[This Transcript is Unedited]
Washington Terrace Hotel
1515 Rhode Island Avenue, NW
Washington, DC
Agenda Item: Call to Order, Introductions, Review Agenda - Dr. Cohn and Mr. Blair
DR. COHN: Good morning. I want to call this meeting to order. This is the first day of three days of hearings of the Subcommittee on Standards and Security of the National Committee on Vital and Health Statistics. The Committee as you all know is the main public advisory committee to the U.S. Department of Health and Human Services on national health information policy. I’m Simon Cohn, I am Chairman of the Subcommittee, my role when I’m not here in Washington is as National Director of Health Information Policy for Kaiser Permanente. Obviously I want to welcome those here, HHS staff and others in person. I do want to indicate that today unlike most of our hearings, we’re actually not on the internet, so I think we want to apologize for that, apparently the server is down and I believe we’re going on the internet tomorrow again, so just a piece of information for everyone.
Now this is obviously being taped and thus I would remind everyone to speak clearly and into the microphone so we can tape the proceedings and make them available on the internet. I would also ask those from the audience if you have questions to make sure to come to a microphone and identify yourself as part of that process.
Obviously we have a lot to cover over the next three days. Today we continue our work on health care terminologies. The focus of today is really on the issues of drug and device terminologies and Jeff Blair the Vice Chair, will be leading those sessions. Tomorrow morning we will be reviewing and updating as necessary, a PMRI terminology recommendation letter and a companion report with the hopes, and certainly with an eye towards bringing this forward in September for Full Committee consideration, and we’re thankful that Walter Sujansky, who has been our consultant on that process, is here in attendance today and will be working with us tomorrow during that morning session.
Tomorrow afternoon we shift to an update and discussion of the federally sponsored ICD-10-CM cost impact analysis being undertaken by RAND. We’ll also hear a report from the AHA and AHIMA on their ICD-9-CM pilot study that they, I believe, have completed now or at least is underway.
Thursday morning starts at 8:30 a.m., our third day, we begin with a brief update on the claims attachment NPRM and then move into an update on the Combined Healthcare Informatics Initiative and a discussion of several of their current recommendations. This is followed by a HIPAA update and discussion with an expert panel on issues related to the implementation. Finally we’ll take a couple of minutes on Thursday to talk about Subcommittee business, review dates and likely topics for the remaining hearings this year and make sure that we have in place everything that we need to come to successful conclusion on a number of our issues in a reasonable timeframe.
I also do want to warn everybody that we are shortly going to be scheduling 2004 hearing dates, at least for the first half of the year, so we’ll be, while not querying people about dates on Thursday, I just do want to warn everyone.
Now for everyone in the audience and we’ll say this obviously every day sort of going forward, this is clearly an open session. Those in attendance are welcome to make brief remarks if you have information that’s pertinent to the subject being discussed and we have time. We will also try to make time at the end of each session for brief comments by those in attendance. Obviously for those in the audience and anyone who is listening to this afterwards, we obviously also welcome comments via the internet and via email as well as letters on any new business coming before the Subcommittee.
With that let’s have introductions around the room, around the table first of all and then around the room, so those obviously on the Subcommittee if there are any issues coming before the Committee today for which you need to recuse yourself please so state. Jeff, would you like to lead off?
MR. BLAIR: Hello, I’m Jeff Blair, I’m with the Medical Records Institute, I’m Vice Chair of the subcommittee, and I’m a member of AMIA, ASTM, HMSIS, HL-7, and there’s nothing that I know 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. ZUBELDIA: Kepa Zubeldia with Claredi Corporation, member of the committee and subcommittee.
MS. GREENBERG: Marjorie Greenberg, National Center for Health Statistics, CDC, and executive secretary to the Committee.
DR. BLUMENFELD: Hi, I’m Barry Blumenfeld, I’m Associate Director for Clinical Informatics at Partners Healthcare, and also head up our Medication Services Group there.
DR. LEVIN: Randy Levin with the Food and Drug Administration, Center for Drug Evaluation Research.
MS. 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, senior science advisor for information technology to the director of the Agency for Healthcare Research and Quality, liaison to the National Committee and staff to the Subcommittee on Standards and Security.
MS. BEBEE: Susie Bebee, ASPE, staff to the subcommittee.
DR. MCDONALD: Clem McDonald, member of the subcommittee and of the committee, Indiana University and Regenstrief is my home. I’m chairman of the LOINC Committee but that’s not under discussion today so I don’t think I have to recuse.
DR. HUFF: Stan Huff with Intermountain Healthcare and the University of Utah in Salt Lake City. I’m co-chair of the HL-7 Vocabulary Technical Committee and also a co-chair of the LOINC Committee, but again I don’t think those subjects will come up today.
MS. SQUIRE: Marietta Squire, CDC, NCHS, and staff to the subcommittee.
DR. MCNAMARA: Tim McNamara, I’m with Cerner Corporation.
MR. NELSON: Stuart Nelson, National Library of Medicine.
MR. BROWN: Steve Brown, Department of Veterans Affairs.
DR. LINCOLN: Mike Lincoln from VA.
MR. GRIFFITH: Stan Griffith with Indian Health Service.
MS. FRASIER: Judy Frasier, American Optometric Association.
MS. ECKERT: Karen Eckert with Medi-Span.
DR. BICKFORD: Carol Bickford, American Nurses Association.
MS. HABER(?): Margaret Haber, National Cancer Institute.
MS. WILLIAMSON: Michelle Williamson, National Center for Health Statistics, CDC.
MS. PICKETT: Donna Pickett, National Center for Health Statistics, CDC, and staff to the subcommittee.
MS. FRIEDMAN: Marie Friedman, CMS, staff to the subcommittee.
MS. LESH: Kathy Lesh, the Kevver Company.
DR. SPACKMAN: Ken Spackman, I’m with Oregon Health and Science University, and I’m chair of the SNOMED International Editorial Board.
MR. DUBOIS: Mark Dubois, Wolters Kluwer Health, the Medi-Span product line.
MR. ROBINSON: George Robinson, First Data Bank.
MR. BIZARRO(?): Tom Bizarro, First Data Bank.
MR. KOMEZI:: Said Komezi, medical informaticist, VHA.
MR. RANDINA(?): Mark Randina, health informatics specialist with CMS.
MR. BOOTHWAITE(?): Bill Boothwaite, consultant in Washington, D.C.
DR. SUJANSKY: Walter Sujansky, independent consultant and advisor to the NCVHS on PMRI terminology standards. I’d like to add in the interest of disclosing other of my consulting engagements that may be related in any way to the NCVHS standards activity I’d like to make the following statements. I’m 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 my work at the NCVHS, I’ve also been engaged by Amacore, Incorporated. an electronic medical record vendor that has licensed SNOMED CT for use in its products. In this capacity I’ve also interacted with Health Language, Inc., 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. And lastly, during this project I was also engaged by a company named Innovion, whose chief technology officer, Keith Campbell, is a member of the SNOMED editorial board. Innovion is not a SNOMED licensee and my work there was unrelated to SNOMED. Thank you.
DR. COHN: Thank you all, oh, two more.
MS. LEE-HESSEIN(?): Marilee Lee-Hessein, American Hospital Association.
MS. PROFITT-BOWMAN: Sue Profett-Bowman, American Health Information Management Association.
DR. COHN: I want to thank all of you for joining us. In the interest of full disclosure I actually don’t think that I have anything to recuse myself about though I obviously should indicate that my organization, Kaiser Permanente, is a user of clinical terminologies, and obviously they’re involved in the development and implementation of a number of terminologies that we have, that we’ll be hearing from today and I’m sure tomorrow. Jeff, would you like to make a couple of introductory comments?
MR. BLAIR: Certainly. To put the hearings today in context the Health Insurance Portability and Accountability Act directed the NCVHS to study issues related to uniform data standards for patient medical record information and the electronic exchange of standards for PMRI. In August of 2000, the NCVHS set forth a report on PMRI standards which was a framework for how we would proceed over the subsequent years, where we would put our priorities, and it also set forth the guiding principles that we would use for the selection of PMRI standards.
Health and Human Services indicated that they had a level of comfort with those guiding principles for the selection of standards. We modified those guiding principles in the selection of message format standards, PMRI message format standards, which were recommended February 27, 2002 and which the Secretary announced publicly I guess it was March 21st of this year, as CHI, Consolidated Health Informatics Initiative standards. August of last year we began the process of evaluating, selecting, and recommending PMRI terminology standards.
At that point in August we heard from a number of testifiers from the industry in a day and a half that gave us guidance on how we should again modify the guiding principles to make them appropriate for the selection of PMRI terminologies. We also got guidance from them on what should be the appropriate scope. We modified that in October of last year. We then crafted that into a fairly comprehensive 14, 16 page questionnaire that was distributed to as many organizations that identified themselves as PMRI terminology developers. Those questionnaires were sent out in February. Dr. Sujansky compiled and analyzed the results from those questionnaires and reported to us March 23rd I believe it was of this year, which was our first feedback.
As part of that process one of the things that we decided to do was focus on those terminologies that could form a core set of PMRI terminologies at the advise and direction of those folks that had testified to us back in August. And as part of the core set clearly there were certain technical requirements for terminologies that would allow them to serve as reference terminologies, so we wound up narrowing the list of those folks who had filled in the questionnaire to the candidates for the core set. We then took that information and distributed that report back to the terminology developers to make sure that we had interpreted their information properly, and we wound up getting a second revision of our analysis in April.
In May we wound up hearing from users of terminology developers, and Walter proceeded with a third revision and update as a result of feedback from the terminology developers to make sure that the technical analysis was right, the technical analysis that would indicate that terminologies would qualify for the core set, being that they would be reference terminologies.
The users and vendors of the terminologies testified to us in May, and at the end of the May hearings it became clear to us that while there were certain areas of comfort that the Subcommittee had in terms of recommending terminologies, we still felt we needed additional information related to drug terminologies and medical device terminologies, and Steven Steindel who’s sitting to the left of me, has prepared the agenda for today and arranged with all of the folks who are testifying today, and also set forth the questions, gave us guidance on the questions, for the terminology developers for today. So I want to thank Steve for really doing everything to set up our day today.
So we’re going to be hearing in the morning initially from Barry Blumenfeld and Randy Levin who are going to wind up giving us kind of a drug terminology 101 and basic information on the FDA directions. We’re then going to be hearing from the terminology developers from NLM and VA with respect to RX-Norm and NDF-RT, and then in the afternoon we’re going to be hearing from the private sector drug developers. And then later in the afternoon we’ll be hearing from medical device folks and then at the end of the day we’ll try to pull together our ideas and all.
So our first testifier is Dr. Barry Blumenfeld and he’ll be giving us the drug terminology 101 for Steve tells me about 20 minutes, is that correct Barry?
DR. BLUMENFELD: Yes, hopefully.
MR. BLAIR: Thank you and we’re ready.
Agenda Item: Overview on Drug Terminologies - Dr. Blumenfeld
DR. BLUMENFELD: Well, I guess I’d start out by saying really it’s an honor to be here and although Jeff portrayed this as drug terminology 101 I’d say it’s probably given that most of the folks here know as much about drug terminology or more than I do it’s perhaps advanced beginners drug terminology. Also by way of disclosures I probably should have said something about this initially, I have in the past worked for both First Date Bank and/or Cerner Corporation which offers Multum, so although I do not currently work for either and do not consult for either I am certainly familiar with their terminology offerings for drugs. And with that let’s get a start.
So today I thought in my rather brief time and really the biggest challenge is taking what is usually a much longer discourse and cutting it down to 20 minutes or so, I want to try to cover the need for drug terminology, a few definitions and some requirements, some important concepts from my perspective, and again my perspective here is as someone working in the context of a large integrated delivery system, talk about a few challenges that I’ve come across over the years and finally a rather brief summary.
So really what is the problem here? The problem is that we have a limited ability to exchange drug information electronically. There certainly is some information that we can exchange and do exchange every day, for instance package level information in the form of NDC codes, that information is going back and forth between various entities all the time. On the other hand, if I wanted to send a patient’s medication list from a hospital in New York to a hospital in Los Angeles, and I wanted to note that the patient was taking Vioxx 50 mg. PO once daily, there really is not a readily available standard terminology that allows me to make that sort of statement and to send it.
Now that’s beginning to change. Certainly the increased focus on health care quality and safety, on automating decision support, on adverse event monitoring, HIPAA itself, even consumerism folks wanting to have access to their information in an online fashion, and of course bio-surveillance post 9/11, all of these things are really pushing us to improve our ability to send this information electronically. Also keep in mind, and I don’t need to tell anyone here, that there are terminologies available, it’s not that we don’t have terminologies. The problem in a sense is that there’s so many of them, in many cases they can be expensive, they can be specialized, and very often they’re incompatible.
So where do we use drug terminology? Just a few use cases. Well, if you are a consumer you might want to access your online medical record and either review or add medications yourself. Of course, if you’re a physician in ambulatory practice you’re going to want to be able to write prescriptions and send them possibly electronically to a pharmacy. You’re certainly going to need to get information between plans and DPM’s back and forth to the pharmacy. If you’re an inpatient setting and you’re a physician you’re probably working, or will be increasingly working in the future in the context of computerized physician order entry, and there you need terminology to write medication orders. Of course you’ve got to send those orders to the pharmacy and you probably would like to do it in a way that’s electronically mediated so that that information doesn’t require any hand retyping or recoding. And subsequent to that you’ll probably want to send that information on to the administering nurse.
And then finally, in all of these cases as we move towards an overwhelming electronic health record, you’ll want to be able to save this information in all the differing use cases and levels of granularity into that health record. And there are many more use cases as well, certainly things like designing new drugs and drug trials, registering products, reimbursement, there are many different cases beyond this where you need to send drug terminology.
So stepping back for a moment just to be slightly didactic, when I talk about terminology I’m referring here specifically to a set of terms representing the system of concepts of a particular subject field, and I’m just drawing this directly from the ISO standard of core terminology.
I’d like to say a few words about the qualities of good terminologies and these are really very broadly accepted criteria at this point. They were first presented in a paper by Jim Semino(?) in 1998. I think most people would not argue with these at this point, things like completeness in content coverage, a concept orientation, the fact that concepts are permanent and are not replaced, the idea of non-semantic or meaningless concept identifiers, poly-hierarchy in the terminology, formal definitions, multiple granularities, especially important in drug terminology, multiple consistent views, context representation, graceful evolution, and recognized redundancy. And these, as I said, requirements are now broadly accepted, that doesn’t mean though that all terminologies, both existing and planned, either do it here to them or necessarily always need to adhere to them.
I’m presenting today some additional requirements for drug terminology, and I use the word requirements here in a less absolute sense, but these are all criteria or requirements that have been advanced by myself and by others, some of whom are in the room at this time, and I’d like to list them. One is that a drug terminology should be maintained by a primary code assigning authority. Another is that it should be available for use at little or no cost.
The content should be updated and made available quickly as drugs enter the market. Drugs as you know, change very, very, very quickly, and in a clinical situation such as I find myself in at Partners, if we don’t have that information available quickly it can be very difficult for us to deal with. It should be backwards compatible with the NDC system. It should be designed for compatibility with evolving international standards. It should adhere to an accepted common information model, and by that I mean concepts in relationships should be organized into formal classes for query and presentation. It should support abstraction at multiple levels of detail, there are multiple users and their need to be multiple use and settings as I showed you in my previous slide. We should define how statements are composed and recorded. You should build from atomic concepts whenever possible. And finally, the terminology should support synonyms and lexical variance.
So from here I’d like to actually turn to some core concepts, or things that I believe are core to drug terminology. Again, this is from my own perspective, it’s not all of the concepts, and I should also add that given the time constraints it was impossible to cover everything one might want to consider. Some of these concepts are ingredients, the concept of a routed generic, the concept of a clinical drug, a manufactured drug, and a packaged product. There are also additional concepts that I think are very important, including therapeutic classification, forms, routes, and strength. So let’s start with ingredients.
Typically when we talk about ingredients we’re really talking about active ingredients or the chemical that are in the drug. An example would be Rofecoxib or Vioxx. Systems usually identify allergies at the level of ingredients. Now this isn’t the only thing sufficient for identifying allergies, we also have to group allergies and one of the questions that may come up later today is whether in fact these various groupings and classifications should be part of the terminology.
Ingredients can be further specified and in fact in the RX-Norm worldview you do distinguish between ingredients in the base and the precise mode where one might specify a salt. Many systems represent ingredients in a multi-ingredient drug as a specific entity in their data. For example, First Data Bank has the concept of the HICL, or hierarchical ingredient code list.
One ingredient would be, or in one to many routed generics, and a routed generic is the combination of an active ingredient or ingredients plus a route. So in this case we’d be talking about Rofecoxib, I should have picked a better drug, this is hard to say, PO. Now this concept is very key in order entry or in order entry systems. It’s certainly key at Partners where much of our drug dictionary for inpatient usage is organized around this concept. It also is very useful for some decision support functions like drug/drug interactions where you really want to consider interactions not at the level of ingredients, because they don’t include the route, but rather at the level of a routed generic so that you can distinguish where for instance a topical drug might not interact with another drug whereas its systemic version would.
Moving from there, and again this is a one to many relationship meaning that one routed generic can be in multiple clinical drugs, we would add things to define a clinical drug, things like strength, dosage form, excuse me, and dosage form, and some examples of that would be Rofecoxib 25 mg. tablet, or 25 mg. per 5 ml. oral suspension. Now this does not typically include the route of administration, although the concept of a formulation does, and we frequently move back and forth between those in clinical systems.
The clinical drug is probably the most important concept from the perspective of clinicians, particularly physicians, both in order entry systems and in outpatient prescribing, and in fact prescriptions are usually specified at this level. And just to compare and contrast for a moment between orders and prescriptions as concepts, you’ll note that in the case of an inpatient setting we’re really looking at an absolute amount of a drug with a routed generic, in a sense we’re depending on the pharmacist to assist us and to further specify what we’re actually going to give the patient. So a perfectly legal order would be amoxicillin PO 250 mg. TID, and you’ll see that in many order entry systems. Keep in mind that the physician there has not specified a strength and a form, that’s been left up to the pharmacist.
On the other hand if you’re working in the outpatient world you really do need to specify those things. In that case you need to say something like amoxicillin suspension, 250 mg. per 5 ml. one teaspoon PO TID. So there really is both conceptually and in terms of the database and terminology needs some differences between inpatient and outpatient ordering.
Okay, so moving up from there we have the concept of a manufactured drug, and again, this adds additional information to that of the clinical drug and it is also a one to many relationship. Now we add things like trademarks, inactive ingredients for the first time, manufacturer. Many folks also refer to this as a dispensable drug because it is in fact what is dispensable. An example of this would be Merck Vioxx 25 mg. per 5 ml. suspension, and note that we now can identify inactives like citric acid, strawberry flavor and so forth. I should add also that inactives is a little bit of a misnomer, in fact in some cases inactives can be very active when there are things like lactose or dies or latex or any number of other things. The only level at which we can begin to identify them is at the manufactured drug level and therefore in most cases the physician ordering the drug really is not aware of and cannot specify that information. This is a key concept in the idea of an electronic medication administration record. Most nurses when they are administering drugs will do it at this level.
And finally we come up again in a one to many relationship the concept of a packaged product. Now we’re adding all of the information that lets you literally pull a package off of the shelf, things like size, units, free of, package specific devices, the one or more packaged components. An example of this would be Vioxx 50 mg. tablets, bottle of 100. And note that I’ve included an NDC here because in fact packaged products are what we apply NDC codes to. Most pharmacy systems are very closely aligned with the idea of a packaged drug because that’s what they need to pull off the shelf.
Okay, so having taken this whirlwind tour through some of the important concepts, I should mention that there are many, many challenges. One is in fact conflicting views on key concepts, at what level of granularity we need to place them, and in fact what they are. For instance, should the concept of a clinical drug which does not include the route or the concept of a formulation be considered to be a key concept. Also knowing where the terminology stops and the information model starts, a little bit of a technical statement but it’s very often hard, it’s hard to tell where we’re using the terminology itself to organize our concepts versus where we’re actually using them if you will as the actual items that we’re talking about.
Certainly there’s a lot of discussion around supporting vocabularies. I mentioned that routes and dose forms were important, but there are many questions to answer there. For instance, when we talk about a route are we talking about a method versus a site, i.e., injection versus epidural. And granularity, are we talking about IV or should we talk about a continuous IV infusion as the lowest level. Likewise with dose forms, are we talking about physical forms or form plus side of administration, i.e., a tablet versus an oral tablet. Dispensable form. Inhaler versus spray. And very often, and this is something we really struggle in order entry systems with, is it enough to say that it’s a tablet or do we need to have an increased level of granularity in the classification around that to refer to things like sustained released tabs and Enteric coated tabs and any number of other sub concepts.
Operationally, and I included this slide because per se it’s not a question about terminology, but it’s very important to those of us like Partners where we are clinically users, there are a lot of considerations that I think move beyond just the actual structure of the terminology. Things like the frequencies of the updates, the available data in database formats, the degree to which customization is possible, the degree to which legacy systems are supported and backwards compatibility is maintained, the degree to which the data is available for multiple platforms, and most importantly, customer support. When we use an institution like Partners, when we use drug terminology, the suppliers of that data often need to respond very quickly to concerns we have or problems with the data, and that becomes a large concern when we move to a standardized terminology.
So in summary, and hopefully fairly on time, I’d like to say that there certainly is a pressing need for drug terminology and drug terminology standards, as well as an information model for that terminology. There are some clear requirements and that any solution must address both conceptual and operational challenges. As you listen today think about how the various speakers discussing their terminology perspectives in their terminologies, how they meet these criteria, how they meet these challenges, both for now and in the future. Of course we all make mistakes and we will continue to make mistakes. I like to say though that in the words of Yogi Bera, we don’t want to make the wrong mistakes. Thank you.
MR. BLAIR: Barry, thank you very much. Our next presenter is going to be Randy Levin from the FDA, he’s going to be speaking to us for about 30 minutes and I’m mentioning that because we will have about 25 minutes after that for questions before we hit our break.
Agenda Item: Overview of FDA Drug Directions - Dr. Levin
DR. LEVIN: Thank you. I’m going to be going over drug terminology from what the FDA has been working on and we’ve been working with other government agencies on a collaborative, this is a collaborative effort so I’m going to go over some of the building of the drug terminology and this collaborative effort, but then I’m going to be discussing maintaining a drug terminology. And specifically a discussion about leveraging existing processes, that exist within the FDA that will be helpful in maintaining any drug terminology, which I think is probably one of the most difficult parts. And I’ll be discussing activities with the FDA on ingredient terminologies, terminologies for finished dosage forms, drug products, drug product packages, and structured product labeling. I’ll mention something about medical devices, a little brief note about medical devices, though most of, we’ll have more discussion on medical devices in the FDA this afternoon.
First I’m going to talk about the collaboration that the FDA has been involved with drug terminology, and the collaboration focuses in on the concept of the clinical drug and the RX-Norm that the NLM has been working on. The core of the terminology is the medical reference terminology, the NDF-RT from the VA, and some more details added on to that from the FDA on drug products. And you’ll be seeing more of this later when Stuart Nelson talks about RX-Norm and Steve Brown talks about the NDF-RT. But starting with the clinical drug concept in RX-Norm with the active ingredient drug component in clinical drug with a dosage form is the first part, and again Stuart will follow-up on discussion on this aspect.
Added to that then is the VA NDF-RT, where they start to take information, tagging it to the active ingredient, things like drug class, mechanism of action, physiologic effect, chemical structure, therapeutic intent, clinical kinetics, and also tag things to the clinical drugs, including therapeutic intent and kinetics and dosage form. And all these things are linked to the packaged drug and the national drug code. And then the details, the final part and this is from the collaborative effort, is some additional things that are added on, the details from what the FDA has and that includes codes for active ingredients, the unique ingredient identifier codes. Also information about finished dosage form and drug product, information about the indications, the official approved indications for the drug products, and information about clinical effects for active ingredients and this includes adverse events, interactions, those types of effects where you’re describing, these details are coming from the FDA from the labeling. So a lot of this information, what I’m showing you right here, is a combination of pieces from the FDA and from the labeling that we currently have with each drug product, along with additional detail information, what we gather from listing and additional information that’s added on after the fact from things that are relayed from the other collaborators.
The building of the terminology is one thing, maintaining the terminology is another, and that’s where a lot of the difficulty comes in. And Barry was mentioning some of these things as far as keeping up to date, making sure the information is available when the drug comes to market, various aspects along those lines. And this is where we can leverage existing processes that the FDA, is either ongoing in the FDA or the FDA is under their regulatory authority as part of regulating all drug products marketed in the United States. And the FDA through three processes, one is their drug product applications, where we receive, they’re all now approved on drugs, they have drug applications. We view all labeling for all these products and labeling changes, and every drug product in the United States has to be listed, and this is for the regulations through the Food and Drug Administration. Other leveraging issues related to National Library of Medicine with the RMLS(?) and the distribution of medication information and I’m sure we’ll talk more about that later.
So for a drug terminology similar to what Barry was demonstrating, there is an active ingredient, and if you had strength, there’s drug component, different then what Barry was talking about, his adding route of administration, but here you have strength, you have drug component. If you have form then you have the clinical drug concept, it can be one or more drug components. Inactive ingredient in appearance then leads to the finished dosage form, and one clinical drug can have multiple finished dosage forms as the manufacturers add different inactive ingredients in different appearance, but you can put one or more finished dosage forms together to have a drug product. Usually there is just one finished dosage form per drug product but sometimes there’s more than one. And then the drug product can be packaged in multiple ways to have a packaged product. So those are the basic terminologies and what I’m going to do is go over some of the ways that the FDA is involved with these codes of different drug terminologies.
First I’ll talk about the ingredient. We have a system, the Substance Registration System, that the FDA uses to provide, assign a code to each ingredient, and we have been doing this internally but now we’re working on a system that we would be able to share this externally and have it provide a unique ingredient identifier for every ingredient, that’s both active and inactive ingredients. Now these ingredients will be defined by either chemical structure or by a process, we have worked out business roles for how to define this so it’s consistent. And we have a committee, oversight committee for identifying these substances. Initially our focus has been on drugs, but we are also involved with biologics, dietary supplements, and even substances used in devices.
Here’s just an overview of a system where there is a lot of our products are identified by a structure alone, and we have a system now where we take the molecular structure and by assigning a, we have software that will allow us to identify each structure uniquely by its structure. Then we have that linked, if there is not a unique structure to a substance then we define it by other methods using a process or some other definition that our working group puts together, an oversight group will put together. So this is for substances where there may be more than one structure or the structures aren’t defined, such as conjugated estrogens where there are multiple structures or herbal products or botanicals and the like, so it can either be defined by a structure or a process. It provides a unique code, which is tied into our current existing process, internal process where we tie this to our applications and other information about the drug products. So that’s our substance registration system. This will provide codes for all ingredients, for active ingredients will provide codes for both the active (?) and the substance. And this is, the codes are provided through two processes, one is through our investigational and marking applications, and through drug listing. So if either company provides their ingredient information to us through their applications, even during their investigational phase, or when they list a product with us. And the information we’re working with the NLM, we want to work with the NLM to disseminate this information. So when an IND comes in, that’s an Investigational New Drug application comes in, we have the ingredient structure and names, we put that into our registration system and have the information, the codes put into our ingredient dictionary.
But on the other side, the NLM can also add ingredients as they see, that they find, so if they find an ingredient in the public domain that’s not part of something that we have already released then they can provide the structure to us and we can provide a code to them. Since investigational drugs we cannot reveal anything, any information about any investigational drug, we will not provide any codes for that but if it’s in the public domain the NLM can ask us for a code and we’ll provide a code for any substance. So this will allow us to provide codes for things that are either investigational or things that are not, we are not currently marketed in the United States, so on substances that are not marketed in the United States. They will put this into their databases and of course when we release something or something is marketed, it will also be made available. So this will allow ingredients to be coded and the information to be made publicly available. Currently the codes are just internal to the FDA.
The next piece is with the finished dosage form, this is when you take the active ingredient and now you’re adding the components that make it unique to a manufacturer, so the inactive ingredients, the appearance are all involved. We have codes for the finished dosage form that we currently apply for every finished dosage form. There’s also imprint coding on every solid dosage form, this is for our regulations that no two drug solid dosage form can be, they have to be distinguishable or uniquely identifiable by the imprints code that’s in conjunction with size, shape, and color, so whatever is placed on the solid dosage form, whether it’s numbers or logos or something, along with the size, shape, and color, have to make it uniquely distinguishable, so this is imprinting information that is currently required and that we track in our listing process.
Finally, the other terminologies are for the drug product and the packaged product, and this is when you take the finished dosage form, one or more finished dosage forms, put them together and then you’re going to package them in different packaging as Barry went over. And there is a National Drug Code for the packaged product, if you take the initial portion of the National Drug Code this is what’s called a labeler code, and this identifies, this is what has been given to the manufacturer to create their national drug codes. But if you take the labeler code and the next section that is a code for the drug product. If you take all three sections of the National Drug Code that is the code for the packaged product. So here the drug product is those first two sections and the packaged product is the entire code.
Now we’ve been having some problems with drug listing, the way we currently do drug listing. It is an inventory for, or is intended to be an inventory for all the drugs marketed in the United States with a National Drug Code but there are some problems with that because the manufacturers currently are allowed to assign the National Drug Code without the control. We do have in our regulations published the rules on how you apply the National Drug Code but we find that numbers are being assigned to non-drug products. We find that numbers are being changed without changes in products, and numbers are being reused, even though the last piece is allowable by our current rules. We also find that the listing information is not the best quality, that there’s many times we get incomplete listing information, that the companies will come back to us and not tell us all the details about the product including the ingredients and the appearance, or that the information is inaccurate, or that there is no information provided. And finally there are instances where there’s inaccuracy in the drug entry.
And a lot of this is related to our current process where a company comes in and requests the labeler code from us and we return a labeler code to them. We do not track this in any way, that the companies, the only way we track it is that we only give out the labeler code once, so that it’s assigned to a company but companies can have more than one labeler code. Then the company uses the labeler code to assign their NDC numbers and they put their products to the market. They then are asked to come back to us after they’ve assigned an NDC and give us the listing information, and this is where we find a problem that the companies don’t come to us in a timely fashion, they do not give us the complete listing information, and various other aspects. So that’s our current process. We are looking now to improve the drug listing process so that we can improve our the National Drug Code and the information we collect and we want to make the National Drug Code a robust identifier and publish the information that we gather from drug listing, publish up to date comprehensive information. We have heard a lot of recommendations from our stakeholders, people who use the National Drug Code, of ways that we can improve the efficiency of monitoring and processing this information, to establish and enforce business rules on the drug listing. That’s something the FDA is currently working on, with working on to propose some regulation changes.
Another piece of the drug terminology is the structured product labeling and this is where we want to take the, we’ve been talking about the terminology for the products like the ingredients and the finished dosage form and drug package, but now we’re talking about the information that we current know about the products that are in our labels. The labeling right now is in a paper format and is very hard to access, but we’re looking to take the label and structure it into a more structured format so computers can find information and we’re looking at structuring certain labeling, elements in the label, so that computers can find that. And we’ve been working on a standard in HL-7 for structured product labeling based on the clinical document architecture. Basically we have divided up the sections of the labeling into blocks of text so each one of these sections would be labeled so that a computer can find each section, such as the dosage administration section, how supplied, contraindications, etc. These are all text pieces of the labeling. But then the second part is to define, also structure pieces of the labeling, such as things related to mechanism of action, the clinical effects, these are the adverse events, the drug interactions, the drug/food interactions, et cetera, drug class, indication, these are some of the things that we’re looking to structure as well as unique ingredient identifiers, the active ingredients, the finished dosage form, the drug product and the packaged product.
To get the information to the public we partnered with four different partners for this. One the manufacturers who provide us with the labeling changes. Then the FDA partners with them to review the labeling changes, approve them, or they come through the FDA. We’re working on a system to automate that process to improve the way that we review those changes. Then we pass those changes to the National Library of Medicine which will add additional published information to that through the Daily Med, which then gets passed to the health care suppliers, and those are the people that have currently taking the information and providing it to the public with their various access systems and various applications and software and publications, et cetera.
Currently these groups have to collect that information themselves but through this partnership we would be able to with the manufacturers collect the information directly from them as per our usual processes and then disseminate them through the National Library of Medicine so they can provide that to the stakeholders and customers and the public.
So this is the component, this is the collaborative effort that would come from, part of this is from the FDA which is the ingredients and the finished dosage form and drug product, and mechanism of action, adverse events, interim actions, clinical effects, indications, but also adding other things on there that might be outside the scope of FDA, such as the RX-Norm, which would be done in the National Library of Medicine, or some other therapeutic intents that might not be approved indications so that the extra information can be added to the product that goes to the public. The information would be in a computer readable format, it would be comprehensive including all the U.S. marketed products, U.S. marketed drug products. It would be reliable. The information is coming directly from the manufacturer and FDA. It would be up to date and changes would be on a daily basis, and it would be distributed free through the National Library of Medicine. So those are some of the qualities that Barry was discussing with drug terminology.
I just want to mention some things about drug devices. We have similar regulations and in our laws about listing and maintaining inventories of drug devices, but our uses of that and what we’ve heard at least from the stakeholders are different then what we’ve heard from drug. At this point we don’t really have all the details of what people want for drug devices and what’s the information that they want.
I’m just going to go over some of the things that we currently do, we have listing for devices, just like we do for drugs, though we don’t have the same type of information that’s been listed. We just take the classification name, the device, which is based on our regulations, we have a code for those classification names which you can find on our website at the device, Center for Device and Radiological Health, and we tie that to the brand and commonly used name. That’s the extent that we have listed for devices at this point.
There’s also something called the National Health Related Items Code. Back in the ‘70’s we provided certain labeler codes to device manufacturers so they can create National Drug Code type coding, which is called the National Health Related Items Code. Very few products have these codes, but this was set up as a way to provide codes to the devices. And then what we’ll hear later on this afternoon is about the global medical device nomenclature, which is an international effort for classification of medical devices that provides definitions and names for medical devices.
I just wanted to show you these things here is that there is potential for adding more information or getting better terminology for medical devices, because we have some of this process in place but we have not had a mandate on the outside at least that we know about of what kind of information people want to know about devices, so that might be something of interest for people on the outside and I know the Consolidated Health Informatics is looking into that for supplies, medical supplies and other issues.
So those are the areas I wanted to cover today so thank you for your attention.
MR. BLAIR: Thank you, Randy and thank you, Barry, and I’m going to turn it back to Dr. Cohn to facilitate the questions.
DR. COHN: Barry, why don’t we have you sit in front here since we’re going to be talking and asking you questions. I actually wanted to start out and obviously others from the table will be asking you questions and I’m sure we’ll be hearing some comments. As I listened to you I have trouble telling the tense sometimes of what you’re describing and I do know that sometimes I’m sure that not all of this is present or past tense in your descriptions so maybe you can just reflect for me for a moment what is here now, what is your vision, what is, can you help describe that.
DR. LEVIN: One of the problems is a lot of the information we have now, currently have, is not publicly available, that we use it internally for our own purposes and it was not publicly available, so that’s why there’s some confusion because we have certain information that we have that we’ve been using for many years internally that now we want to make publicly available and so we have to separate that from what we have, our confidential information, to separate what’s releasable and what’s not releasable. So for ingredient codes and for structures we currently have structures for the products that we regulate, the drug products, for everything that’s been approved, that’s what we currently have. We are in, what we’re doing now is taking no structures for the future to put them in the unique ingredient, the substance registration system, so that is a future piece where we would take all of our codes that we currently have for our internal codes and put them in the substance registration system and so that is a future piece that will, should be available very soon for our ingredients. This is for, so ingredient codes we have and we’re going to make them publicly available, that is going to be a future piece. That’s one.
The other piece is the National Drug Code piece, which is both the product code and the package code. We currently have both of those, those are in existence, the problem is that we feel that because the drug listing process is not ideal that it’s not comprehensive and not complete. We want to change both of those things and that’s why we want to work on our regulations to update that so we will have both of those complete, so that’s the future piece. We do have some of those in the present but we want to make that more comprehensive and complete, so that’s a future piece, the drug codes.
The finished dosage form, we currently have that internally again, that’s an internal code, we don’t have that externally and we’re going to be working on moving that to an external format when we are providing information in a structured format.
That brings up the labeling. Labeling, we currently have labeling for every product, it’s currently in paper. What we are working on now in the future is that the structured product labeling format is currently in HL-7, we have validated once, we’re validating again in the committee level, so we’re working on that. We’re working on the first phase of the structure product labeling, which will be both the sections of the labeling, which I showed you all the sections, and the basic listing type information, that’s all the ingredients, the route of administration, the dosage form, the different codes, the unique codes, the administration form codes, that will be coming. So that is coming, we’re looking at changing our regulations to allow us to do that, and those regulations have been proposed and should be final very soon. So a lot of this we currently have internally but it’s not external, to make it available externally those are the things that we’re doing for future. Does that address --
DR. COHN: Yes, I’d actually love to see that maybe on some sort of a communication, just because obviously you’ve sort of quickly run through things and I’m still sort of left with the --
MS. HUMPHREYS: What’s the definition of very soon?
DR. COHN: Yes, thank you, Betsy.
DR. LEVIN: Very soon is very difficult in the regulatory world --
MS. HUMPHREYS: No, I mean the stuff like the ingredients, that’s not regulation, right?
DR. LEVIN: No, the ingredients, right, the ingredients we have this system already done for that, it’s a matter of taking the codes and putting it into the system. By the end of this year we should have the first round of those ingredient codes. The other pieces where we, one for drug listing where we do actually have to change the regulations and that will take time to do. The more that you’re interested in having us do this, and the more you’re interested in that support, you would support our regulation changes the sooner those things would happen, but that’s just, regulations do go through the public comment and we’ll get feedback from there and it depends on what the comments are how they go.
DR. COHN: The Subcommittee has had some experience with the regulatory process, I can appreciate what you’re saying. Questions? Michael.
DR. FITZMAURICE: Randy, it seems to me that the biggest barrier is not having the regulatory authority to do what you want and get the information packaged just right, but yet the drug companies do sent it in in paper form. It would seem to me that you could have a little section that says let me type it in myself in a word processor, let me arrange it and put it into the forms that we think make sense, use some focus groups so that you gain some knowledge in this and then maybe persuade them to give it to you in electronic form, those that will. That’s just kind of a description of what could be done, but I have two questions and I’ll give them to you right away. What vocabulary is used for product information or package insert information? Does FDA plan to insist upon a particular vocabulary for describing what’s in the package insert information? And then secondly, do the drug manufacturers participate in the FDA HL-7 quest for a standard in this area?
DR. LEVIN: First thing you talked about the electronic labeling piece. Actually we proposed a rule to require the manufacturers to submit the labeling electronically to us. That was proposed, it was well received, we have addressed the comments and we’re looking for that rule to go final soon. That’s the first issue.
DR. FITZMAURICE: Will they give that to you voluntarily even without a regulation, in electronic form?
DR. LEVIN: There are some companies that do provide us electronically, but in order to do, and we have some experience then from the electronic labeling. But in order for us to do this right we need everyone to provide the labeling, we have thousands of labeling changes every year that we need them to provide us with the labeling in electronic format so it will work right. If we have to translate things over and retype them it would be just too burdensome. So right now we’re confident that the labeling, we’ll be able to get the labeling in electronically.
The HL-7, one of the other issues you talked about was HL-7, how if the manufacturers are cooperating with us at HL-7, there are few pharmaceutical companies that participate in HL-7 at this point. We do meet with the pharmaceutical companies and describe this information to, describe to them what’s going on in HL-7 and a few pharmaceutical companies do participate. But what we have done so far in HL-7 is talk about the standard for disseminating the labeling out to the community, which is not as much as what they’re involved in, the regulation would require them to, we would come up with a format for them to send in to us, so that will go through our guidance and they’ll be comment making --
DR. FITZMAURICE: So essentially they’re willing to accept whatever you come up with rather than trying to participate in the process through these kind of developing organizations.
DR. LEVIN: Well, we’re hoping that we will make it such that there would be something that would be easy and advantageous for them to submit it to us in that way.
DR. FITZMAURICE: They’re hoping that you read their minds, rather than telling you their minds.
DR. LEVIN: We have asked them to, or we have offered, or there is availability for them to become involved with HL-7, though they just have not come out in force to that and get involved.
The last thing about the vocabulary, that’s a much more tricky question. We do have vocabulary that we do require for labeling, for things like dosage form and there’s standards for that, names of products, the established and the brand names, we do regulate that, we do control that in labeling, so those things are controlled. The National Drug Code of course is controlled, but for terminology on other things like indication, adverse events and things like that at this point we have not come up with standards for vocabularies for that, though as we become, as we structure these elements we’re going to look for that and going to be moving forward toward that goal. So the first phase of the structured product labeling is more dealing with descriptive aspects of the drug and that we do have a lot of control terminology for -- and things like that, route of administration, that we do have a vocabulary for.
DR. COHN: I think Walter, Kepa, and then Clem.
DR. SUJANSKY: A few questions for Randy and one for Barry. I think I got a little confused when you were describing Daily Med, that process, and I wanted to make sure I understood correctly what comprised in its entirety Daily Med. One slide you showed Daily Med at top and then basically all of the contents of RX-Norm, NDF-RT, and the FDA additions to those, is that what Daily Med is?
DR. LEVIN: Well, Daily Med is, our portion of the Daily Med, the FDA portion related to the labeling and listing information, so this is the, so much of the things that are described in the NDF-RT come from labeling, are already in labeling, so that we, we’re looking as far as again for maintenance of that that we would provide that kind of information. So for example mechanism of action, that’s something that’s in labeling that can be provided from labeling, so just to get that sort of straightened out a little bit.
Then there are certain things that, so we have our structured product labeling that has all this information structured that we will provide to the National Library of Medicine, the concept of Daily Med though is that there could be other things that are not from the FDA labeling and listing that could be added to this, and that would be something as a separate piece that you can say this is the FDA labeling, the manufacturers labeling information, but here’s some extra information that comes from published literature, from something else, that would be something in addition to the structured product labeling.
DR. SUJANSKY: But the research that you described as Daily Med and that’s distributed by the information suppliers that you mentioned, that’s ultimately distributed by, that does not include and will not include RX-Norm and NDF-RT, is that correct?
DR. LEVIN: The Daily Med piece is the data as I view it is the information in a structured way, so it includes RX-Norm, includes the things from our labeling, includes the codes, and then the information suppliers though will use it in a way that they populate their systems so that they will be able to supply this information out to the public in different formats. So here for example Poison Control Center, they’re interested in knowing things about the appearance of a product because they get a lot of calls about someone took this product or I got this from the pharmacy, what is it type of thing. There’s going to be systems that will be able, you’ll be able to query on the identification and appearance piece of information and provide that information. Now they’ll be able to get that data from this feed, from the National Library of Medicine, which will have that data in it. Does that make sense?
DR. SUJANSKY: Yeah, maybe I can follow-up with the RX-Norm and then the other folks when they, I’m still a little confused about the exact contents of it, but I don’t want to take up too much time right now. The second question to you is you mentioned problems both with existing regulations not being complied with and absence of sufficient regulations, specifically with NDC system. What guarantee or how will the FDA ensure that any new regulations are complied with given that existing regulations aren’t being complied with?
DR. LEVIN: We’ve had a lot of input from the outside about ways that we could do that and that’s what we’re working on in our proposed rule, to have a system that would be more likely, not more likely but provide that we would be able to have a comprehensive list of the National Drug Codes, so we’re working on that from the information that we’ve, from what our experience that we currently have with the National Drug Code and with information we’ve heard from --
DR. SUJANSKY: Last question to you Randy is very mundane. The NDC code example you showed was a ten digit NDC code. Many folks use an 11 digit NDC code. From the FDA’s point of view what is the standard, if we’re going to have a standard and we’re going to use NDC’s and we’re going to require everyone to comply with this system, what is really going to be the standard NDC?
DR. LEVIN: Well, we just had this large discussion, I was at the NCPDP meeting and there was a lot of discussion about that. The NDC is ten digits, ok, no matter what people tell you, it’s ten digits, but what people do with that is add leading zeros and to normalize the code because it can come in three different patterns. Labor codes can be five digits or four digits, product codes can be three digits or four digits, package codes can be one or two digits. So to normalize it you have to add a leading zero in one of those three areas and that makes it the 11 digits, but when you go to a barcode it’s ten digits, that’s what the NDC is is a ten digit code.
DR. SUJANSKY: That will continue to be the FDA standard for the NDC code?
DR. LEVIN: That’s the things that we’re working on, when we come out with a proposed rule you’ll see what --
DR. SUJANSKY: I have a question for Barry, I know I’ve taken up a lot of time. Just a quick clarification on your comments regarding the requirements for drug terminology, and you mentioned specifically low cost. Were you referring to a standard drug terminology in specifying those requirements as opposed to a drug terminology, for example a proprietary drug terminology that may be created by one of the vendors to support their specific --
DR. BLUMENFELD: Yeah, I guess I can answer that from two perspectives. Certainly at Partners we pay, and actually pay rather dearly, for a drug terminology at this point because it is so critical to operation. However, we also know that in many cases, in some cases actually other systems that we’re working with because it is so expensive to purchase that proprietary terminology they don’t use it and of course that creates a challenge. So my feeling is that at least a core terminology, and one can debate what core means, but at least a core terminology should be available to the public at very low cost, and that core terminology might be enhanced, it might be specialized, it might any number of different things as a commercial venture, but at least we would have a common low cost core terminology.
DR. COHN: Kepa and then Clem.
DR. ZUBELDIA: First an observation and then a question. It’s interesting to hear that the NDC code is a ten digit code when HIPAA has adopted the NDC as an 11 digit code as published by the FDA, and the FDA doesn’t publish such. But the question is what are the rules or the standards for retiring or reusing NDC codes? Because I heard you say that the NDC codes are reusable. Are there rules for use?
DR. LEVIN: First, back to the ten digit, you’ll see from the responses that it is still a ten digit from the FDA, but going back to your question about the reuse, our current regulations on the rules on creating NDC, which the company manufacturers are supposed to follow, you’re allowed to reuse codes, so for a product that’s been discontinued, there are rules for that, if a product has been discontinued, that code has not been used for five years, then you go ahead and you can reuse that. Now whether people, how many codes are reused and how that falls out in reality, it’s one thing, but the rules allow it to happen.
DR. COHN: Barry I think you wanted to make a comment on that.
DR. BLUMENFELD: Well, just stating the obvious, that for many years the fact that there have been different formats to NDC’s, the fact that they are occasionally reused and so forth, has been certainly cause for significant concern in clinical settings. So for instance we’ve always made sure that we don’t use the NDC as an identifier in the clinical setting because otherwise you’d be faced possibly with a situation where you had saved information in a patient’s medical record about a drug, you call it up five years later or ten years later and you actually based on the NDC think that they were taking a different drug then they current are taking, or were taking. So many of the past issues in terms of reuse, in terms of possible duplications in NDC codes have caused significant concern in the clinical side. On the other hand, Randy’s fairly up front about that and the FDA is working to change that. I can’t stress enough, though, how important it is to change that.
DR. LEVIN: If I can make one more comment on that that we’ve heard both from HL-7, Stan with the Vocabulary Committee and other places about what makes a good code and those things so that we can consider that in our proposed rulemaking.
DR. COHN: Clem?
DR. MCDONALD: I had sort of some comments and some questions, and the comment is that this is great, I mean I’d like to cheer this on. And it’s kind of an example of all this information is there and all they’re saying is there’s a format to make the printed document, all they’re saying is give it to us in something we can read, this format, it’s not going to bother anyone I don’t think, and then we can reuse it in so many rich ways, the community, the industry, etc., will invent all kinds of neat stuff. So I just kind of want, this is tremendous because a relatively small amount of effort although it must seem monumental, the leverage is so great once this stuff can get connected.
Regarding the NDC, one other question about it, one of the challenges we face trying to map this to that, because the NDC is nice because at least it’s there, and in fact I think the reason why the commercial, I mean the community pharmacy industry is so standardized I think is largely because they had this code they could hook onto, maybe other using it, but that really was an enabler. But in the inpatient setting NDC codes are mostly not in anybody’s, I mean 30 percent of our NDC codes we have in our pharmacy are not in anybody’s database because they’re these so called contract codes or something, I don’t know what they really are. But they invent them for good business uses, they’re just not useful. Can you speak to those things where they just kind of invent these codes for each contract?
DR. LEVIN: That companies invent codes? Because the company, because the manufacturer has the labeler, the first five digits or four or five digits of the code, they can do whatever they want with the rest without fear that someone else is going to use that same number. So this is where they, from what we’ve heard that they need codes for reimbursement, for whatever they need, they need something so that they make up numbers and then what happens is they start assigning what looks like an NDC to things that are not supposed to have NDC’s, and that would be dietary supplements, that might be used in --
DR. MCDONALD: These are more often like ampicillan but it’s a special deal.
DR. LEVIN: If ampicillan, ampicillan has a National Drug Code, is supposed to have a National Drug Code.
DR. MCDONALD: I understand. We take real drug codes and try to match them up to say First Data, or I think it was Medi-Span, they’re not in there.
DR. LEVIN: That’s one of the problems that we see, too, that they’re not in our database, so it’s not that it’s not comprehensive, our database is not comprehensive. What we’re looking at in the proposed rule is to fix those problems.
DR. MCDONALD: Well, I want to bring that up, I just wanted to basically make you aware of an issue, but it probably is something the industry needs now because they’re using it, and is there a way to accommodate the industry need without blocking the goals of the standardization or sort of making contract numbers that are special?
DR. LEVIN: As far as getting it, if it’s a drug, to get the National Drug Code, we would do nothing, we want to, they would have to have that, we want them to have that and we want to put that in the database so you will not, you’ll always find something associated with that National Drug Code. That’s two things that the areas where we have heard problems with are people want a code for things that are not drugs, and they want codes for dietary supplements, one, and two, for devices or supplies, and we need to hear, if that’s a need we need to hear about that so that we understand that that’s something that needs to be done and find out ways that we can help to address that. As I was pointing out for devices, we do have mechanisms in place to do things like that. For dietary supplements it’s a little different but if there is incentives for that and if there’s a need out there we’d like to hear about that.
DR. MCDONALD: And the second question really is regarding what you mentioned as this route related drug, and I’d like to keep that alive and prominent because in ordering processes to find the best fit for the ordering you’d like to not have to over specify it. For example, and even in the outpatient, prescribe 500 mg., let the pharmacy decide what pill size to use depending on what they have is a great convenience and is a common way people write prescriptions. So we’ve talked a little bit about this but just to keep that other node in there alive, it gets complicated because there’s different, it’s like a fractal with different stair steps for different, I mean it’s not uniform how you really need it. But there is this thing out there in the ordering world that’s not quite the same as the inventory world, or what would look like the most elegant regular world. So I’d guess I’d push, I hope that you can keep pushing that part of the structure.
DR. LEVIN: Clem, I’ll keep pushing it.
DR. BLUMENFELD: I actually had a quick question for Randy. Randy, you mentioned the finished dosage form code and also the existence of imprint codes. Now imprint codes have some of the same problem as NDC’s, for people that are familiar with them, because they take into account the size and shape and color and so forth. It’s very possible and in fact does occur that the actual imprint part of it is duplicative and it’s sometimes difficult for people, if a patient hands them a pill and says what is this, doc, as commonly happens in clinical practice, it can be very difficult to ascertain what that pill actually is. Has there been any thought given to the idea that the finished dosage form code and the imprint code should in fact be the same code?
DR. LEVIN: We heard, in the early ‘90’s we changed our imprint regulations, in that time we said that the two solid dosage forms have to be distinguishable. But we did not come down and say that you had to put, it had to be something, an alphanumeric that was printed on the finished dosage form had to be the thing that would distinguish them. And we have heard comments that there should be something like that where the alphanumeric on there should be what distinguishes them, not just using alphanumeric with a color, shape and size, because now you can have just if you are the only one that makes that pill and you put your logo on there, then that makes it distinguishable from every other product because they’re not going to use your logo. But that’s very hard to put into a database to read and so we’ve heard issues along those lines and so that’s, we have heard that kind of problem. But one of the things that we have, that we don’t know, is what the extent of this problem. If we knew, if we had some more information about the extent of the problem and how it relates to safety issues and things like that it might be something that might drive a regulatory change more than knowing theoretical.
The other thing that we are not sure of is how much of this is related to things on the market that are not drugs but have a drug form, like a dietary supplement or a food that might be in the form of a drug, a pill or tablet or something like that that people are having trouble with those pieces. So there is a lot of dietary supplements out there that are not covered under this imprinting regulation. Is that where the issue is and should that be expanded to cover everything that has a -- in the shape of looks like it would be a medication? So those are some of the issues that we’ve heard and we did just change, well, just, maybe about ten years ago, changed our regulatory, regulations on imprinting but we are interested in that and again we need to hear comments from the outside about the issues that we’re facing with that and ways that we can address it.
DR. COHN: Clem, do you have a final question?
DR. MCDONALD: Yes, an imprint question, can you really fit the whole darn number on a pill? I mean is that a possibility?
DR. BLUMENFELD: I’m going to answer first, Randy might not agree with me. I think from a technology standpoint it’s certainly not a challenge, even the smallest pill out there with micro embossing and other things you should be able to fit some sort of code on it. However, there certainly are challenges on there and I think the pharmaceutical industry objects on that basis where anytime they change that sort of information it can be quite expensive for them in terms of changing the way they formulate the drug, codings, inks, you name it. On the other hand, I guess from the clinical perspective, I should consider mounting a funding campaign where every physician whose ever had a patient bring him a pill that they can’t identify should call the FDA to let them know that it really does happen.
DR. COHN: Randy, do you have a comment on that?
DR. BICKFORD: Carol Bickford, American Nurses Association. I have a question for the FDA in relation to a coverage for complementary and alternative therapies. Are herbals and some of those other components that are becoming mainstream for care delivery, is that encompassed in your coding system? Is it another component of your manufactured pharmaceuticals? How does that fit in your initiatives?
DR. LEVIN: The things that I’ve been talking about are what we consider drugs by our regulations, so that would not take into account things that are not drugs, so a lot of the botanicals and other products that you’re talking about are not drugs from the FDA’s regulations. But again, we regulate those products, those products you’re describing are regulated by the FDA and it’s the issue about whether this is a need, we need to hear that if this is a need that you want those things included or they should be, there are issues around those products. Those are issues that you should bring to us and let us know about that. But those products are regulated by the FDA but they are not part of the products what I’ve been described as related to drugs that we consider drugs.
DR. COHN: Final comments? I think we need to take a break at this point, actually running about 20 minutes late which I’m not surprised at. So we will take a 15 minute break and then we will come back together. I just want to thank our speakers for a very interesting first session.
[Brief break.]
DR. COHN: This microphone certainly works. I want to apologize to everybody, we are running significantly late. Let’s see if we can make up the time but we probably won’t, but we’ll survive anyway. Anyway our next panel is Stuart Nelson and Steven Brown, I think Stuart you are leading off the discussion.
Agenda Item: Panel of Public Sector Drug Terminology Developers - RX-Norm - Dr. Nelson
DR. NELSON: Thank you. I’m supposed to talk about RX-Norm and in fact, answer some very specific questions that Steve Steindel sent me and I’m going to try to address that during these comments.
RX-Norm is a project within the UMLS to address the problem of naming of clinical drugs. One of the important things to do is try and define exactly what is the scope and we are talking about putting names to a standards format of all approved drugs in the United States, that includes prescription drugs as well as the over the counters, other clinical drugs that showed up in the UMLS for one reason or another, including the Read codes. It includes the names from commercial sources and as an added feature this year we’re putting in or addressing the problem since we’ve begun to address what happens when drugs are withdrawn from the market, or there are ingredient changes and reformulations.
An example of this might be contact tablets that at one point had phenypropanolamine in them and no longer have even though they’re still sold with the same name. So we have undertaken some steps to address that so that retrospective if one knows that someone received a given medication coded in a certain way at a certain time what exactly what those ingredients were. So this is a concern that we are facing.
Just a historical note, the clinical drug interoperability is really what we are concerned about and it started, the whole effort grew out of discussion at HL-7 where most of the commercial vendors were present and started talking about how they were going to achieve a common model and then interoperability at the level of what was prescribed. And the first step that we did is we said ok, we’re going to call anything that’s an ingredient --
MR. BLAIR: Stuart, could I just mention something? Commercial vendors, I think many of us understand what commercial vendors means, but there’s been a little bit of confusion by some folks, so could you just identify what you’re meaning by commercial vendors?
DR. NELSON: The commercial vendors, pharmacy knowledge bases which were at that time Multum, Micromedix, Medi-Span, First Data Bank. All of them had vocabularies at one level or another included in the UMLS. Noting that on the first step we did in the UMLS was to identify what was a clinical drug, which was defined for us as a drug, an ingredient name with either a dose or a form or both. And when we did that and we ended up with 81,000 clinical drugs in the UMLS. There was only 10,000 approved by the FDA, we thought we probably had a significant problem with missed synonymy, and that are traditional lexical methods of recognizing missed synonymy were not working.
So then we proceeded to define this RX-Norm form that we’re going to discuss, and then out of that also has arisen an effort that’s one of interagency cooperation between the VA, the FDA, and the National Library of Medicine, and this cooperation between agencies has been done by defining areas of responsibility and working to share information and cooperate with one another in a way that we think is a model of how the government should work. So you see this slide from Randy and I think, I suspect you’ll see it again from Steve, that kind of illustrates this extent of cooperative efforts so that we will start at the top of the slide, in the center we’ll see that the ingredient codes, the structural codes, the chemical structure from ingredients on down to the packaged NDC drug with information about indications, dose forms, kinetics, therapeutic intent, all of these different areas of information need to be addressed by this interagency cooperative effort. The RX-Norm portion of that responsibility is really the linking from the active ingredient to the clinical drug to the dosage form, and as an additional feature we’re now having some links to the NDC codes.
The approach was to develop a standardized notation for ingredients, strength and dose forms, a standard list of dose forms, and a standard method for representing strength. So the plan was to create a normalized form for these clinical drugs and establish concepts in the UMLS for the ingredients, the drug components, which we called the ingredient and the strength, the dose forms, and the drug formulations, which were the component and the dose form, and to establish formal, defining what relationships between those concepts.
Our goal at that time was to have that every concept with a semantic type clinical drug from whatever source, but especially from the VA, from Multum, Micromedix, First Data Bank, Medi-Span, and now SNOMED-CT related to an RX-Norm name, with a set of possible relationships. And I’ve shown this example before of Cetirizine as being a form of, the base form of Cetirizine hydrochloride, which is traded under the name of Zyrtec, that there’s component, Cetirizine 5 mg. oral tablet of Cetirizine 5 mg., the formalized generic name be the Cetirizine 5 mg. oral tablet, the formalized brand name is Cetirizine 5 mg. oral tablets [Zyrtec] and then it has a dose form of an oral tablet. And that Cetirizine tablets have an is/or relationship to the Cetirizine 5 mg. oral tablet.
What we don’t see here that, so where are we today, there’s a question about how complete we are. In the latest release of the 2003 AB UMLS we have 4,647 ingredients. We have 16,000 components, that is ingredient plus strength, we have 20,000 generic drugs. Interesting that this was much higher then the estimate of 10,000 approved drugs by the FDA, and 11,000 branded drugs. The formal definitions of the relationships between them is a process that’s ongoing, but at this point we can see that we have 11,000 relationships of trade names to generic names, 32,000 ingredient relationships, 1,000 relationships between a specific form and a base form of a drug, 30,000 relationships to the dose form, so that one could argue that the RX-Norm, for the areas that it has addressed is remarkably complete.
What are the newer features? Well, in this version of the UMLS what we have done is we have downloaded from the FDA their publicly available listing of NDC codes and parsed them out and put the drug names for those into the normalized form, and then taken those NDC codes that are there and used them and listed them in the UMLS as an attribute of the drug name. We are not trying to represent NDC codes as separate concepts in the UMLS because we think right now it’s a little too hard to track, as Randy might have observed to you.
We have identified certain areas where we’re problematic. The Orthonovum 77728 tablets, known and loved by all, is the dispenser of oral contraceptives, or sequentially oral contraceptives. Now several of the drug vocabularies ended up trying to fit things into their model by defining a tri-phasic tablet, which was a tablet whose component changed according to where you were in the sequence, and we chose not to follow that and said well, these are actually dispensed as in a device which dispenses them and lets call these things drug delivery devices.
There are also things which have lots of medications on them which where the patient does not necessarily get a very specific dose of that medication, like a medicated swab, and we wanted to call those drug delivery devices rather than calling them drugs themselves. And then formalizing the relationship, these contain, these drug delivery devices contain the drugs that they, and have a listing for the drugs and formalized names for those drugs.
We have also devised an algorithm to come up with drug form names so that we can have a set of normalized drug form names. So that for example we could have a name for doxycycline capsules and relate that to all the strengths of doxycyclene capsules that we could have. Now I know that Dr. McDonald and Dr. Blumenfeld both feel that routed generics are a very important thing to have at hospitals and I’m very sympathetic to that. I suspect that we can algorithmically derive routed generics out of a drug form name. I think that the way the drug forms are essentially with very little, pretty predictable about what the route is, so that for the possibility of use in an order entry system where you want to have that sort of thing that might be information that could be available. And I think that in future versions you will probably see more of that.
DR. MCDONALD: You showed something called Cetirizine tablets, it looks to me like not a generic but I didn’t see it in any of your tables. Is that the real thing, or that comes pretty close to --
DR. NELSON: That’s where we’re going to be, we are generating these names algorithmically, yes. Once we have the formalized name at the bottom, Cetirizine hydrochloride 5 mg. tablet, we’re deriving algorithmically the Cetirizine tablets and those are going to be separate concepts in the UMLS. And I think the routed generic can come out of that.
DR. MCDONALD: That’s very close.
DR. NELSON: It’s very close, yes.
What does it mean to be complete with a drug vocabulary? We’re dealing in a world where there are about 4,000 NDC codes a month that are changed, there are about 30 new molecular entities a year. I think that anybody who talks about a static picture of completion is kidding themselves, it’s kind of the Heisenberg uncertainty principle, you can say where you were yesterday. I think what we need to do is we need to get the NDC codes faster, right now they’re being posted on that website quarterly. I think if you talk about use in a clinical system you need weekly or better availability of all the drug information, so in order to do that you have to have good reliable data feeds, we are looking to the FDA for some assistance in that regard.
We have some other areas that we see are problematic, contrast media look a lot and act a lot like drugs they’re naming, but are not quite so easy to deal with. I mean what’s the dose, what’s the concentration of a barium sulfate solution. Multi ingredient over the counters are, medications are problematic just because the names are so long and difficult and it’s so hard to track all of their ingredients at the present time, but I see these are problematic areas that are areas that need to be addressed, but it’s what we’re working on right now.
Would I recommend RX-Norm for use in the computerized order entry system today? I’d say honestly it’s not tested yet. It needs a lot of feedback. Every vocabulary improves with feedback, it needs feedback, and we need much more involvement with developers. We need better documentation. And when you’re talking about how information models and terminology models work together, because they have to work together, you have to have a system model for how the whole thing works together, when you do that what you find out is that there are always areas that are problematic. And the problematic areas for the most part can be dealt with by setting up a set of business rules and documenting them, so that everybody knows exactly what’s being done, and how if not why, at least what the decision was about how things were dealt with so it’s predictable.
I think the VA is standing ready to become one of the first test environments for the suitability of RX-Norm for order entry, and I observe that reimbursement is often based on NDC codes and I think that’s why we put the information in passively but hoping that someday it will be a lot more complete. By the way, out of around 80,000 NDC codes from that FDA site that we were able to parse and looked like drugs, we got 79,000 of them in to the RX-Norm form, so we did, got a pretty good yield out of that.
Another thing that we addressed is when we’re talking about standardized clinical vocabularies, how do they interlock, where do they hand off responsibility with the successful negotiations and contract with SNOMED-CT, which we see as an important clinical vocabulary, where are the logical areas for the handoff of responsibility from SNOMED-CT to RX-Norm and similarly from SNOMED-CT to LOINC? We need to avoid overlapping efforts. Duplicative efforts just waste time and energy. And we need to prepare for dealing with new areas, one of the new areas that it seems like it’s coming up are these custom designed biological preparations that are approved by the FDA where it’s a vaccine but it’s based on something about the patient and the patient’s genetics, or the patient’s tumor or whatever. How do you deal with those? Who is going to deal with them? How does it work in the model? When you sit down and start talking about having a number of clinical vocabularies that are standard, one of the things you have to be concerned about is what’s the process for deciding who’s going to do what, and I think that that’s something that we’re going to need to be addressing. Well, we’ve already started addressing that.
Randy gave you his view of the Daily Med, this is kind of my view of the Daily Med from the NLM side, what it is that we want to have. I think one of the things we see that we want to do is we’re going to be able to disseminate and distribute all this different kinds of information. We want to have the RX-Norm names to support interoperation among vocabularies, possibly to support order entry. We want to have all the ingredients. Most of the ingredients are already in the medical subject headings, and that provides a link into the literature, Medline. Many of you are probably not aware of the chem ID system at the National Library of Medicine which offers a way to search through structures, it has structures of over 100,000 compounds of interest that we would like to be able, and one of the interesting features about that it allows sub-structure searching, and I think those who get very involved in chemical names of things understand that chemical naming is problematic, it’s difficult. Being able to search on structures and being able to do things like sub-structure searching is very interesting and I think that the system that the FDA is building for the unique ingredient identifier will support interoperation there.
We sort of expected in the Daily Med to have NDF-RT information. The FDA when it approves a label for release we’d like to have that. We’d like to have the NDC codes, to make all that available at one site.
And I think that’s all my comments.
DR. COHN: Stuart, thank you. Steve, I think you’re on.
Agenda Item: Panel of Public Sector Drug Terminology Developers - NDF-RT - Dr. Brown
DR. BROWN: What I thought I’d do is give you an overview of what we’ve done with NDF-RT, a little bit of perspective of where we’ve come from and where we intend to go with it, and also try to incorporate answers to the questions that we delivered in written format as we go and if it’s not sufficient then just go ahead and feel free to ask those questions directly.
What we started with with NDF-RT was this thing we called our National Drug File. The National Drug File is a centrally maintained sort of formulary/terminology that’s used across VA, depending upon how you count medical centers with mergers and all, roughly 103 major medical centers, probably 130ish implementations of our information system, all mapping local codes to our National Drug File, largely for the purposes of decision support, so sort of class based decision support, and also for transmission to our mail-out pharmacies.
So it’s a way to take whatever happens to be in the local drug name, and since there are only seven mail-out pharmacies a number of medical centers would have to use the same pharmacy so it’s a normalization of the name for that. And that happened roughly 75 million times in 2002, so it’s out there and it’s working.
One of the issues we have with NDF is every drug is in one drug class, and VA’s drug classes are pretty similar to the drug classes that you might have seen in just about anyplace else, sort of a mixture of things, like we have a drug class of anti-hypertensives, we have a drug class of ace inhibitors, we have a drug class of beta blockers. There are things that have sprung up over time as being useful for whatever reason, have been instantiated in our National Drug File. Although it’s possible I suppose to have more than one drug class per drug at the end of the day what we have is one. So we might get to know that something is a beta blocker but then we don’t get to know it’s an anti-hypertensive, for instance, with the current NDF system. That said, 75 million prescriptions and having it working day in and day out is good. It is coordinated and versioned and all that kind of good thing.
The areas that we felt we might improve on by this project, really we’re trying to do a better job in the arenas of patient safety and clinical decision support, and as I mentioned it doesn’t make a heck of a lot of sense to know that something is an anti-hypertensive but not a beta blocker, I mean there’s a whole lot more we could do with that kind of information, and that’s really the main driving force. In addition we’re looking at trying to reduce our operational costs and pains of terminology anyway, so we’re looking for by formalizing to look to see if there’s maintenance, reduction, if we can algorithmically map between sites or our partners, for instance, by using a formal approach. And interoperation for us is becoming increasingly important, and as I see they’ll be now a transitional drug benefit, so guys who show up with medications from their outside providers will for some period of time we’ll honor those prescriptions, so it makes an even more pressing case for this type of interoperation.
What we did was we took an ingredient centered approach using description logics where we wanted to spend our modeling effort on ingredients and then inherit that up through for the biggest bang we could get out of that. We’ve taken a clinical perspective and also a deliberately collaborative perspective so that we can try to reuse others efforts and share what we’ve done.
So examples of the kinds of things we’re looking for, don’t give drug X with an MAO inhibitor or an organic nitrate, which I took from TV, it’s an interesting source of drug information. But I also then included with other negative inotropes to try and give the idea that sometimes these warnings come through for things that we may have had in a single hierarchy, but ideas of chemical structure or physiologic effect or mechanisms of action, that really do from our view need to be teased out rather than being all glommed together.
Another issue that we anticipated and has come up even faster then we expected was the connection of drugs for conditions, specifically in VA for service connected conditions, which are depending upon the year and level of funding, a different class of animal, but also for just linking drugs to problems in general, which is hot and heavy right now. To me it was a bolt out of the blue but there you go.
Where we’re using reference terminology is in the sense that I think Heath and Kent referred to back in ’97, where it’s computer usable definitions for aggregation and retrieval and not just something we refer to.
The process we followed has been one of using use cases sort of as a lightening rod for reality testing, otherwise you could end up modeling forever and get nothing done. So use cases, semantic models that are driven by the use cases, reference taxonomy creation, about defining the terms using those reference taxonomies, first in an algorithmic way and then actually with human expert review. So you’ve seen this before and you’ll see it again.
And the areas that we’re working with primarily in NDF-RT are the blue triangles, and those triangles are meant to represent hierarchies, and how we’ve attached hierarchies at different levels to try to maximize our modeling effort and deliver some of the smarter alerts and different aggregations and slicing and dicing of our data. Semantic model has been created, we created the reference taxonomies, for the most part that you see in blue it’s loaded with NDF, I’ve told you about that before, we algorithmically initialized our term definitions, we’ve just finished a pass of clinical form D’s going through and reviewing those definitions, getting rid of things that weren’t quite right or weren’t specific enough, adding some that weren’t in there. They did a really fantastic job -- that just finished up end of June. We did a refresh with another version of NDF so we had to load it with NDF to get going and then did the modeling, so an NDF refresh in July and also we exercised our new drug transaction approach and did a refresh from FDA in July also and pulled in some 23,000 new medications, and that was to get us a better slice of information, not just what’s in NDF, but also to try to start gearing the machinery for an automated handoff from FDA through NLM to us, to anyone but we’re mostly interested in to us, as well as sort of automated ways for authoritative input from whatever sources.
So as of the July version we’re at about 108,000 NDC level concepts. Orderable products around 11,000, and that’s basically the RX-Norm SCD level. Drug components, the SCDC level, again around 10,500, 11,000, and 4200 active ingredients.
This is a breakdown of some of the modeling work that got done, so the physiologic effect hierarchy that we created has 1664 elements in it, and when you percolate that through that ends up being around 120,000 assignments of physiologic effects. Mechanism of action, again about the same numbers although the hierarchy is relatively smaller. We used MESH initially as our basis for disease names, and we’ll probably use CT as that becomes available to us, we did MeSH because we had MESH. So the named roles that we have involving disease states are may treat, may prevent, may diagnose, or induces, and you can see the counts for those kinds of role assignments.
Dose forms are based on HL-7 hierarchies and kinetics is somewhat underdeveloped at this point, it’s mostly whether it’s liver or kidney excretions, we tend to do more with that and I’ll talk about that in a bit. There are a whole bunch of other properties that aren’t roles in our formal sense. I didn’t want to list all 39 of them but things like NDC’s, the MESH definitions which gives us links to literature, VA drug classes we kept because we have software that’s built on it that needs that sort of thing.
These are just two screen shots, or pieces of screen shots from the Apelon tool. The one on the left is meant to show some of the properties that might be assigned, so we have information from FDA, we have the NDC’s for instance, we have a VA search name, these kinds of things and that’s sort of in the purple. And then some of the modeling that’s been done by the clin form D’s is over in the right hand window, and you can see we have mechanisms of action, physiologic effects, and a bunch of things that you might want to use to plan it all, or you could use to plan it all to treat somebody for. I don’t know if you can read it or not, it might be a little too small.
Regarding NDF-RT completeness, as part of the VA’s ERT strategy we’re going to be doing formal IVNV’s, so we have sort of contractors to do the work, and we’re going to have other contractors go around behind them and check and make sure that they’ve done the right thing. So just as a general process and we’ll learn from it what we learn from it and fix up what we need to fix up.
I wasn’t here last time when this conversation took place but there was some issue about oncology drugs that came up, and so I talked to the guys from Mayo about that coverage and asked them what had gone on, I had no idea. I got basically the data sheet from them. Based on the new update basically admit there are some trade names that weren’t in the version they had used, which is seeded well over a year ago, a year and a half ago data, and so we’ve got obviously a whole bunch of those trade names in now and it’s upwards of 97 percent by my calculations and I’m going to ask them to rerun that, but we got a lot more trade names, not largely but some with the FDA input through the new drug transaction mechanism. But the important thing is we’ll be assessing completeness with an independent evaluation.
So where is VA going and how does this fit in with what we’re doing? NDF-RT is becoming merged in with our initial part of VA’s enterprise terminology services, so there’s an effort in VA to formalize where necessary or where valuable anyway terminology and have abstracted terminology services that feed our reengineered applications. We’ve had enough problems with variation from site to site and difficulties with aggregation, so this is a major effort that’s underway, and NDF-RT is on a first year deliverable path for that. It also is fitting into our reengineering initiatives and I’m sure many of you have heard Gary talk about HealthePeople and how we like to share things as well, so it fits in at all three levels.
DR. LINCOLN: So a little about the VHA Enterprise Terminology Service. It’s a national initiative to standardize our terminology content and services across VHA, and with some recent developments in our health data repository project, the scope of Enterprise Terminology Services is even bigger then it was six months ago. The Enterprise Reference Terminology, which as Steve mentioned starts with NDF-RT, is to have a formalized content and include a number of domain areas. In year one, this year, we’re developing NDF-RT, continuing to develop NDF-RT, working on our laboratory results using LOINC as a foundation, and our document titles for our clinical documents, all of the notes that clinicians write in the VA are written on the computer and currently they’re titled in an idiosyncratic way and we’re using the LOINC and HL-7 clinical document architecture to normalize those titles before we start diving down into the structure of the documents below the title level.
The ETS, or Enterprise Terminology Services, has been very sufficiently resourced and is currently budgeted through the year of 2007. It’s based in the VA Office of Information under Dr. Rob Claudner(?) and specifically in systems design and development under Dan Bishop and Cindy Kindred. The main effort is located in Salt Lake City, Steve is the architectural co-director of the ERT project through the VA architect’s office based in Albany, and I’m the sort of the OI co-director of the ETS. And I think that’s enough about ETS, that’s not the primary focus of today’s talk.
DR. BROWN: So basically with NDF-RT we’re on the hook for delivering this year and we’re supporting, this year of HDR, which is sort of OI’s biggest project to try and have a national database that can then aggregate all data from all these different sites. So we’re on the hook for this and it will be delivered on time.
So the other things that are going on in VA are pharmacy reengineering project, so the code that we’re using for all our prescribing and all that is in the process of being rewritten, we’re in the design phases for that. As a matter of fact next week we have a meeting with them face to face for how is pharmacy reengineering, ERT, NDF-RT going to all work together. And we can’t just change from NDF to NDF-RT, I mean there has to be a process, this is a live operation, and so that’s the rewrite and reintegration is on the books right now, that’s an ongoing project. Additionally we suppose they’ll be a reengineering with CPRS, that’s sort of some of the decision support angle. And the integrated billing as I was alluding to with the hooking up problems to ordered medications, that’s the one that sort of came out of the blue at us.
DR. LINCOLN: That’s very important by the way because the billing office in VA has this dream that somehow clinicians would enter the reason for every order they write in the electronic record, which they’ve received feedback that that might be an over ambitious goal and that possibly they should try to get some decision support using enterprise terminology and NDF-RT to assist that.
DR. BROWN: Given the number of prescriptions we write we figured that if it were 15 seconds per it would be 100 FDA.
DR. LINCOLN: We actually did the calculation, if it was 15 seconds to attach a problem to every order, the number of orders written, it would be 100 FTE in the VA.
DR. BROWN: We’re hopeful that some of these may treats may be able to automatically load up some of that for the docs and save them the trouble.
In terms of fitting into the bigger picture within the government, you’ve heard how VA and NLM and FDA are playing nice together, NCI is also playing nice with us. We’re working with them, DOD, and NIGMS has also had a role to play.
DR. COHN: What is NIGMS?
DR. BROWN: National Institute for General Medical Sciences.
DR. COHN: Thank you.
DR. BROWN: And you’ve probably seen some of the goals but basically what we think is if we can work together to integrate the kinds of data that each of us has and put it in a useful format and make it available to the public, that’s a good plan. That is sort of the short version of this, that the public has paid for this information and to the extent that we have it we’d like to make it available. There is RX-Norm, NDF-RT, and the overall sort of what we view as the potential end game, all nicely fitting together with pretty clearly delineated areas of who’s doing what.
In terms of other examples of collaboration, right now with NCI we’re working to integrate our formal drug models, so NCI has some ideas about using description logic roles, and we have some ideas about roles, and for the most part they fit reasonably well, they have, it’s a lot more basic science bent to it and an investigational bent to it. That’s okay. So those will be integrated, we have an MLU with them, we’d like to get data about anti-cancer therapies from them and if they want to use NDF-RT for basic drugs so they don’t have to maintain it, great.
Regarding NIGMS, as you heard Diane Oliver last time testify is that NDF-RT is being used by PharmGKB, Chris Shoots(?) spearheaded an effort to get NIGMS funding for Mayo and Vanderbilt, and to look at how we might use NDF-RT and extending the NDF-RT to reflect pharmacogenomics data and advances. We’ve been busying away on that and there’s some nice extensions, we’ve learned some lessons along the way, and we’ll see how that gets packaged up as a final deliverable. I know Chris is working hard to organize all that and he did a great job. And DOD also will be contributing information modeling to the overall effort. There are probably other examples you could come up with but these are three big ones.
So I guess what I should finish up with is we are trying to work with others to establish this sort of pipeline of information coming from the manufacturers through the FDA so that we can get information rapidly, disseminate it through our ERT mechanisms out to the field rather than having these long cycles of updates, which with NDF now on the order of months. I mean we really do respect the Daily Med and have hopefully daily updates and not make people discover new things at each site independently but to let the first one find it, provide a rapid update, and distribute it out to the field without folks even knowing at most sites that they even missed this medication yet.
So we have some pretty I guess difficult goals to achieve just in the general for updates to terminology but we feel that’s necessary if we want to have comparable data across our sites.
DR. LINCOLN: The only other thing would be that as part of the larger Enterprise Terminology Services effort, VA is strongly committed to interlocking SNOMED, NDF-RT, LOINC, and the HL-7, CDA, and other sources to try to get away from our local variation that we’ve had for a long time. The HDR strategy that we’ve adopted won’t really allow that anymore and I think there’s a firm commitment both at the Rob Claudner level and at the OI level to standardize office of information level, the software design and development people.
DR. COHN: Thank you. Questions? I know Jeff has a question, I have a question, also. Looks like Stan has a question.
MR. BLAIR: Stuart and Steve and Mike Lincoln, thank you very much for the testimony. I am a little confused by the terminology used. Not being a clinician I have an excuse, see? And here’s my struggle. RX-Norm, the description of RX-Norm, seems to fit very closely what I perceive and understand as a reference terminology. And when I read the documentation for NDF-RT it seems to be very similar to what the commercial vendors refer to as drug knowledge bases. Am I completely off-base or is my perceptions correct and the names just don’t fit, or, help me out with this.
DR. BROWN: Well, there are a couple issues there. First, regarding reference terminology using Heath and Kent’s definition I think it is. It depends on the definitional reference terminology, but using the one I gave with formal definitions for aggregation and retrieval, yes, it’s reference terminology. The issue of knowledge base versus terminology model versus information model is one that you can go around and around on forever. There’s probably never an answer to that. We’ve decided to be sort of use case driven on it and might it be some in the knowledge base, well, yes, it might be, depending upon your point of view. As it stands now we’ve broken down our existing drug classes in order to help us slice and dice better -- so from that perspective probably not. And at the end of the day I’m not sure it matters. It will help us meet our needs.
DR. COHN: Stuart, did you have a comment?
DR. NELSON: I think that, I’ve had many of the same concerns that Jeff has had at times but I think you can say that both RX-Norm, RX-Norm is a reference terminology in the sense that it has a formal definition, it has a reference taxonomy that it refers to, which is the taxonomy of dose forms, and it has lots and lots and lots of primitives --
MR. BLAIR: And it does the mapping.
DR. NELSON: And it does the mapping, it has formal definitions and formal relationships, and that’s really what it’s intended to do. It does not represent very much other than that, I mean there are many ways that taxonomies can look at medications, for example, Kent is standing up and I’m sure Kent is going to say some stuff, too. When you start talking about what are the defining properties of a certain kind of medication, an antibiotic for example, you’re already talking about it’s therapeutic use, so it’s reasonable to talk about drugs with other viewpoints in mind than just saying that an ingredient is a primitive or we don’t know anything at all about it in our terminology. I think there’s much more we can say, that’s what I think the NDF-RT does, I know SNOMED has lots of information about classes of medications. Kent?
DR. SPACKMAN: Kent Spackman, chair of the SNOMED Editorial Board. I think in all of the work that we’ve done we’ve tried to differentiate between what we call terminological knowledge, and that’s the knowledge that’s true about a term, it’s always and necessarily true about a term, versus assertional knowledge, which may change depending on regulation or realm or over time these things may change. So from my perspective all of the information that Steven presented that talks about may be, may be used for, that’s all a social knowledge, and from the way we use the term in SNOMED that’s not terminological knowledge and it’s outside the scope of what we would include in SNOMED as reference terminology. Now whether you want to use reference terminology or not to differentiate those I think there’s a real value in differentiating between the aspects of a concept that are always and necessarily true about it, and those things that are assertions about the particular uses and context in which that concept is used. So I think those are really valuable pieces of information.
I would agree with what Jeff suggested earlier, that that’s knowledge based, that that’s assertional knowledge, and that the core piece of RX-Norm is really what I would call the reference terminology part of it. If you don’t want to use that term for it at least I think it’s useful to make a distinction between those two different kinds of knowledge, from SNOMED’s perspective because that forms the boundary over which we don’t pass. We don’t start to model all of the may be used for things as terminologically always true about a concept.
DR. NELSON: Can I cross-examine you for a moment?
DR. SPACKMAN: Sure.
DR. NELSON: You do assert that tetracycline is an antibiotic.
DR. SPACKMAN: Well, sure, because it’s --
DR. NELSON: Is it that, some people might consider that assertional.
DR. SPACKMAN: So we can have a debate about whether what we’ve got there now follows those rules vigorously, but that’s a different question. But we do have --
DR. NELSON: It’s a tricky thing and I think Barry put it very nicely when he said you have to coordinate your information models and your terminology models because this is exactly the type of thing that you can get into.
DR. SPACKMAN: If tetracycline kills organisms and thereby, and always unnecessarily kills those organisms, then that’s definitional. So it is in some sense --
MR. BLAIR: I really don’t need a precise definition, and there could be differences and that’s fine, I just wanted to make sure that I didn’t misunderstand these issues and apparently there could be some debate on the definitions and that could be separate.
DR. BROWN: There can be endless debate on these issues and at the end of the day you have to do what your system needs.
DR. SPACKMAN: I think it really is useful to make that distinction, to make the distinction between what’s assertional and what’s not, at least from a terminology development standpoint.
DR. COHN: Okay, Ken, thank you. Jeff, do you have any other questions? I think I will pass with my question and that’s only because I think much of it was included in your question. Do you have a comment? Come up to the microphone. And then we’ll move on to Stan’s question.
DR. BLUMENFELD: Really two quick comments, and one question. The one question is around NDC’s and I’m wondering whether you’ve looked or plan to look at also mapping or in some way assessing a base of NDC’s that are available from the commercial vendors. I know, and I’ll probably get corrected on this, but I know in our database at Partners I think we have something like 150,000 NDC’s, so clearly there’s a broader universe of NDC’s out there.
DR. NELSON: When I get them from the commercial vendors I’ll put them in. And I actually, I view that as really I would like to do that if only as a test to see how well we compare with what we got out of the FDA and what they have to see whether or not those things ended up in the same RX-Norm form. I think that would be an important test of the validity of the model and then we can look at the model and say, well we really believe this is a good valid model, but that is not like proving it.
MS. HUMPHREYS: Well, the other thing, too, is that we could actually potentially even if some of the vendors were not interested in having us incorporate that data and re-release it in the UMLS we might look at the notion of getting the data in order to do the validation even if we didn’t release it.
DR. BLUMENFELD: The two quick comments, one, just to restate again the route prospective of this. I think that there is some correlation between dosage forms and route, but keep in mind that not all tablets are taken by mouth, and even worse, when you get to things like suspensions, for instance, it’s very hard to say whether a suspension is for oral usage or whether it’s going to be given IV.
DR. NELSON: Our dosage forms that we are using do have this formulation specification in them so it does say injectable suspension versus oral suspension. That’s why I said I think I can do this.
DR. BLUMENFELD: Okay, last comment, really I guess I’m going back to what Kent was saying but I do believe both for reasons of achievable scope and also in terms of its importance in terms of how we use it that whether we call things definitional versus assertional, the words we use at my joint are descriptive versus clinical, there is some concept there around the idea that what I’ll call descriptive information is typically physically about the thing and is used in transactions and tends to be more definitional, I think is a great term, whereas other information really is used in a clinical context and it’s very valuable information but it opens up a whole additional universe of terminology.
DR. NELSON: I sit at home, at my bedside reading is a textbook on the philosophy of language because I ponder this issue all the time and sometimes I really think that it’s very hard to tell the difference, at a philosophical level I agree. Philosophically what we’re trying to do here is impossible, this is engineering, and what it is is we’re going to say we’re going to some engineering approaches to try to represent this sort of stuff in a way that humans can understand it and computers can use it. And fundamentally there’s no perfection, the only thing we can do is we can build systems that would get better and better and would make them work together. That’s why I thought it was so important that you said that this morning about making terminology model and information model fit together because I absolutely believe that, but fundamental it’s engineering.
DR. COHN: Stan, question, then Walter, then I have a final comment and then we’ll adjourn for lunch.
DR. HUFF: So a couple questions. Just practically what is the availability of NDF-RT? Specifically I’m wondering are there cost considerations, since it’s a government product it’s free for use for everyone. And practically how are you going to distribute it or how would it, have you thought outside of the VA, if people wanted to use it outside of the VA?
DR. BROWN: Right now what we’re doing if you want to have a look is just asking folks to recognize that it’s a project that’s ongoing and we send them copies, no cost. We even buy the CD. For longer term, I think we’d like to distribute it via the Library. So yes, there should be no costs associated with it. Our major concern is how to get the right kind of user agreement. I think one like Clem has done, LOINC is probably a nice model just to prevent malignant working but I think we’d like to get it out to whoever wants to use it for whatever purposes they want.
DR. SUJANSKY: Can you expand on that a little bit? This is an important point, I’m sorry to jump in, just to follow-up on your question Stan, regarding the intellectual property considerations because in the response to the survey that the VA sent they indicated that redistribution of NDF-RT is not allowed and some of our analysis and evaluation was based on that data. Is there any change to that? Is there any update in terms of official policy?
DR. BROWN: I think the intent of that statement was to prevent forging, someone to take it, make a few changes and call it the same thing and redistribute it, which is clearly not what we want. What we’d love is for people to do what they want with it, reuse it, my thought was well, why would you want to redistribute it, why would that matter if you could have it for free anyway? Well, if you’re a vendor and you want to put it in your product, so based on that I think that’s reasonable. And so what we’ll try to do is, I just got the agreement from LOINC and we’re in discussion with the VA powers that be and I think that that’s our intent, is to try to get it out there in as useful a way as possible, but I don’t consider allowing renaming, changing concepts and keeping the same name to be useful. That’s our intent and we’ll be moving on that, we are moving on that right now.
DR. SUJANSKY: So just to restate that, your intent would be to allow a vendor to incorporate that, NDF-RT into a product and redistribute with their product, that wasn’t what you were trying to prevent.
DR. BROWN: No, no, that’s perfectly fine. It’s just if they want to go through and change all the work that we did and make things that we didn’t put in and then say, and then blame the VA for it.
DR. COHN: I think there’s a follow-on from Steve, Clem is there a follow-on?
DR. HUFF: I had two questions.
DR. COHN: Oh, you had two, I’m sorry, go ahead.
DR. HUFF: This was a clarification for either Betsy or for Stuart, is the drug and ingredient portion all or part of SNOMED-CT included in the government contract for the use of CT? How is that negotiated?
DR. NELSON: I believe the interlocking boundary, SNOMED-CT has ingredient names, SNOMED-CT in their core vocabulary, which is what is licensed for use in the United States, goes down to the level that’s called a virtual medicinal product. Now the virtual medicinal product is very close but not identical to the clinical drug at a generic level, a generic clinical drug. There’s some slight differences but I think that that’s where the boundary lies, and that’s, I think that answers your question, doesn’t it?
DR. HUFF: So SNOMED-CT, for instance, doesn’t have components the way that --
DR. NELSON: SNOMED-CT does not have components. It has routed generics in its names, right, you have oral tetracycline --
DR. SPACKMAN: What do you mean by components?
DR. NELSON: The component in RX-Norm, ingredient plus strength.
MR. STEINDEL: Stan, we’ll be hearing about this I believe in the afternoon session specifically. I’m just saying that in the interest of time.
DR. COHN: Steve, you have a follow-up on this?
DR. STEINDEL: Yes, I had a clarification question and this is basically to the Library. We’ve heard several times mentioned this concept of something called the Daily Med that’s going to be fed from various components and it’s going to be maintained by the National Library of Medicine. Is that looked at as another product of the National Library of Medicine? How will that be available?
DR. NELSON: My boss doesn’t want to touch it, so I have to touch it. I would think you could look at it as a product or a service, because I think it would probably be something that we would have an interface to to provide information about single things. I think the two most common questions for consumers coming to Medline Plus is my doctor told me this is wrong with me, what is it, or my doctor gave me this, what is it, so I think that there is that kind of level of information seeking that we want to be able to address, but also I think what the other part of it is that we need to be able to disseminate all of this kind of information.
MS. HUMPHREYS: My looking perplexed here was only because I suspect that the information, you saw that there was this range of information in Daily Med, and much of this information obviously is terminological information and it would be part of the UMLS. Some of it relates directly to drug information, ingredient information, that is already part of the medical subject headings and might be used to update the medical subject headings itself. We have this chem ID and we will be rethinking perhaps all of our products in this area, so I think that essentially pieces of the daily, the Daily Med may very well be there as a queriable database to the public or whatever but it also will be feeding into these other information systems and other products that we disseminate, like the UMLS, I think, so it isn’t sort of this is the only way the information will be available. And I can also imagine cases where it would become something which would be another database that was linkable in the entrée framework so if you were in Medline and there was a drug name in the middle of an abstract you might be able to click on that and go over and read what was in the Daily Med about that drug. So I think we are, our eyes are filled with fascinating possibilities about how we can make this information available to the public and we haven’t worked them all out yet by any means. First we need to get the first feed from the FDA, then we’ll really be able to work on it.
DR. COHN: Walter, did you have an additional question?
DR. SUJANSKY: I have a lot of questions but I’m only going to ask one of them, and this is addressed to both of the testifiers. We heard this morning from Dr. Blumenfeld and Partners that timely updates are very important in this domain. We’ve also heard from each of you that your maintenance process is very much dependent on data from the FDA. Correct me if I’m wrong after I finish. And therefore, I guess I would posit this and get your response that timely updates, given the resources available to your organizations and the processes you have for updates, the changes in regulation and in accurate and timely updates to the FDA’s own databases of drug listings are in the critical path to timely updates to your databases. How would you respond to that? Is that true?
DR. NELSON: I would say it’s not quite that bad. First of all we have this large health care provider called the VA that has to deal with that update problem today, and they do that. So when the FDA approves a drug they’ve got to put it, the clinician wants to prescribe it, they’ve got to put it into their vocabulary servers, so that part of it, right now what we’re looking at is what are we going to do until the FDA can feed us all of this information and certainly I have to say that I’ve had great cooperation from the commercial vendors of pharmacy knowledge base about being willing to give frequent updates. I think the VA is very concerned about that and we’ve already started working on modeling new drug transactions, it’s been initiated by them, so I would say we’re not totally dependent on the FDA.
I think what we are dependent upon the FDA is when they have this information available, like the NDC codes, we are going to want them to do it. I’d say that right now we probably wouldn’t have added in an NDC code unless we had a good reliable source of that NDC code on a regular basis. If the FDA puts it up quarterly all we can do is add it in quarterly.
DR. SUJANSKY: That’s what I’m saying, that’s the problem.
DR. NELSON: I understand that, so that information may not be available. I never said I was an NDC provider.
DR. SUJANSKY: No, I understand, but if an NDC code comes out with a new strength for a medication that one of the manufacturers has created and it takes a long time for that NDC code to get into the FDA, what is the effect on --
DR. NELSON: The effect is going to be seen earlier by the VA and by the other sources of information for the UMLS who are going to be feeding that into the format.
DR. SUJANSKY: I was under the impression that the VA was getting its drug, core drug knowledge from RX-Norm and then it would --
DR. NELSON: Not right now, I mean where we’re looking at is we’re saying this is what we have to do today, and how are we going to deal with it tomorrow --
DR. SUJANSKY: -- dependent on the VA --
DR. NELSON: We’re dependent upon our drug name sources, I wouldn’t just point the finger at Steve, I’d say First Data Bank, Micromedix, Multum, Medi-Span --
MS. HUMPHREYS: In the case of NLM the literature.
DR. NELSON: And the literature. So we have all of those things.
DR. SUJANSKY: I guess we have this sort of weird co-dependence thing going on.
DR. BROWN: We’re already maintaining it, we’re paying people to do it and we’re paying them to use really 1970’s tools and hunt and peck, so it’s not, sure, we’re looking forward to getting in an automated way and the more, the faster, the better, great. But right now we’re maintaining our operations and it’s working pretty well, so it’s really sort of a non-issue. What we’re trying to do is look forward to how can we do a better job of using our pharmacists for their pharmacy knowledge and their typing skills, so that’s really where we’d like to go with that. And regarding what we’re doing, you probably saw the paper where the first version of RX-Norm before it was RX-Norm was a collaborative effort so that was sort of the initial seeding, and then we’re taking things back and forth, and I think we’re helping each other there.
In terms of getting things updated also rapidly, we have a couple of other projects we’ve been working on, you’ve heard alluded to the new drug transaction, which is a transactional update based on XML tag messages. We also have within VA a project we call our new term rapid turnaround, which is a way we will reach out to the field and invite input requests for new terms from the sites, rapidly incorporate them, reclassify, and then put back out to the servers, so really only the first person has to find the problem, or the first site, if there’s a request and if there’s something new that comes up they want to put on their formularies, and we’ll be able to much more rapidly respond to their needs by doing that general approach. We piloted it using drugs but we’ll also use it for other aspects of the Enterprise Terminology Service.
DR. COHN: Clem, I think you’re our last question and I would obviously if possible ask you to be brief so we do have some lunch time to take.
DR. MCDONALD: You didn’t ask those guys to be brief.
DR. COHN: I guess I was just wishing up until now.
DR. MCDONALD: I just want to again applaud the general effort. I guess this is really an extension of what we heard this morning, the other branches of it. And a suggestion about the vocabulary, I think part of the things is you can specify what can be changed and what can’t be changed, it would be nice to be able to intercalate, if some other place had the NDC code, in fact it would be nice if you could get other partners who might be willing to contribute the NDC code space that they have and you don’t have, some way to make that work better because they’ll have to do some of that work anyway to get the vocabulary work. But I think, mostly just sort of applause, and hurry up I guess is the other thing.
DR. COHN: Any questions?
DR. MCDONALD: That was actually a comment, I had a lot of questions but you’ve kind of suppressed them.
DR. COHN: We do have an afternoon session. Obviously I’m glad that Randy Levin is joining us for this as well will be around this afternoon because I suspect that as we delve into this more and more recognizing that I think much of what we’re talking about is based on a high quality drug database that comes through the FDA, I think we may have to reflect on that a little bit in terms of our recommendations.
Anyway, it is right now at least according to my watch 12:30, we’re running about a half an hour late, I’m going to suggest if we can, let’s try to hold our lunch to 40 minutes, get back at ten minutes after 1:00, that way we’ll make up some of our time.
[Whereupon, at 12:41 p.m., the meeting was recessed, to reconvene at 1:36 p.m., the same afternoon).
DR. COHN: Well, my apologies to the panel for getting started a little late, I’m sure that we’ll be seeing Dr. McDonald and Dr. Zubeldia coming in any second but I do think it makes sense for us to get started. Now, Jeff, I know you wanted to do some introductions and set the stage for this session.
MR. BLAIR: This is our panel of the commercial drug knowledge base vendors, and we actually have four presenters that are going to be part of this panel, I understand only three are here but the fourth, which is SNOMED and Kent Spackman will probably be joining us, you’re there, sneaked in, so to keep on schedule if we can each of the drug knowledge base vendors will have 15 minutes and Kent you’ll have ten, and then we should have time for questions. And we have First Data Bank and George Robinson will be presenting, and followed by Tim McNamara from Multum, and then we’ll have Medi-Span and then we’ll have Kent, and please try to stay to the 15 minutes if you can and we’ll have questions when we’re all done, except for Kent, and you’ll do ten.
MR. ROBINSON: I’m George Robinson, I’m a pharmacist by background, I am the vice president of knowledge based development for First Data Bank. My colleague, Tom Bizarro, is also a pharmacist, he’s back there waiting to bail me out with any market or sales related questions, he’s our VP of sales and marketing.
First Data Bank just as a little background, our core business is providing concepts and related knowledge bases that are useful at the point of need. We’re also addressing the need for drug concepts and related knowledge bases really throughout the continuum of care, so ranging from profiling patient allergy or patient meds to creating order entries to ambulatory prescribing through order fulfillment, adjudication of claims and possible retrospective analysis, all those types of applications we support millions of times a day. And we have done for several years.
My personal goal of the day that I would ask from the Committee is I would really like some guidance on the scope of what we mean by drug terminology, and we’ll go into that a little bit later. I guess my interpretation of what I think the scope of the terminology should be, but I know some of us are having some reservations and concerns over where the line is drawn between a public domain drug vocabulary and the business that we’re in. That being said we’ll move on.
What we recognized through time is that no single concept really meets all the needs in the continuum of care, so over time as our business needs have changed as far as responding to the requirements of our customers First Data Bank has evolved as far as introducing new types of concepts and identifiers and related supportive databases. And what I’ve put up here is simply a snapshot of at a high level of the structure that we have. What drives everything in our database today is something that’s been packaged and identified with an NDC or universal product code, or health related item code, so it’s something that the pharmacist or the health institution can buy, touch, smell, feel, so that’s the real world that we deal with. What we do then is we categorize the attributes related to that and recognizing that other than pharmacy, physician, nurses, patients think in terms of drug names or attributes of a drug name such as a drug has a clinical route, drug has a dosage form, we’ve output other concepts to support these other needs. So we’ve got an example here of gentamicin sulfate when we worked through a basic name. We know that gentamicin sulfate can be administered through several different routes, including intermuscular, intravenous, ophthalmic and oral. The relationship to those types of clinical routes really are apparent through our dosing modules, that’s where that information lives.
What you’ll see at these other levels of abstraction, we move up next to this routed med ID, which is analogous to what Dr. Blumenfeld had mentioned earlier as far as the routed med, or I’m sorry the routed generic. At that level we have gentamicin sulfate injection, and you’ll notice injection is a useless route of administration in clinical terms, but what it implies is as we provide connectivity from the concept that’s useful at the point of prescribing through the point of order fulfillment, this injection represents a formulation in which the route can be administered through different parentral means. So a single formulation perhaps may be administered both intermuscularly and IV.
The next level of abstraction that we maintain is a routed dosage format ID so that we have the example Gentamicin Sulfate injection solution, solution being the dosage form. And the next level is the med ID, which is really compatible with ambulatory prescribing where you need the additional elements of the dosage form strength and additional elements of the strength people measure plus the dosage form. So we have an example, Gentamicin Sulfate 40 mg per ml injectable solution. That concept then is related to again package products. We also have a relationship that takes that generic entity to brand level information, so we have a relationship back to Gentamicin 40 mg per ml injectable solution, and then we can traverse all the way back to the level of simply a name, Gentamicin, Genoptic, Genosol, all these different names that are proprietary in nature.
We also have the world of generic formulations ID’s, which are more akin I think to the level of clinical drug that’s been discussed earlier, so that’s the path that we take to individual ingredients and also dosage forms and routes of administration are defined through that path.
Over time, again, reacting to what we’re seeing as first an increased awareness of good vocabulary practices and market requests, as we introduce new concepts into our vocabulary we introduce the notion of status, so if a track of a concept is active, if it’s been retired, made obsolete or if it’s replaced. If a concept has been replaced we’ll provide a path from what it previously was to what the replacement concept here.
Here’s an example for a med ID, it’s Tylenol Sinus 500-30 mg oral capsule. Upon further review over time we discovered that the proper name for that should be Tylenol Sinus Maximum Strength 500 mg 30 mg oral tablet. So what we did is we replaced that concept with a new ID. But that ID remains in the database and from now until the end of time, our promise to our customers is to provide you a description for that identifier.
Moving on, I’m going to address some of the specific questions that were addressed or asked by the Committee. One is what aspects of drug orders dispensing or clinical research are not supported by our drug terminology currently. We’re not cataloguing pre-market released drugs for investigational drugs. We may be doing that in the future but our current editorial policy, again, what drives it is the fact that a product has been released into the U.S. marketplace. For the sake of this discussion we’re focusing on U.S. drugs.
The next question, what is the approximate user base of the drug terminology? We estimate that over 50 percent of the hospitals in the United States, over 35,000 retail pharmacies, 47 state Medicaid programs, in one form or another the Department of Defense and a majority of pharmacy benefit managers and claims processors use our content. So we have a very large installed database that’s using our content day in day out.
The next question is what is the approximate user base of the drug terminology by update frequency, which is page two of that, you can take a look. Medi-Span and Multum are taking notes right now. We have a considerable number, well, first of all just so you understand our metrics, each entity that we count here can be a pretty major player, so for example McKesson or Siemans, which can represent a number of installations is counted as one entity. So we have 17 entities that are receiving a daily update of our files, that’s primarily pricing, but as we introduce or as new NDC that we catalogue are captured we’re also making those available on a daily basis. We have around 350 month customers of our content, about 160 quarterly. Semi-annual we have 14, semi-monthly 32, and weekly we have 155. What we’re seeing as far as a trend in the industry is people are gravitating toward weekly updates of their content.
How does a drug knowledge base differ from drug terminology? From my perspective a drug knowledge base offers added value, it’s through relationships that we provide to pricing, uses, side effects, drug interactions, and it’s also we have several indicators that are used throughout the industry determining the generic nature of a drug. That’s used quite commonly as you get a prescription filled to determine if you’re paying the branch co-pay or if you’re paying the generic. Just to go into detail, the types of knowledge that we aggregate around our drug concepts, it’s indications, drug disease contraindications, side effects, drug/drug and drug/food interactions, allergy, the list goes on and on.
How does your product fit in with the proposed changes in vision by the FDA? Well, at the time we addressed this question we really didn’t know what the proposed changes were officially, so I know we’ve participated in the discussions previously while we’ve heard discussions of a normalized public domain ingredient identifier, which we would wholesomely endorse and add into our databases as quickly as possible once that’s available. We’re concerned about discussions of changing the national drug code, we would be much more in favor if an alternative number was offered to augment the NDC, we’re very concerned about the impact of the current industry, how claims are processed, how inventory is handled by messing with the current NDC, but generally our approach, at least their internal systems is that we recognize that a packaged product is just that and we’re prepared to associate to that packaged product an external ID, whatever that may be over time.
I look forward to that happy day when labeling is electronic using a codified format. Right now in lieu of that we have a staff in excess of 20 people whose sole reason for life is to catalogue NDC’s with entities ranging from pricing to the generic formulation to the ingredients, the list goes on. So I know we have that goal of a drug terminology being available inexpensively, well, the simple fact is it costs a tremendous amount of money to capture that information in order to make it available on a daily basis.
And if NDF-RT and RX-Norm are selected as the PMRI drug terminology what use would your drug terminology still need to perform? I still believe there’s a need for First Data Bank content to provide the range of knowledge bases that we discussed up here, ranging from indications to mediation images, and this is part of my --
MR. BLAIR: George, could I just give you one piece of guidance on that, if I could do something a little bit on the fly and this would apply to each of our other testifiers as well. The question is saying if NDF-RT or RX-Norm is selected, please don’t interpret that as both, look at it as one or the other, or both, ok, and if that alters the way you’d answer the question then ok.
MR. ROBINSON: Well, basically what I saw today, RX-Norm is a little more compatible with my idea of where I’m thinking the drug terminology needs to be. I welcome an interoperability layer through the clinical drug. Based on our costs at this point in time of capturing brand name and proprietary information in lieu of feed from the FDA that has codified format I would be hesitant to see in the public domain normalized medication names that mimic what we offer today. I would see that as a proprietary competitor to what we do today. I do embrace the notion for disparate systems talking to each other and communicating allergen information and communicating patient profile information, I think that can be handled at the generic level and perhaps communicate brand information as text. But again, I guess my concern in that area is that it is quite an expense to capture that information and I don’t really see how an outside entity could do that without again an automated feed from the FDA. Frankly what I saw about NDF-RT was a little concerning in that I think it’s crossing the line of offering knowledge base content. Now if we carved out from that just simply the terminology component that’s less threatening, so I could see a role for that.
The uses I think we will continue to perform is I think we have a consistent view of codifying drug content and that our customers find useful and I think that would continue to be something that our customers would need. I know we’re running out of time.
As far as licensing terms, licensure of NDDF+ requires a payment of fees but the amount is really dependent upon how frequently the customer gets the content and the application that’s used for the data.
In the event that NDDF+ would be selected as a drug terminology we would continue to require licensure of content by parties implementing our information, and again, for the sake of discussion, what I’m talking about is more just the terminology component of NDDF. However, we certainly would be open to discussion that if a third party wanted to compensate First Data Bank for the placement of that information in a more open forum we would consider that.
Other factors that we would like the Committee to consider in their deliberations is one, I really think we need to get our hands around the types of health care applications that are intended for support through this terminology from the perspective that forms should follow function. Until we really define and nail down the applications that we want to support there’s going to be a lot of discussion as everybody views terminology from the perspective of meeting their needs. What somebody views as an order entry need is much different then what you need in an ambulatory prescribing setting, order fulfillment, or in claims processing. So I think we need to get our hands around that. I think we need to address how important the timeliness of content, of updates and the frequency of database distribution is to the group and how much energy needs to be dedicated toward providing that.
The comprehensiveness required of drug concepts, we catalogue concepts ranging from prescription drugs, over the counter, ancillary items, a limited amount of medical supplies, we do catalogue some herbal items, some homeopathic items, so depending on again the application we’re trying to support we need to get our hands around what the breadth of the terminology needs to be.
Lastly, I’d really like the Committee to consider what the hidden costs are associated with the use of a standardized drug concept dictionary. One is there’s a cost of implementation, what I firmly believe based on our experience is the cost of programming resources can far exceed what you pay for the terminology itself, so we really need to think through that. Along with that is what type of technical support documentation and actual product delivery is expect to accompany the terminology. Secondly, what the majority of knowledge base providers are doing now is building wraparound software to again make the implementation of the content a much quicker endeavor.
Secondly, maintenance, I think we need to understand what the expectations are on the maintenance of the content. This hits your pocketbook and your local institution if you’re getting an update every six months of a terminology you need to account through creating local codes, creating codes to meet your application needs between update cycles. So again, I think what we should be looking at for a standard terminology is something where within practical limits the delivery of updates are as close to real time as possible. That would be the ultraistic goal.
Lastly we really need to consider the impact on preinstalled drug databases, the plethora of different hospital systems, pharmacy systems that are out there and where do we the standardized vocabulary sitting there, I see it as an interoperability layer, but someone is going to have to cross reference between the existing knowledge bases and what this terminology is. Somebody is going to have to code for using the content and passing it throughout different enterprises of health care. So again, what I’m looking for guidance is what do you expect this information to do? And then from that point we could probably better address what the terminology needs to be and who’s in the best position to provide that content.
Thank you.
MR. BLAIR: The next speaker is Tim McNamara of Multum.
DR. MCNAMARA: I appreciate the opportunity to be here, my name is Tim McNamara, I’m the vice president of content development at Cerner Corporation and I’ve had an opportunity to speak before this Committee before so I welcome the opportunity to return. In brief, the discussion points I’d like to try to cover this afternoon, although time is short, I was told that there would be 20 minutes and now there’s 15, so I’ll try to go a little bit faster or cut some of the material out, but I’d like to describe briefly Cerner’s position regarding drug terminology standardization. Give you a brief introduction, and an even briefer introduction then I planned to Multum and some of the content and products that we provide. I’ll give you a brief response to the NCVHS questionnaire that we completed, and then spend I hope a little bit of time on a proposal to the Subcommittee and the FDA that you may find interesting.
Regarding Cerner’s position about drug standardization, we fully endorse federal initiatives to standardize drug terminology, we think it’s good for the safety of the public, we think it’s good for our clients in particularly, and it’s good for business. And we believe that the FDA should have the primary responsibility for standardized terminology, which has been neglected we think somewhat historically but we’re heartened I think to see the progress that the FDA reported earlier today. We believe that 13 desideratum, what people have called the 13 desideratum, that we proposed to this Committee back in 1998 for drug terminologies, is something that ought to be incorporated into any federal initiative and it looks like some of those thoughts are being incorporated. Dr. Blumenfeld gave you a fairly good overview of some of the more important points from that desideratum and I have those available for those who may be interested. And lastly, we’re prepared to contribute our intellectual property and our resources to FDA or any other agency that’s appropriate under a proposal that I’ll try to present briefly later in this discussion.
A brief overview, Multum was founded in ’92, our mission was to reduce medication errors by delivering patient specific drug information, primarily clinical information at the point of care, expanded the suite of content to include a lot more administrative information about drugs and their action characteristics, for example, and in so doing developed a fairly extensive drug vocabulary system in the mid 1990’s. Multum was acquired by Cerner Corporation in 1998, it continues today as a wholly owned clinical knowledge subsidiary based in Denver, Colorado.
Multum has a variety of products, I’m not going to discuss any of them other than the first two in just a passing bit of detail. The Multum Lexicon is a product that I think is relevant to today’s discussion as I hope you’ll see, the VantageRx Database may be as well. We also supply a variety of other clinically oriented products to the market and also have a number of products that are relevant to, used in Canada that are probably not appropriate for this conversation.
All Multum’s products are based on a lexical foundation that at the highest view entails links between packaged drug products, generic formulation concepts, things that we call Main Multum Drug Codes, MMDC’s, and a generic concept identifier, all of which are grouped into various categories. This gives you just the highest level of how the category system works, it’s actually substantially more complicated and there are more levels here then are actually demonstrated in the slide. At a basic level we link drug packaged products coded with NDC’s to exactly one Main Multum Drug Code. Each Main Multum Drug Code is in turn related up to a generic concept identifier for drugs, and you see some examples of what those drugs are at those levels of stratification.
In terms of the drug names, we keep our concepts very separate from the names that describe those concepts, so we have a large table that includes about 38,000 different drug names at all different types of levels of granularity, so you see some examples here of erythromycin as well as specific brand products of erythromycin, some forms, etc., each of these different drug names are explicitly related to the concepts like drug ID’s, Main Multum Drug Codes, a subset, and even a smaller subset down to NDC codes. Using that stratification system or hierarchy we’re also able to hang appropriate clinical information at various levels, so dosing information, allergy side effects are coded at the level of Multum drug identifier, some content like dose range checking and certain administrative content like controlled substances acts classification are coded at the Main Multum Drug Code, and then a variety of more detailed information like package data or manufacturer data or obsolescence data are coded down at the bottom at the NDC level.
With that bit of background on the structures that we use to encode data let me give you just a quick comparison of the Multum Lexicon and the VantageRx Database. Multum Lexicon is a collection of about 17 tables, it’s a minimal dataset that’s available under a no fee license agreement for anyone who wants to download it on the web, and we currently have in excess of 5700 licensees who have taken that data and potentially used it. In contrast, the VantageRx Database is 144 tables, it includes everything that’s in the Multum Lexicon and a lot more in the way of drug categorization, expanded synonymy, additional clinical content. It is available for a fee and a license, and is currently used by an excess of 150 paying clients, that excludes by the way those clients that license this content through Cerner and our relationships with our clients.
Here are some examples of some of the clients, some fairly large EMR/HIS suppliers like Eclipsys, Cerner, hospital enterprises, pharmacy benefit management organizations, web portals and others. This is a quick screen shot for the Multum Lexicon, it gives you a feel for what those 17 tables actually look like so it’s got NDC related information, formulation related information and drug names. This is the VantageRx Database, the commercial product, that includes a small subset, a larger set of tables, in this case you can’t see all of the tables. I will make reference just as an aside for where we’re going that a number of the tables that you see here are appended with a UPD, those are the update tables that contain metadata about the time which and the circumstances at which particular items have been added or updated to the product.
Now changing gears a little bit, we were asked to complete a questionnaire for this organization. Multum was late in responding to that questionnaire as was discussed at one of the previous meetings. The questionnaire was mailed out I understand on January 6th, 2003, it was addressed to a Multum associate who fell off a ladder and sustained a substantial injury requiring intensive care admission among other things on January 1st, so unfortunately the questionnaire sat in her mailbox for some time and when it was ultimately found we completed it, unfortunately it was after the date at which the Committee was looking to have the material returned. We apologize for that delinquency, however we think it perhaps is understandable and we appreciate the flexibility of this Committee in asking us to complete the questionnaire and have it evaluated.
On July 22nd of this year we received notice from Dr. Steindel, who’s here today, that the Multum Lexicon failed to meet one of the required four criteria. We responded by the way on the questionnaire with respect to the Multum Lexicon, not to the VantageRx Database product. Specifically we were told that the issue of concept permanence was not reflect in the Multum Lexicon. We actually respectfully disagree with the conclusion of that evaluation on the basis that concepts permanence actually is well supported and we apologize for any misunderstanding that may have occurred in our response. Specifically, obsolete drugs and their identifiers are retained in the Multum Lexicon. Obsolete drugs are specifically marked as obsolete with the date in which they became obsolete. Multum has complex rules for trickling up the obsolescence of specific concepts. Concept identifiers are unique, concept identifiers are not reused, they do not change in meaning, if there’s a need to change meaning we retire the code and put in a new code. Multum drug names are also explicitly marked as obsolete are still retained, and then we maintain record level controls for each modified record and expose those metadata to our clients.
A couple examples of that, Baycol is an example of a drug that was removed from the market a number of years ago, cervistatin(?) is its generic identifier in the Multum scheme, this is from the most current version of the Multum Lexicon that you can download, cervistatin is still there, the Main Multum Drug Code is reflected for cervistatin drug 4140, it’s still available --
MR. BLAIR: Tim, maybe I could save you a little bit of time if our consultant, Dr. Sujansky, looked through what you’re telling us and he agrees that you do meet that criteria.
DR. MCNAMARA: Dr. Sujansky and I discussed this issue in detail last week and perhaps he can comment now.
MR. BLAIR: I kind of can save you time from that technical standpoint of your meeting the criteria so you do have time in your presentation if there’s other things that you wish to convey, you’ve won this argument.
DR. MCNAMARA: Ok, well we’ll jump forward. So we believe concept permanence is represented and we appreciate the Committee’s agreement on that.
Let’s get to the heart of what I think may be the most interest to you all, and that is a proposal from Cerner to this Committee. A little bit of background, for years Cerner has publicly advocated the creation of a central code assignment authority for drugs and drug products and we’ve testified to that effect before this Committee before. We believe that the continued use of disparate generally proprietary drug terminology systems is a potential threat to the safety of individual patients and of public health and limits the ability of organizations like care provider organizations, payers, and others, to perform their work effectively, efficiently, and safety. For the same general reasons that the federal government has sponsored and subsidized the use of SNOMED-CT we believe that the federal government, and FDA in particular, should establish and maintain drug terminology standards and code sets, and it may be that FDA is already well along the path to doing so.
Our assumptions are, however, that the creation of a drug terminology standard is expensive and time consuming. Multiple well-developed drug terminology systems already exist, and we think that the FDA should profit from the experience of commercial vendors of drug information systems and select one or more as a starting point for that effort. So our proposal is that FDA should invite proprietary drug information providers to contribute drug terminology data in an open source format to FDA, and if that occurs then we make the following commitment in the next three bullets. First, Cerner will make the complete VantageRx Database available for evaluation by the FDA or any other appropriate federal agency. Secondly, FDA may determine which tables, fields, and data it may need in order to achieve the goals of creating a national drug terminology standard from that set and use that code set. Finally, Cerner will provide to FDA any tables, fields, and data selected by FDA for use by the FDA and the U.S. public under what we’ll call a liberal open source agreement, in other words unlimited right for FDA to distribute, redistribute, use, and modify the data with no restriction. Effectively we’ll cede control of those tables and data that the FDA may select if it agrees to this particular proposal.
The benefits. We believe that such an approach creates a fast track and a low cost path to a comprehensive national standard for drug terminology. We believe that it encourages multiple vendors to participate and contribute to a core set in an open transparent process. And we believe that this approach will ultimately facilitate interoperability as all the vendors and other parties that are interested with have access to the selected data. And at that point, that’s the extent of my prepared remarks and I’ll be happy to take questions now or at the end.
MR. BLAIR: Well, we’re going to save the questions for after everyone has testified.
MR. DUBOIS: Good afternoon. My name’s Mark Dubois and I work for the Clinical Tools Division of Wolters Kluwer Health, which includes the Medi-Span product line. With me today is Karen Eckert, she is in our knowledge base development department and can help support and answer any questions we may have later on. I certainly appreciate the opportunity to be involved in today’s meeting and to address this Subcommittee, especially due to the fact that the Medi-Span products did not initially receive an opportunity to complete the PMRI questionnaire. We were not aware of this initiative until April of this year, so thank you for letting us come and talk.
I don’t know how much most of you know about the Medi-Span product line so I’d like to give you a brief overview of what has transpired since 1977 when Medi-Span was originated. It originated as a monthly drug pricing publication and quickly drove into a portfolio of drug terminology, drug pricing, clinical screening data sets and clinical screening application software. In January of 1998, we were acquired by First Data Bank and subsequently during 2001 the FDC decided that the drug information market was made less competitive due to this acquisition and mandated that First Data Bank divest the Medi-Span product line, which occurred in December of 2001. Wolters Kluwer purchased the Medi-Span products and the Medi-Span products themselves are closely aligned with an organization by the name of Facts and Comparisons, who provides referential content to the health care industry and has for over 50 years. Wolters Kluwer itself is a multinational organization which is the third largest publisher in the world, with interests ranging from health to tax and legal. And I think it’s important to note that throughout this series of events we’ve been successfully licensing the Medi-Span databases and software throughout the health care industry to thousands of customers who literally use our terminology millions of times a day.
The Medi-Span dataset support a wide range of applications throughout the health care industry. What you see here is a brief list representing some of those and they include of course integrated clinical drug and disease screening, patient profiling, computerized prescriber order entry to include e-prescribing, billing, claims adjudication, drug dispensing, drug formulary creation and maintenance, retrospective and data warehouse drug analysis, and generally any application that requires drug information can be supported by our wide range of products.
One of the questions we were asked to address was what aspects of drug orders dispensing or clinical research are not supported by your drug terminology. In our opinion the Medi-Span data sets are timely and complete, and certainly meet the market demands as they are defined today. However, while we have some investigational drug information in our data sets, due to lack of customer demand as it is today we have not made a concerted effort to make their inclusion comprehensive. However, if the market asks us to do that we will.
As asked in the PMRI questionnaire we included some base numbers on the quantity and types of organizations using our terminologies. Obviously since we’ve been providing this data and knowledge to the health care industry since 1977 we have a wide range of customers within the health care industry. As you can see here we have thousands of health care organizations that utilize our content and knowledge, to include hundreds of software developers with hundreds of thousands of end users, and again, millions of transactions daily use our knowledge to facilitate those. Frequencies of our data range from daily to weekly to every other week to monthly to quarterly, basically whatever our customers need as far as frequency is concerned we are prepared to provide those.
In our opinion drug terminologies only define what a drug is, this include route, route and form, a clinical drug or generic packaging drug. It does not include other attributes and these could include but are not limited to pricing information or data elements that identify AWP’s and wholesale acquisition costs and their histories, ingredients, indicators such as RXOTC and a host of other items that define an individual drug. We believe a drug knowledge base includes those other attributes. Some examples in the Medi-Span data sets, which include over 100 drug attributes, include DEA schedule, RXOTC indicator, up to 99 active and inactive ingredients and strengths per drug product, six different types of pricing and their histories, therapeutic classification, brand generic indicators and a host of others. A drug knowledge base will also enable clinical screening and can provide integrated and referential knowledge that covers interactions, allergies, dosing, precautions, indications, adverse effects, and patient education.
How does our product fit in the proposed changes envisioned by the FDA? Other than what we’ve learned today it’s a little difficult to answer that question due to the lack of readily available detailed information on these initiatives. From what we understand the FDA intends to work with drug manufacturers to acquire and publish basic new drug data in the Daily Med database. While our having access to that information could be helpful we feel it would still be necessary for us to make direct contact with those manufacturers to gain additional information. With the VA’s NDF-RT and RX-Norm, Medi-Span’s drug files and clinical products will continue to include additional value added fields as well as the key data links to the clinical screening and referential database content available through Wolters Kluwer Health. As other initiatives are implemented we will rely on the partnership of our customer base to determine what additional products, mappings, or enhancements are necessary.
Another question that was asked is to what degree are RX-Norm and NDF-RT complementary or competitive to your drug terminology knowledge base. While we certainly support terminology that would help systems communicate with each other, based upon what we’ve heard today and what we know and the potential scope of the project we are concerned that these initiatives directly are competitive with ours and jeopardize our business. The FDC mandated Medi-Span be divested from First Data Bank to ensure a competitive environment and we feel that it’s potentially true that the NDF-RT and RX-Norm efforts could remove that competition. With RX-Norm Medi-Span initially provided our proprietary concepts for the UMLS and UMLS used the concepts to create their own values that have evolved and potentially now can compete with ours. The NDF-RT is also competitive to both the drug knowledge bases and some of the clinical content that we’ve been developing, maintaining, and supplying to our customers since 1977.
If NDF-RT and RX-Norm were selected as PMRI drug terminologies, what uses would your drug terminology knowledge base still be needed to perform? Again, other than what we’ve learned today it’s difficult to answer this question, however, the Medi-Span products have always been guided by market demand and what our customers ask us for, and we will certainly continue along that process while supporting the need for a terminology that will help systems communicate between them.
What are your current licensing terms? How would those terms change if part or all of your products are selected as a PMRI drug terminology? Part of this is somewhat of a moot question since the Medi-Span terminology did not make the first cut for the same reasons that Multum did, and of course we respectfully disagree with that as well. However, if we have been selected or if upon further consideration we could be included in this initiative, we certainly would be willing to create a drug terminology database as a separate deliverable that minimally included the NDC and drug concepts, with their descriptions or unique identifying attributes. The licensing terms and costs would certainly need to be negotiated and we suspect would be determined in a manner similar to the recent negotiations with cap for the use of the SNOMED disease terminology and we would be certainly willing to work with all interested parties to ensure an equitable licensing solution was reached.
Are there any additional factors that we wish you’d consider during your deliberations? Certainly. Since we’re close to our customers, on behalf of them we would like to recommend that the VA NDF-RT drug file use the NDC billing unit standard for defining packaging. The NDC billing unit standard is widely utilized in the marketplace and allows billing and packaging concepts to be shared more easily with those applications using a drug file other than the VA NDF-RT. This was not done in the implementation of the Medicaid rebates and has led to a time consuming and expensive reconciliation process to deal with the resulting discrepancies.
In summery we believe that for these initiatives to have widespread success there needs to be more communication and collaboration between public initiatives and the private sector, especially the clinical knowledge base providers. I’d also like to state that the Medi-Span databases and Wolters Kluwer Health are open and willing to participate in any way that we can to become a valuable partner to this initiative.
Thank you.
MR. BLAIR: Thank you. Dr. Spackman from SNOMED is next.
DR. SPACKMAN: First let me just say that I did provide this written document, I hope the members of the Committee have it, if not we will get you copies, I sent it in a little bit late --
MR. BLAIR: We actually received it before the weekend with almost everyone else’s stuff, so we’ve had a chance to read it through.
DR. SPACKMAN: Okay, great, so what I’ll do is go to the presentation then, and go straight to the main point. The HHS license for SNOMED includes SNOMED core drugs. I’ve heard SNOMED referred to as a disease terminology, SNOMED is not just a disease terminology, SNOMED is a comprehensive clinical terminology. Enough said. It also includes all of the hierarchical and logical integration between SNOMED clinical drugs and the rest of SNOMED. It doesn’t include any of the data that would be a in a U.S. realm specific extension, and that kind of extension is needed by users, so there is stuff that SNOMED doesn’t do that’s definitely needed. A U.S. realm extension that complements the SNOMED core should be produced, possibly utilizing RX-Norm and NDF-RT plus perhaps other additions, and should be distributed free for use. And obviously the FDA and HHS should be responsible for at least some part of that whole process.
The second point is that SNOMED currently provides and maintains core drug integration, it has codes, hierarchies, and logic links between generic ingredients and their classes, therapeutic drug monitoring tests, which is integrated between SNOMED procedures, LOINC codes, and SNOMED ingredient codes, although the LOINC mapping needs to be, LOINC integration needs to be updated. Administration of medications adverse effects, allergies, poisonings, indications, and contraindications, all of these things, all of these areas contain concepts that refer to drug concepts. For example,benzodiazepine overdose or benzodiazepine dependency. That would be a problem on a problem list, but you need to link that to benzodiazepine, SNOMED does that already.
There are three main components that we see that are needed support end users. First there are the concept identifiers, and from a SNOMED perspective those split into two groups, the SNOMED realm independent drugs plus clinical concepts, those would be the red triangles in the following diagram. Then realm specific drugs and devices and other realm specific extensions in areas like administrative codes, and those are the blue squares in the following diagram. And then knowledge bases, and this was the point I was making earlier today in my comments and I think I was distressed a little bit that the RX-Norm and NDF-RT folks seem not to clearly differentiate this because we think that’s really important. And these knowledge bases will probably come from both the commercial sector and the local public sector, that’s the model that we have at least from a SNOMED perspective, so this is sort of a diagram that I’m going to use a little bit. So SNOMED-CT drugs and SNOMED-CT clinical is in those triangles the pyramid. Then extensions, proprietary devices, proprietary drugs and other information that’s necessary to go with those, the blue rectangles at the bottom. And then knowledge databases, both from the commercial sector and the local public sector.
And not surprisingly this is how the model works in the UK, and that’s partly why I’m presenting it to you because we felt like all of the work and effort and energy and time that has gone into a design for a well-integrated clinical terminology to support the National Health Service came up with a very workable model. So they have, in the UK they have a primary drug data dictionary, which is managed by the Prescription Pricing Authority, PPA, has a secondary care drug dictionary, which is managed by the National Health Service Information Authority, and they also have a medical devices dictionary managed by the NHSIA, so they have different people responsible for different pieces. They also still will rely on knowledge databases coming from the commercial sector FDB and other suppliers who provide those knowledge bases into the UK market. There is probably some economies of scale to be achieved if the U.S. and the UK both follow a similar kind of model, so there’s something to be said both for suppliers who serve both markets and also for terminology developers such as ourselves.
Just a little bit more detail here, there’s a SNOMED-CT core and then a UK realm extension, and within that realm extension you’ve got these three different dictionaries. And this integrated model for drug actually use different names but the names are very similar to the names you’ve heard already, there’s what they call a virtual therapeutic moiety, which is basically drug containing some ingredient. And then a virtual medicinal product, and that’s the level to which SNOMED goes, and then below that they’re all extensions. So the red ones are SNOMED core and the blue ones are extension.
So here’s an example, cephalosporin would be a category in SNOMED, Cephalachlora would be a particular, cephalosporin, Cephalochlora 250 mg capsule would be the virtual medicinal product or clinical drug, and then below that level would be the extension concepts. SNOMED core would also have a concept called allergy to penicillin and then where the patient is allergic to penicillin you have to caution in use of sephalysporin’s, that would come from a knowledge base, either First Data Bank or Cerner or some other, or VA NDF-RT, would be a local public sector knowledge base. The connection would have to be made to a SNOMED code that says allergic to penicillin, connection back to the knowledge base, connections to these other codes that indicate sephaclora is a sephalysporin and therefore a patient allergic to penicillin needs to have that message then. That message, that whole process, takes place using a knowledge base.
There was a question about use cases, I’ve added this slide since the one I distributed to you, but there’s a long list of use cases supported by SNOMED-CT core plus realm extension in the UK, prescribing, dispensing, administration, pricing, decision support, electronic prescribing guidelines, word stocks, protocols, formulary development, a large number of very detailed use cases have been examined and models have been developed for those in the UK.
How much time do I have left here? I’ve got maybe two more slides.
I think it’s really important to recognize a couple of distinctions here. One is the distinction that we made between what’s terminological knowledge, what’s always true about a drug, and things that are specific to a particular realm. So the blue boxes here we think largely are specific to a realm and these are boxes that are populated by the prescription pricing authority of the NHSIA in the NHS, so control drug legal category, adverse event reporting recommendations, drug tariff categories, reimbursement information, product prescribing specifics, NDF codes in the U.S., those are all specific to a particular realm whereas for a clinical drug the ingredient strength and the form, those are going to be universal, those are going to be common for that clinical drug no matter which country they’re used in, that’s why we stop at that level with the SNOMED core.
Just a little bit about RX-Norm and SNOMED, the models are similar but not identical throughout. They appear to be identical for ordinary name, strength, dose form kinds of drugs, and elsewhere there are logical places to connect the models. Given the need for core SNOMED to support country specific drug extensions there will always be some overlap between SNOMED-CT and RX-Norm, but it should be possible to coordinate their development to reduce duplicate effort by an NLM and CAP. NLM is proceeding now to insert SNOMED into the med-thesaurus which should help determine the exact degree of overlap.
Here’s a little diagram that helps to illustrate where there is overlap and where we have slight differences. If you look at Lasiks 20 mg tablet, at the bottom are 20 mg per 2 ml injection ampule(?) and go up the hierarchy, the green boxes are in SNOMED, the ones circled in red are the ones that are in RX-Norm, so furosomide(?) 20 mg tablet, with its British name frusomide(?) 20 mg tablet, will be in SNOMED core. Furosomide 10 mg per ml injection solution 2 ml ampule is the way the virtual medicinal product exists in SNOMED core, and as I understand it furosomide 10 mg per ml injection solution would be the clinical drug in RX-Norm. Obviously these cover very much the same ground and need to be harmonized. The levels above, probably there’s a whole lot of overlap in things like oral furosomide product, furosomide tablet and so on.
With that, I’ll think I’ll stop and leave the rest for questions.
MR. BLAIR: Thank you, Kent, and Simon will be continuing to facilitate the questions.
DR. COHN: Well, sure, I think Clem had his hand up earlier so Clem, rather than trying to have you be the last this time we’ll let you be the first.
DR. MCDONALD: I’ve got a number of questions. I can’t read my writing, though, it’s been so long since I wrote them down. Let me ask about, Medi-Span used to have and probably still does, a ten digit hierarchical clinical drug code --
MR. DUBOIS: The generic product identifier?
DR. MCDONALD: And that was going to be, when you guys married, one of the children of that marriage was going to be a new one that was even better. Did that ever happen and do you still have the old one?
MR. DUBOIS: Yes, we certainly support the generic product identifier, which is proprietary to the Medi-Span data sets and continue to be used by thousands of customers. As far as a new enhanced attribute that will identify therapeutic classification, it is not part at this point in time of the Medi-Span database.
DR. MCDONALD: In terms of Cerner’s product, I didn’t realize that one can now download the core stuff without a license.
DR. MCNAMARA: That’s been the case for about six years.
DR. MCDONALD: Is it still the same as it was when you were at --
DR. MCNAMARA: The only thing that has happened in the six years is we’ve removed the therapeutic categories and you have to be a non-profit entity in order to get the free access to those, but everything else besides the therapeutic categories remain unchanged in terms of its structure since we originally posted it up there circa 1997 I think.
DR. MCDONALD: And then some general, what the NCPDP billing units, is that literally, give an example of what, is that like eaches(?) and bottles, is that what you mean by units, or is it something more general?
MS. ECKERT: I’m Karen Eckert with Medi-Span, also one of the co-chairs for the Billing Unit Standard Workgroup in NCPDP. The Billing Unit Standard is a common definition of what the drug files will call either an each, a gram, or an ml.
DR. MCDONALD: So it’s a standard for units.
MS. ECKERT: Yes, and then the philosophy behind it allows the data to be shared across the different databases.
DR. MCDONALD: Does NCPDP allow it to be used widely or with licensing or what’s the constraints or limits on its usage?
MS. ECKERT: Tom, can you answer that?
MR. BIZARRO: I’m Tom Bizarro with First Data Bank, and with Karen I’m a co-chair of Workgroup Two at NCPDP. The NCPDP standards are available for use to members free of charge, they do highly promote and widely promote the use of the NCPDP Billing Unit Standard, and it is really a Billing Unit Standard to make sure that electronic claims transmissions are then billed and adjudicated using a common unit of measure. So I guess no it’s not free but with a membership to NCPDP the standards are downloadable and available free of charge.
DR. MCDONALD: And then the last question is of the different groups, I got mixed up, I was trying to track them, some of you have a sort of a thing like ampicillan pills versus ampicillan injectable as a definite defined unit, some don’t. Is that right or you all do? It’s sort of the route combined with the drug generically, maybe a little more than the route.
MR. ROBINSON: From a First Data Bank perspective we maintain unique identifiers for all those different levels of drug abstraction that I presented on the slide, I don’t know if you want to address Medi-Span.
MR. DUBOIS: Yes, as part of our application software that concept is available as well.
DR. MCNAMARA: Our generic drug identifiers include the concept of route administration separately systemically intended systemic effect versus localized effect --
DR. SPACKMAN: SNOMED has those, like oral penicillin as a concept.
DR. COHN: Bill Yasnoff then Walter Sujansky.
DR. YASNOFF: I just had a very quick clarifying question for Kent. In your model where you, that model points to a place where it says local public authority or something like that, local public sector. Is that what most people would call national public sector? I realize your view of the world may be a little different.
DR. SPACKMAN: Actually this --
DR. YASNOFF: What do you mean exactly by local, do you mean --
DR. SPACKMAN: The words came here from Paul Frosnick(?), so from the UK perspective they considered there to be at present multiple different local public sector knowledge bases. Now there might also be a national public sector knowledge base but this was explicitly to mean local meaning, I mean VA isn’t very small but it is local in some sense.
DR. YASNOFF: That’s what I’m getting at, in other words in UK when they say local are they talking about regions or are they talking about every practitioner?
DR. SPACKMAN: Yeah, regions, it would be regions.
DR. YASNOFF: Thank you.
DR. COHN: Walter?
DR. SUJANSKY: A few questions. First about some of the new proposals that we heard. Dr. McNamara, regarding your proposal of offering VantageRx to the FDA for distribution and so forth, here on the record does that include, would that entail allowing other vendors such as Multum’s and Cerner’s, competitors to use VantageRx in their products and to distribute it to their customers?
DR. MCNAMARA: That would entail redistribution of any of the content that the FDA selects for drug terminology to any entity unrestricted by Multum or by Cerner, there may be restrictions imposed by FDA but we wouldn’t purport to provide any restrictions to FDA in how it redistributes it.
DR. SUJANSKY: So that’s a yes it would allow that.
DR. MCNAMARA: For the drug terminology content, not for all the contents of the VantageRx Database, but for the parts that the FDA views as terminologically relevant.
DR. SUJANSKY: You would put forth all the content including terminology and non-terminology content?
DR. MCNAMARA: For evaluation, and then for those parts that are deemed by FDA or other relevant federal agency to be terminologically relevant as opposed to referential content and clinical content, we’ll make that part open source. And we’re open to discussion about what is terminologically relevant versus what is clinical information.
DR. SUJANSKY: But you would be involved in those discussions. If the FDA decided that it was all terminologically relevant would you be fine with all of it being available to everyone for any --
DR. MCNAMARA: No, it wouldn’t be fine, because there’s a lot of content in there that it’s obviously not terminologically relevant, so we wouldn’t be comfortable with that response, but we are --
DR. SUJANSKY: What would be the purpose then for the evaluation?
DR. MCNAMARA: Because there are substantial portions of the content that Multum produces that are not available in the Medi-Source Lexicon that may be of interest to FDA and may further the goal of standardization, and we’re amenable to placing those content areas into the public arena for distribution by FDA.
DR. SUJANSKY: I’d like to ask you the same question about you provisionally mentioned that Wolters Kluwer would be willing to put some of the terminology into open source I believe you called, in some open source status. If that were the case also would it be available to any vendor, including First Data Bank and Cerner and Multum and anyone else for any purpose?
MR. DUBOIS: Certainly anything that we decided to put forth in an open forum would be available to whoever would access it. Certainly we need to know more about the scope, we need to understand this whole initiative much better then we currently do before we could make any kind of a commitment as far as what that data or content or attributes would be, but we’re certainly open and willing to be participating in this endeavor.
DR. SUJANSKY: Dr. Robinson, you mentioned 20 FTE’s were used at First Data Bank to maintain the NDC database and incorporate NDC related information into the other tables and so forth. If it’s possible for you to do so how would you divide that between those FTE’s that are involved with just tracking down NDC changes, additions and changes and retirement of NDC’s and so forth versus then once you have the NDC incorporating it into all of the other types of information that you provide?
MR. ROBINSON: I believe we have 12 to 13 research associates, basically pharmacy technicians as a background, their function is based on labeler, we split up labelers alphabetically, they track new drugs that are added for those particular manufacturers and then also send turnaround documents to those manufacturers asking for updates. We also pay attention of course, to updated labeling as we get it from the FDA or from the manufacturers themselves. Those same people then turn around and also do the data entry. But just for simply tracking the changes and doing the data entry that’s about 13 people. Then we have a staff of about four to five who maintain more of the structured content, that would be the therapeutic classification, maintain the generic formulation identifiers, maintain, or attempt to maintain, a normalized just medication name for use in prescribing areas, and then of course there’s the supervisory level over that. That of course is not addressed in resources required for packaging, distribution, documentation and all of the things, so this is not an inexpensive proposal. But again, this area of sending information to the FDA are not, I would get real nervous about the FDA putting out a drug database dependent upon information from a pooled set of providers, I think the role of the FDA is to properly ask for the information from the manufacturers and use the labeling information. I’m amenable to a third party perhaps doing that, but just from so many cross referencing projects I’ve been involved in you always look to the core, I’m sorry, the source information to see where the truth lies and then amongst the different contributors there’s slightly different interpretations as to which drug form to assign, which route of administration, what are the units of measure for the strength, and I am less than optimistic about the FDA taking on that type of responsibility. And again, what I’d rather see them is fast track providing labeling in a codified format and I think we all would benefit from that. But that’s somewhat, the thing we wrestle with is that’s in the future and we need to deal with business needs now.
DR. SUJANSKY: Dr. Spackman just to follow-up on that point, needing things now and the challenges of updating drug terminologies, what are the processes in place now, today, for updating the drug content in SNOMED-CT? Specifically what update frequency is available today or in the near future for that? What personnel are available? Are there pharmacists on the Editorial Board who are available to review information? And what is the source of the new drug information that is incorporated into SNOMED-CT?
DR. SPACKMAN: The update frequency for the extension is different then the update frequency for the core, so the extension actually gets, I think it’s once every two weeks, it might be once a week, released to the users, and so we would see that a U.S. drug extension would need the same kind of frequency or maybe even more frequent depending on exactly what uses were being made of it.
DR. SUJANSKY: One to two weeks is what they do in the UK.
DR. SPACKMAN: In the UK, right. SNOMED core is updated once every six months and we’re prepared to go to more frequent updates if it’s necessary for the core to be updated more frequently then that, we’re watching that carefully.
DR. SUJANSKY: If a new generic medication came out or a new, even a new strength came out, that would require an update to the core, correct?
DR. SPACKMAN: That’s right, that’s right. And so in the UK they have a set of pharmacists at NHSIA, they also have people at the Prescription Pricing Authority, whenever a new package comes out they update their terminology and then there’s a ripple effect back to us and so then we update those. For the U.S. right now what we’re doing is we’re maintaining a U.S. drug extension that basically is used to feed the core, but this U.S. drug extension that we’re maintaining is less grandiose then what’s done in the UK, but it could certainly expand beyond that depending on whether there was a need. But basically we’re using that process basically looking at new manufactured drugs that come out and looking at information that comes from the FDA in order to identify U.S. specific as opposed to international clinical drugs.
DR. SUJANSKY: For the extension and the core them presumably.
DR. SPACKMAN: Then we have pharmacists both in the UK and at CAP who are responsible for that task.
DR. COHN: Jeff and then Steve.
DR. SUJANSKY: Actually I did have one other question or comment to Mr. Dubois’ comments about respectfully disagreeing with the assessment of the questionnaire responses, and we can talk about this offline, or if you want to state why you disagree on the record now with the assessment. I wanted to give you an opportunity to do that since Dr. McNamara had an opportunity.
MR. DUBOIS: I appreciate that.
MS. ECKERT: The reasons we were respectfully disagreeing is the fact that what we maintain of the terminology with the descriptions does not change the intent when we were either clarifying something and in the name of the thing we didn’t change it from an apple to an orange, we may have just fixed a typo and said orange has an e at the end, and that was what we had included and we were saying the description has changed. Additionally we don’t output all the inactive terms to the customers because the customers are wanting to know what the values are to date so when they’re creating pick lists they will just see the current values, whereas if we output the values that are inactive and that we’re storing historically that we can still tell customers what they meant, we don’t output it in the file because it’s an additional step for them to restrict them in all their applications and in all their displays. So that was why we disagreed with that.
DR. SUJANSKY: Okay, thank you.
DR. COHN: Jeff and then Stan.
MR. BLAIR: I’d like a little bit more help with some clarification on the issue I touched on earlier, and it seems to me from Kent, from what you were indicating that there’s quite a good deal of overlap between the SNOMED drug codes and RX-Norm but since that’s under the National Library of Medicine now there could be harmonization and you didn’t seem to be overly concerned with that. So in terms of that being something you’d feel uncomfortable with I didn’t detect a great deal of discomfort on that.
DR. SPACKMAN: The reason for not being uncomfortable is because we’ve initiated discussions and I’m confident that we can talk about what’s the best way to proceed, but I think we have to have some very serious heart to heart discussions about exactly how we’re going to do it but I don’t see a major barrier about finding a pathway forward.
MR. BLAIR: Since I didn’t, I wasn’t detecting that this is a major issue of concern at this point, let me kind of switch my focus to our other three testifiers and it wasn’t clear to me, there was some concern, degrees of concern, with respect to whether RX-Norm is selected to be in the core set of PMRI terminologies, and/or NDF-RT, but there wasn’t a distinction. So could each of you clarify individually for RX-Norm versus NDF-RT the degree to which you feel that would duplicate, overlap, or be a direct competitor to what you offer, separately for each.
MS. ECKERT: Karen Eckert with Medi-Span. I believe the RX-Norm and those drug concepts and those relationships we feel would be a good use into the realm that we wouldn’t mind mapping to and we don’t see that as being competitive. The VA’s NDF-RT, due to all of the clinical overlaps, is where we see that extremely competitive and have a lot of concerns.
MR. BLAIR: And you’re speaking for Medi-Span.
MS. ECKERT: Yes, I’m speaking for Medi-Span.
MR. ROBINSON: George Robinson, First Data Bank. I have degrees of concern, so I think where I have less anxiety is really almost analogous to the core level that I saw in SNOMED where that is for the interoperability purposes so supporting allergy screening and communicate the patient med, I really see the usefulness of that. I start to get a little anxious on the level of an extension where now we’re starting to provide proprietary names and packaging information related to the core concepts. My concern --
MR. BLAIR: Those are the words that you used before but you didn’t, are you talking about NDF-RT when you say it?
MR. ROBINSON: I’m sorry. Well, this level of anxiety applies to both because both have, what I’m seeing in both there’s a level of ingredient definition, a clinical drug definition, then it may span the manufacturer drug where a clinical drug such as amoxicillin 250 mg capsules has a brand name Amoxil, it may be packaged in a bottle of 100 by Beecham, that type of level. So the extension just on the drug names themselves plus the packaging is an area of concern. Lastly, the last area of concern is really the knowledge bases. Now if bundled with that was a knowledge base of uses, side effects, drug interactions, that is directly competitive to our business as it is defined today.
Now the biggest area of concern in just the extension level is again, because we dedicate tremendous numbers of resources toward doing that and the way we bundle those drug and identifiers or make them available is we’re meeting health care application needs today. Now if that information was available in a codified format from the FDA that changes everything, in that situation what we see is like an FDA extension to the core of NDF-RT or to RX-Norm. So again, hopefully it’s not an all or nothing premise here, so if we look at RX-Norm my hope would be lets define what the scope of that public terminology is, if we look at NDF-RT, perhaps if you exclude the knowledge bases that are VA centric and just deal, and that again goes to that local concept that Kent was mentioning with SNOMED, then I see less concern there.
MR. BLAIR: Thank you.
DR. MCNAMARA: Speaking for Cerner and Multum we are not threatened at all by the RX-Norm initiative, in fact we’d like to see it or something like it succeed because we want to support that kind of standardization around drug terminology. With respect to the NDF-RT I again do not feel threatened, I don’t believe my organization feels threatened, because the core drug vocabulary part would likely be the same as RX-Norm. In terms of the drug knowledge base part of it I think we can effectively compete, I think we’ve got a better product in terms of the clinical data that we supply around things like interactions, allergies, therapeutic duplication dosing, you name it and I think we’re going to have a better product then what it is that the VA is likely to produce and we’d be happy to address that if you will in the marketplace.
MR. BIZARRO: From a sales and marketing, this is Tom Bizarro from First Data Bank. From a sales and marketing perspective I think that Kent has brought up a good point. Would First Data Bank feel comfortable competing against other knowledge bases? Yes, we do it now. Would we feel comfortable competing against knowledge bases available for free? That’s a whole other market effect that is very difficult to compete against when you’re talking a database that’s available in the public domain free of charge. People will be willing to maybe accept something not as robust, comprehensive, or maybe even as good when the price is free. It’s very difficult to compete in an open marketplace against free.
DR. LINCOLN: This is Mike Lincoln from VA. I wonder if VA could be permitted to make a comment? Thanks. Steve had to go to the airport, but this is something Steve and I have talked about quite a bit over the course of developing NDF-RT and also integrating it with our enterprise terminology strategy. A couple of caveats. First of all, anything the VA develops is available for free under FOIA. NDF-RT is based on our NDF and right now you can go to www.va.gov/vista and download our NDF including our drug classes and all the other information that’s in NDF, our pricing and so forth. That’s currently available to the public and NDF-RT would not change that at all. At the same time the people have talked about whether NDF-RT or RX-Norm are somehow competitive with or would somehow displace commercial knowledge databases and Kent Spackman brought up the issue of what’s the dividing line between knowledge base information and reference definitional information, I think he and Stuart were sort of saying well, that’s a soft line. And it might be soft a little bit. VA has taken the approach driven by use cases and those are our use cases that we have to deal with that there’s certain information we will put in our NDF-RT which may fit some other people’s definitions of a knowledge base, but the reason were putting that, the reason we have put that information in is because our current clinical systems depend on it.
At the same time we anticipate the need to buy quite a bit of commercial drug database technology and information in order to supply other needs. VA, for instance, I was writing an email about this a couple of weeks ago trying to explain this to our managers and they were saying well, gee, if we have NDF-RT why do we need say First Data Bank or Multum or Medi-Span, and I was trying to explain that well, VA may not want to be in the business of maintaining, for instance, and this is just one very small example, the patient information handout sheets. We would want to buy that from somebody else and have through RX-Norm a mapping to somebody else’s drug terminology so that when a patient got amoxicillin 250 mg oral tablet and RX-Norm form of drug dispenses, that we could find the right amoxicillin handout sheet to give the patient, and that would extend to 20 or 30 other items of interest beyond handout sheets as well.
So we in VA don’t view RX-Norm as precluding in any way purchase of commercial drug knowledge databases, in fact our model, although again compared to Kent, we may draw the line between a knowledge base and a definitional attribute a little bit differently partly because of our legacy that we’ve had to comply with, we have a model very much like Kent has on the screen right now where we have NDF-RT mapped to SNOMED-CT, and Kent’s right, we have to have serious discussions about how to do that. We have commercial sector knowledge databases, the cylinder next to SNOMED-CT there, and then there are also other knowledge databases that are in the public sector and those are some of the legacy NDF, our current legacy drug system knowledge bases, that we’ve imported into NDF-RT because we have to because otherwise our systems break. It’s not our goal to extend NDF-RT like an ameba until it encompasses all of drug information, in fact that’s explicitly not our goal, and I thought it would just be helpful if VA clarified that. Thanks.
DR. COHN: Stan’s next.
DR. HUFF: So this is a question for Medi-Span. I was interested that you included as part of the knowledge base and not as part of the reference terminology the composition of the medications in terms of their active and inactive ingredients, and I just, especially in the case of active ingredients, I wondered how you maintain, how you recognize synonymy and substitutability if that’s not a part of the definition of the concept.
MS. ECKERT: We included in the definition of the concept but then we also make them available as a complete listing of what all the ingredients are and that was what we included on that slide as the attribute. But we take the summarization of how that product is comprised when we create the therapeutic, well we look at the therapeutic classes but when we create the GPI and when we’re saying this is a group of ingredients that make this up we do take that into, I mean that’s our source of the information but we also make it available separately. Are you saying that you think it should be part of the terminology that’s available?
DR. HUFF: I’m saying, in the slide that was presented, the question was what is a reference, what’s the reference terminology part you provide versus a knowledge base, and what you said was part of your knowledge base and not part of the reference terminology was the active ingredients, they were part of the drug. So I guess you clarified and said no, well, I’m not sure I understand yet. In a reference terminology would you expect to see the active ingredients for a clinical drug or a generic prescribable drug as part of the definition of that concept or not? And if not how in the world can you recognize synonymy and substitutability?
MR. DUBOIS: I think perhaps the slides were a little bit misleading. Certainly we look at an ingredient whether it’s active or inactive as an attribute of a drug item, and that’s the bottom line, the key to it being is that then we use that information as Karen said, to define that generic product identifier that we talked about which is the proprietary therapeutic classification.
DR. HUFF: Thank you.
DR. COHN: I guess I have a question, it’s maybe a little bit off the point, I’ve sort of reflected throughout this whole conversation today that really the quality of all the drug terminologies is really based on the number of drugs it has in it, that isn’t the only characteristic but obviously comprehensiveness is an attribute that I think we all seek. We heard this morning that I guess the RX-Norm is based on, I guess it was 80,000 NDC numbers that appear to be, at least as I understood it, all that the FDA had, to basically supply for that. I heard later on somebody referenced that many of you because of the fact that you’re working very closely with the industry have upwards of 140,000, 150,000 drugs and formulations with unique NDC codes that you have in your systems, is that true?
MR. BLAIR: Could I just mention something?
DR. COHN: Did I misunderstand?
MR. BLAIR: Well, I think if Stuart is still here, he indicated yes, you are correct, that they got 79,000 out of the 80,000 NDC codes.
DR. NELSON: I think you misunderstood. RX-Norm was not built out of the NDC codes and I wanted to correct Mark. Mark, we do not have any packaging information in RX-Norm, so I don’t, I think that’s a misapprehension on your part. RX-Norm was built out of the drug terminologies, what we did is we took off of the FDA site in this last go around, we took off of the FDA site the ones that they had in their database. We took that as a separate drug terminology and then fused it all in to the rest of RX-Norm. The coverage of RX-Norm includes all of the VA NDF, all of Multum, all of Medi-Span, and all of First Data Bank at the generic level. Okay, now some of these sources have included in addition to that, some brand name information which we’ve also incorporated. And then what we did is we looked to the site. Now Steve Brown at the VA, they have 120,000 NDC codes, with probably very little overlap with the FDA’s database because the VA doesn’t report its NDC codes to the FDA. Does that make it a little bit clearer?
DR. COHN: I asked the wrong question to the wrong person.
MR. BLAIR: Did you also pick up the drug codes that had been from the Read classification system as well? I thought you did.
DR. NELSON: We do have the Read codes, yes, and the drug names from the Read codes. They don’t have anything like an NDC but we do have those. We have not viewed them as something that was primary for our purposes of trying to parse out and represent in a standard format the name of a drug, it was one of those things that we said, the low hanging fruit if you will was the NDF, was Multum, was First Data Bank, was Medi-Span, that what’s in SNOMED, what’s in the Read codes, what’s in MeSH for example, there are clinical drug names in all of those, there are clinical drug names in many of the different, the 100 different vocabularies of the UMLS, a lot of them have some amount of clinical drug names. All of those we view as targets of opportunity to --
MR. BLAIR: Stuart, if we were to look at things, if the NCVHS were to look at things in terms of the broadest possible use, is public, private, but the CDC, the Department of Defense, and to some extent the Department of Defense and CDC really has to look at things internationally so the fact that you have those codes from the United Kingdom I think is a positive consideration. Am I overstating that or is that incidental or is that not --
DR. NELSON: I think at this point, Jeff, I’d have to say that’s incidental, we have, the Library has a great deal of concern and every two weeks I’d meet with Donald Limburgh(?) about UMLS and it almost inevitably comes up, he says what are you doing about foreign drug names, and I have to answer Donald Limburgh every two weeks about that, so I’ll show you my scars.
DR. COHN: And I see Tim wanted to comment.
DR. MCNAMARA: I just wanted to question if I would the assumption of your question, you seemed to indicate that the more NDC’s perhaps the better the product or the better the coverage, and I think that that may not necessarily be an assumption that I would agree with. There are a lot of what I would call fake NDC’s, ten digit numbers, 11 digit numbers, running around that may or may not have anything to do with drug products, or may also represent non-drug products that have been included into some organization’s databases. I think it’s probably not the problem that we have too many NDC’s, we’ve got too many, what we really need to do I think is reduce and improve the quality. So for example at Multum we try to make an editorial policy where we only include an NDC if we can verify it in two different locations, or if we can find a price for it, because it would assume that if there’s a price there’s probably a good chance it’s a real code rather than a made-up code as so many of our clients actually do, a lot of our clients have made up a number of codes. So I’d really challenge the assumption that the number is probably not necessarily indicative of quality.
MR. ROBINSON: Just a quick comment on that, it’s the use case again that I’m driving home that we need to define, if you’re trying to do retrospective analysis and you’re going back several years then you may need a historical database that may be upwards, I think First Data Bank 240,000 NDC’s, but a lot of those are historical that have not been updated for years of course because they’re off the market. So again, what is this animal that we’re trying to define as a standardized vocabulary and we can go from there.
DR. COHN: Well, thank you all for correcting my misunderstanding. Michael and Clem and I think we’ll wrap this up.
DR. FITZMAURICE: This has really been interesting because I see in front of us a group of people who recognized a need in the marketplace, maybe even before the government did or maybe simultaneously, but they went out and they did something about it, took a risk and invested their own time and finances and developed new products. As I understand the testimony you wouldn’t mind if the government identified drugs, put out an identifier, and even mapped it to your identifiers, that’s part of terminology and that could be useful. But when the government starts adding overlapping clinical information that you also develop that can get threatening, and I kind of like the word threatening rather than anti-competitive or whatever else. So let me pose a scenario, lets suppose the FDA has a drug identifier, the FDA receives a package insert information --
PARTICIPANT: FDA?
DR. FITZMAURICE: Food and Drug Administration, receives the package insert information from a manufacturer, and links it to the drug identifier with FDA, NLM. And then to improve patient safety and physician education puts it on a website. Would that be threatening to the kind of information that you develop?
DR. MCNAMARA: I think that would serve our clients very well and we would be very happy if that occurred. We’d like to be part of that process if possible.
DR. FITZMAURICE: How about the rest?
MR. ROBINSON: I think it would be useful to us, again, the concern from the FDA if they’re doing that data collection and making it available to us in a codified format, then that would make our job and collectively all of our jobs easier. The additional value we add to that is perhaps we would add attributes such as frequency, severity, additional value to these concepts that would allow an institution, for example, to determine if there was going to be a hard stop with this potential side effect or this drug interaction. I think part of my concern right now is that we know that in lieu of this information being available from the FDA we’re spending a tremendous amount of resources to capture the information the hard way and what we would not like to see would be the government to subsidize someone, other than perhaps the FDA or someone to produce the information, make it available as a competitive product, so it’s one thing if it’s collected in a format that everybody can use it and perhaps we’d have a more efficient operation.
DR. FITZMAURICE: Do you feel the same way?
MR. DUBOIS: I feel exactly the same way, we would like to be involved in the mapping portion of it --
DR. SPACKMAN: We’d like to see the FDA, SNOMED code, indications, contraindications, and other things and distribute those to everybody as well.
DR. FITZMAURICE: Thank you, that’s really very helpful for me to understanding what the government role is and what your role is.
DR. COHN: Any final questions? Oh, I’m sorry, please.
DR. MCDONALD: I have two questions and observations and forgive me if I present these, what I’m about to say. As I understood it we just about had a standard when you two guys merged because you’re basically, the SEC didn’t like it because you’re so big, so as I remember the numbers you did and you probably still do represent 80 to 90 percent of the drug code usage in the country on this side of the room, I don’t want to smear anyone on the other side, and so you have sort of different interests, but you also have been in the business a long time. I guess the other side, and I want to throw this together with your suggestion I thought should get more discussion, you really threw out a fairly radical thought I thought, and I just wondered whether we could stimulate anything that would get everyone jumping into that pot, namely finding that core whatever it is to make this, as I say it I know all the complexities and maybe the hopelessness of this, but a core whatever it is, that there would be one of code, but not owned by anyone, that people could hang onto and do in their communications. Just a little, if there’s any, it’s a tricky question or maybe a trick question.
DR. SPACKMAN: Because we don’t see ourselves actively doing that particular piece right now, I mean I see it as something that is really important. Again, I hate to harp on this theme but we’ve seen what’s happened in the UK with what they need to do and the set of drug classes that they’ve developed and the drug information that they’ve developed were designed specifically to serve all these use cases that we’re talking about, pharmacy and inpatient and outpatient and so on, and I don’t see anybody, including SNOMED right now, providing a single public identifier for those pieces, right, they’re all commercial. My sense of things is that if we go beyond the RX-Norm level down to these package levels what we’re doing today publicly isn’t enough, so I would see it as a very positive thing if FDA would look at that and say oh, we could maintain it. So I think it’s an excellent idea.
DR. MCDONALD: And then one other question, it’s actually for you Kent, is that when you described the RX-Norm, actually -- should get in that, RX-Norm and SNOMED plan a happy sort of life together, what would the advice be, we get messages, we get probably 200,000, 300,000 messages a day from various systems in Indianapolis, and the codes count, I mean it’s a pain in the butt dealing with all these messages. What advice would you give the people of which code to use between the two? SNOMED or RX-Norm for the drug entity, if they were equal?
DR. SPACKMAN: I think that’s one of the hard questions we have to answer. I think the best solution, the optimal solution would be to have one selected, a sort of sub-optimal one would be to say well, you can use either one and we’ll translate between them. Obviously from a SNOMED perspective we want SNOMED codes to be used, but there are ways of talking about dividing up the world that possibly wouldn’t result in that, but we have a whole set of other considerations that would come down on us if we did it that way, so we have to talk about it.
DR. COHN: I think we have on actually the agenda Randy, were you going to make a couple of comments? I actually see you on the agenda here. I presume you are our FDA representative. Do you want to come to a microphone? Actually, I want to thank our panelists, it’s been a wonderful presentation, and certainly you don’t have to leave, you may have some comments after Randy.
DR. LEVIN: They can or cannot make comments after me?
DR. LEVIN: Well, it’s very nice to have everyone want us to do so many things, so that’s good. I guess the things that I’ve been hearing is that people would be interested in the FDA becoming the source of this information, collecting it from the manufacturers, the terminology for ingredients and things that we talked about this morning, and labeling information. It sounds like the First Data Bank hires 20 plus people to do this activity, maybe they would lend them to the FDA to help with that, but it sounds like that is just something that they would want from the FDA and this is basically what we’re interested in doing and this is what I had proposed this morning is that we would be, as part of our process, that’s part of what we do, is that we interface with the manufacturers to collect information about every marketed product in the United States. And that’s what we do in what we call drug listing is we take, get the information about all the active ingredients, inactive ingredients, appearance, all those things that are now, that First Data Bank now collects and the other groups collect and put in their systems, and put in their knowledge bases to feed to their customers, and what we were interested in doing is that if we collect that information then provide that to the information suppliers to provide that to their customers. We’re not interested in providing that information directly to meet all these customers needs, that’s what these people who were just up here, that’s their role to do that, but we want to be able to provide the information, and we think that that would be more our role. That’s part of our process for drug listing, for labeling, that’s what we do for labeling, we review the label, we get the label, we work with the manufacturer to get the proper label, and we put that information in, we want people to use that information.
Right now it’s in paper, people are not using that information, they go off and they start to collect it from other sources, and I hear it from people here that if we had it in some way that they could easily access that information that that would be very beneficial, and that would be beneficial to us as well because then we will be, people would be using the information that we work with the manufacturers to produce about every drug. So this was, the comments were very encouraging and I think will fit with what we were, our idea that we would be supplying this information as part of our process of what the FDA is doing with regulating drug products and these companies, as well as other people, would be using that information to support the various uses that are needed throughout the health care community. So I don’t know, people might have specific questions about that?
DR. COHN: Comments from the Subcommittee?
DR. MCDONALD: I would just say go to it.
DR. COHN: I guess maybe a question I’d have for you, and this is maybe just a little bit of confusion with the tenses from this morning, is this all dependent on the completion of new federal regulations to allow this to happen efficiently?
DR. LEVIN: Well, there are two processes that, our main process, one is what we call the drug listing and the other is drug labeling. For the drug listing part, for the drug labeling we’ve already put forward a proposed regulation so that we could handle that more efficiently, that means the companies would come in with a labeling in electronic format and then we would be able to more efficiently put it out in electronic format for us, so that part we’re on our way, but that needs a regulatory, that regulation has to go final and our processes have to be put in place for that.
The second piece is drug listing where there was a lot of discussion here about some of the problems with the NDC and confusions about what’s the proper NDC’s, why the made up numbers, I have X number, you have this number, we feel that there should be one set of NDC’s that we should really, we should know all the NDC’s and that there’s a problem with the way that the NDC’s are being managed at this point, but to change that would require changes in regulations, so that would be something that we’re working on and that would have to move forward and be done in order for the NDC’s to be under control. So the rules and what is an NDC and how is it changed and things like that would be properly managed, but that would require regulatory change.
DR. SUJANSKY: Randy, that last piece you were describing, the regulations about the new listing process, the desired process, in the best case scenario, in the most optimistic scenario, how long would it be before those regulations are in place?
DR. LEVIN: You probably know, some of the people on the panel here know more about the problems with the regulations or the process. A lot depends on the need and the priority because there it goes to the process of going through regulations that’s in a line, a queue with other things, and whether it’s controversial or very much needed and people want to push for it, there’s a lot of factors that can move regulation through. And one of those just recently was the bar coding, a regulation to require barcodes on products, that by the way would code the NDC, which is one of the reasons why you would want to code NDC on that barcode, and that went through relatively quickly. I mean as far as was proposed and now is out, for a proposal that went fairly quickly.
DR. SUJANSKY: How long did that take?
DR. LEVIN: I don’t have the exact number of months but it was not a long time.
DR. SUJANSKY: In any case there’s 18 months after the final regulation for, typically --
DR. LEVIN: There’s an implementation phase that would have to go, like for the electronic labeling implementation is six months, so after the regulation is approved, it’s finalized, it’s six months and then it’s implemented, so that’s relatively quickly. So it has to be approved so if it goes through and it takes a year for a rule to be finalized then there is an implementation phase. Depending on how burdensome it is to implement -- it can extend that, it can be very long. If it’s not a burden, though, for industry or people to follow through with the regulation then the implementation is not as long.
DR. COHN: I think we need to wrap up here pretty quickly. Are there any burning questions? Ok. Randy, thank you so much, I think we need to adjourn, we’re running about a half hour late at least as I call it, we’ll come back in 15 minutes.
[Brief break.]
DR. COHN: Now we shift to a discussion around medical devices.
MR. BLAIR: We should indicate that after the medical devices, we’ll be talking about medical devices and then the Subcommittee as a whole after that will have a period when we will be discussing what we’ve heard from all the testifiers today and see to what degree we could come to consensus as to what we should be doing, so that remains for the day. Are our testifiers here? Let me let you introduce yourselves and just go ahead and proceed.
DR. HEFFLIN: I’m Brock Hefflin, I’m a medical officer at the Food and Drug Administration and the Center for Devices and Radiological Health.
DR. KESSLER: I’m Larry Kessler, I’m the director of the Office of Science and Technology, and Brock and I are going to present together.
DR. HEFFLIN: Okay, so this presentation is a focus on the Global Medical Device Nomenclature, which is a relatively new system of names and definitions for medical devices that was developed primarily through established terminology. I’m going to get you a brief history of --
MR. BLAIR: Maybe there’s one thing and I’m afraid it’s because I can’t see, so I would hope that our folks who testified on the drug terminologies would remain as late as they can today because after we go through the medical device testimony we’re going to be considering what we should be doing on all testimony for today, and if there are questions it would be helpful if the testifiers are still here.
DR. HEFFLIN: Okay, again, the Global Medical Device Nomenclature or the GMDN as we call it, I’m going to give a brief history of the GMDN project and then describe its development and its current status, and then Dr. Kessler is going to come forth and discuss the issues that you see listed here.
Planning for the project took to develop the Global Medical Device Nomenclature was initiated in 1993 by the European Committee for Standardization to meet the European Union Medical Device Directive requirements for product certification and registration, vigilance reporting, and product recall. In 1996 the International Organization for Standardization became involved with the project, expanding its international scope, and at that point in time a draft standard was developed that was for a medical device terminology. And here you see the references for that standard. And the GMDN project itself was actually initiated in 1997.
The GMDN standard provided for the basic structure, the basic naming convention, and the data file requirements of the nomenclature. Regarding the structure there was to be several device categories populated by many generic device groups. The standard also describes device types however the nomenclature doesn’t get that specific, it stops at the generic device group level.
The GMDN was created through the merger of six established medical device terminologies, the Universal Medical Device Nomenclature from ECRI, the Norwegian Nomenclature, the medical device nomenclature system used by the Food and Drug Administration, the nomenclature used in Japan, and two specialized nomenclature systems, the European Diagnostic Manufacturers Association’s In Vitro Diagnostic Product Classification and ISO’s Technical Aids for Disabled Persons Classification.
Now when the six source nomenclatures were merged it resulted in 13,000 terms for generic device groups, a generic device group being a set of devices of similar intended use or technology. And as you can imagine when the six nomenclatures were merged there were multiple terms for a number of generic device groups. Take for example Carbon Dioxide Absorber, there was an ECRI term, and NKKN term, and an FDA term, all for the same concept and we needed to identify these equivalent sets of terms, so that was done using the National Library of Medicine’s Unified Medical Language Systems Specialist Lexicon, which is essentially a computer that groups similar, that groups terms based on similar words. In addition to that a manual view needed to be performed to make sure that that grouping was accurate.
Now all the 13,000 terms in the candidate list were distributed among these 12 medical device categories. Every term was assigned to a primary category, and some terms were also assigned to a secondary or tertiary category. Now assignment to a medical device category was based on the intended use of the device and the specificity of the category. For example a dental probe could be placed in both a dental devices and the reusable instruments categories. However, dental devices is more specific so this became its primary category and reusable instruments became its secondary category.
Here you see a scheme of the project organization, there was a project council, an expert advisory term, and 12 device expert task groups, one for each of the medical device categories you saw on the previous slide. The device expert task groups were the workgroup, the expert advisory team provided technical oversight, and the project council was the administrative group. All of these groups were international. FDA staff, there were six FDA staff involved in the product overall, and FDA staff was involved at each level of the project organization.
Now the device expert task groups or the workgroups were instructed to select a preferred term to represent each generic device group in the set of terms that they had to work on. If no satisfactory term existed then they were asked to create a preferred term. They were asked to link the remaining terms in the equivalent set to the preferred term as a synonym term, and we defined synonym as a common use or familiar term for users so that they would be able to navigate the system, or an equivalent term, and an equivalent term was as term that was essentially identical to a term in the nomenclature, it didn’t add any additional information, so it would not be part of the functional GMDN. Then they were also asked to create synonyms as they thought necessary.
Here we see an equivalent set of terms, each term coming from a different source, and this example, the device expert task group decided to make the ECRI term the preferred term, and they linked one of the remaining two terms to that term as a synonym term, again, so for individuals who are familiar with the Fundus Camera would have an entry into the system and be able to find the preferred term Camera, fundus, and the remaining term they linked as an equivalent term because it’s identical to this and it doesn’t provide any additional information.
Here we see another equivalent set of terms, each term coming from a different source, however in this example the device expert task force group wasn’t happy with any term as a preferred term so they created a preferred term and linked the remaining, the terms on the candidate list as synonym terms, in addition they created a synonym term, so that users of laser, image, or multi format would have an entry into the system and they’d be able to locate the preferred term.
The device expert task groups were also asked to create definitions for each of the terms, each of the preferred terms that they worked on, and those definitions were to express the device intended use, the target area of intended use, they were to describe the device technical principal or working method, to describe materials or components involved, and to describe the device form, shape, and physical state.
And here we see an abbreviated version of such said definition for Camera, fundus, it provides a brief description of the device, a box-like device that holds photographic film, and the intended use of the device, is used specifically to focus and record magnified images of the ocular fundus viewed through the pupil.
Now the work of the device expert task groups was passed on to the expert advisory team and it was the work of the expert advisory team to reconcile inconsistencies in their work. In this example here we see a list of terms all with the same base concept, of course part of the term, the base concept being camera, however you see it coming from three different device expert task groups, electro medical mechanical group worked on camera terms, the radiological group worked on camera terms, the ophthalmic group worked on camera terms, and there were inconsistencies in the way that they developed the terms and the definitions and those inconsistencies need to be reconciled. So the device expert task group did that. In addition, when three or more terms had the same base concept another type of term needed to be entered into the system, which we called a template term which you see there at the top, it’s identified by the specify in brackets.
And here you see an abridged version of the GMDN. In this column we have the template terms, which provide a simple hierarchy for the system. In this column we have the preferred terms, and in this column we have the synonym terms linked to the preferred terms.
So the database that was used to create the GMDN was inactivated in mid 2001 and at that time there were deficiencies noted in the nomenclature system. However, in September 2002 that database was reactivated and since that time we have worked to standardize the base concepts of all the terms in the GMDN so that they reflect the intended use of the device. Before that time they didn’t, there were some that did, some that didn’t, but we’ve worked to correct that. Thousands of definitions have been improved so that they incorporate all the aspects of the standard that you saw previously, especially definitions from the IVD terms. Many new synonyms have been linked so that users can navigate their way through the system, and preferred terms have been added for new devices, all in preparation for the second version of the GMDN which is due for release later this year. So the GMDN currently is comprised of about 17,000 terms, that’s 6,400 preferred terms that are all defined, 10,000 synonym terms, and 600 template terms, which help to create simple hierarchy in the system.
I’m going to turn things over to Dr. Kessler.
DR. KESSLER: I’m going to go back for just a second because a lot of people, when they think about devices especially after they’ve heard a presentation about drugs, they think they’re talking about the same animal, and devices is very different, the kind of devices we deal with span a broad range of things that don’t look anything like each other, don’t act anything like each other, don’t get approved like each other, don’t get priced out like each other, don’t get dealt with anything like each other, and they’re all over the map because we regulate things as simple as Band-Aids and gauze and condoms and latex gloves and as complicated as MRI machines, implantable cardio converter defibrillator, and trying to find these things in various places is a trick. So you need to make sure when you’re thinking about devices, if you think about it with a drug paradigm you’re going to wind up in the wrong place, so I just want to start reorienting you for just a second, because we’re talking about active implantable devices, reusable instruments, technical aids for disabled persons, a table in an operating room is a device, but so is everything the clinician touches from the minutes he or she walks into that room to the minute they walk out, so it’s a wide range of things that we deal with in very strange ways, and the FDA doesn’t see many, many, many, many of the products that are used in hospitals and doctor’s offices, we don’t see them, we don’t see paper about them, they’re manufactured by manufacturers, they are provided to hospitals, the clinicians, and they’re used, and we don’t see anything about them other then when we inspect the facilities once every seven to ten years. So think carefully about the level of regulatory control that FDA has about devices, it’s very different then drugs.
So I’ll talk real briefly because I think the slides are self-explanatory about the maintenance agency which is how the Global Medical Device Nomenclature group has been set up to run and operate the GMDN. There’s a policy group, which consists of regulators, ISO and CEN members, these are the standards organizations that supported the building of the GMDN largely, industry representatives, the Commission, that’s the European Commission, the Global Harmonization Task Force, which is comprised of industry and regulators around the world that harmonize regulations about devices. There is experts, FDA, NKKN, Health Canada, Academia, and ECRI has recently been reinvited to step back into GMDN, they originally helped us build GMDN, there has been a hiatus in their participation, we’re hoping to fix that. And the secretariat, up until recently, has been the British Standards Institute, and we’re transitioning to a commercial company just this year to help manage GMDN.
The roles, the policy group provides oversight and sets policy, the experts will develop and amend term, and frankly they will control the terminology, as Dr. Hefflin and his team have done over the past year, and the secretariat will maintain/publish the GMDN, receive proposals, develop user guidance, publicize the program, try and make it work.
GMDN use will require licensing through the maintenance agency. We’re hoping to get voluntary sponsorship by regulatory agency to assist in maintenance costs, some money has come from the FDA, in different ways the GMDN, as well as from other regulatory bodies around the country. You can see more information about the license, how we’re going to distribute the data file. The regulatory bodies, which have statutory responsibility to provide terms so that industry can function will allow people to step in and use a GMDN term essentially one at a time. We will not allow under things like FOI the pillaring of the GMDN, so an industry can’t call us up and send us the whole GMDN, it’s a proprietary product licensed by the Commission and by the standards organizations in Europe.
Data maintenance. The maintenance agency must be contacted to modify the data, we will not allow modification by anyone who has it, it will be controlled through the maintenance agency. The data will be continuously updated by the expert term because the device world changes very rapidly. Requests for new terms will be dealt with as received and responses will be provided within a few weeks. And hopefully there will be at least one new version per year, and we also hope not more than that.
I mentioned regulatory body voluntary sponsorship, there will be licensing fees for other users and the secretariat will use GMDN as a trade name in association with the same of other products, that is other software companies we hope will incorporate GMDN under license from the secretariat so they can generate some money to support the way the secretariat will have to operate.
There’s some issues related to UMDNS and I want to point out that there’s an error in the handout that you have from an earlier slide, so we have sent and we’ll send again a more accurate version of this and it’s in the slide somewhere. So GMDN is an international nomenclature, it provides generic descriptors for devices, it’s based on an international standard, ISO 15225. The UMDNS was one of the original nomenclatures used to develop GMDN and at its inception, when we started GMDN UMDNS did not cover all devices, so we had to expand the scope of GMDN to make sure we covered everything. Since that time, and Vivian Coates will tell you, that UMDNS has expanded considerably, probably quite parallel to GMDN, although we think we still have a wider scope in GMDN but Vivian will talk about that when she gets to her presentation.
We’re attempting to develop plans to merge with ECRI’s UMDNS over the next three to five years, I don’t think it’s going to take any shorter than that, so that we hope in the next period there will be one indeed truly global medical device nomenclature.
We’ve been investing in the GMDN structure not only by personnel time but by our agreement to map the FDA system of product codes to the GMDN, it’s beginning this year, we’re going to convert our systems to the use of GMDN next year, and then develop a web interface for industry and consumers, be able to find product codes using GMDN terms and searching the FDA database in 2004 and 2005.
Here’s some information, relatively recently from the maintenance agency in Europe GMDN has used 70 experts from 16 countries to develop and maintain it. It has been adopted for use by the European Economic Association, it includes 18 members plus the 10 Eastern European countries. Japan has adopted GMDN and is translating it. The Australians have put it in their regulations, GMDN is required as part of the Australian registration system. Both South America and the Asian groups have begun using it and the GMDN has been adopted by the Global Harmonization Task Force as a means for establishing generic device descriptors principally to exchange information about devices among countries. And I can talk about the importance of that if you have questions later.
So that’s the end of our presentation, I don’t know whether you want to do questions or turn to Vivian, it’s up to you.
DR. COHN: Let’s turn to Vivian.
Agenda Item: Panel of Medical Device Terminology Testifiers - UMDNS - Dr. Coates
DR. COATES: Okay, well thank you for inviting me to come here today to talk to you about ECRI’s Universal Medical Device Nomenclature System, and as Larry Kessler has aptly said the world of devices is very different from the world of drugs. I like to think of it as subsuming things you can pour down the sink, things that you can walk inside of, and things that get put right into patients and many other things besides. I’m ECRI’s vice president for information services and technology assessment.
Just a brief word about ECRI, we just don’t develop terminology, we also create a lot of the knowledge databases on devices that utilize the terminology. We’re a non-profit, we’re a collaborating center for the World Health Organization with quite unique terms of reference. I believe at this point we’re the only WHO collaborating center with a specific purview for patient safety and health technology assessment. And we’re also an evidence based practice center as designated by the U.S. Agency for Healthcare Research and Quality. We have about in our headquarters about 250 full time interdisciplinary staff and we have offices around the world.
Since 1977 we’ve operated an international voluntary device problem reporting system, and to this day we continue to do onsite accident investigations all over the world, so we’re right there when there are mishaps and things going wrong due to medical error and other causes with medical devices. And we conduct a lot of applied research onsite, if you came to our facility you’d see laboratories, and we have a consumer reports like activity where in those laboratories we’re testing and evaluating medical equipment every day. And of course we’ve maintained a universal medical device nomenclature system since 1971, so we’re the old kid on the block.
Our WHO terms of reference, I mentioned the purview for patient safety, but also specifically to develop, maintain, and promulgate worldwide a universal medical device nomenclature system, facilitate its translation into various languages in order to enhance information exchange on device safety, procurement, and management within the global harmonization efforts.
And our approach to device nomenclature is that we’ve established a formal hierarchical system for organizing all kinds of device related information, it’s not simply a list of terminology, it’s a database, it’s dynamic. As Larry and Brock have mentioned, this kind of domain has to be reflected, it has to be supported by terminology that must be maintained and updated continuously to reflect the constantly changing world. And it needs constant monitoring to maintain and improve quality based on user feedback.
It’s been part of the Unified Medical Language System since 1991 and since that time we’ve maintained mappings and linkages to other device terminologies such as FDA’s original terminology, and as Larry and Brock have said GMDN, and also to the eCommerce world’s UNSPSC, the United Nations Standard Product and Services Codes. We’ve also mapped from UMDNS to other medical vocabularies in UMLS, including ICD, CPT, MeSH, SNOMED, and Read. And as of the 2004 edition, which we just put to bed this month, it contains 17,221 medical device terms, of which 6,745 are preferred terms at the same device group level in the data model of these standards that GMDN is, and we have 10,476 entry terms, so we respectfully disagree with our GMDN colleagues that we’re less broad in our coverage.
In addition to the preferred terms and codes and entry terms we have a hierarchy. We also encode and maintain intra-nomenclature relationships, parent/child as well as sibling. We have attributes and definitions in addition to the mappings and linkages. We use a five digit unique numeric code/identifier that corresponds to each preferred term. These are assigned randomly and sequentially, and the codes intentionally do not carry inherent meaning. This is an example of a little grouping on stimulators and how we have many, many, many broad terms that get expanded out to be a very precise level of specificity that we feel is absolutely essential to capture important device information.
We take a concept orientation and by that I mean that a given medical device concept with given identifier corresponding to it is then described by the word terms, such as shown here for stimulators, electrical, neuromuscular, gait, implantable, in the form of a plural noun that’s modified by a string of adjectives or adjectival phrases. And this particular example is a child of apparent concept with its own identifier code and described by the term given here. And this precise concept orientation leads to non-ambiguity of concepts.
The entry terms we consider to be extremely important because they provide user friendly entry points to the terminology, and the entry terms include not only synonyms but quasi-synonyms, lexical variants, initialisms, etc., and the entry terms can include imprecise but commonly used terminology and/or extremely broad commonly used terminology, or extremely narrow, very specific concepts that in and of themselves are not appropriate for preferred terms, such as ophthalmic lasers, MRI units, Q-tips, etc.
Now we encode the hierarchical and other relationships between terms, as I said before, it’s not just a question of synonymy, entry terms or preferred terms may point to one or more related preferred terms, here’s an example. We also precisely define the sibling relationships among all of the preferred terms and from the entry terms to the preferred terms.
With respect to the hierarchy we’re developing a higher level series of categories that’s broader then the level represented by the preferred terms, so for example similar use devices breaks down into a number of different things including catheters, and catheters in turn break down into a number of different types of devices. And we have several hundred of these middle level categories currently under development.
We also maintain formal definitions for each preferred term or coded concept. We construct the definition in a standardized way that describes the device and its intended purpose. Older terms created before 1990 did not have definitions so we’re in the process of adding definitions for all of these. Our database tracks many, many attributes and characteristics for particular device concepts and each preferred term is linked to multiple attributes in the database, such as whether the device is implantable, whether it’s capital equipment, multiple clinical specialties it may be used in, it’s price, and FDA regulatory class.
We also believe very, very strongly that we do indeed adhere to concept permanence. We were notified by Dr. Sujansky earlier this summer that UMDNS fails on concept permanence, and we respectfully disagree with that. UMDNS concepts are very removed, but they may become obsolete or retired, so it’s an issue of their status changing. Codes for retired concepts and terms are never reused for new concepts and in most cases a retired preferred term becomes an entry term that points to the new preferred term or terms, and we maintain an archival list of retired preferred terms in the master database.
Here’s an example. We originally created a whole grouping of laser terms to represent the devices clinical function so we had lasers, ophthalmic, lasers, surgical, etc., and as the technology evolved over a period of years we retired these terms as preferred terms in favor of a classification approach that really is based on the structure of the lasing media, so we have lasers, homeum(?), itrium, aluminum, garnet, and that breaks down to lasers, Ho:YAG for dental, for ophthalmic and for surgical applications. The natural language expressions ophthalmic lasers and surgical lasers are now entry terms to the appropriate preferred terms. Annual releases of UMDNS include the list of preferred terms retired since the previous version, so we don’t automatically disseminate a list of all archived retired terms when we disseminate the new release of the terminology.
As we said before, the terminology is broad in scope covering all medical devices and supplies, clinical laboratory equipment and reagents, hospital furniture, systems, test equipment, and since 2001 we have expanded the in vitro diagnostic area, adding thousands of new concepts for this area, but also to expand in other areas, disposables, aids for the disabled, emergency preparedness and emerging technologies. So right now we also maintain a related dataset that includes unique standard names for more than 17,300 device manufacturers and suppliers and service companies, and their associated ECRI assigned six digit codes. And one of the things that we’re able to do is to link the clean dataset of companies that produce devices with what they produce.
I’ve been asked to just comment on where UMDNS sits relative to SNOMED’s device descriptors, device concepts, in looking at SNOMED’s coverage of the device area. We believe that UMDNS is much more comprehensive and specific, and that SNOMED’s device domain is not adequate to capture the kind of precise device concepts that you need both for the purposes of the electronic medical record and also for patient safety and medical error reporting. And this Insulin pump example is just one small example where the concept represented in SNOMED to us is ambiguous and that you really need to have some related and far more specific concepts.
In terms of how we license and distribute UMDNS we license it free to regulatory agencies, individual hospitals, manufacturers, but charge a fee for eCommerce sites and commercial software companies. The free version is available electronically as an ASCII download from the UMDNS web page and it’s updated annually, and we provide user support by fax, email, and telephone, and also support for translation efforts. The commercial license fee that we charge is really dependent on the level of support and also data that the licensee wants. We find a lot of times that application developers want very specialized formats, they want monthly updates, and they usually want our data on suppliers and their associated product lines. Commercial licensees do not pay a per user fee, but really a per institution fee and within the hospital, for example, there’s unlimited use. Right now two examples of this would be the agreements that we have with the application developers for the DOD, defense medical logistics support system, and the VA procurement system where they wish to load UMDNS directly into their software application.
Licensees are also free to purchase our web and print annual health devices directory database, that costs $500.00 a year and includes UMDNS preferred terms and codes, entry terms, definitions, attributes, and data on the 17,000 plus companies, encompassing about 3,000 pages of data in the print version. And we have published this annually since 1979, so we look on this as one of our other channels of distribution for UMDNS, and we’re evolving. We are now licensing the Apelon terminology development environment, which is the environment that SNOMED uses to maintain UMDNS, and we have a lot of additional mapping efforts underway to trade names and manufacturers’ identifiers at the make and model specific level to support eCommerce, data cleaning of hospital data and barcoding applications. And we’re adding additional terminology to represent the kinds of devices that are used in counterterrorism/emergency preparedness disaster planning, but also for facilities planning, for forecasting and emerging technologies and patient safety.
And we have a partnership agreement with FDA to work together to map respective nomenclature systems and use the same guidelines in establishing new terminology. And the status of this effort has a lot to do with the GMDN project, just as an aside, ECRI was a member of the GMDN project council and expert advisory team for five years from 1996 when we had signed memorandum of understanding with the European standards organization to use UMDNS as the interim standard while GMDN was developed. And the notion was that there would be one, at the end of the day, one device nomenclature and ECRI would use that and everyone else would use that. And so we very much devoted a huge amount of resources and effort to getting the GMDN project off the ground, we were the ones that did the mapping of the six original parent vocabularies to each other and contributed mightily to the effort. But at the end of the day when the current version of GMDN was released in 2001 it did not meet our quality standards and so respectfully we withdrew from the project feeling that we could no in good conscious adopt GMDN and use it to replace UMDNS given all of the contractual obligations that we had for providing a high quality nomenclature, but I’ll just go on from that.
So what is our current user base? At this point in time we’ve got 1922 signed license agreements for users in 95 countries, regulatory agencies for post market surveillance, vigilance reporting, data storage and retrieval, but some of these agencies are needing to adopt GMDN. I should say when a country, when there’s an agreement at the level of a regulatory agency to adopt the GMDN that does not mean automatically that all of the device nomenclature users in that country are going to be using GMDN, in fact they’re not, they’re using UMDNS, and I’m thinking now of the thousands of hospitals all over the world where that is the case.
Within the U.S., I mentioned that the U.S. Department of Defense and the VA is using UMDNS, but we have a host of private health systems, health care facilities using it. Outside of the U.S. similar kinds of organizations use UMDNS. What do they use it for? They use it for inventory control, for hazard and recall tracking, for equipment planning, procurement, technology management, for monitoring emerging technology, emergency preparedness, and they often are using it in conjunction with our software and knowledge bases. Manufacturers use it to classify their product lines, to facilitate communications with hospitals. I mentioned application developers -- since UMDNS is one of the UMLS vocabularies there are a lot of application developers who are using it in that context. The World Health Organization is using UMDNS, it’s imbedded in their Essential Health Technology Package software tool for resource planning for population based care in developing countries. And the eCommerce exchanges, such as global health care exchange and others, are using it to link ECRI data used by hospitals to eCommerce transactions, and we facilitate that via our mappings to UNSPSC.
In patient safety the Australian Patient Safety Foundation has imbedded UMDNS in their software system for coding adverse events reports and analyzing medical errors from all Australian Hospitals, and this APSF system is also being used now in the UK and other countries. And these are just some of the countries where we have users.
We’ve had a lot of experience with high quality technical translation and what it takes to achieve that. The Spanish version of UMDNS is maintained at ECRI and it’s endorsed both by the Spanish Ministry of Health and for Latin America by the Pan American Health Organization, and that was a difficult thing to achieve because there are a lot of language variations in a given spoken language and everyone believes that their version of the language takes precedence. These are some of the other translation efforts. I’ll just say one word about the German translation. Originally funded by the German Informatics Agency DIMDI, we’re now looking at updating that translation, again with the EMTEC organization. And we’re working now also on a new Russian edition and a Mandarin Chinese, edition. A word about translation, that it’s labor intensive and difficult, and it needs, best done using an interdisciplinary approach.
That’s just the cover of how we publish the Spanish edition. So I think with all of that that’s all I’m going to say at this point and I thank you for your attention.
DR. COHN: Questions? Betsy?
MS. HUMPHREYS: I have a couple questions about the maintenance plans for GMDN, first off, do I understand since responses are provided within a few weeks if we people request updates and you’re planning to release the new version once a year and that’s your preferred, that basically the updates that occur between those releases would be then what, published as updates on a website or something?
DR. KESSLER: Probably, the updates that we’re talking about are new terms. When FDA gets in a new product that we’ve never seen before we need to turn, we just can’t wait for a yearly update, so we will have to ask, the way other regulatory agencies will have to ask GMDN, give us a new term for this new camera or this new biopsy forceps, so we’ll need to get a new term for a while and that update will go probably into the database and be available through the web, and then the formal updates will happen every few months or year. We could not update as fast as we need, terms we need, terms every week.
MS. HUMPHREYS: So presumably then those will be published so everybody else who’s using it could get them, too.
DR. KESSLER: Sure.
MS. HUMPHREYS: Do I understand that you’re transitioning to a commercial company for maintenance of it so maybe some of this is the business about the signed license and the sponsorship and the validity of the license and everything, is this real bedrock or is this sort of the where you’re going? Is this in effect right now I guess?
DR. KESSLER: Is it signed? Is the agreement with United signed yet?
DR. HEFFLIN: It’s being formalized as we speak.
MS. HUMPHREYS: Ok, well, I don’t want you to announce something before it’s formal. The question that I have here is it sounds like your model for the maintenance of this is a combination of direct sponsorship by the FDA and other device regulatory bodies around the country, and then license fees and it is the combination of those two that will support the ongoing maintenance of this, is that the plan?
DR. KESSLER: Yes.
MS. HUMPHREYS: And I’m very suspicious of voluntary sponsorship as opposed to committed sponsorship when it comes to something that a lot of people might be relying on around the world and certainly across this country, I mean if it were designated as a standard for use here. It makes me kind of nervous.
DR. KESSLER: It makes me kind of nervous, too.
DR. HEFFLIN: There’s one more component. The secretariat, the commercial company that we made reference to, is to develop the business aspect of the GMDN, so that would be also a part of its funding as well.
MS. HUMPHREYS: Well, this sounds like this model is similar to the model for Medra(?), I mean is it sort of modeling on the --
DR. KESSLER: Not dissimilar, it’s not dissimilar, but we just think that the GMDN is going to have to be supported much more by regulatory agencies then Medra is.
MS. HUMPHREYS: Yes, I was going to say that an analysis of the usefulness of that model anyway might be good, but I mean just how well it’s working because I don’t know that anybody in the Committee or just generally knows whether it is even.
DR. KESSLER: They have some information, it’s not working so hot.
MS. HUMPHREYS: Well, I had heard those rumors. I do think that I would agree with you that having some committed core support for something like the maintenance of a terminology that we would could eventually become, I mean if it were designated for broad use in this country anyway could become something that lots of people relied on, that without some believable plan for ensuring that that takes place, that to my mind unfortunately puts a question around this entire activity.
DR. KESSLER: Let me just say two things. First of all we are negotiating that arrangement now along with the Europeans. Second, the way this whole thing was built was on money from the European Standards Organization and basically on the blood, sweat, and tears of people from FDA, the European regulatory agencies, and frankly the folks at ECRI, so that’s the way device nomenclatures have been built. They’re just not going to be big money makers for anyone, I mean ECRI gives it away relatively free to everybody, it’s not generating money.
MS. HUMPHREYS: There’s nobody who believes less than I do that a terminology is a big money maker, but I also believe that a terminology is expensive if it’s going to be something that’s really heavily used, so I’m not arguing with your model, I like the model, I just feel that the issue is what kind of an agreement can you reach in terms of ongoing support of this that would give everyone who might throw their hat in a particular ring some feeling of assurance going forward.
DR. KESSLER: We are negotiating an arrangement with ECRI to map all three nomenclatures together, FDA and our product code system, GMDN and UMDNS, over the next three to five years. I want you to link them, I don’t want to depend on the commercial venture of a maintenance agency for GMDN, if they fall apart tomorrow we need something at the FDA to rely on and our hope is that it will be, our negotiation and relationship with ECRI that will carry us in the future if this thing doesn’t work.
DR. COHN: Steve, do you have a question?
DR. STEINDEL: Yes, I was very interested in hearing the two discussions and Vivian, thank you, several of my questions you did clear up and I’m hoping that the FDA is in line with the thinking, but some of those questions had to do with what do you see as the reasons for a device terminology and you extended it much beyond the 510K(?) approval process and just classifying devices into things like product safety and recall notification, etc., and I’m hoping that the FDA is in agreement with that sort of use of device terminologies. Because I think that’s something that a lot of people are unclear about, they just think of a device terminology as just a list of products and it goes much more than that. So I was thankful for that.
I also, and this is in preparation to something that I’m going to be doing, I think it’s Thursday, with regard to CHI and I’d just like this to confirm, both of your terminologies do include an area where you can list clinical laboratory tests and in each for diagnoses. I don’t know what level of specificity because that is going to be a recommendation from CHI that that be developed some place, that it really doesn’t right now.
Now to get to my question, I couldn’t help but observing an extreme amount of similarity between these two efforts. And I’m concerned about that, and I’m concerned about efforts being spent in the two areas. I’m very thankful to hear, I was thankful to hear two things, number one, ECRI’s comment that they were involved in the development of GMDN initially and also I had heard about this about a year or two ago for the first time and had a similar reaction, that there was some quality issues associated with it and I assume that those are being solved at this time. And what I’d like to hear a little bit better sense for is that the two products, or the two terminologies, are going to merge, they sound just very, very similar, there was a similar number of terms in the two, they were both involved internationally, they both seem to be covering the same areas, and it would seem very difficult for this Subcommittee if it were making a decision on devices to say anything except lets wait and see if they do merge. Do you have any comments to that?
DR. KESSLER: I’ll let Vivian go first and then I’ll add a few things.
DR. COATES: Let me comment, I mentioned the memorandum of understanding that we signed with the European Standards Organization back in ’96 with the hope, the intent, and the optimism that there would be one terminology in devices that we could adopt and therefore put all those resources into helping develop. But at the end of the day we really had two problems, one were the quality issues which arguably Brock Hefflin has been working, has addressed, he’s certainly worked very hard at it and we’re going to be able to examine what he’s done over the next year. But our second concern was the incredibly bureaucratic process for maintenance and updating GMDN and dissemination of it that we didn’t feel served the needs of our constituencies. And there as Larry Kessler has said only time will tell. I know that they have taken it to heart but I got to tell you, any of you who’ve had the experience of sitting on European standards bodies with representation from 18 member states and all of the regulatory and industry perspectives, that’s not a way to maintain a terminology in practical real time. So that was a real problem for us and it just didn’t serve the needs of our constituency.
DR. KESSLER: Vivian is being kind. I suspect this is on the record and I’m going to try and be as cautious as I can, but I want to make sure that there’s no mistaking what happened over the past four or five years so that you get a real clear picture. FDA and ECRI spent a lot of time together trying to get the Global Medical Device Nomenclature to be truly global, and then we hit some really serious bumps in the road and there was a lot of gnashing of teeth and a lot of serious debate about how things were working. Some things worked well, a lot of things didn’t work well. And when ECRI withdrew their support from GMDN it was a very wise decision. The product that had been generated had problems and they could not migrate legitimately to their product, and we couldn’t at that time move FDA’s product code effort to GMDN either, and it’s one of the reasons we’ve spent, Brock full time, over the past couple of years trying to wrestle GMDN back toward something that made more sense, and frankly looked a lot more like ECRI’s terminology.
The political problems of getting that done, frankly, were daunting, and you could lay some of the blame at my doorstep by not hammering the heck out of some of the folks over in Europe. But we’re working within the constraint not only of the GMDN organization, which was employed by CEN and ISO, but also with the Global Harmonization Task Force. And for those of you who’ve run international standards groups or tried to coordinate the harmonization of regulations around the world, it takes time and it takes patience and you wind up doing some things in ways that you might not otherwise do. So we could have been quite peremptory and taken our ball and bat, the FDA, and walked away with ECRI but it wouldn’t have served the purposes of global harmonization, and we’ve got a lot of global harmonization efforts on the one side, and nomenclature is only a small part of that.
So I’m trying to harmonize pre-market regulations, post-market regulations, compliance audits around the world, and pull those people along and do nomenclature at the same time and we’re having serious fundamental disagreements about how things work. It was a real balancing act and ECRI’s lack of participation the last couple of years has been difficult. Since I’ve come back to the FDA after a brief period of time, taking a brief holiday, we’ve gotten the GMDN folks to reinvite ECRI back to the table with the understanding that we’re going to try and merge this over the period of the next three to five years to make it work. And the reason FDA is going to work with ECRI is because their approach and their system makes a lot of sense and I want to migrate as much of this as I can in a careful way so we have one nomenclature in a few years. But trying to give you the political picture of this, it ain’t easy and it’s going to take several years to do. Does that help?
DR. STEINDEL: Just a very quick comment before my colleagues from CAP in the back might make a comment on this, on your slide where you showed the SNOMED term for Insulin pump, they’ve removed NOS from it. But other than that the term is still a solitary term like you indicated.
DR. COHN: Maybe I’ll ask a question and then I’ll let Michael. I actually want to thank you both for coming and Larry I really appreciate you being as frank as you have been given that this is being recorded. I guess as I’ve listened to this obviously it seems like that we are dealing with sort of an opportunity that shouldn’t be wasted or lost because I think as Steve commented these are clearly terminologies doing if not exactly the same thing about as close as one could imagine to the same thing. I guess I was wondering, this is just sort of an issue around the secretariat, has there, I mean obviously it sounds like we may be beyond the opportunity for ECRI to become the secretariat for this, but it would seem given their track record that would have been a likely spot or a place to take it.
DR. COATES: Brock you can clarify this but it’s my understanding that the secretariat isn’t really doing the technical maintenance of the nomenclature, they’re managing the licensees and the business relationships.
DR. COHN: Is that correct?
DR. HEFFLIN: Primarily, there’s an expert, when we showed you the maintenance agency, the stratification for the maintenance agency, there’s an expert team there that’s responsible for all the technical aspects of the nomenclature and updating and maintaining it.
DR. COHN: Steve, do you have a comment about that?
DR. STEINDEL: No, this is a comment going, if you have some other comments on organization or questions, because I want to drift a little bit away from that.
DR. COHN: Steve and then Mike.
DR. STEINDEL: My question may have something to do in line with Mike’s but I’m just second-guessing him. What are the plans for device nomenclatures that are going to be used internally within FDA and in particular internally with respect to the various patient safety movements occurring in the government, in particular the one that AHRQ is leading?
DR. FITZMAURICE: Yes, that was one of my questions, Steve.
DR. KESSLER: What do you think the answer is, Brock?
DR. HEFFLIN: We would continue to use, we would use the nomenclature that we developed as FDA uses the nomenclature currently. If there was a push towards using it for patient safety issues I believe that it’s well enough developed --
DR. STEINDEL: When you say you’re going to use the nomenclature that was developed, you mean the FDA nomenclature or GMDN?
DR. HEFFLIN: As we migrate to the GMDN, GMDN would be used in the same capacity that the current FDA system is being used. And I think it’s developed to the level that it could be used for other purposes, for example, patient safety, etc.
DR. KESSLER: One of the reasons we developed GMDN was for exchange of reports for post-market surveillance and vigilance around the world. One of the problems we’ve had in the past is understanding patient safety problems occurring in one country, trying to read the reports from other countries, you don’t know what product they’re talking about because there hasn’t been a nomenclature. So right now we use the product code system in FDA, we’re going to map and merge to GMDN in the next year or two, and hopefully map at the same time to the ECRI UMDNS.
DR. HEFFLIN: Again so that we’re all speaking the same language internationally.
DR. KESSLER: But our current coding systems have to depend today on FDA’s antiquated and very problematic proco(?) system, it’s why we spent so much time and money trying to build the new nomenclature.
DR. FITZMAURICE: Two questions, one is just a curiosity. Since this was built in good part with government standards money, particularly European standards organizations, is this an ISO standard or a CEN standard?
DR. KESSLER: It is indeed.
DR. FITZMAURICE: Which one or both?
DR. KESSLER: Both.
DR. FITZMAURICE: So that becomes mandatory for many of those countries to use it.
DR. KESSLER: Absolutely. Another one of the reasons I’m really hoping that we can merge systems, we can make the ECRI UMDNS system and GMDN the same used worldwide, makes life a lot easier if it’s a CEN and ISO standard and all the European countries have to use it. If we walk away, talking to them about devices becomes harder.
DR. FITZMAURICE: And then somebody makes money doing mapping when mapping might not have been necessary. A tradeoff. The next question is is there a patient safety database that goes along with GMDN? That is is there an adverse event reporting database that will be built?
DR. HEFFLIN: Currently efforts are being made to develop such a database, in fact some of the individuals involved in the GMDN project are part of that group. It’s in its initial phases.
DR. KESSLER: I chaired the group for the Global Harmonization Task Force that started to build that system.
DR. FITZMAURICE: What about within the United States, there are about 15 states that do have adverse event reporting, a lot of that is drug, but it could very well be device. Do they use the ECRI device identifier in their adverse event reporting? I’m looking to see how strongly this is currently being used in patient safety event reporting and how should it be used in the future?
DR. COATES: I have to research that, I don’t want to just answer that off the top of my head. I’d be happy to do that and get back to you on it. I know that we have our own adverse event database that we’ve maintained for more than 30 years using UMDNS, and disseminate that and sell that very broadly to hospitals and others, not only in this country but worldwide. And that’s a very, very well supported activity of ours. With the state reporting efforts, I don’t know, but I’m looking over there at Kathy Lesh from Kevric, Kathy probably knows the answer to that, because I think you had to research that didn’t you, for the AHRQ safety net?
MS. LESH: We have not looked at the state programs yet, we only looked at the six federal systems and of which the only thing that’s currently being used is the FDA standard product nomenclature.
DR. COATES: And also ECRI just got the contract to establish a medical error adverse event reporting database for the state of Pennsylvania, and there the state was very specific about what they did or did not want and sometimes it’s very hard after the fact go to a state and say look, you were shortsighted, you should have done this, that or the other. So I’m not sure to what extent state usage of a coding system or the lack thereof really factors in to where you want to be going nationally.
DR. FITZMAURICE: Well, what you can see is that there is a bill in Congress, a couple bills in Congress, that would mandate a voluntary, use those two words together in the same sentence, but voluntary reporting of adverse events, and it would behoove the reporters and the analyzers of that to get some consistency in what’s being reported. We have the Institute of Medicine looking at that and they’ll be providing us with good advice in the report in September. But if there are things out there we want to make sure that we know about them, we wonder why the states didn’t use them or know about them, and what should we advocate for the future. That’s why I’m asking the question.
One last follow-up and that is let me look at it in reverse, not reporting adverse events but lets suppose you know a device has a problem, you need to get that information out to people who have those devices or maybe have them implanted, is your database used in that reverse method to notify people that this particular valve may go bad?
DR. COATES: All the time we send out weekly alerting bulletins that do exactly that.
DR. KESSLER: You need to make sure though that you understand that most nomenclatures, and Vivian correct me if I’m wrong, even the ECRI nomenclature don’t go as detailed as make and model number, which is usually what you need related to a recall. You want to link the nomenclature to the product, but --
DR. COATES: As I said before our nomenclature sits at a level above the make and model level, but we have it linked, because of our knowledge databases that link from that level to the specific manufacturer identifier level, when we send out our alerting, our hazard and recall bulletins, we absolutely link it to the very specific make and model level.
DR. FITZMAURICE: It seems clear that we need the device nomenclature and more.
DR. COHN: I actually think it’s probably not so different then our discussions about drugs since you the nomenclature, you need the knowledge base.
DR. FITZMAURICE: Another question would be does FDA have the same regulatory authority for drugs as it does for devices, and I gather from what Larry and Brock said earlier that they don’t, but I don’t know.
DR. KESSLER: It depends on the device, it’s very different.
DR. SUJANSKY: This is getting into the area of questions that I had for you both regarding the role of these terminologies in the thing that we’re discussing here, why we’re here talking about patient medical record terminology standards, or really terminologies that are used in electronic medical records, patient medical records, as they exist now or may exist in the future. So first given your list of current licensees and applications and so forth I didn’t recall seeing any vendors of electronic medical records systems or hospital information systems or clinical data warehouses and so forth. Are there such licensees who are using it in those ways? If so, how are they using it? If not, why do you think there are not?
DR. COATES: You’re asking a lot of questions in one. You have to understand that the people who are hands on with devices, buying them, keeping track of them, figuring out who they’re connected to in the hospital, taking them out of service if there is a problem, are largely the materials managers and clinical engineers in the hospital. And to the extent that they have systems for inventory control and work order tracking and QA and so on and so forth UMDNS is in those systems. But to this point in time I think largely that medical records developers are kind of oblivious to devices, I mean they’re well aware that they have to encode clinical concepts for diseases and drugs and so on and so forth, but I don’t think they’ve given the kind of thought and analysis to what you need in the device area that they have in these other areas. From our perspective where we sit, and I think I speak for my colleague at FDA, we feel, and we know this from huge experience that the kinds of problems that you can get into with ambiguity and confusion if you can’t precisely and encode to the right level of specificity what the patient was connected to, what came near the patient or what was put into the patient, so I think there’s been a disconnect in the way that medical records developers have been thinking about their domain and the domain of devices. Larry, do you want to --
DR. KESSLER: Couldn’t agree more.
DR. SUJANSKY: Is it possible that the benefits of having that level of detail is already being met in the materials management functions that you describe within hospitals and other institutions? Or is there, what’s the gap where this type of detail terminology actually belongs in a patient medical record?
DR. COATES: I think there’s a gap because these systems right now are not well integrated, the materials management clinical engineering systems are out there, the medical record system is here, and they don’t talk to each other necessarily. I don’t even think in the VA, which is probably at the forefront of information systems development for hospitals, I think there is a disconnect even there between the procurement system and the patient safety systems that they’re building.
DR. SUJANSKY: So mapping may be required between the concepts and the patient medical records and in these materials management systems.
DR. COATES: The term I think you’d like to use is interoperability may be better then just mapping.
DR. KESSLER: You can see bigger examples of that even in clinical records systems built by groups like the American College of Cardiology or the Society for Thoracic Surgery, you look in their databases and you’re trying to find out what heart valve was put in with what, you can’t find it, you can’t find a correct name and you can’t find the make and model, so when there’s a big recall these databases are of no use to anybody. So Vivian has said something that’s really important, the role of medical records in the way we’re building electronic systems, for some reason have left the devices world out of the discussion and it makes a big problem for us. We try to do post-market surveillance, we try to do alerts, and we face enormous problems and so does ECRI, it’s really hard to work with these systems. But I think there’s a comment behind, someone might want to object to --
DR. MCDONALD: I just want to maybe clarify a little bit and I’ve followed ECRI, I lost track three or four years ago, it’s a foundation wasn’t it, non-profit foundation at least at the beginning?
DR. COATES: What ECRI is? We’re not a foundation, we’re a non-profit research organization.
DR. MCDONALD: I thought originally it was a foundation, but that’s not important. I just wanted to clarify that there is device content medical records in a number of places and the problem, there’s two kinds of problems. Firstly, in pharmacies, because we write orders for advantage, I mean a very specific devices to stick in fingers and all the rest. We also in the surgery systems and the operative systems there’s very detailed fields about implantables and devices used, and the management of the surgery devices is very specific. I think the problem is they often get even more specific then what the classification of the ECRI has, I mean they’re talking about specific model numbers and managing them. So I don’t think it’s exactly true, I think the problem is there’s not enough communication between these standards groups and the needs. I don’t think they’re using the ECRI standards but they have some kind of pretty detailed records about model numbers and serial numbers, it might be local. It’s a good thing, we need to do more of it, but it’s just not that it’s not being captured, it’s captured in a different level, probably more detail and not standardized.
DR. COHN: Clem, you’re speaking of the gap.
DR. MCDONALD: I don’t know what I’m speaking to, I think that more mixing of those two things would be valuable and more awareness of both sides of the business would be valuable.
Kepa, you mentioned I thought, a long time ago that there was a barcode standard which got down to great detail, hooking or purchasers standard about supplies and equipment. Where has that gone?
DR. ZUBELDIA: The UPCEAN barcode is being used especially in South America and Central America and Europe to identify essentially anything you want, it can be a device, it can be anything you want. There’s a well structured barcode but I don’t know --
DR. COATES: I’d just like to comment that I think barcoding is very important and it should be used at that very highly specific level. It’s not in place of the nomenclature, the nomenclature serves other kinds of purposes, it enables you to retrieve information about like devices where if all you had is at the level of a make and model identifier you aren’t going to be able to do that, you’re constrained if you’re at that very highly specific level.
DR. MCDONALD: I don’t think it’s either/or, it’s maybe the same kind of model as we see in the medication world, that in the clinical environment they go to know which electrocardy(?) machine and it satisfies a specific template that one surgeon wants which is they use the other category, and then they have to record that, and they do record that typically someplace to manage all that staff. The question really is is having all this stuff linked together so that it can be used at all the levels.
DR. COATES: I think there’s a very patient lady waiting.
DR. COHN: I know, I was waiting for Clem to finish here, please, Carol.
DR. BICKFORD: I’m responding to the discussion about why we don’t have the medical devices tracking connected with our clinical piece and that’s in relation to how we have looked at our care practice and that’s diagnoses and procedures. Many of us are concerned about the context, which is in the environment, and now we’re beginning to realize that has a significant factor in the outcomes of our care delivery. So until we change our framework to look at it as a system with multiple entities, not just diagnosis and not just procedures, we’ll never be able to put all the pieces together to look at it correctly.
DR. COHN: Okay, I know that Jeff and I think others probably want us to sort of spend some time trying to put all this together. Are we ok with, burning questions are handled for the moment? Well, I want to thank our panelists and this has been illuminating to put it mildly. I’m sure we’ll have some sidebar questions for you as we go along. Now Jeff, how would you like to handle this last parade, do you want to lead it, do you want to have Walter to lead it, Steve lead it?
Agenda Item: Subcommittee Discussion
MR. BLAIR: Well, first of all, do we still have many of our testifiers from this morning and from the commercial drug vendors here because number one, I wanted them to be able to hear our discussion as well as if they could, if we wind up making an assertion that is catastrophically incorrect they could jump in and wind up saying that we’re incorrect.
My thought, and let me see if the Subcommittee feels comfortable with this, if we were to look at our recommendations first for the drug terminologies. The thought that I had was that we’re trying to identify the drug terminologies that would be able to play the role of a reference terminology for drugs and fit within the core set of PMRI terminologies. So I think I’m going to go ahead and make a proposal, I’m going to make a proposal, maybe it’s a straw person to go out there, and from what I have heard I have the feeling that RX-Norm appears to be well suited to perform that role and that the degree to which there’s overlap with SNOMED-CT, it looks like within NLM they are in the process of reconciling whatever that is, but that if we were to recognize RX-Norm as a PMRI terminology, part of the core set, that that would play a positive role and that the other part of that is that there are many other terminologies that where we heard from the terminology developers today, NDF-RT, First Data Bank, Medi-Span, Multum, and it appears as if they’re all meeting direct patient care needs or will be soon if they’re not already, and they may be slightly different but they’re meeting needs out there. And I don’t know that it would serve us well to pick from among that group and recognize one, which could have a negative impact on many of the others, they could still all go ahead and serve the domains that they serve, intend to serve, they could map to RX-Norm, so that would be the proposal I would set forth, and lets see if there’s agreement or disagreement.
DR. HUFF: I would just add to that and invite Randy Levin or others from the FDA to comment, but I think we want to say more than just RX-Norm, I think we would like to add some supports specifically for the FDA’s activities in establishing the identifiers for ingredients, clinical drugs, etc., that whole part, which is I think is activity that’s separate per se from RX-Norm, or it’s coordinated with but not funded exactly to the NLM the same way RX-Norm is. So I agree with the recommendation for RX-Norm but I would say we want to add specific wording that supports and promotes the FDA to do it’s role in that regard.
DR. STEINDEL: Stan, I’d just like a little bit of a clarification there. When you say support for the FDA’s role, the Subcommittee could support the FDA’s role as a comment in the letter because the FDA is not really developing terminology in this area, they’re developing a system in this area and the Subcommittee could add that support in the form of narrative information. Would you agree with that?
DR. HUFF: Well, I agree that they’re making a process, but I mean they are going to make identifiers for ingredients, for components, for the clinical drugs, for the branded drugs, for the other stuff, and to the extent that they do that I want them, I mean I want them to do that. So whatever we need to say in the letter to have them do that I would like to add our weight to that.
DR. COHN: Maybe we could have Randy join in for just a second and help us clarify this.
DR. MCDONALD: I mean I think that’s separate, it’s not separate because there really isn’t an engagement, but the thing that will make this work the best is actually getting those codes as part of the vendors submission process, so that’s all part of RX-Norm but if we don’t specifically say that the FDA has a role in it --
DR. LEVIN: But those aren’t part of RX-Norm, though.
DR. MCDONALD: I’m not saying it right. It’s correct, that’s maybe all the more reason, it’s an important, it would be wonderfully great if your process of accepting applications and all the rest, of getting NDC codes that are consistent and nice and the rest of that, also ties together, the RX-Norm process will work better.
DR. LEVIN: The codes that we’re, that Stan was talking about, the ingredient identifiers, those support RX-Norm and that’s right, the ingredient identifiers and the structured labeling, both those, because in the structured labeling they have things along the lines of dosage form, route of administration, those are things that also will be helpful for RX-Norm.
MS. HUMPHREYS: It would seem to me that if the FDA’s plans to, even if there were no RX-Norm, the FDA’s plans to clean up and centrally distribute and have a mechanism where they actually can give you a complete and correct file of NDC’s is certainly important to the work of the overall NCVHS Committee because the NDC has already been declared to be a HIPAA code set. So I know it’s off the topic of your PMRI terminology but it does seem that supporting the FDA’s effort to clean up this act and be able to have a complete and correct set available to everyone of the NDC codes is certainly something NCVHS should get behind whether it has to do with PMRI terminology or not. Of course it does also.
DR. LEVIN: But the other thing is the unique ingredient identifiers, the unique codes, with the NLM, those are going to be also important, too.
MS. HUMPHREYS: Oh, yes, that leads directly into the RX-Norm.
DR. COHN: We already have the unique identifier codes, it’s just that they’re private --
DR. LEVIN: That’s right, we have them, we need to publish them and we need to combine with the NLM so we can pull in codes that are not, that we can’t disseminate on our own or that are coming from out, products that are not marketed in the United States.
DR. COHN: Jeff, you had a comment?
MR. BLAIR: Yes. I personally feel very comfortable with the suggestion that Stan just offered, and I’ll paraphrase it so that Stan could correct me if I’m misstating it, but it’s that we also support the efforts of the FDA to move forward with its improvements, it actually has an array of improvements in identifiers for drugs and the role that it plays as it supports the development of RX-Norm. Is that correct?
DR. HUFF: Yes.
MR. BLAIR: Now given that restatement there was some separate concern when folks were aware of the fact that we were having this testimony today and we were considering the selection of terminologies to be part of the core set of PMRI terminologies. And it related to whether we were going to be hearing from pharmaceutical manufacturers, and the statement that I made was that we were hearing from commercial drug vendors and we were hearing from the folks that were here, but I didn’t feel that it was necessary to hear from the pharmaceutical manufacturers because there was nothing, at least I thought there was going to be nothing, that we would be recommending that would alter the FDA’s role in the NDC codes, the NDC codes are picked as a HIPAA terminology and there’s no alternation of that. There’s nothing that we would do that would recommend a change in the way the FDA and the pharmaceutical manufacturers relate to each other, that’s between the FDA and the pharmaceutical manufacturers in terms of how they make those changes if they do make any changes. So if there’s any individuals that feel as if adding this phrase of support for the FDA would in any way compromise the position that I indicated, which was the rationale for why we didn’t need to hear from pharmaceutical manufacturers, could they please express whatever discomforts they might have, if I don’t hear a level of discomfort then I will assume that you feel like this is ok.
DR. COHN: Bill Yasnoff is moving forward here.
DR. YASNOFF: I certainly don’t have any objection to the general statement of support. I think if you’re going to say anything in a letter to the Secretary that specifically discusses anything related to NDC codes that you might want to reconsider. As you stated what Stan said I don’t see any problem with that, but if we’re going to say something about NDC codes I think that’s, I think that more information needs to be gathered and the folks who are directly involved in that process need to have a chance to say something. I’d be interested in Randy’s thoughts on that.
DR. HUFF: Why wouldn’t we say, I mean they’re obviously broken and if we said please fix them, I mean without saying how, I mean I have the same concern, if we were trying to mandate from this Committee how they would be fixed, but why can’t we say we’d like them to be fixed?
DR. YASNOFF: I think before you even say they’re broken you have to hear from the people who use them everyday, and I think you will hear from those people that they are broken, but I think it would be prudent to let them say so because I suspect that when you collect information about this different people think it’s broken in different ways that are incompatible.
DR. SUJANSKY: One way to look at it if I can just offer a quick comment on that point is that what we’ve heard here today is that from the point of view of supporting RX-Norm, which is the topic of the recommendation, there’s a need to get the support from the FDA with respect to NDC codes, but we’ve heard from RX-Norm in that regard. And we’ve also heard from the FDA most directly in regards to the need for better NDC support in order to, the need for better NDC management in order to support RX-Norm. So in the context of a recommendation regarding RX-Norm as the terminology standard, I would just throw this out, have we not heard from all of the users of NDC codes that whom are relevant to that, that specific role of NDC codes.
DR. COHN: Steve and then Betsy.
DR. STEINDEL: I think what Jeff was suggesting is that when we put modifying language, and I believe Stan supported this, is that we add the text to the extent that we support the FDA’s efforts in what they need to do to support RX-Norm, and we don’t necessarily have to specifically say anything about NDC codes with regards to that. Now that’s specific to this terminology letter. Now with regard to the Subcommittee as a whole, as it was pointed out, NDC codes are part of the HIPAA process. If FDA puts in a regulation that’s going to change NDC codes this Subcommittee will be hearing from all parties at that point in time. So I don’t think we need to go much further than just say we’re supporting the FDA’s efforts in this particular letter.
DR. SUJANSKY: Efforts to make regulatory changes that support --
DR. STEINDEL: There are regulatory and non-regulatory.
MS. HUMPHREYS: I was going to say that I think that what we’re hoping to get as the future best way of maintaining RX-Norm is this data feed from the FDA about new products. So what we want is support for whatever the FDA has to do to have an electronic feed to the National Library of Medicine of new approved products, I mean of new information about approved products because that way we will be able to update RX-Norm and distribute it and we won’t have to wonder, wandering around wondering whether we’ve got all the FDA approved products in there or not.
MR. BLAIR: And maybe what Stan was really trying to bring forth is that the FDA has a key role, an essential role, in the comprehensiveness of RX-Norm, the completeness of it, the timeliness of it, and I think it would be very appropriate for us to make that statement of support for the FDA’s role and indicate how important it is relative to RX-Norm. And does that help, Bill, avoid the problem?
DR. YASNOFF: Yes, I think it’s much better to state the functional requirement then to say anything about even specific things that are broken. To say this is clearly a requirement for RX-Norm, we support this, the development of whatever is needed to meet that requirement, and lets not assume that we know what that is, because I don’t think we do.
DR. LEVIN: And can I just make one clarification about the issue about approved products? Remember there are a lot of products out there that are not approved that we still list, so if you want those products as well --
MS. HUMPHREYS: Yes, all the products that you list.
DR. COHN: Walter, are you ok at this point?
DR. SUJANSKY: I’m fine. If we’re done with that immediate topic I did have a couple other comments. I’d like to raise two other issues related to this to make sure we address them. One is given the need that we just described of supporting the FDA efforts in order to enable the use of RX-Norm and the timely maintenance of RX-Norm as a terminology standard I think we need to consider what is the status of RX-Norm as a terminology standard prior to the time that those changes are made. And is it ready to be a terminology standard? We heard today, going back to our testimony, that there are 20 FTE’s, in the absence of an electronic feed that one of the vendors was using to keep track of NDC codes and 12 of them are used just for tracking NDC codes --
MS. HUMPHREYS: Yes, well I think this is a big distinction because the RX-Norm concept level goes down to the clinical drug and we connect brand names and what have you, and we are carrying NDC codes as available to us as attributes of a clinical drug. Like we even have one I guess in our latest thing, there are 1800 valid NDC codes that connect to the same clinical drug. But we are not generating concept level information at that low product NDC level, and that is exactly what First Data Bank is doing in order to do all the added value services. And the RX-Norm is not going to have identifiable separate concepts for every NDC level, so we should all know that.
DR. SUJANSKY: Yes, I’m aware of that, but the whole reason there’s a need for the changes at the FDA are because RX-Norm is using NDC level information, as well as labeling information and so forth which is near the NDC level, in order to populate RX-Norm and to maintain RX-Norm. And my only point was that that’s a big job in the absence of an electronic feed from the FDA. Maybe given the more limited scope, if not 20 FTE’s but we’ve heard that it’s significant.
MS. HUMPHREYS: The scope of RX-Norm is very much more circumscribed because we’re not establishing concept at the NDC level. That is a vast difference, that’s the difference between 80,000 and 15,000 or 20,000.
DR. SUJANSKY: But then the question is does the NLM feel that, we’re kind of having a conversation here because I want to explore this issue and disclose it, does the NLM feel that given existing processes in the absence of the improved FDA feed that RX-Norm can be maintained on a weekly to monthly basis that’s required, that we heard is required for a drug terminology standard?
MS. HUMPHREYS: At the level at which RX-Norm is produced it can be maintained that way. And I think that the issue is yes, would we want the whole thing totally connected and that’s a value added that in fact the drug vendors could provide, down to the NDC level? Of course we do, I mean people want to be able to take their NDC’s, look this up, of course they want that. But in the absence of having that if we distribute RX-Norm and RX-Norm connects, has a standard clinical drug level and is connected to ingredients with these other relationships that Stuart described this morning and also connect to the clinical drug level terminology that is used in Multum and Medi-Span and First Data Bank and whatever, we already have something that is actually quite useful for operational places where they need to use these various drug terminologies, and also something that is potentially useful as saying hey, when you send your HL-7 message in it has to include a clinical drug in it, send it at the standard level because then we’ll all be able to understand it theoretically irrespective of whether we have our local First Data Bank or our this or our that or other thing in our local system. So it would be bigger service to people if we had something that connected every single level of drug terminology that we need but I believe that RX-Norm as we are able to maintain it in NLM is a service that is needed. It solves part of the problem.
DR. SUJANSKY: Just to really make sure we’re on the same page here, in order to be adopted within the government perhaps and certainly outside the government by those who are currently relying on the vendors to provide this drug terminology for them for just a coding fee, in order to be adopted the frequency of updates will need to meet their needs.
MS. HUMPHREYS: Yes it will.
DR. SUJANSKY: And we’re hearing that it will meet their needs.
MS. HUMPHREYS: I think that you, the NCVHS in terms of making a recommendation has to decide what they’re going to state about that subject because they can’t say it meets everyone’s needs today because it’s currently going quarterly, so the question is what do you want to put in, or what do they want to put in as a recommendation of how frequently it should be distributed. I believe we can probably handle what they do in one way, shape, or form, I mean whatever they recommend, but I can’t tell you it’s issued weekly today because it isn’t.
DR. LINCOLN: Clem is anxious to comment, too, so I’ll just say what she said.
MR. BLAIR: Could you identify yourself?
DR. LINCOLN: Mike Lincoln, sorry, from VA. Yeah, what Betsy said I think is essentially true and the other point is that RX-Norm at this point does not depend on labeling level or NDC level information, that was sort of a misnomer that I’ve heard said a couple of times today and it’s not true, so we’re not dependent on a flow of NDC’s from an authoritative source in order to maintain RX-Norm.
DR. MCDONALD: I was just going to say that I think everyone assumes that these commercial vendors are going to have a very important continuing purpose and that RX-Norm is not going to be the root thing under most scenarios to what they’re going to be using. But it would have a tremendous value even if it doesn’t get down to the product line to have these things to which everything could connect and --
MS. HUMPHREYS: Yes, I think that what we have found obviously in the work that Stuart and the folks who’ve built RX-Norm is that in all of the commercial drug systems at the various level of pieces that are in RX-Norm, there generally is a corresponding level in the commercial systems. If we match those up then we have this thing and they’re all connect to it which provides a level of interoperability perhaps within health systems, where these people can be using, I mean these systems have various values, I know of many people who use more than one of them in the same facility. So it seems to me that this actually is adding in a level of operation that they don’t themselves have to maintain, and it has value irrespective of the extent to which it does or does not or ever will have attributes so that you can actually go from an NDC up to this level within RX-Norm itself, I mean you may be using other services --
MR. BIZARRO: This is Tom Bizarro from First Data Bank. The one comment I would make as far as the frequency of updates and the timelines is that our customers, and this is true I think for all the compendia, hate getting a database that has not yet assigned as a value for a particular data element. So what we would want to do is to make sure that we have those type of codes used only when absolutely necessary, we’re all going to be on our own schedules. The update that we provide to our customers coming from First Data Bank is not going to be delayed if we don’t have an RX-Norm code because our customers have other responsibilities, so the more often that you can supply it, I think that you are going to have to supply it at least weekly, but that’s the goal that you should have and the would meet the majority of the needs of our customers.
DR. COHN: And that’s maybe a suggestion about, in the document about moving to a more frequent update.
DR. SUJANSKY: But that’s my point, how far do we want to go in this recommendation? Do we want to say today, now, we’re saying RX-Norm is an adequate drug terminology standard so please start adopting it? Or do we want to say when the frequency gets to the point that there are weekly updates RX-Norm will be an adequate drug terminology standard? Or do we not want to say anything about that at all?
DR. COHN: I guess my view is that it’s important, this major first step that needs X, Y, Z additionally, but --
MS. HUMPHREYS: The other thing, too, is that you have to think about what is the update number and frequency of new clinical drugs, not of new products, but of new clinical drugs. And fortunately for us, since that’s what we’re committed to working on, it is a lot less frequent then new products.
DR. HUFF: I just wanted to say that I think it’s useful today, and it will get more useful in the future, and that we ought to say that with the goal of enabling more support to make it even better. All of our clinical systems, it would be wonderful if we could get daily updates on new codes and products, the truth is that we don’t have anybody on call over the weekend so we know we can go at least two days without making a new code, and we have ways within our system, in fact we go months at a time making new codes locally until the national code is available. So yes, I would encourage and hope that we could support more frequent updates but I would state it that way, not withhold endorsement until that exists because I think it’s absolutely useful and practical in the form that it is even if it only came out once a year, it would have added value. And the more frequent the updates and the better supported it is the more value it would have.
DR. COHN: But Stan I think there’s a difference between added value and a core terminology, I think we better make it as a stronger statement then as an added value.
DR. MCNAMARA: Quick question, or quick comment. I’m concerned about the direction that the Committee is going in this dialogue for the following reason. If I understand correctly what we’re proposing, this is Tim McNamara from Cerner, if I understand correctly what we’re talking about we would have RX-Norm be sort of an intermediary that would serve as a conduit for mapping potentially between Multum and First Data Bank or Micromedix or Medi-Span or other providers. The trouble is, as I see it, is that this represents a patient safety issue because there is no obvious way for a sender of a Multum encoded term to know for sure how that’s going to be represented if it’s communicated ultimately into the First Data Bank iteration or corresponding, there’s no good quality assurance step that would necessarily assure that that communication is effectively mapped for the purpose for what it’s intended. So I strongly advise the Committee to consider a uniform standard with one set of codes rather than continued reliance on multiple coding systems.
MS. HUMPHREYS: I think that from our perspective what RX-Norm is is certainly its purpose is to have a uniform standard, the issue is that in order to populate the uniform standard and be sure that we had coverage of heavily used drugs we are using multiple sources to say well have we covered all the drugs that are currently being used. And as a result of that process we have created this map between the RX-Norm forms and forms in the other systems. But that to me is not incompatible with the fact that RX-Norm is being developed in a standard way and is a standard nomenclature and would be a standard nomenclature for any new clinical drug that we created it for irrespective of whether we knew it was in any of the other systems or not.
DR. SPACKMAN: I just wanted to just make sure that in the Committee’s consideration and recommendation that you explicitly say that there needs to be a coordination between the SNOMED clinical terminology, the SNOMED drug terminology, and RX-Norm. And that’s a task to be done, it’s not a fait accompli, just acknowledge that. The fact that we’re feeling comfortable moving ahead with certain recommendations is sort of contingent on that and I’d like us to explicitly state it if that’s alright.
DR. SUJANSKY: Before we take the next one, is everyone comfortable with that in the Subcommittee?
DR. MCDONALD: I don’t know why that’s necessary any more than with CPT or ICD-9 --
DR. SUJANSKY: Well, because SNOMED and RX-Norm may both be in the core terminology as ICD-9 and CPT won’t be in the core terminology.
MS. HUMPHREYS: If you are making information about the core terminologies then I think that probably what you want to do is bump, I mean I’m not disagreeing with Kent’s point, I’m just saying I think you would bump that up and say having designated in whatever they are as core terminologies and we see as a task coordination, identification of any overlaps, and if the overlaps have to persist some method of communicating, you know making it clear to people, if they’re following the CHI standard, not the CHI standards, but your recommended standards then how are they going to deal with the overlap, I mean which one are they going to pick?
DR. COHN: -- we agree, actually we think there’s a section in the letter that sort of needs to address that, so we should talk about it at that point.
MS. HUMPHREYS: And I think that dealing with it generally is better then otherwise --
DR. SUJANSKY: We should address whether we’re just talking about mapping overlap exactly.
DR. STEINDEL: Walter, we’re going to spend four hours talking about the letter tomorrow so we can defer the specifics.
MR. ROBINSON: George Robinson, First Data Bank. A comment about cross-referencing -- what I view RX-Norm from the FDB perspective is that RX-Norm would provide a defined set of clinical drugs. I see it as our responsibility to take that set and apply it to our identifiers, our GCN signal level as we deem fit and most likely we would introduce the notion of is it an equal to, broader than or narrower concept, so that I think as the drug knowledge base providers I think we need to take ownership of reviewing the concept and maintaining those cross references. At the same time, within the RX-Norm project, I’m not quite sure how you want to handle the notion of cross referenced entities within UMLS, because they’re still proprietary in nature that within UMLS you may provide the link to an external vocabulary but again the use of that link is limited by the licensing of the proprietary content.
MS. HUMPHREYS: Of course that’s handled by the UMLS license where basically we provide the codes for people to exclude from their copy of the medical --
DR. COHN: Okay, can we move on from this?
DR. SUJANSKY: I don’t know how much longer you want to discuss, I do, I’m aware that we’re going to discuss this tomorrow as well though not everyone may be here tomorrow. So the other point, not everyone in the audience. The other points I wanted to make, point I wanted to make is about talking about potential gaps in RX-Norm that we may or may not want to include in our recommendation for filling. So for example allergen classes, this is an issue we discussed earlier. It was under consideration whether allergen classes in the interest of interoperability and exchanging standard representations of patient medical record information whether allergen classes should be part of the standard so that you can say this patient is allergic to sulfanomide or Alicilytes(?), etc. I don’t know that RX-Norm currently has that.
MS. HUMPHREYS: Well, I’m sorry Stuart isn’t here and he could tell you definitively, I don’t think it does at the moment, but if we’re talking about the notion of minimizing overlap that would certainly be a place where we would be creating it I believe, because I do believe being allergic to things is sort of something that is reasonably well covered in SNOMED.
DR. SUJANSKY: Ok, we can discuss that one tomorrow. Some of the other things were, we discussed a little bit whether we want a comprehensive mapping to NDC codes. Also a therapeutic classification, is there a role for that in the standard? RX-Norm doesn’t, RX-Norm through its UMLS link has that but as far as being part of the core terminology, that may again be something where mapping an integration with other elements in the core terminology can address that. Is that something that we should include in our recommendation that we can accommodate a standard therapeutic classification system for drugs as part of our recommendation? These issues were raised earlier without resolution.
DR. COHN: Given that we’re at 5:30 p.m. I guess I would, we’re not going to spend the next half an hour going through your list. How many more of these do you have?
DR. SUJANSKY: Well, that’s why I asked how much more we wanted to discuss today and if not that but what else do we want to discuss?
DR. COHN: I think if there’s some pertinent issue on the devices, I think the rest of these can be uncovered as we go through the letter or be the first topic of conversation tomorrow.
DR. STEINDEL: It’s getting late and we have the device people here, I mean they probably won’t be here tomorrow and instead of drilling down I think we need to see if there’s something we can conclude with respect to devices at least from an overall level and perhaps drill down tomorrow if necessary in the same area.
DR. MCDONALD: I’d also suggest that it’s probably, in general if you could make fairly constrained decisions, we’re going to be better off then making an extremely broad one. It will be less likely to run into unknown connections and links and trouble, so I would suggest not trying to push that into RX-Norm, at this time in history anyway.
DR. COHN: Do we have a direction that we want to discuss around devices?
DR. STEINDEL: It’s my personal observation looking at it over the last two plus years that I think it would be very useful for the Subcommittee to come out with a very strong statement that there is a place in PMRI terminology for devices. I think as was pointed out in the closing moments of the device testimonies that is an area that people seem to be somewhat mixed on at this point and I think all of us are aware that the need to specifically identify devices in PMRI terminology is important. I think at the first level the letter should point that out. I would like to see the letter go a little bit further and based on what I heard today I’m reluctant to recommend one of the two terminologies because I think they’re the same terminology. And I think the letter should say that they should be the same terminology, that the efforts of the ECRI and the FDA and ISO CEN groups to merge the terminologies should be encouraged, supported, and perhaps accelerated.
DR. COHN: Jeff, do you have a comment?
MR. BLAIR: The letter was distributed by Walter on Friday to the Subcommittee members, the draft that we’ll look at for tomorrow, and Steve, you may recall one of the sections we have in the letter is for open issues and concerns and things we would be looking at in the future. And given your suggestions it sounds as if the statements you made about medical devices, that they are important, that they do have a role to play in PMRI terminologies and the other statements, I’m not going to repeat them all, would fall into that section? Okay.
DR. SUJANSKY: I think we should again put some general language in there regarding that, but no more at this point.
DR. MCDONALD: Is there any way that we as Committee members could get our hands on these nomenclatures just to be more informed? The device ones, I think what everybody said is right, that it’s certainly a very key area in health care and is important.
MS. HUMPHREYS: The UMDNS vocabulary is part of the UMLS so you can --
DR. MCDONALD: Can you get the ECRI one?
MS. HUMPHREYS: Yes, you can see it there, and I spoke to Larry about when would be the appropriate moment for putting GMDN into the UMLS and he said they were incorporating all these changes that Brock was describing that they had been making over the past year or so, and that there would be a revised edition that incorporated those within a few months. And so it sounds to me as if you wanted to look at GMDN you would want to look at this revised and enhanced version that will be available in a few months and I was talking to him about when NLM could get a copy of it in terms of looking at it to put it into the UMLS.
DR. COHN: I guess I would make one other comment, I’m not going to necessarily propose this but I want everyone to think about it, and it can indeed be future direction proposing is that where these are two very similar terminologies that ought to be one, and at the very least they’re already talking about merging and/or mapping so tightly that you can’t tell the difference. And given that ECRI is actually the one terminology we’ve discussed in the last several months, it’s actually been around for a while to mature, one might want to consider that that’s actually something we could actually recommend with all of these additional provisos, and I’m not saying we should put that in a letter but I do want everybody to at least sleep on that because that would actually be something that somebody could do tomorrow, and would actually be likely to be a relatively high quality product and implementation. So just something to think about. Steve?
DR. STEINDEL: In response to that, Simon, I would just like some comments from the FDA about how if this would have any impact on their efforts with GMDN.
DR. HEFFLIN: We would plan to continue in our efforts to support and develop the GMDN, that’s been the goal from the beginning and it’s something that David Fivo’s(?) committed to and the Center in general, so we would continue to put our efforts into the development of the GMDN for global harmonization purposes.
DR. STEINDEL: But would a statement like this help you in the efforts to merge the two terminologies?
DR. HEFFLIN: That you would like to see the terminologies merged?
DR. STEINDEL: The sense that I have is that they basically, they should be the same terminology. I mean if they’re not they’re pretty close.
DR. HEFFLIN: So we’re interested in that sort of activity in that merger, so that we have one system.
DR. STEINDEL: The problem that I see in this recommendation is if something like that did not occur we could be in the situation where we’re recommending a PMRI terminology that is different then the one that FDA is using for regulatory purposes if they choose to use GMDN, and if they choose to go with the one that we’re recommending for PMRI terminology and they’re choosing to use a regulatory terminology that’s different from the rest of the world.
DR. HEFFLIN: Correct.
DR. STEINDEL: So that’s why I suggested it might be a way to accelerate that process.
DR. HEFFLIN: The merger?
DR. STEINDEL: The merger.
DR. HEFFLIN: I think we would be in very much agreement with that.
DR. FITZMAURICE: Along with that, but even apart from that, if we think that the principals behind the ECRI terminology are good principals then we may want to come out with a statement of support for the ECRI terminology, and if the world wants to merge GMDN with ECRI or not that’s probably beyond our scope of influence, but if we influence the national choice that might aid FDA and might also lead to a merger. But I would do it because they were good principals not because it influences the politics in some other country.
DR. COHN: I want everybody to sleep on this one, this is an important issue and I think we all see what we want the end game to be, the question is is whether we want to defer comment, have it be an issue to be determined, or whether we want to come out a little stronger with something that could be useful now.
DR. COATES: I just want to comment. It’s so interesting to me how, Vivian Coates, ECRI, it’s so interesting to me how the arcane world of terminology in the device arena is mirroring the large view political situation. Without wanting to rain on the parade too much ECRI and FDA can’t just hold hands, march off and merge with respect to GMDN without bringing in the Europeans, who have a vested interest in it as well. So while ECRI and FDA have an agreement in principle to work together to make this happen we still have to let it play out with happens with CEN and the Europeans.
DR. COHN: Thank you, Vivian, I think we agree.
DR. HEFFLIN: And that being said I don’t, having worked with them over the past several years I don’t think they’d have a great objective if we’re going to end up with a good product, I don’t think this merger would be a problem for them.
DR. COHN: Other comments? We’re only running about half an hour beyond where we thought we were going to be since it’s still early, but Jeff I think you were going to try to make a few closing comments and then we’re going to adjourn until 9:00 a.m. tomorrow.
MR. BLAIR: My closing comments is a profound thank you to Steve Steindel for pulling together a very difficult agenda to try to set it through, make sure that all of the right individuals have realistic expectations before they testified, that the questions were appropriate to help us drive us to try to come to some conclusions out of this complex maze, and I think Steve just really made it possible for us to come to some conclusions today. Thank you.
The other piece is there’s a lot of folks that are, they may be in the room right now, I’m not sure, who are very, very interested in what will happen tomorrow morning with what we include in this second draft of the PMRI terminology recommendation letter. Is Marietta Squire? Marietta, if there’s any possibility that you could make extra copies of the recommendation letter for the folks that are in the audience so that as the Subcommittee goes through its deliberations they will also be able to --
MS. SQUIRE: I have copies right here.
MR. BLAIR: You do, great, ok. And then the last thing is what, I’m going to turn it back to you in terms of --
DR. COHN: Well, all I’m going to say is I think it’s time to adjourn, we will reconvene tomorrow at 9:00 a.m. See you in the morning.
[Whereupon, the meeting was recessed at 5:45 p.m. to reconvene the following day, Wednesday, August 20, 2003, at 9:00 a.m.]