August 19-21, 2003
Washington, D.C.
Meeting Minutes
The Subcommittee on Standards and Security of the NCVHS was convened on August 19, 2003 at the Hubert H. Humphrey Building in Washington, D.C. The meeting was open to the public. Present:
Simon P. Cohn, M.D., Chair
Jeffrey S. Blair, M.B.A., Vice Chair
Stanley M. Huff, M.D.
Clement J. McDonald, M.D.
Kepa Zubeldia, M.D.
John P. Houston, J.D.
Vivian Auld, NLM
John P. Burke, III, OS
Suzie Burke-Bebee, ASPE
J. Michael Fitzmaurice, Ph.D., DVA
Maria Friedman, CMS
Marjorie S. Greenberg, NCHS, CDC, NCVHS Executive Secretary
Stanley Griffith, M.D., IHS
Betsy Humphreys, NLM
Donna Pickett, NCHS
Steven Steindel, Ph.D., CDC liaison to NCVHS
Marietta Squire, NCHS
Jared Adair, CMS
Nelly Leon-Chisen, AHA
George Arges, AHA
Diane Aschman, SNOMED Int., Coll. of Amer. Pathologists
P. Aspden, IOM
Richard Averill, 3M HIS
Michael Bebee, AMA
Carol J. Bickford, ANA
Tom Bizzaro, First DataBank.
Amy Blum, NCHS, CDC
Barry Blumenfeld, Partners Healthcare
Irene Brahmakulam, RAND
William R. Braithwaite, Health Policy Consultant
Steven Brown, VA
Michael J. DeCarlo, Health Ins. Assn. of Amer.
Vivian Coates, ECRI
Craig Cochra, Program Assoc.
Marck Dubois, Wolters Kluwer Health/Medi-Span
Irene Dworakowski. Amer. Coll. of Surgeons
Karen C. Eckert, Wolters Kluwer Health/Medi-Span
Jack Emery, AMA
Christine Frasier, Amer. Academy of Professional Coders
Judy Frazier, Amer. Optometric Assn.
Joanne Glisson, Amer. Clinical Lab Assn
Norbert Goldfarb, 3M HIS
Kimberly Greco, Hogan and Hartson
Margaret W. Haber, Natl. Cancer Inst.
Brockton Hefflin, FDA
Cathy Hollinger, FDA
Larry Kessler, FDA
Mary Kofnar, AMA
Said Komezi, VA
Nelly Leon-Chisen, AHA
Kathy Lesh, the Kevric Company
Randy Levin, Ctr. for Drug Evaluation Research, FDA
Martin Libicki, RAND
Mike Lincoln, VA
Timothy McNamara, Cerner Corporation
Nick Macarelli, FDC Reports
Barbara Marone, Amer Coll. of Emergency Physicians
Reed Miller, The Gray Sheet
Cory Mitchell, CAP
Thomas Musco, HIAA
Stanley Nachimson, CMS
Stuart Nelson, NLM
Todd Omundson, AHA
Sue Prophet-Bowman, Amer. Health Information Mgt. Assn.
Mark Rendina, CMS
George Robinson, First DataBank
Dan Rode, Amer. Health Information Mgt. Assn.
Virginia Saba, Georgetown Univ.
Kent Spackman, Chair, SNOMED Int. Editorial Board
Paul Speidel, MGMA
Walter Sujansky, independent consultant and advisor to NCVHS
Robert Tennant, MGMA
Maureen Thompson, ASHA
Irene Torakasky, Amer. Coll. of Surgeons
Nancy Trenti, Amer. Psychiatric Assn.
Syed Tirmizi, VA
Tim Van Zoudt, Adva Med
Shelley Walworth, Unicor Med.
Kathryn Wilber, AAHP
Tom Wilder, ASHP
Michelle Williamson, NCHS, CDC
Julie Wolcott, IOM
The Subcommittee on Standards and Security of the National Committee on Vital and Health Statistics held three days of hearings August 19-21, 2003. The Committee is the main public advisory committee to the U.S. Department of Health and Human Services on national health information policy. Work continued on the health care terminologies with 12 presentations focusing on the issues of drug and device terminologies. The second day focused on review and update of a PMRI terminology recommendation letter and companion report, discussion of the federally sponsored ICD-10-CM cost impact analysis undertaken by RAND, and a report from American Hospital Association (AHA) and the American Health Information and Management Association (AHIMA) on their ICD-10-CM pilot study. The third day members heard updates on the claims attachment NPRM and the Consolidated Healthcare Informatics Initiative (CHI), a discussion of current CIHI recommendations, a HIPAA update and panel on implementation issues, and conducted Subcommittee business.
OVERVIEW ON DRUG TERMINOLOGIES
Dr. Blumenfeld discussed the need for drug terminology, definitions, requirements, concepts, use cases and challenges faced working with large integrated delivery systems.
OVERVIEW OF FDA DRUG DIRECTIONS
Dr. Levin discussed FDA’s collaboration with other government agencies and how leveraging existing processes makes it easier for information to be available as new drugs are marketed. He described FDA’s actions to resolve problems with the drug listing process for drugs marketed in the U.S. with a national drug code (NDC).
Discussion
Dr. Levin noted steps to make key information used internally available publicly. He said FDA needed to hear about issues and needs for code. FDA expects a final rule soon requiring manufacturers to submit labeling electronically. Participants emphasized NDC duplications and the difficulty mapping with the current ratio of “contract” codes.
PANEL OF PUBLIC SECTOR DRUG TERMINOLOGY DEVELOPERS – RXNORM
Dr. Nelson explained RxNorm as a project within the Unified Medical Language System (UMLS) to address the naming of all prescription and over-the-counter (OTC) drugs approved in the U.S. in a standards format. He described the cooperative effort and RxNorm’s responsibilities for linking active ingredient, drug component, clinical drug, and dosage form.
Dr. Brown explained how NDF-RT developed from the National Drug File and focused on patient safety, clinical decision support, and reducing operational costs and “pains” of terminology. NDF-RT is being merged with VA’s enterprise terminology services in order to formalize terminology and feed reengineered applications.
Dr. Lincoln said VA’s Enterprise Terminology Service (ETS) is a national initiative to standardize terminology content and services across VA. VA is committed to interlocking SNOMED, NDF-RT, LOINC, HL7, CDA, and other sources to eliminate local variation.
Discussion
Participants noted that RxNorm resembled a reference terminology and NDF-RT seemed similar to drug knowledge bases. They discussed a correlation between dosage forms and route, viewed the Daily Med as a service, and reported that timely updates to their databases were dependent upon FDA, commercial sources, and the literature.
PANEL OF PRIVATE SECTOR (COMMERCIAL) DRUG TERMINOLOGY DEVELOPERS
Mr. Robinson discussed First DataBank’s consistent view of codifying drug content, various levels of abstraction, and how drug knowledge bases offered added value.
Dr. McNamara said FDA should have primary responsibility for standardized terminology and that the 13 “desiderata” for drug terminologies should be incorporated into any federal initiative. Cerner offered to contribute intellectual property and resources to FDA under a “liberal open-source agreement.”
Mr. Dubois said Medi-Span data sets supported a range of applications throughout the industry, were timely, complete, met market demands and generally could support any application requiring drug information. He expressed concern that NDF-RT and RxNorm were directly competitive.
Dr. Spackman clarified that SNOMED was a comprehensive clinical terminology that included hierarchical and logical integration between SNOMED clinical drugs and the rest of SNOMED. He advocated producing and distributing free, with FDA and HHS responsible for some of the process, a realm extension complementing the SNOMED core and possibly utilizing RxNorm and NDF-RT.
Dr. Levin noted that in their drug listing FDA interfaced with manufacturers to collect information about every marketed product in the U.S. and everything First DataBank and others fed to customers. FDA proposed a regulation that companies provide labeling in electronic format in order to handle drug labeling more efficiently.
Discussion
Concern was expressed about FDA having a drug database dependent upon pooled providers. Medi-Span found RxNorm useful and NDF-RT competitive. First DataBank recommended defining the scope of public terminology with RxNorm and excluding knowledge bases VA-centric with NDF-RT. Cerner and Multum supported RxNorm’s standardization and believed they could compete with the drug knowledge base side of NDF-RT. Testifiers indicated it could be useful if the government identified drugs, issued identifiers, and mapped, but said it was threatening when clinical information was overlapped. They wanted to be involved in mapping and see FDA, SNOMED code, indications, contraindications, and other components.
PANEL OF MEDICAL DEVICE TERMINOLOGY TESTIFIERS
Global Medical Device Nomenclature (GMDN), created through merging six established medical device terminologies, has 13,000 terms for generic device groups. The project works to standardize base concepts of all terms and reflect devices’ intended use. Definitions incorporate all aspects of the standard; new synonyms were linked and preferred terms added.
Mr. Kessler explained that GMDN is an international nomenclature that provides generic descriptors for “everything clinicians touch in the operating room” and is based on ISO 15225. The Office of Science and Technology is mapping the FDA system of product codes to GMDN and will convert next year and develop a Web interface.
Ms. Coates reported that ECRI, whose formal hierarchical system for organizing device-related information has been part of the UMLS since 1991, has a partnership agreement with FDA to use the same guidelines to map respective nomenclature systems and establish new terminology.
Discussion
Concern was expressed about voluntary sponsorship of a designated standard. FDA and ECRI were “back at the table” negotiating to map GMDN and the universal medical device nomenclature system (UMDNS). Efforts were underway to develop an adverse event reporting database. Participants noted a disconnect in how medical records developers thought about domains. Dr. Bickford cautioned that “the pieces” could never be assembled until their framework encompassed more than diagnosis and procedures.
SUBCOMMITTEE DISCUSSION
Members considered that RxNorm was suited as a reference terminology for drugs and fit the core set of PMRI terminologies, emphasizing their confidence was contingent on coordination between RxNorm and the SNOMED drug terminology. Members discussed potential gaps in RxNorm, mapping to NDC codes and accommodating a standard therapeutic classification system for drugs. They emphasized the need to identify devices in PMRI terminology and support ECRI, FDA, ISO, and CEN to merge terminologies.
REVIEW AND UPDATE OF PMRI - TERMINOLOGY RECOMMENDATIONS
Members noted that drug knowledge bases served different markets and users and that terminologies FDA would support were needed to sustain medication terminologies and impart RxNorm. Noting that RxNorm was a necessary element still in development and would be revisited when in production, members discussed how to augment it and support of resources and timeliness for NLM, FDA and medical information system development.
ICD-10 IMPACT ANALYSIS - STATUS REPORT
Dr. Libicki presented work done by the Science and Technology Institute of RAND on the costs and benefits related to switching from ICD-9-CM to ICD-10-CM and ICD-10-PCS.
Discussion
Participants said potential gain of a new coding system exceeded potential cost but cautioned about possible errors and suggested incentives were needed. Ms. Pickett noted that now that AHA and AHIMA had finished their pilot, impact on the classification could be gauged and the crosswalk finalized.
ICD-10 IMPACT ANALYSIS - OPEN MICROPHONES
Ms. Fraizer discussed why the American Academy of Professional Coders (AAPC) supports enhanced standards for coding health care information and why it takes partnering for a successful transition.
Discussion
Participants discussed a variability range for implementation cost estimates and the incentive for providers facing cost shifts, the expense of staying with an outmoded coding system, and calculating that cost as an investment in benefits. They noted NCVHS’s vision to collect information once and use it for multiple purposes, the government’s obligations to use certain international classifications, and ICD-10’s longer shelf life and incremental expansion.
ICD-10-CM PILOT STUDY
Ms. Prophet-Bowman and Ms. Leon-Chisen reported on the AHIMA and AHA field test to assess the functionality and utility of applying ICD-10-CM codes to medical records in various health care settings and the level of education and training professional credentialed coders would require. They reported that coding professionals generally supported adoption and felt the system was an improvement over ICD-9-CM.
Discussion
Dr. Libicki clarified that he estimated a week’s training for hospital inpatient coders; a day for those coding part time. Dr. McDonald suggested doubling the time if doing ICD-10-PCS. Presenters reported that subjects could have coded at least twice as fast if a tool had been available for a fair comparison.
SUBCOMMITTEE DISCUSSION
Members concurred that their potential plan for the coming months was aggressive but doable and discussed future agendas. Considering whether CHI should ease its vigorous schedule, they reiterated their commitment to being appropriate partners and said they could accomplish what CHI wanted in the timeframe suggested, but in a different order.
CHI UPDATE – OVERVIEW
Ms. Adair said CHI, in their effort to adopt standards for use in the federal system, sought initial and concluding input from members and the private sector on inventories and standards available in each of 24 domains they strove to complete by December 31, 2003.
CHI UPDATE - PRESENTATION ON LAB RESULTS CONTENTS AND UNITS
CHI’s primary recommendation is to use a set of hierarchies of SNOMED CT for alpha-numeric Lab results contents. Another recommendation is to use National Center for Biotechnology Information (NCBI) Taxonomy codes maintained by the National Cancer Institute (NCI) for living organisms.
Discussion
CHI sought the Subcommittee’s confirmation of identified gaps; Ms. Adair said CHI would work with SDOs on gaps that were a priority. A motion to accept the report and endorse the recommendations with a caveat that references to LOINC be clarified was made and seconded. Realizing that, technically, they did not have a quorum, members noted the spirit of the group. The vote will be re-visited as a quorum during a Subcommittee conference call. The Executive Subcommittee will be asked to consider how reports can be made available to the full Committee for endorsement without presentation.
CMS HIPAA UPDATE
Mr. Nachimson reported that ten percent of electronic claims submitted to Medicare in late July were in the HIPAA standard format. Since April, percentages rose one percent per month. He noted major issues raised by health plans and said the agency continued to monitor and support the readiness of state Medicaid plans. He encouraged people to follow the guidance in making contingency plans and bring issues to the agency.
Discussion
Mr. Nachimson was complimented on an outstanding guidance letter. He clarified that the regulation that claims submitted after the October implementation deadline must be electronic is only a requirement for Medicare. Rather than set particular guidelines for good faith efforts, he said they needed to look at each situation individually. He emphasized that CMS’s office charged with enforcement was separate and strove not to treat Medicare and Medicaid differentially.
CLAIMS ATTACHMENT
The SDO has presented a proposal using a subset of HL7's clinical document (CD) architecture for contents of attachment data exchanged between trading partners. Either codified, structured data using LOINC values or scanned images can be used for sending attachment data. The Department will consider the proposal in the context of HIPAA, CHI and the electronic health record initiatives.
Discussion
Participants expressed concern that plans or providers might send scanned copies of the entire record and numerous different pieces of information might be required as attachments, losing standardization. Mr. Nachimson said hopefully the definition and explanation of use will include business rules. He said a strength of the proposal was the ability to implement now and later move to a more sophisticated version. Time permitting, the Subcommittee will hold a public session before release of the NPRM.
REACTOR PANEL - HIPAA IMPLEMENTATION
Mr. Wilder said AAHP’s members were pleased with guidance that provided flexibility to briefly continue processing transactions under legacy systems. He noted the guidance indicated CMS and NCVHS would look closely at what worked and did not work to get a code set or standard approved or modified as well as a process for handling issues on an emergency basis.
Mr. Arges urged CMS to outline specific performance expectations and respond quickly and decisively to critical issues not addressed in recent guidance. He said the focus should be on developing specific contingency plan components that ensured that hospitals and other caregivers continued to receive comparable payments.
Mr. Emery reported the preliminary findings of AMA’s recent online survey determining physician’s familiarity with the HIPAA transactions and code set standards, and their preparedness for implementing the standards. He cautioned that a substantial portion of the health care community were likely to encounter problems unless there was additional flexibility on enforcement.
Discussion
Noting tension in HIPAA between meeting business needs and developing a uniform set of rules, it was suggested that technical experts from the health plans identify situations where they believed a batch was appropriate to reject. Mr. Wilder cautioned about regulatory pressure that signaled that paper was a way to avoid HIPAA. Mr. Nachimson said his office would primarily look for good faith efforts. The Subcommittee noted the need to ensure that CMS was aware of the issues and see what could be done without regulation.
The Subcommittee will hold a conference call after Labor Day to ensure that the ICD-10 cost-benefit analysis is adequate to distribute to the full Committe and to approve the CHI recommendations. Members will meet during the September 23-24 full Committee meeting for a HIPAA update and discussion of outstanding issues, hopefully including how CMS addresses them. If any extraordinary CHI issue must be resolved prior to the September full Committee meeting, they will meet September 22. The next full Subcommittee meeting is another three-day agenda October 28-30 that includes discussions on both CHI and ICD-10, and possibly issues on HIPAA stemming from events on October 16. Hopefully the PMRI report will be ready for final revisions and presentation to the full Committee on November 5-6. The Subcommittee will meet December 9-10. Assuming a quorum, they will discuss and hopefully ratify the CHI lab results recommendations on their conference call. Members will be queried about hearing dates for the first half of 2004.
The details of all presentations can be found in the meeting transcript posted on the NCVHS Web site, www.ncvhs.hhs.gov
- DAY ONE -
OVERVIEW ON DRUG TERMINOLOGIES
Dr. Blumenfeld discussed the need for drug terminology, definitions, requirements, concepts and challenges faced working with large integrated delivery systems. Increased focus on health care quality and safety, automated decision support, adverse event monitoring, bio-surveillance post 9/11, HIPAA and consumerism pushed for improving ability to exchange drug information electronically. Terminologies were available but numerous and could be expensive, specialized and incompatible.
He noted use cases for a drug terminology including: (1) consumer access to one’s online medical record for review or to add medications; (2) physicians in ambulatory practice prescribing and sending prescriptions electronically to a pharmacy, transmitting information between plans and DPM’s to-and-from the pharmacy; and (3) computerized physician order entry in the inpatient setting along with communication to the pharmacy and administering nurse. He emphasized the desirability of saving information from all these cases and levels of granularity in an electronic health record.
Dr. Blumenfeld discussed broadly accepted criteria for the quality of terminologies: completeness in content coverage, concept orientation, recognition that concepts are permanent and not replaced, non-semantic or meaningless concept identifiers, poly-hierarchy, formal definitions, context representation, graceful evolution, recognized redundancy, multiple granularities and consistent views.
He noted that a drug terminology should be maintained by a primary code-assigning authority, be available for use at little or no cost, and content should be updated and available as drugs enter the market. Terminology should be backwards compatible with NDC, compatible with evolving international standards, and adhere to an accepted common information model (i.e., concepts in relationships organized into formal classes for query and presentation). It should support abstraction at multiple levels of detail; define how statements are composed and recorded; support synonyms and lexical variance; and, when possible, build from atomic concepts. Core concepts include: ingredients, routed generic, clinical and manufactured drugs, and a packaged product. Additional concepts include therapeutic classification, forms, routes and strength.
Dr. Blumenfeld noted conflicting views on: key concepts, appropriate levels of granularity, where terminology stopped and the information model began, and determining supporting vocabularies. He said it was important to clinical users that considerations move beyond the terminology’s structure and address frequency of updates, data available in database formats, legacy systems supported, backwards compatibility, and the level of customer support as well as the extent customization is possible and data available for multiple platforms. He also emphasized the need for drug terminology, standards, and an information model, noting there are clear requirements and any solution must address conceptual and operational challenges.
OVERVIEW OF FDA DRUG DIRECTIONS
Dr. Levin said FDA’s collaboration with other government agencies focused on the clinical drug and RxNorm, beginning with the active ingredient drug component, dosage form, and additional details ( e.g., adverse events, interim actions,) from FDA on drug products. Utilizing VA NDF-RT, information (e.g., drug class, mechanism of action, physiologic effect, chemical structure, therapeutic intent, clinical kinetics) is tagged to the active ingredient and clinical drugs (including therapeutic intent, kinetics and finished dosage form). All link to the packaged drug and NDC.
Dr. Levin explained how leveraging existing processes (FDA drug product applications and labeling change reviews, NLM and the UMLS) made it easier for information to be available as new drugs were marketed. In processing investigational new drug applications and listings, FDA assigns a code to both active and inactive ingredients. FDA works with NLM to disseminate these unique identifiers. Initially the focus has been on drugs, but FDA is also involved with biologics, dietary supplements, and substances used in devices. NLM can provide the structure of ingredients found in the public domain, so FDA can code. In collaboration, they can code and make publicly available information on investigational substances and others not marketed in the U.S.
Other terminologies are finished dosage form (active ingredient, manufacturer’s unique components), imprint coding on solid dosage form, drug and packaged products.
Dr. Levin reported problems with the drug listing process for drugs marketed in the U.S. with a national drug code: numbers were assigned to non-drug products, altered without changes in products, and reused. FDA will propose regulation changes. FDA is working to: (1) reformat the structured product labeling so information (e.g., dosage administration, contraindications) is computer readable and (2) develop an HL7 standard for structured product labeling based on the clinical document architecture. FDA is working to automate how information is collected from manufacturers through usual processes and disseminated to NLM who passes it on with additional information through the Daily Med to health care suppliers who inform the public. Reliable, comprehensive information coming directly from the manufacturer and FDA will include all U.S. marketed drug products. Changes will be made daily. Distribution will be free via NLM.
FDA has listings for devices tied to brand and commonly used name. The Center for Device and Radiological Health Web site has a code for classification names. Dr. Levin said some of the process for getting better terminology for medical devices is in place but FDA did not have a mandate about information people wanted to know about devices.
Discussion
Dr. Levin noted that: much information used internally was not available publicly, NDC had to be more comprehensive and complete, the finished dosage form had to be moved to an external format when providing information in a structured format, and the structured product labeling format was in HL7. The first round of ingredient codes will be done in 2003; the drug listing involves changing regulations and will take longer. FDA expects a final rule soon requiring manufacturers to submit labeling electronically. He said FDA advised pharmaceutical companies about what happened in HL7 and some companies participated. There were standards for NDC and vocabulary required for labeling, but FDA had not “come up with” standards for other vocabularies (e.g., indication, adverse events.) Dr. Levin explained that the FDA portion of the Daily Med related to labeling and listing information that would now be available to the public. FDA’s proposed rule aimed at a comprehensive list of NDCs.
Dr. Blumenfeld said a core terminology might be enhanced as a commercial venture, but a common core should be available to the public at low cost. Noting concern on the clinical side, he emphasized working to eliminate duplications in NDC codes. Dr. Levin said FDA would consider comments about good codes in their proposed rulemaking.
Dr. McDonald noted the challenge faced trying to map when 30 percent of the NDC codes in pharmacy were “contract” codes that were not in a database. Dr. Levin said the proposed rule aimed to make the database more comprehensive. FDA heard the need for a code for dietary supplements, devices and supplies. Mechanisms were in place for devices. FDA sought to hear more about the need for dietary supplements.
Asked if FDA had considered that the finished dosage form code and imprint code should be the same, Dr. Levin noted that in the 1990’s FDA changed imprint regulations so that the two solid dosage forms were distinguishable. He said FDA needed to hear about the extent of the problem and ways they might address it. He said many botanicals and other components that were becoming mainstream were not considered drugs by FDA but were regulated by them and they also needed to hear about those issues.
PANEL OF PUBLIC SECTOR DRUG TERMINOLOGY DEVELOPERS – RXNORM
Dr. Nelson said RxNorm was a project within UMLS to address the naming of all prescription and OTC drugs approved in the U.S. in a standard format. RxNorm included other clinical drugs in the UMLS, names from commercial sources, and historical entries (drugs withdrawn from the market, ingredient changes, and reformulations since the inception of RxNorm). The clinical drug interoperability effort grew out of HL7 deliberations about achieving a common model and interoperability at the prescribed level. Only 10,000 of the 81,165 clinical drugs (ingredient names with dose, form, or both) identified in the UMLS were approved by FDA, underscoring the need to overcome missed synonymy and refine traditional lexical methods.
Interagency cooperation between VA, FDA, and NLM, honed by defining areas of responsibility and sharing information, addressed the chemical structure and ingredient and structural codes with information about indications, dose form, clinical kinetics, and therapeutic intent. RxNorm’s responsibility was linking active ingredient, drug component, clinical drug, and dosage form. Linking to NDC codes was an additional feature.
The approach was to develop: a standard notation for ingredient, strength and dose form; a standard list of dose forms; and a standard method for representing strength. The plan was to create a normalized form for clinical drugs; establish UMLS concepts for ingredients, drug components (ingredient, strength), dose forms, and drug formulations (drug component, dose form); and formal relationships between concepts. The goal was to have: every concept with a semantic-type clinical drug (e.g., from VA, Multum, Micromedex, First DataBank, Medi-Span, and SNOMED-CT) related to RxNorm and with possible relationships (equivalent, mapped to, trade name, and inverse).
Dr. Nelson noted that the latest release of the 2003 AB UMLS had 4,647 ingredients, 16,983 clinical drug components, 20,318 generic drugs, and 11,326 branded drugs. Currently there are 11,021 relationships of trade names to generic names, 32,688 ingredient relationships, 1,057 relationships between a specific and base form of a drug, and 30,866 relationships to dose form. Newer additions included NDC codes, drug delivery devices, and drug form names.
He recommended steps for “completion”: regular updates and distribution (getting NDC codes faster, weekly or better availability), reliable data feeds, and addressing problematic areas (multi-ingredient OTC, contrast media). Remarking that RxNorm was untested as yet for use in computerized order entry and needed feedback and involvement with developers, Dr. Nelson noted that reimbursement was often based on NDC codes and VA’s readiness to be a test environment. He emphasized the need to address how clinical vocabularies interlock: logical points for the handoff of responsibility from SNOMED-CT to RxNorm, avoiding overlap, and preparing for dealing with new areas.
NLM’s perspective of the Daily Med emphasized having available at one site: RxNorm names to support interoperation among vocabularies and order entry, ingredients as links to the literature though the Medical Subject Headings (MeSH), structure in the Chem ID system, the unique ingredient identifier (UNII) to support interoperation, NDF-RT information, FDA-approved labels, and NDC codes.
Dr. Brown said NDF-RT started with NDF, a centrally maintained formulary/terminology used across VA for class-based decision support and transmission to mail-out pharmacies. Targeted areas for improvement were: patient safety, clinical decision support, and reducing operational costs and “pains” of terminology. A study underway evaluated any maintenance reduction from algorithmically mapping between sites or partners. He noted a transitional drug benefit (honoring prescriptions from outside providers) made a pressing case for interoperation.
Dr. Brown described inheriting up through an ingredient-centered approach using description logics and utilizing a clinical, collaborative perspective. Semantic models are driven by use cases. Reference taxonomy are created using taxonomies algorithmically followed by human expert review. Hierarchies attached at different levels maximize the modeling effort and deliver “smarter” alerts and other aggregations. A semantic model was created. Term definitions have been reviewed and pruned.
The new drug transaction approach and NDF refresh pulled in 23,000 medications. Dr. Brown reported 108,000 NDC level concepts, 11,000 orderable products, 10,500 drug components (both basically the RxNorm SCD level) and 4,200 active ingredients in the July version. The physiologic effect hierarchy has 1,664 elements that “percolate” into about 120,000 assignments of physiologic effects. Mechanism of action is comparable in size, although the hierarchy is relatively smaller. MeSH is the basis for disease names. CT will be used as available. Named roles involving disease states are” “may treat,” “may prevent,” “may diagnose,” or “induces.” Dose forms are based on HL7 hierarchies.
Noting an issue about oncology drugs, Dr. Brown reported an increase in trade names in the update, due partly to FDA input through the new drug transaction mechanism. An independent evaluation will assess completeness.
NDF-RT is being merged with VA’s enterprise terminology services in order to formalize terminology and feed reengineered applications.
ETS is a national initiative to standardize terminology content and services across VA. The terminology will have formalized content and include numerous domains. ETS will deliver NDF-RT in 2003. Laboratory results will utilize LOINC. Clinicians’ notes will use HL7 clinical document architecture.
ERT and NDF-RT collaborate in a VA pharmacy project to rewrite and reintegrate code used for prescribing. CPRS will reengineer decision support components. Enterprise terminology and NDF-RT will link integrated billing and ordered medications. NLM, VA, FDA, NCI, DOD, and the National Institute for General Medical Sciences (NIGMS) collaborate to integrate data in a format available and useful to the public. NCI is working to integrate formal drug models. NIGMS will fund Mayo and Vanderbilt to analyze how NDF-RT might be extended to reflect pharmacogenomics data and advances. DOD will contribute information modeling. The goal is for RxNorm and NDF-RT to fit together with clearly delineated areas and functions. VA is committed to interlocking SNOMED, NDF-RT, LOINC, HL7, CDA, and other sources to eliminate local variation
Discussion
Participants noted that RxNorm resembled a reference terminology (i.e., it had formal definitions and relationships, reference taxonomy, and mapping) and NDF-RT seemed similar to drug knowledge bases. Kent Spackman, chair of the SNOMED Editorial Board, said SNOMED differentiated between terminological knowledge (what is always and necessarily true about a term) and assertional knowledge which might change over time depending on regulation and realm. Assertional information was outside the scope of what SNOMED included as reference terminology.
Dr. Blumenfeld emphasized a correlation between dosage forms and route. Dr. Nelson reported that the dosage forms NLM used had a formulation specification stating injectable or oral suspension. Dr. Blumenfeld observed that descriptive information usually was physically about “a thing” and tended to be definitional. Other information was used in a clinical context and opened up another universe of terminology. Concurring with the importance of making terminology and information models fit together, Dr. Nelson contended that fundamentally it was a matter of engineering.
Dr. Brown clarified that there should be no costs associated with NDF-RT. NDF-RT’s main concern is getting the right kind of user agreement. He suggested LOINC as a model for preventing malignant working, but stressed that they sought to get NDF-RT to whoever wanted to use it for whatever purposes. Their intent was to allow a vendor to incorporate NDF-RT, and redistribute it with their product.
Dr. Nelson explained that the Daily Med could be viewed as a product or service of NLM that provided information about single things. He noted that the most common questions consumers asked Medline Plus was, “My doctor told me this is wrong with me - What is it?” and, “My doctor gave me this - What is it?” Ms. Humphreys observed that much of the information was terminological information and part of UMLS. Noting some related directly to drug and ingredient information that already was part of MeSH and might be used to update them, she pointed out that the Daily Med also fed into other information systems and products disseminated (e.g., UMLS).
Noting they heard the importance in this domain of timely updates and that the maintenance process was dependent on data from FDA, Dr. Sujansky asked if changes in regulation and accurate, timely updates to FDA’s own databases of drug listings were in the critical path to timely updates to the testifiers’ databases. Dr. Nelson said they “were not totally dependent” on FDA: the VA dealt with the update problem, commercial vendors of pharmacy knowledge bases cooperated, and new drug transactions already were being modeled. But he noted that they were dependent upon FDA for the availability of information (e.g., NDC codes) and on First DataBank, Micromedix, Multum, Medi-Span, and the literature. Dr. Brown said VA already maintained it, using 1970’s tools and looked forward to automation. He noted another VA project, New Term Rapid Turnaround, invited the field to input requests for new terms and rapidly incorporated, reclassified, and released them to the servers.
PANEL OF PRIVATE SECTOR (COMMERCIAL) DRUG TERMINOLOGY DEVELOPERS
First DataBank recognized that no single concept met all continuums of care needs and evolved new concepts, identifiers, and related supportive databases. Everything in their database is driven by something packaged and identified with an NDC, universal product code, or health-related item code. Attributes are categorized (e.g., clinical route, dosage form). Relationships are apparent through dosing modules.
Another level of abstraction, routed med ID, connects concepts useful in prescribing through to order fulfillment. Other levels are the routed dosage format ID, the med ID and brand level information, proprietary name, and generic formulations. Routes of administration are defined through this path of individual ingredients and dosage forms. The notion of status is introduced over time: e.g., active, retired, obsolete replaced (with a path to replacement concepts).
First DataBank does not catalogue pre-market released or investigational drugs. Mr. Robinson estimated that half of U.S. hospitals, 35,000 retail pharmacies, 47 state Medicaid programs, DOD, and most pharmacy benefit managers and claims processors use their content. Seventeen entities receive daily updates. First DataBank has about 350 monthly customers, 160 quarterly, 14 semi-annual, 32 semi-monthly, and 155 weekly. The trend is toward weekly updates.
Mr. Robinson contended that a drug knowledge base offered added value through relationships to pricing, uses, side effects, drug interactions as well as indicators of generic nature. First DataBank endorsed and would add a normalized public domain ingredient identifier into their databases, but he noted concern about changing NDC and proposed instead augmenting them with an alternative number.
He said there would continue to be a need for First DataBank’s consistent view of codifying drug content if NDF-RT and RxNorm were selected as the PMRI drug terminology. He considered RxNorm more compatible with what drug terminology had to be and welcomed an interoperability layer through the clinical drug. But he noted that normalized medication names in the public domain that mimicked what First DataBank offered were proprietary competitors. He embraced the notion of disparate systems communicating allergen and patient profile information at the generic level, but he expressed concern that NDF-RT crossed the line and offered knowledge base content.
Licensure of NDDF Plus requires a fee (based on how frequently content is received and the application) and would be required for their terminology component if NDDF Plus became a drug terminology. First DataBank is open to discussion if a third party chose to compensate for placing that information in a more open forum.
Mr. Robinson urged NCVHS to consider: types of health care applications intended for support through this terminology and that form should follow function; addressing timeliness of content, updates, frequency of database distribution, and the energy committed to providing them; the breadth of the terminology and the comprehensiveness required of drug concepts; and hidden costs of a standardized drug concept dictionary (e.g., implementation, maintenance, and impact on preinstalled drug databases).
Cerner endorsed federal initiatives to standardize drug terminology, noting it was good for public safety, clients, and business. Cerner believed FDA should have primary responsibility for standardized terminology and that the 13 “desiderata” for drug terminologies presented to NCVHS in 1998 should be incorporated into any federal initiative. Cerner was prepared to contribute intellectual property and resources to FDA and Dr. McNamara presented a proposal.
Multum’s products are based on a lexical foundation that, at the highest view, links packaged drug products, generic formulation concepts, main Multum drug codes (MMDC), and a generic concept identifier, all grouped into categories. Drug packaged products coded with NDC are linked to a MMDC. Each MMDC relates up to a generic concept identifier. Concepts are separate from names describing concepts; 38,000 drug names at different types of levels of granularity are included. Clinical information is hung at various levels within that hierarchy (e.g., MMDC, NDC).
Multum Lexicon is a minimal dataset of 17 tables available under for-fee license agreement (5,700 licensees) and as paying clients. The Vantage Rx Database has 144 tables including everything in Multum Lexicon plus drug categorization, expanded synonymy, and clinical content. Cerner respectfully disagreed with the Subcommittee’s conclusion that concept permanence was not reflected in Multum Lexicon. Mr. Blair confirmed that, after discussing the issue with Dr. McNamara, Dr. Sujansky had agreed that, technically, MMDC met the required criteria.
Noting that: creating a drug terminology standard is expensive and time consuming, multiple well-developed drug terminologies exist, and FDA can gain from experiences of commercial vendors of drug information systems, Dr. McNamara proposed that FDA invite proprietary drug information providers to contribute drug terminology data in open source format to FDA. In turn, Cerner would: make the Vantage Rx Database available for evaluation; determine which tables, fields and data it needed to create a national drug terminology standard; use that code set; and provide FDA with tables, fields, and data selected by FDA for it and the public’s use under a “liberal open-source agreement.” FDA would have unlimited right to distribute, redistribute, use, and modify data.
Dr. McNamara said this approach created a fast-track and low-cost path to a comprehensive national standard for drug terminology, encouraged multiple vendors to contribute to a core set in an open transparent process, and ultimately facilitated interoperability.
Medi-Span data sets support a wide range of applications throughout the industry. Mr. Dubois said Medi-Span data sets were timely, complete, and met market demands. Generally, any application requiring drug information could be supported. Given a lack of customer demand, Medi-Span had not made investigational drug information in their data sets comprehensive, but would if the market asked. He reported that thousands of health care organizations utilized Medi-Span’s content and knowledge to facilitate millions of transactions daily. Frequencies of update ranged from daily to quarterly.
Medi-Span believed that drug terminologies defined a drug (including route, form, clinical or generic packaging drug) but did not include attributes found in a drug knowledge base that enabled clinical screening and provided integrated and referential knowledge.
Mr. Dubois agreed that access to new drug data in the Daily Med database could be helpful, but he noted that Medi-Span would still have to contact manufacturers for information. With NDF-RT and RxNorm, he said Medi-Span’s drug files and clinical products would provide additional value-added fields and key data links to Wolters Kluwer Health’s clinical screening and referential database content. He expressed concern that these initiatives were directly competitive. FDC mandated that Medi-Span divest from First DataBank to ensure a competitive environment, and he contended that NDF-RT and RxNorm could remove that competition. Mr. Dubois said Medi-Span products were guided by market demand; if NDF-RT and RxNorm were selected as PMRI drug terminologies, that process would continue.
Like Multum, Medi-Span respectfully disagreed that their terminology did not meet required criteria. If, upon further consideration they were included, Mr. Dubois said Medi-Span would create a drug terminology database as a separate deliverable that minimally included NDC and drug concepts with descriptions or unique identifying attributes. Licensing terms and costs could be negotiated.
Mr. Dubois recommended that the VA NDF-RT drug file use the NDC billing unit standard for defining packaging, noting it was widely utilized in the marketplace and allowed billing and packaging concepts to be shared with applications using another drug file. Emphasizing that more communication and collaboration were needed between public initiatives and the private sector, especially clinical knowledge-base providers, he conveyed Medi-Span’s and Wolters Kluwer Health’s commitment to this initiative.
Dr. Spackman clarified that SNOMED’s HHS license included core drugs and that it was a comprehensive clinical terminology that included hierarchical and logical integration between SNOMED clinical drugs and the rest of SNOMED. He emphasized that a U.S. realm extension complementing the SNOMED core and possibly utilizing RxNorm and NDF-RT should be produced and distributed free, with FDA and HHS responsible for some of that process.
SNOMED provided and maintained core drug integration; had codes, hierarchies, and logic links between generic ingredients and their classes; therapeutic drug monitoring tests integrated between SNOMED procedures; LOINC codes and SNOMED ingredient codes (LOINC mapping and updated integration were needed).
Main components needed to support end users included: concept identifiers (SNOMED realm independent drugs plus clinical concepts); realm specific drugs and devices including other realm specific extensions in areas like administrative codes; and commercial and local public sector knowledge bases.
Dr. Spackman noted that the U.K. had a SNOMED-CT core and a U.K. realm extension with dictionaries for primary drug data, secondary care, and medical devices. Dictionaries were managed by different entities. The commercial sector FDB and other suppliers also provide knowledge bases to that market. The integrated model had a virtual therapeutic moiety (drug containing ingredient) and a virtual medicinal product. SNOMED-CT core plus realm extension supported numerous use cases.
He pointed out the distinction between terminological knowledge (what is always true about a drug) and things specific to a realm. He noted that RxNorm and SNOMED models were similar, but not identical. Given the need for core SNOMED to support country-specific drug extensions, there would always be overlap, but coordinating development could reduce duplicate effort.
Noting that people were interested in FDA collecting terminology for ingredients and labeling information from manufacturers, Dr. Levin reiterated that in their drug listing FDA interfaced with manufacturers to collect information about every marketed product in the U.S. including everything First DataBank and others collected. He said the comments were encouraging and supported FDA supplying this information as part of what FDA did regulating drug products. FDA already proposed a regulation that companies provide labeling in electronic format in order to handle drug labeling more efficiently. He said improving drug listing would require regulatory change about an NDC. Need and priority would determine how long it took to put regulations in place.
Discussion
Mr. Dubois said Medi-Span supported the generic product identifier, but the database did not have an enhanced attribute that identified therapeutic classification. Dr. McDonald explained that nonprofit entities had free access to everything but the therapeutic categories in Cerner’s product. Mr. Robinson reported that First DataBank maintained unique identifiers for all the levels of drug abstraction he presented. Medi-Span and Cerner’s generic drug identifiers included the concept of route administration.
Karen Eckert and Tom Bizarro, Co-Chairs for NCPDP’s Billing Unit Standard (BUS) Workgroup, said BUS, available free to members, ensured that electronic claims transmissions were billed and adjudicated using a common unit of measure.
Dr. McNamara clarified that Cerner’s proposal to offer Vantage Rx to FDA for distribution entailed unrestricted redistribution of any content FDA viewed as terminologically relevant for drug terminology (not referential and clinical content). Mr. Dubois said anything Wolters Kluwer offered in open forum would be available but that they needed to know more about the scope and initiative before making a commitment.
Ms. Eckert believed Medi-Span met the required criteria because what the descriptors maintained did not change the intent, and inactive terms were not outputted to customers.
Mr. Robinson explained that about 12 research associates at First DataBank tracked NDC changes and entered data. Five others maintained structured content. He expressed concern about FDA having a drug database dependent upon pooled providers and suggested instead that a fast-track provide labeling in codified format.
Dr. Spackman said the update frequency for SNOMED-CT’s drug extension that fed the U.K. core was every one-or-two weeks. The U.S. core was updated every six months. SNOMED is prepared for more frequent updates. Discussions about overlap between SNOMED drug codes and RxNorm were initiated. He saw no major barriers.
Ms. Eckert said Medi-Span found RxNorm and its drug concepts and relationships useful, did not consider RxNorm competitive and would map to it. Medi-Span found NDF-RT extremely competitive. First DataBank viewed interoperability (e.g., supporting allergy screening, communicating the patient med) useful, but Mr. Robinson was anxious about providing proprietary names and packaging information related to core concepts. A knowledge base of uses, side effects, and drug interactions would be directly competitive. He recommended defining the scope of public terminology with RxNorm and excluding knowledge bases VA-centric with NDF-RT. Cerner and Multum supported RxNorm’s standardization around drug terminology and believed that they could compete with the drug knowledge base side of NDF-RT. Mr. Bizarro, Vice President, Sales and Marketing, said First DataBank felt comfortable competing against other knowledge bases, so long as they were not available in the public domain free of charge. Dr. Lincoln noted that anything VA developed was available free under FOIA. NDF-RT is based on NDF (downloadable at www.va.gov/vista). Noting they had discussed the soft line between knowledge base and reference definitional information, he said VA’s approach was driven by use cases and information NDF-RT contained because current clinical systems depended on it, though others might define it as knowledge base. Dr. Lincoln said VA did not view RxNorm as precluding purchase of commercial drug knowledge databases.
Asked how Medi-Span recognized synonymy/substitutability when composition of the medications in terms of active/inactive ingredients was not in the concept definition, Mr. Dubois said that an ingredient was seen as an attribute of a drug item. That information defined the generic product identifier that is the proprietary therapeutic classification.
Dr. Nelson clarified that RxNorm built out of the drug terminologies and did not have packaging information. Coverage included of NDF, Multum, Medi-Span, and First DataBank at the generic level, including NHS Clinical Terms (Read codes) and their drug names.
Noting a proliferation of “fake” NDCs, Dr. McNamara questioned that their number was indicative of the quality of product or coverage. He suggested that the quantity of NDCs be reduced and their quality improved. Multum’s editorial policy included an NDC only when verified in two locations or containing a price, indicating an actual code.
Dr. Fitzmaurice noted that testifiers indicated it could be useful if the government identified drugs, issued identifiers, and mapped, but said it was threatening when clinical information was overlapped. Testifiers said it would be useful if FDA received package insert information from a manufacturer, linked to the drug identifier, and posted on a Web site. But they did not want the government to subsidize someone (other than perhaps FDA) to produce and make available this information as a competitive product. Testifiers wanted to be involved in mapping and see FDA, SNOMED code, indications, contraindications, and other components.
Cautioning that enough was not being done publicly if they went beyond RxNorm to package levels, Dr. Spackman supported FDA’s maintenance. Asked whether to use SNOMED or RxNorm (if they were considered equal), he said either; but having both and translating would be sub-optimal.
PANEL OF MEDICAL DEVICE TERMINOLOGY TESTIFIERS
Dr. Hefflin noted that the GMDN was created through merging six established medical device terminologies and resulted in 13,000 terms for generic device groups (sets of devices of similar intended use or technology) distributed among 12 medical device categories. Terms were assigned to a primary category; some were also assigned to a secondary or tertiary category. Assignment was based on intended use and the specificity of the category (e.g., “reusable instruments” is a secondary category for a dental probe; “dental devices” is more specific and primary).
The project consists of a project council, expert advisory team, and 12 device expert task workgroups. Groups are international; FDA staff are active at all levels. Workgroups selected or created a preferred term to represent each generic device group and linked remaining terms to the preferred term as synonym (common-use or familiar terms users could utilize to navigate), or equivalent (an essentially identical term that did not add information and was not part of the functional GMDN terms. Task groups created definitions for each preferred term conveying intended use; target area; the device’s technical principal or working method; materials/components; and device form/shape/ physical state. When three-or-more terms had the same base concept, the advisory team reconciled inconsistencies and entered template terms, providing a simple hierarchy.
Dr. Hefflin said the project worked to standardize the base concepts of all terms in the GMDN and reflect the device’s intended use. Definitions were improved to incorporate all aspects of the standard, especially definitions from the IVD terms. Many new synonyms were linked and preferred terms were added for new devices. The second version of the GMDN is due for release in 2003. GMDN contains about 6,400 preferred terms, 10,000 synonym terms, and 600 template terms.
GMDN is an international nomenclature that provides generic descriptors for devices and is based on ISO 15225. GMDN is developing plans to merge with UMDNS (a nomenclature used to develop GMDN over the next three-to-five years). Mr. Kessler said the devices GMDN regulated encompassed everything clinicians touched in the operating room, ranging from Band-Aids to implantable cardio converter defibrillators. The GMDN policy group consisted of regulators, ISO and CEN members, industry representatives, the European Commission, and the Global Harmonization Task Force (GHTF). After a hiatus, FDA, NKKN, Health Canada, Academia, and ECRI were recently invited back into GMDN. Management is transitioning to a commercial company.
A proprietary product licensed by the Commission and the standards organizations in Europe, GMDN’s licensing and modification will be controlled through the maintenance agency. Data will be continuously updated. Requests for new terms will be dealt with as received. A new version is intended annually.
GMDN has been adopted by the European Economic Association, GHTF, and Japan, Australia, and South American and Asian groups. The Office of Science and Technology is mapping the FDA system of product codes to the GMDN and will convert to GMDN next year and develop a Web interface. In 2004-2005, industry and consumers will be able to find product codes by searching the FDA database using GMDN terms.
Ms. Coates explained that ECRI is a collaborating center for WHO with a purview for patient safety and developing, maintaining, and promulgating worldwide a UMDNS. ECRI also facilitates translation to enhance information exchange on device safety, procurement, and management within global harmonization efforts. ECRI’s formal hierarchical system for organizing device-related information has been part of the UMLS since 1991. Since then ECRI has maintained mappings and linkages to other device terminologies, the United Nations Standard Product and Services Codes and other medical vocabularies in UMLS (e.g., ICD, CPT, MeSH, SNOMED, Read). Respectfully disagreeing that ECRI was less broad in its coverage than GMDN, Ms. Coates reported that the 2004 edition contained 17,221 medical device terms (including 6,745 preferred terms/concepts equivalent in device group level to GMDN and 10,476 entry terms.
UMDNS also has a hierarchy, encoded and maintained intra-nomenclature relationships, attributes and definitions as well as mappings and linkages. A unique five-digit numeric code/identifier assigned randomly and sequentially, corresponds to each preferred term, codes carry no inherent meaning, and UMDNS has concept orientation.
Ms. Coates explained that codes for retired concepts and terms were never reused for new concepts. Usually, retired preferred terms became entry terms pointing to new preferred terms. An archival list of retired preferred terms is in the master database.
Terminology was broad in scope, covering medical devices and supplies, clinical lab equipment and reagents, hospital furniture, systems, and test equipment. Ms. Coates said UMDNS’ coverage of devices was more comprehensive and specific than SNOMED, and that SNOMED’s device domain was insufficient to capture the precise device concepts needed for the electronic medical record, patient safety, and medical error reporting.
UMDNS is licensed free to regulatory agencies, hospitals and manufacturers. ECRI charges a fee for eCommerce sites and commercial software companies.
ECRI has a partnership agreement with FDA to use the same guidelines to map respective nomenclature systems and establish new terminology. ECRI currently has 1,922 license agreements in 95 countries and regulatory agencies for post-market surveillance, vigilance reporting, data storage and retrieval. DOD and VA use UMDNS. UMDNS is used for: inventory control, hazard and recall tracking, equipment planning, procurement, technology management, monitoring emerging technology, emergency preparedness and in conjunction with ECRI software and knowledge bases.
Discussion
Noting the maintenance model involves direct sponsorship by FDA and other device regulatory bodies, concern was expressed about voluntary sponsorship of a designated standard. Mr. Kessler said FDA and ECRI were negotiating an arrangement to map their product code system, GMDN and UMDNS over three-to-five years.
Ms. Coates said, despite the MOU signed in 1996 with the European Standards Organization in hope of realizing one device terminology, constituents’ needs were not served by quality issues Dr. Hefflin addressed and an “incredibly” bureaucratic process for maintaining, updating and disseminating GMDN. Mr. Kessler noted that FDA and ECRI spent much time trying to make GMDN global and hit “serious bumps in the road.” The product generated had problems and ECRI could not then migrate legitimately to their product or move FDA’s product code effort to GMDN. He said Dr. Hefflin had worked full time for a couple years trying to “wrestle GMDN toward something that made more sense,” and was closer to ECRI’s terminology. Political problems were daunting. But FDA and ECRI were “back at the table.” Mr. Kessler said FDA believed ECRI’s approach and system made sense and wanted to carefully migrate all they could into one nomenclature.
Dr. Hefflin said an expert team was responsible for technical aspects of UMDNS nomenclature, updating and maintenance. Asked about FDA’s plans for using device nomenclatures internally to support patient safety, Dr. Hefflin said GMDN was developed enough to be used for patient safety and as FDA migrated to GMDN the nomenclature would be used the same way as the current system.
Mr. Kessler said one reason GMDN was developed was for exchanging reports for post-market surveillance and global vigilance. FDA currently used the product code system and would map and merge to GMDN over two years, hopefully mapping simultaneously to UMDNS. It was noted that GMDN is both an ISO and CEN standard.
Efforts were underway to develop an adverse event reporting database. Asked if ECRI’s database was used to notify that a particular valve might go bad, Ms. Coates noted their weekly alerting bulletins. While the nomenclature was above the make-and-model level, knowledge databases linked to the specific manufacturer identifier level.
Noting a disconnect in how medical records developers thought about their domain and the domain of devices, Ms. Coates emphasized that systems were not integrated and required interoperability. Even VA, which she considered at the forefront of information systems development for hospitals, had a disconnect between procurement and patient safety systems. Dr. McDonald clarified that device content medical records were in a number of places, but captured at another level and not standardized. He stressed that there was not enough communication between standards groups and the needs.
Ms. Coates recommended bar coding at a highly specific level. Responding to a query about why tracking of medical devices was separate from the clinical piece, Dr. Bickford cautioned that the pieces could never be assembled until the framework let them see beyond diagnosis and procedures.
SUBCOMMITTEE DISCUSSION
Mr. Blair proposed that RxNorm was suited as a reference terminology for drugs and fit the core set of PMRI terminologies. Overlap with SNOMED-CT was being reconciled within NLM. They also heard that NDF-RT, First DataBank, Medi-Span and Multum were or would soon meet direct patient care needs and map to RxNorm. Members discussed supporting FDA’s efforts to establish identifiers for ingredients and clinical drugs and its essential role supporting comprehensive, timely development of RxNorm.
Dr. Sujansky emphasized considering whether RxNorm was ready as a terminology standard and if NLM believed RxNorm could be maintained on the weekly-to-monthly basis required for a drug terminology standard.
Observing the lack of a quality assurance step to ensure a sender of a Multum-encoded term that the communication would be effectively mapped for the intended purpose, Dr. McNamara from Cerner advised NCVHS to consider a uniform standard with one set of codes. Ms. Humphreys contended that using multiple sources to secure coverage of heavily used drugs and creating a map between RxNorm forms and forms in other systems was not incompatible with RxNorm being developed as a standard nomenclature for any new clinical drug. Having designated core terminologies, she noted the need to identify overlaps and choices in following the recommended standards.
Dr. Spackman emphasized that many participants’ confidence in the recommendations was contingent on NCVHS explicitly stating that there had to be coordination between RxNorm and the SNOMED drug terminology. Dr. Sujansky noted potential gaps in RxNorm that the Subcommittee could recommend filling and that members had discussed, without resolution, comprehensive mapping to NDC codes and accommodating a standard therapeutic classification system for drugs.
Dr. Steindel suggested that the letter emphasize the need to identify devices in PMRI terminology and that efforts by ECRI, FDA, ISO, and CEN to merge terminologies be encouraged, supported, and accelerated. Dr. Hefflin said FDA continued to support and develop GMDN for global harmonization purposes. Observing that if the terminologies merged they would be recommending a PMRI terminology different than what FDA and the rest of the world used for regulatory purposes, members emphasized that any statement supporting the ECRI terminology that aided FDA and led to a merger would be based on a determination that the principals were good. Mr. Blair thanked Dr. Steindel for an agenda that made it possible to reach conclusions.
- DAY TWO -
REVIEW AND UPDATE OF PMRI - TERMINOLOGY RECOMMENDATIONS
Mr. Blair noted that a year-and-a-half ago the Subcommittee recommended the first PMRI standards: the message format standards formally adopted in March 2003. This second phase selecting terminology standards began a year ago. Guidance from experts on modifying the August 2000 guiding principles for the selection of PMRI standards was folded into a questionnaire sent to 44 organizations that asked to be considered as a PMRI terminology. First results were presented to the Subcommittee in March 2003 and subsequent versions of that analysis elicited further critiques ensuring that interpretations were correct and consistent. Today members would review the second draft of the letter offering recommendations for the core set of PMRI terminologies that would serve as reference terminologies. The letter identified how NLM would map to other clinical terminologies and HIPAA financial administrative transactions, noted related terminologies, and discussed key areas requiring attention.
Members focused on realizing a controlled terminology for clinical drugs and for selected related drug terminology (e.g., dosage forms and components). They noted that drug knowledge bases served different markets and set of users and that terminologies FDA would support were needed to sustain medication terminologies and impart RxNorm. Noting that RxNorm was a necessary element still in development that would be revisited when it went into production, members discussed areas where it had to be augmented and support of resources and timeliness for NLM, FDA and medical information system development.
ICD-10 IMPACT ANALYSIS - STATUS REPORT
Dr. Libicki presented work done by the Science and Technology Institute of RAND on the costs and benefits related to switching from ICD-9-CM to ICD-10-CM and ICD-10-PCS. RAND drew on: prior testimony before the Subcommittee, a literature review, and 87 telephone interviews with 102 experts and stakeholders.
The ICD-10 codes most salient feature remained greater specificity: a mostly two-fold increase in codes in ICD-10-CM, and a 50-fold increase in ICD-10-PCS. Neither was designed to restrict predecessors’ one-to-many expansions but, with its 50-fold increase, ICD-10-PCS was de facto a one-to-many expansion; ICD-10-CM is a one-to-many expansion for two-thirds of its codes.
RAND estimated first-year transition costs at $600-$750 million, plus up to $20 million-per-year thereafter. Estimates do not include the cost of revamping CMS’s information systems and assume a simultaneous switch. Considered separately, RAND estimates up to another $200 million evenly split. Most potential loss in productivity is associated with ICD-10-PCS.
RAND subdivided the cost of training to use both classifications into the cost of training hospital and non-hospital coders, code users, and others (notably physicians). AHIMA estimates that bringing 50,000 hospital coders to proficiency in new codes (mostly ICD-10-PCS) will require several days to a week and cost $100,000,000. Dr. Libicki noted that VA informally tested ICD-10-PCS and concluded that coders with incomplete knowledge of anatomy and physiology might require months of training, doubling the cost. Dr. Libicki noted that there was no evidence that broad training was required in Canada. RAND calculated that ICD-10-PCS and CM were not necessarily harder to learn than ICD-9.
RAND determined it would require eight hours to train people who only did a small percentage of coding and solely with ICD-10-CM, yielding a cost of $100 million to train coders outside of hospitals. Input suggests that the cost of retraining physicians will depend on how different ICD-10 is from ICD-9. RAND estimated that physicians in clinical practice might need four hours of education in the new diagnostic codes, and another four hours of additional sessions. (Estimated cost: $200,000,000). Dr. Libicki noted that professional coders were extremely wary of doctors doing their own coding.
Assuming a six-month learning curve and that, in some cases, information must be added to the patient record, annual loss in productivity might be $20,000,000. Noting the small sample and the extent that most hospitals used automatic encoders, Dr. Libicki said the actual loss might be lower. Systems upgrades account for half the cost of transition.
AHA and FAH were primarily worried about covering training costs. Testing and integration are major hospital chains major cost. Software vendors said they needed up to three years to transition. RAND estimated $100,000,000 for the sector, adding that the cost of adopting software to accommodate ICD-10 will run $100,000,000-$250,000,000 if extrapolated to the entire payor base. Dr. Libicki noted, however, that fiscal intermediaries that handle claims for CMS will also have to expand the code lengths they accommodate because of HIPAA regulations. Not including CMS, hospitals are likely to bear 40 percent of the cost, outpatient facilities 25 percent, and payors 35 percent. No systematic shift in payments was predicted.
Possible benefits include: more accurate payments; fewer erroneous, rejected or fraudulent claims; better understanding of health care; improved disease management; enhanced ability to evaluate providers; more timely intervention for emerging diseases. Rand was in the process of estimating these benefits.
Dr. Libicki said simultaneous switching: made it easier to mentally prepare coders, was more efficient, meant testing only once, and long-awaited DRGs (linked to diagnosis and procedures) would not be further postponed. He emphasized that at nearly a trillion-and-a-half dollars a year, the U.S. health care industry is enormous and a slight shift in efficacy could make an economic difference that easily surpassed these estimates.
Discussion
Dr. Libicki emphasized that potential gain exceeded potential cost, but he cautioned that there was considerable room for error, especially in the assumption of benefits. Dr. Fitzmaurice said incentives were needed for using a new coding system. Participants clarified that ICD-10-CM is in the public domain. ICD-10, which is only used for mortality in the United States, has a copyright, but is available to anyone who needs it for that purpose.
Ms.. Pickett said, now that AHA and AHIMA had finished their pilot, impact on the classification could be gauged and the crosswalk finalized. Dr. Libicki emphasized that, with the logic of the old crosswalk and a clean one, they could translate codes and, in protocol terms, translate, process and “work the number out the back end.” As the DRGs changed, the logic needed to be fixed, but initially the crosswalk would reduce cost.
Ms. Greenberg noted that the target was to get the final report to the Subcommittee members and staff a week before the full Committee meeting.
ICD-10 IMPACT ANALYSIS - OPEN MICROPHONES
Ms. Fraizer discussed why AAPC supports enhanced standards for coding health care information and why it takes partnering for a successful transition. Referencing Dr. Libicki’s white paper, “The Role of Technical Standards in Today's Society and in the Future,” she noted a healthy process depended on: letting the market and different communities decide; partnerships that presume a public interest, timely standards, and that neither split the user community into disparate camps or biased the marketplace; and partners that collaborate in the voluntary regulatory process delivering a shared message to a diverse audience. AAPC offered help in enhancing the ability to standardize and advance information technology among the coding community.
Discussion
Robert Tennant, with the Medical Group Management Association, pointed out that, although costs could be categorized, payors and vendors would cost shift to physician practices that had to pay more for premiums and software upgrades. He said including the variability range for cost estimates in the report would give a more accurate presentation of potential costs. He asked what the incentive was for providers facing cost shifts to move forward with coding changes.
Ms. Leon-Chisen commended Dr. Libicki on doing the tough job of quantifying these numbers. She suggested that calculating the cost to implement spread out over 10 or 20 years might give a better idea of what it meant to invest in benefits. She emphasized being clear about the expense of staying with an outmoded coding system that could not represent today’s conditions and services.
Dr. McDonald questioned that these changes would be of much value to the whole system. He recommended looking at CPT codes for details on procedures and patients. He questioned why they did not use SNOMED for coding diagnoses and be done, rather than use a system without internal structure that had to be changed every decade.
Ms. Greenberg observed that there was a fair amount of recognition that FDA and other parts of the government had obligations to use certain international classifications. ICD-10-CM is a clinical modification of an international classification, and ICD-10 has an alphanumeric code structure that provides opportunities for incremental expansion rather than wholesale replacement. She noted that ICD-10, which 10-CM was based on, had a longer shelf life by being alphanumeric and that there is an international process by which ICD-10 can be modified incrementally. The full Committee stated on record that the word terminology applied to the whole continuum. The PMRI report and WHO viewed very detailed clinical terminology and more aggregated classifications with different applications for each. The goal was that data collected in systems at the more granular level recommended for electronic health records could be used to map to other classifications (e.g., statistical and reimbursement applications, reporting to state health departments). One way to implement 10-CM when doing that kind of map is to implement SNOMED CT.
Cautioning that PCS would run out of code space and that different hierarchies could not be mapped because it was so complicated, Dr. McDonald warned against “putting themselves in a box.” Recalling that NCVHS’s vision was to collect information once and use it for multiple purposes, Ms. Greenberg noted an ongoing need for classification.
ICD-10-CM PILOT STUDY
Ms. Prophet-Bowman said over the summer of 2003 AHIMA and AHA field tested ICD-10-CM to assess both the functionality and utility of applying ICD-10-CM codes to medical records in various health care settings and the level of education and training professional credentialed coders would require. Some 187 health information management professionals participated. Almost every state was represented.
She said the study’s primary limitation was the lack of user-friendly tools. Locating a diagnostic term is the first step in determining the applicable code. Terms are located and appropriate codes and code categories identified with the ICD-10-CM alphabetic index.
Both ICD-9-CM and ICD-10-CM diagnosis codes were assigned for each medical record. Participants were instructed to assign codes as completely and accurately as possible, according to existing medical record documentation. Participants completed three surveys: demographic, record, and follow-up.
All participants had a general health information management or coding credential; 85 percent was credentialed as a registered health information administrator or registered health information technician. Over half the participants worked in a short-term acute care hospital; 18 percent worked in an integrated delivery system. Average bed size for short-term acute care hospitals was 365 beds. Size ranged from fewer than 50 beds (13 percent) to greater than 400 (43 percent).
Ms. Leon-Chisen reported that 6,191 medical records were coded including: short-term acute care hospital inpatient (42 percent), short-term acute care hospital outpatient (39 percent), post-acute care setting (8 percent), and physician practices (6 percent). Ninety-six inpatient and outpatient encounters in a behavior health facility were coded. Only 34 records were reported in the “other setting” category. Participants were required to submit a problem identification form for each record that could not be coded because of conflicting instruction entry or diagnosis. Forms identified specific ICD-9-CM and ICD-10-CM code(s) assigned and gave narrative explanations.
More than 60 percent of the 158 participants that completed the survey indicated being at least somewhat familiar with ICD-10-CM. Over 75 percent (participants only had two-hours training via the Web) said three-days-or-less ICD-10-CM training would be adequate for experienced coders. Nearly 71 percent said ICD-10-CM’s clinical descriptions were better than ICD-9-CM’s. Despite difficulty in locating diagnostic terms in the index, 64 percent considered the notes, instructions and guidelines clear and comprehensive. Over 76 percent felt ICD-10-CM was an improvement over ICD-9-CM; 83 percent supported migration to ICD-10-CM.
Respondents uncertain about supporting migration to ICD-10-CM noted problems with the tool used in the index and poor physician documentation that could stifle reaping benefits from ICD-10-CM’s greater specificity. The most significant comments concerned: the tool (Ms. Leon-Chisen emphasized that this was unrelated to the ICD-10-CM structure and would not be a factor when ICD-10-CM was implemented), the cost of change, and availability of resources to assist coders. Other concerns involved the cost of implementation, the shortage of coders, and that systems would have to change.
Some 78 percent of the respondents said ICD-10-CM should be implemented in three years or less. Others expressed the need to implement “soon,” but lacked a definitive time frame. Respondents emphasized the need to improve the tools and implement as soon as vendors and payors could accommodate them.
Next steps include reviewing the validation sample and identifying mismatches and code assignments between participant and validator. Reconciliation will be via review of the affected code or codes and relevant index entries, instructions and guidelines. Data from problem identification forms will be compiled by issue and relevant ICD-10-CM codes and analyzed. Identified problems will be verified and a compilation will be submitted to NCHS for appropriate modifications. A formal report will be issued.
Ms. Leon-Chisen reported that coding professionals generally supported adoption and felt that the system was an improvement over ICD-9-CM. Participants did not report more difficulty using ICD-10-CM in any setting. Although training was minimal and most said additional training would be advantageous prior to implementation, in most cases preliminary results indicated participants appropriately assigned ICD-10-CM codes. She said an analysis of the problem identification forms would be shared with NCHS so enhancements might be made prior to implementation.
Discussion
Dr. Cohn noted that, on ICD-10-CM alone, this self-selected group of “brightest and best” felt they needed four-days-or-more for training. Ms. Prophet-Bowman said responses were “all over the map,” but the majority of this group that received extremely minimal training estimated that experienced coders would need one, two or three days training. Dr. McDonald suggested doubling the time if also doing ICD-10-PCS. Asked how the group compared to coders in two-person doctor's offices, Ms. Prophet-Bowman reiterated that these were all credentialed health information management professionals with formal ICD-9-CM coding training. Dr. Libicki clarified that he estimated a week’s training for hospital inpatient coders and a day for those who coded part time.
Ms. Prophet-Bowman said collaborators at Ohio State University were quantifying an estimate of the impact from not having a tool available for a fair comparison. Ms. Leon-Chisen and she estimated that, conservatively, participants could have coded at least twice as fast if they had the equivalent level of the manual codebook/electronic tools used for coding with ICD-9-CM.
Ms. Prophet-Bowman will present a short report at the September meeting. The full written report is targeted for the end of September and will be posted on their Web site. Ms. Greenberg expressed NCHS’s appreciation to AHIMA and AHA for this voluntary effort and the field for their interest in 10-CM and this process.
SUBCOMMITTEE DISCUSSION
The Subcommittee will discuss the PMRI letter further at the October meeting, aiming to bring it to the full Committee in November. Dr. Sujansky will include elements agreed to that day in a draft he will present to the full Committee in September. CHI will report on their medication domain recommendations for terminology in September or October. Within two weeks, members will receive and review another version of an ICD-10-CM and PCS impact analysis in a conference call and release as a draft public document. Members will also assemble a panel to react to Dr. Libicki’s presentation at the September full Committee meeting. Panelists will provide further information on cost and impact. Time permitting, new NCVHS members will be updated on what has been evaluated about CM and PCS. Dr. Cohn noted that the September agenda was transitional. Members will consider the impact analysis, integrate it with other information, and move toward next steps and recommendations.
The Subcommittee outlined another three-day hearing October 28-30 focused on: (1) completing the PMRI recommendations and letter to the Secretary, (2) more testimony on ICD-10 and cost-benefit issues as well as comments about what the Committee might recommend regarding ICD-10-CM and PCS, (3) time sensitive CHI issues including the medication report, (4) and other PMRI issues (e.g., claims attachments). Members hope to bring forward the letters to the Secretary related to PMRI and ICD-10 recommendations for discussion with the full Committee in November.
December hearings will include security, CHI, the DSMO yearly update, and possibly CDA and CCR. Dr. Cohn noted that, due to problematic scheduling issues for a joint session on security with the Subcommittee on Privacy, members had decided to schedule a separate hearing on the security final rule, progress made, and issues identified. Members of the Privacy Subcommittee will be invited. Dental and electronic signatures are among the issues for early next year.
Members concurred that the Subcommittee’s potential plan for the next several months was aggressive, but doable. However they considered whether CHI should ease its vigorous schedule and commitment and only meet with NCVHS after the council’s deliberation on each domain unless they encountered a problem and had a specific reason. Reiterating their commitment to being appropriate partners, members said they could accomplish all CHI wanted in the timeframe suggested, but in a different order. If time-sensitive issues had to be handled in September, the Subcommittee will consider adding another day before the September full Committee meeting.
- DAY THREE -
CHI UPDATE - OVERVIEW
CHI, a federal effort to adopt standards for use in the federal system, had asked to discuss with the Subcommittee the inventories and standards available in each of the domains they will look at and CHI’s thought processes and preliminary findings. Noting a need to balance this movement forward with prudent adoption, ensure they had valuable input, and step back amidst controversy and determine what else they ought to consider, Ms. Adair said CHI sought input from members and the private sector and to consult again before making any final recommendations. CHI’s goal is to complete all 24 domains by December 31, 2003.
CHI UPDATE - PRESENTATION ON LAB RESULTS CONTENTS AND UNITS
The Work Group on Laboratory Result Content defined a laboratory result as: the result, results units if applicable, normal ranges/indicator flags, and associated comments. He noted the result is a complex entity with four characteristics: numerical result, ordinal results, results versus cut-off value, and alphanumeric results. Only alphanumeric results need a terminology. The group looked at: MeSH, NCI NCBI taxonomy for organisms, SNOMED CT, LOINC, other UMLS category 0 terminologies, and HL7 Version 2 and 3 Unit Coding.
CHI’s primary recommendation is to use a set of hierarchies of SNOMED CT for alpha-numeric Lab results contents. Another recommendation is to use NCBI Taxonomy codes maintained by NCI that list 100,000 codes for living organisms (SNOMED CT lists 17,000). Noting overlap and a high degree of synonymy, Dr. Steindel said they related fairly clearly. CHI recommended transmitting information on an organism as both a SNOMED code and UMLS Concept Unique Identifier, facilitating switching between terminologies. CHI also recommended HL7 for the Units domain, noting it was heavily-based on the general standards for units in the U.S: ISO/ANSI.
Despite gaps that “always exist,” he said the SNOMED domains were complete for all areas of laboratory results of interest to CHI. Noting that two associated areas (standard comments and abbreviations) lacked standard terminology, he advised NCVHS to urge SDOs to undertake their standardization. Pointing out that it was difficult to interpret an ordinal result without knowledge of the cutoff value comparison or how ordinal scales were derived, he recommended that a method list be available as part of the transmission. Dr. Steindel also asked NCVHS to endorse strengthening the non-human content area.
Discussion
Ms. Adair said CHI would differentiate between priorities and work with SDOs on gaps that were a priority for adoption. Dr. Steindel said CHI sought the Subcommittee’s acknowledgement that they were not aware of anything that filled the identified gaps. Asked how collaboration on resolving gaps with implications for other standards of work and industry components could be conducted at a higher government level, Ms. Adair said conversations with SDOs would be the same, but expectations would be different. CHI would emphasize prioritization within the context of scarce resources and seek a shorter timeframe. She emphasized that CHI received great receptivity from the SDOs. Dr. Steindel noted that representatives from federal agencies that were not specifically involved with human health care were part of this group and their agencies were becoming aware of the CHI process and areas applicable to their domains.
A motion to accept the report and endorse the recommendations with a caveat that the references to LOINC be clarified was duly made and seconded. Observing that the Subcommittee had a new member on the roster (John Paul Houston who participated primarily for security matters) and so actually did not have a quorum, members noted the spirit of the group and moved forward. The vote will be re-visited to re-confirm their approval, based on this day’s presentation, when as a quorum they hold the ICD-10 conference call. Dr. Cohn noted that the full Committee could not commit time to focus on 24 domains, but that their endorsement would be valuable. Ms. Greenberg said the Executive Subcommittee would discuss how reports might be made available without a need for presentation, unless other issues or concerns were identified. Dr. Cohn emphasized that there were no caveats or concerns by anyone present and apologized for parliamentary issues. Ms. Adair said they would accept “in their minds” that there was no barrier to their recommendations, and come back for confirmation. Dr. Steindel will e-mail the minimal revision of the report; after the Subcommittee’s formal vote, it will be included in the background materials for the full Committee meeting.
CLAIMS ATTACHMENT
Mr. Nachimson, lead of the Office of HIPAA Standards’ team at CMS that developed and interpreted HIPAA regulations, noted that the administrative simplification provisions of HIPAA required adoption of a national standard for the claims attachment transaction. The attachment conveyed certain additional information (e.g., lab results, x-rays, emergency room notes) that a health plan might need in order to decide on the adjudication of a claim. Unlike other transactions for which HIPAA required standards, there was no existing standard nor were many attachments exchanged electronically when HIPAA emerged in 1996. In 1998, NCVHS recommended that the Department use HL7 standards to convey clinical information within the attachment transaction.
Since then, NCVHS worked actively with X12 and HL7 to develop a standard electronic attachment transaction and a proposal was developed using HL7 Version 2.4 standards. Six standard attachments were created based on the most common sets of attachment information exchanged today. An NPRM adopting a HIPAA standard transaction for claims attachments was developed. In 2002 it was determined that the industry would benefit from a later version of the HL7 standard. The SDO has presented a proposal using a subset of HL7's CD architecture for contents of attachment data exchanged between trading partners. Either codified, structured data using LOINC values or scanned images can be used for sending attachment data.
For a relatively small investment, plans and providers can implement the scanned variation of the standard that expedites the exchange of information and eliminates some manual processing. Plans and providers ready to invest in the technology can choose to take advantage of the more codified structure. HL7 is balloting the documents to support this proposal, general structure, and six specific attachment booklets: ambulance attachments, emergency department, rehabilitation services, clinical reports, lab results, and medications. Approval is expected in September. The Department will consider the proposal in the context of HIPAA, CHI and the electronic health record initiatives. An NPRM is anticipated early in 2004.
Discussion
Mr. Nachimson explained that X12 standards were used as the wrapper for attachment information. X12 and HL7 collaborated to ensure that this revision of the HL7 standards worked within the wrapper. Both X12 and HL7 have to reach final agreement and ballot the final transaction. The more sophisticated way providers could send information to health plans is based on the LOINC value in a structured XML document that is then coded. A health plan could take advantage of the computer processing of that codified information and eliminate most human interpretation. The other variation allowed a provider to send a scanned image by computer. This less sophisticated way simply provides a copy of a paper record. A strength of the proposal was the ability to implement now and move later to the more sophisticated version.
Participants expressed two concerns: (1) if scanned images were allowed, health plans or providers might send a copy of the entire medical record as opposed to the relevant information, and (2) health plans might say submit on nearly every claim with the standard five different pieces of information as an attachment, losing standardization.
Mr. Nachimson said the claims attachment transaction and its appropriate use were being defined. Hopefully, business rules will be included within the definition and in the explanation of when to use a claims attachment transaction.
Dr. Cohn said that, time permitting, after the standards organizations approved the standard, the Subcommittee would hold a public session to get wider comment before release of the NPRM. The current schedule calls for a February publication.
CMS HIPAA UPDATE
Mr. Nachimson reported that CMS continued to work with all segments of the industry in moving toward implementation of the HIPAA electronic transactions and code set standards. CMS issued compliance guidance at the end of July to provide health plans, providers, and other covered entities an opportunity to develop contingency plans while moving towards compliance. CMS indicated they would consider on a case-by-case basis covered entities' good faith actions to achieve compliance in evaluating complaints for non-compliance after October 16, especially regarding contingency plans and operations immediately after October 16. He said reaction to the guidance was generally favorable.
The Medicare program remains in the forefront. Contractors are able to test claim and remittance advice transactions and move providers into production. Issues remain with the coordination of benefits transaction, but most contractors can successfully test them.
Between July 19 and August 1, slightly over 10 percent of the electronic claims submitted to Medicare were in the HIPAA standard format. Since April/May, percentages rose one percent per month. The agency continues to meet with plans, clearinghouses and vendors to identify obstacles and will meet with provider organizations. Major issues raised by health plans were (1) the difficulty for publicly traded companies of announcing contingency plans which might appear non-compliant, (2) many small providers were concerned that they had not been tested or given a test date, and (3) concern about common missing data elements in the transactions.
Clearinghouses and billing services expressed concern about the burden of re-enrolling. Vendors worried about plans that rejected entire batches of transactions when there were errors in only a small number of transactions. Concerns were also expressed about the readiness of state Medicaid plans. Although CMS had oversight responsibility, each state had authority to operate their Medicaid plan and the states were in a wide range of readiness. Mr. Nachimson said the agency continued to monitor readiness and tried to work with those having major problems.
He said the issue heard on both sides of the industry about information required in the standard that 90 percent of the health plans did not need was indicative of standards development to date. Mr. Nachimson noted a concern raised even prior to HIPAA that X12 was heavily plan-and-vendor influenced. He encouraged providers to work with X12 and others so that their voice is heard in the standards development process.
Mr. Nachimson said the move to standardization raised issues regarding how the regulation was written. He suggested that plans and providers agreeing among themselves about what information should be sent, “as long as it’s within the standard” led to standardization. Noting that in several states health plans are required to pay claims when they have the necessary information, he said interplay with other laws and regulations had to be dealt with. Mr. Nachimson stressed evaluating on an issue-by-issue basis.
Emphasizing lessons to be learned from implementation and the flexibility available on the enforcement side, he encouraged people to follow the guidance in making contingency plans and continue to bring issues to the agency so that they could see what could be handled through regulation and what could be handled through guidance.
Discussion
Mr. Blair complimented Mr. Nachimson on an outstanding guidance letter on enforcement that balanced maintaining the integrity of the compliance deadline and providing appropriate flexibility for entities proceeding with the right intent and best effort.
Mr. Nachimson clarified that the regulation published August 15 is only a requirement for Medicare. Claims submitted to Medicare October 16 and after must be electronic, except situations outlined in the regulation (e.g., small providers, demonstration projects, roster billing for vaccination, and others with two-or-more payers primary to Medicare).
Asked about setting benchmarks that delineated patterns of good faith and evidence of non-compliance as well as guidance that health plans might consider a separate charge for non-compliant transactions that served as an economic incentive, Mr. Nachimson said many issues raised were being considered in developing the enforcement regulation. Rather than set particular guidelines for good faith efforts, he said they needed to look at each situation individually and maintain on a case-by-case basis.
Asked if recent regulations could be interpreted as treating Medicare differentially, Mr. Nachimson explained that CMS’s office charged with responsibility for producing HIPAA regulations, outreach, and enforcement throughout the industry was separate from Medicare and Medicaid and strove not to treat those programs differently. A complaint filed against the Medicare program would be investigated the same as any other health plan, looking at good faith efforts and contingency plans and evaluating on an individual basis.
Noting that a significant number of people had not begun to test and the problem if everyone waits until September 28, he said they were making efforts to “increase the numbers.”
REACTOR PANEL - HIPAA IMPLEMENTATION
Mr. Wilder said AAHP’s members were pleased with guidance that went as far as CMS could within the bounds of the statute and regulations and provided the flexibility needed to continue processing transactions under their legacy systems for a period after October 16. He noted the guidance indicated that CMS and NCVHS were asked to step back “after the smoke cleared” and look closely at what worked and did not work, particularly the process used to get a particular code set or standard approved or modified as well as a process for handling issues on an emergency basis.
Mr. Arges said the guidance was a crucial first step in recognizing the importance of maintaining uninterrupted payment for hospitals and other providers. But he said it failed to create the safety net providers sought to assure that payment continued. He urged CMS to clarify that health plans could process a claim for payment if they found non-material errors that did not affect adjudication. Ideally, the plan should provide feedback about providers’ errors, so corrective measures could be taken.
Mr. Arges noted that a lack of guidance on the handling and acceptance of the transaction standards created a wide variation in how health plans handled batch submission of claims containing an error in one or a few claims. He asked CMS to clarify that health plans should not reject the entire batch.
Pointing out that AHA continued to hear from hospitals about the lateness of testing schedules and their frustration at being unable to test with many health plans, he encouraged CMS to outline more specific performance expectations so covered entities understood the steps to be taken before October 16. AHA believed that plans that did not intend to test with the provider until early October failed to demonstrate a good faith effort.
Mr. Arges emphasized that at this late date the focus should be on developing specific contingency plan components that ensured that hospitals and other caregivers continued to receive comparable payments. At a minimum, CMS should immediately make clear that these components are elements of the contingency plan for the Medicare program, and encourage others to follow suit. He asked NCVHS to urge CMS to respond quickly and decisively to critical issues not addressed in recent guidance
Mr. Emery cautioned that, despite considerable efforts and progress towards meeting requirements of the October 16 transaction rule, a substantial portion of the health care community were likely to encounter problems unless there was additional flexibility on enforcement. He urged CMS to allow covered entities making good faith efforts to continue to rely on legacy claims systems and not be subject to claims denials and/or civil monetary penalty. AMA was hopeful that the exemption process for small physician practices (less than 10 full-time equivalent employees) in the interim final rule would function smoothly. AMA will monitor the process and work with CMS to identify problems or questions.
Mr. Emery reported the preliminary findings of AMA’s recent online survey of over 500 member physicians determining their familiarity with the HIPAA transactions and code set standards, and their preparedness for implementing the standards.
Two months before the HIPAA standard will go into effect, over a third of the physicians report that they do not have a good understanding of the HIPAA transaction requirements. Over two-fifths are unaware of required new patient information. Even more who do not currently collect the required new data elements indicate it would take considerable additional work to collect the information. Most have not received information from Medicare or their private payers about the new data elements they are required to submit. More than one-fourth are unaware that Medicare requires submission of electronic claims by October 16.
Some 41 percent of the physicians had not completed an operational assessment of their practice. Almost a third did not have an implementation plan. More than half had not begun testing for the new standard. More than one-fifth of the physician practices will not be compliant with the HIPAA electronic transaction code set standards by October 16, 2003. Some 65 percent will submit paper claims. Some 58 percent report they do not have enough staff to be compliant, 48 percent need more time for implementation, 47 percent say they need more information. Eight percent state that their Medicare carrier will not be ready to take claims in the format by October 16, 2003. Nine percent said their private payers will not be ready.
Discussion
Mr. Nachimson said the proposed rule would suggest a way to enable the standards organizations to make more minor changes in the standards without going through the entire regulatory process. He said they were discussing this idea with the DSMOs and would see if they also might come up with an emergency change process.
Noting the call for CMS to say rejections had to done by the individual claim, he said there was nothing in the implementation guides, X12 rules or HIPAA law requiring a batch or claim rejection and that, so far, it had been an individual business decision. Medicare gave contractors flexibility for either, but leaned towards ensuring that contractors rejected at the lowest level. CMS hoped health plans made the same business decision. He appreciated the AMA’s offer to help interface with physicians that continued to submit claims on paper and said the comment that physicians did not hear about the requirement to submit to Medicare electronically was valid under the rules. The law was clear that claims had to be submitted electronically, but there had not been a specific date or exceptions until the regulation was published. Currently, Medicare contractors were telling providers what they needed to do to be compliant. Dr. Cohn questioned whether the batch versus single claim issue was a business or legislative/regulatory issue. Participants concurred with the advisory Mr. Nachimson was providing to carriers and financial intermediaries and asked if AAHP needed to advise members to review claims at the lowest level.
Mr. Wilder said the issue of rejecting a batch for a single error pointed out the tension in HIPAA between meeting business needs and developing a uniform set of rules. Noting that, from a payer perspective, there might be business, operational or software reasons, it was suggested that technical experts from the health plans identify situations where they believed the batch was appropriate to reject. Dr. Cohn asked if there was a major business case for health plans to make sure that, as much as possible, claims were dealt with expeditiously. If not, he said the federal government probably had to assist.
Mr. Emery noted anecdotal information suggested that, in contractual relationships with providers of care, some plans required claims to be submitted electronically. Mr. Wilder reiterated that most of today’s transactions were electronic, which were faster, cheaper, and (with real-time connectivity) often more desirable (e.g., in pharmacy claims). Mr. Arges expressed concern about Medicare “closing the door” on paper, suggesting flexibility for at least six months. Mr. Nachimson explained that the formal appeal process could be done on paper, but was a cumbersome process. On the fiscal intermediary side, adjustment of claims could be done on paper or with the intermediary telephoning the provider. Mr. Emery agreed that potentially electronic capability would be more burdensome for small provider and physician practices, but noting that most physician claims came in electronically, he cautioned about signaling that paper was a way to avoid HIPAA.
Dr. Fitzmaurice pointed out that the Subcommittee might be putting people in an uncomfortable position by not having two people from CMS, one to testify on national HIPAA policy and enforcement, another to discuss their problems as a health plan.
Asked what prudent measures would demonstrate that a health plan adopted an appropriate contingency plan, Mr. Nachimson reiterated that the Office of HIPAA Standards stated they would primarily look for good faith efforts to become HIPAA compliant and not rule on the adequacy of contingency plans.
Considering what the health care delivery system could learn from this experience to communicate better to health care providers about what they had do to get ready to make changes in the future, Mr. Emery said physicians looked at HIPAA mostly as a burden, not a long-term benefit that came from standardization. He noted that this was ironic; more than anyone, the physician community had stressed the need for standardization. Asked if professional associations should develop a coordinated communications plan the next time there was a government initiative to encourage standardized electronic communications, Mr. Emery said this administration already did that. Noting frequent open door forums on HIPAA and monthly conference calls with hospital or physicians groups, he said this “new leaf” for the government had begun to pay benefits: people engaged earlier in the process and were more involved.
Mr. Arges stressed a need for coordinated communication that did not “leave the door wide open.” Noting the problem for the hospitals was the inability to conduct testing, he said a mechanism to quickly resolve what was learned from testing would “make the job easier.” But he emphasized that if testing had been given a date certain for completion people could have identified when it had to go live and many of the issues or problems.
Dr. Cohn suggested that they were learning to ensure that there was time between deadlines for small and medium-sized practices and hospitals to gear up for the next project. They heard concerns and unfinished business (e.g., the batch issue) and business reasons why the industry hopefully would “make it all work,” recognizing that on top of everything else, CMS had provided a soft landing memo and option for October 16.
Mr. Arges mentioned required items in the implementation guides (e.g., the tax identification number on the institutional claim) that were not necessary for adjudication but could make it impossible or difficult for providers to comply with those elements. He expressed hope that in the future, if everyone expected to report an item was not capable of reporting it, it would be made situational and usage would define it.
Noting that the NCPDP telecommunications standard for drug claims was a real-time response situation (the retail pharmacist knew immediately whether a covered benefit was in the formulary), Mr. Wilder mentioned AAHP’s ongoing dialogue with CMS about their concern that after October 16 both NCPDP and the 837 would be used in a retail pharmacy setting for supply claims.
Being so close to implementation, Dr. Cohn said they needed to ensure that CMS was aware of the issues and see what could be done without regulation that might “smooth the bumps” with implementation. He noted that CMS could discuss their proposals for fixes or guidance as part of their update on HIPAA scheduled for the September full Committee meeting. Stressing the need to collect these issues, he said Mr. Nachimson’s and his e-mail were always open.
Mr. Arges noted that earlier in August the National Uniform Billing Committee approved recommendations about transition issues and defined steps that serve as guidance on what the date on the claim and the code meant as of that date.
The Subcommittee will hold a conference call after Labor Day to ensure that the ICD-10 cost-benefit analysis is adequate to distribute to the full Committee and to revisit, as a quorum, the motion to accept the CHI report and endorse their recommendation. Members will meet during the September 23-24 full Committee meeting for a HIPAA update and discussion of outstanding issues, hopefully including how CMS addresses them. If any CHI issue must be resolved before then, members will meet September 22. Another three-day agenda is scheduled for October 28-30 and includes discussions on both CHI and ICD-10 and any issues on HIPAA stemming from events on October 16. Hopefully the PMRI report will be ready for final revisions and presentation to the full Committee on November 5-6. The Subcommittee will meet December 9-10. Members will be queried about hearing dates for the first half of 2004.
Dr. Cohn informally asked about any objections to submitting the draft ASCA report for the full Committee’s feedback. Hearing none, he took the prerogative of the chair, agreed it needed to be forwarded for the Committee’s review, and adjourned the meeting.
I hereby certify that, to the best of my knowledge, the foregoing
summary of minutes is accurate and complete.
/s/ June 10, 2004
_____________________________________________________
Chair Date