year?
DR. COHN: I didn't even have it on my list. Do you want me to put that down as an issue for next year? You would like to see it as an issue? You remember, the subcommittee has dealt with that before. It seemed to not be happening.
MS. BEBEE: I had a correspondence with Glenn Marshall, who had been helping to lead that, and you're right, it is status quo. But with the electronic health record moving at such a great speed, I was thinking the electronic signature might benefit from that.
DR. COHN: Obviously, yes, I absolutely agree. In the midst of all this, there are I think some dental issues, if we make it to that part of the letter, that we will have to deal with. I am less optimistic that we will handle that in October than I was. And as I said, there are other issues that we will obviously handle during the first part of next year.
So when we send out dates and availability to people for the first half of the year, please respond.
I guess the question is, does what we are describing as a potential plan for the next couple of months make sense? Is it doable? Are we being overly aggressive or not aggressive enough? Is there something obvious that we are missing that we should be considering? Steve seems to be holding his head in his hand.
DR. STEINDEL: No, it is an aggressive schedule, but I think it is doable. I think that one part of the schedule that is overly aggressive is not from our point of view, but from the CHI point of view.
DR. COHN: You mean their schedule is overly --
DR. STEINDEL: Yes, I think Cynthia and I are going to have to talk about that when she gets back next week.
DR. COHN: Karen, please.
MS. TRUDEL: I need to clarify what the issue is. Is it that you think that CHI is bringing too much before the subcommittee at any one given time?
DR. STEINDEL: I think they are asking for too much time in some subcommittee meetings. As we have seen from this schedule, the subcommittee has a lot in front of it. For the September meeting, they were actually asking for the whole two-hour block, for instance.
MS. TRUDEL: So is the issue that you don't think it will take that long to get through the subject matter, or that it is just too much time?
DR. STEINDEL: I don't see that the subcommittee can spend the full time we have allocated at the September meeting or the November meeting for CHI issues. So some balance needs -- either we need to spend less time presenting things from a CHI point of view, which doesn't seem to be agreeable to CHI, or spread out the time more.
DR. COHN: Let me give a slightly different perspective on this one. I think some of you have actually seen the CHI schedule, which of course is a very aggressive schedule, which I obviously fully support, and I hope the rest of the subcommittee members will do everything we can to accommodate whatever CHI's needs are.
My actual concern is not that CHI is asking too much time from us, but that the amount of time they have allocated or are requesting for the issues they are bringing forward is completely inadequate. So my concern is that -- the time that they seem to be requesting, I am happy to make additional time available as needed, assuming that subcommittee members are willing to go along with me on this one. It is just that if there is any discussion about any of the issues, there is not time in the times that they have requested to discuss it. There is enough time for them to present it. As long as we nod our heads and say yes, that's fine, we can basically handle the time amounts that they have been given, but if there is any discussion, if anybody from the public wants to even ask a question, there is no time in the schedule anymore for that.
So that is my concern. That is why I was asking Steve to work with Cynthia and others to put together a more realistic time frame, so then we can slot it into all of this stuff to make sure that it happens.
DR. GREENBERG: It is not so much that two hours is too much time to ask of the subcommittee; it is that it isn't enough time possibly to cover four areas or six areas.
DR. COHN: Yes, so I want to make sure that we have a realistic time, and then we will figure out how to put it in.
DR. MC DONALD: I think we hope to do better with the same amount of time. I think we have pushed CHI, tell us everything, keep us totally informed, so they have done that. But there are a lot of presentations that really aren't that informative. They are still early.
I don't think we should plan to have all the different groups talk each time. I think it should be when things are right. We can depend on the liaisons to tell us what is right. I sure don't think we need to go over every subject each time. They can give it to us in writing. But the right stuff they communicate pretty well.
DR. FITZMAURICE: You know what would really help too, is to have something written ahead of time, have all the documents, give us a week or two to look them over, so we can come prepared with comments, as opposed to answering things on the fly.
DR. STEINDEL: Mike, they have made that commitment to do that. But it is also CHI's commitment for each one of the domains to touch NCVHS twice on each one of the domains. One is a preliminary presentation on what their thoughts are, and then after follow it after the CHI council deliberates and finalizes the report. What I am hearing from Clem is that unless there is a real problem with the development of a domain and we have a specific reason, perhaps we should touch NCVHS only once. That is after the council has considered it.
DR. COHN: Well, I think we need to reconsider that. Clem, let me just read to you what is being proposed for discussion in two hours somewhere between September 23 and 24. Preliminary reports. Now this is obviously not the final report, but we are getting into this, which I think we need. They are going to be covering diagnosis and problem lists, supplies, anatomy and physiology, nursing and disability in these preliminary presentations, and for final approval they are talking about interventions and procedures, medications and organizations. They are requesting two hours.
Now, I thought they weren't asking for enough time.
DR. MC DONALD: But are they all at the same level? That is the question you have to ask.
DR. COHN: I don't know. Maybe that is a conversation we need to have.
DR. MC DONALD: I think in the presentations we have we have, about half the time is polite time. It isn't information movement time. I'd like to cut down on the polite time, if we could.
DR. STEINDEL: Clem, actually, when I talked with Cynthia about this, I was looking at the schedules. My recommendation to Cynthia would be, as Simon just noted, to take those presentations that we were going to present as final in September, and actually devote maybe a half a day to CHI in October, and present the final ones from September during that period, and use the time that is left for the preliminary ones that are also slated for October. Then we would just use the September time to present the preliminary thoughts on the areas that are listed right now.
I think from that point of view, we would probably use roughly an hour of committee time at the September meeting.
DR. MC DONALD: That sounds good. One other thought, though, I think part of the politeness is, these are things that are all divided into different leadership, and this leadership has to do a slide show, and they all get the same amount of time. Wouldn't there be some way that they could delegate the presentation?
DR. COHN: We actually asked Steve to talk to Cynthia last week, and she hasn't been around. So we need to have them come to a meeting of the minds. We had talked at one point bout having a half day or three-quarters of a day before the September meeting. We have taken that off the plate. We may actually have to have it back on, depending on what comes from the conversation, but I hope not.
DR. GREENBERG: I'm not quite sure what Steve was recommending, but at least on medications, I think if they were prepared to provide the final report in September, or sharing final recommendations in September, I think they should.
DR. COHN: Absolutely.
DR. STEINDEL: The only thing I see about that is that we would be devoting -- for the medication report, I see that as taking a minimum of a half an hour to 45 minutes, and we would devote the whole September meeting to the medication report. I don't see CHI using more than an hour of the two hours. I think the subcommittee has other things to do in September.
DR. FITZMAURICE: But it does tie in with RN Norm, and we are going to have some deliberations of RN Norm. So I think there is a lot of companionship there.
DR. STEINDEL: What I am suggesting is, we do the medication report in October, where we are also going to be discussing RN Norm and the letter, et cetera.
DR. GREENBERG: September is just a breakout session.
DR. COHN: Exactly, for two hours.
DR. STEINDEL: Right, for two hours.
DR. GREENBERG: I'm sorry, I was on another planet.
DR. MC DONALD: Your instincts are right. We should just do that.
DR. STEINDEL: Yes, and I think we can carve out a whole half a day for CHI in October, and then we can give medication its due deliverance during that time period.
DR. GREENBERG: Yes, I was on another planet here, with September being a subcommittee meeting.
DR. COHN: But as I said, if there are indeed time sensitive issues here, this has to be handled in September, then we will reopen the idea of having a day before the full committee meeting in September, on September 22, and we will take care of it. So we are committed to being appropriate partners in moving this stuff through.
I have already heard from most of the members, even though they weren't very happy about it, that if we have to, we will open up September 22 as a day. It is just that I can't tell at this point.
DR. STEINDEL: My sense -- and I will know better after I talk with Cynthia -- is that I think we can accomplish all that she wants accomplished in the time frame between now and December, just in a different order. I don't really see anything that I am aware of from the CHI point of view that requires the order be sped up for a particular domain. There may be something I'm not aware of, but that is my sense.
DR. COHN: Well, in case you didn't think we had enough on our plate, what else are we missing?
DR. FITZMAURICE: We've got an awful lot.
DR. COHN: I think we have got an awful lot, exactly. But I think we can manage it. I really want to thank subcommittee members. We have been adding days to meetings, we have been changing meetings around. This is I think our second three-day hearing in a row, and I really appreciate everyone's fortitude and willingness to stick in there, and obviously the staff, too. Mike, I'm looking at you in particular.
I want to thank you. I'm sorry we didn't get the letter finalized and all of that.
Now, I think the other item is just to talk a little bit about tomorrow. Is there anything else before we reflect a little bit on what is going on tomorrow? Is there anything that is very obvious, that is a process piece, in terms of any other work that we have going on?
I think certainly my idea in terms of the CHI piece is not only that the subcommittee has to think about it, but we actually need to come up with letters on a timely basis to the full committee, supportive of those recommendations that we are supportive of. So that is a piece of all of that.
So, anything else? If there is something else, let me know. And obviously we will be talking more about CHI tomorrow.
Now, tomorrow morning does start at 8:30, I want to warn you, and it actually is here, it is not the Hubert Humphrey Building.
DR. GREENBERG: What was that? I didn't notice that. There is a meeting tomorrow?
DR. COHN: There is a meeting tomorrow. We will start out with a claims attachment update and discussion, which we think will take about a half an hour, and then we move into the CHI update, talking about lab results, which I think we obviously have already put into a letter, as well as contents and units.
Then after a break we will be talking about getting back to our roots with HIPAA, and seeing where we are with that, as well as getting a sense from industry of where we are and the implementation issues that are coming forward. After that, we will obviously take a little bit of time for open mike about any HIPAA issues that we are not aware of or that we need to be discussing. Hopefully we will be adjourning around 1:15.
DR. MC DONALD: Having the meeting in the hotel -- I've said this before -- is really nice, because you don't spend time taxing around and you can go up to your room and check on stuff.
DR. GREENBERG: It does take additional budget.
DR. MC DONALD: Thank you. I would just encourage it. It is really nice. This hotel is pretty good. I guess it was cheap, but it is pretty good.
DR. GREENBERG: Are you staying here?
DR. MC DONALD: Yes. It looks really new, except when you see it is an old apartment building here and there. It is nice.
DR. GREENBERG: I'm glad you're happy.
DR. COHN: Well, just before we adjourn, I actually do want to take a moment. We have a lot on our plate. We would not be able to get anywhere near as far as we have gotten or be able to do this over the next couple of months without all of our staff. I really just want to take a moment to thank you all. Hear, hear.
(Whereupon, the meeting was adjourned at 5:00 p.m.)