The Work Group on Computer-based Patient Records of the National Committee on Vital and Health Statistics held hearings on May 17 and 18, 1999, at the AHCPR Conference Center in Rockville, Maryland.
The NCVHS Work Group on the Computer-based Patient Record (CPR) held two days of hearings on May 17-18, 1999, in an ongoing process to study the issues and advise the Department on the adoption of uniform data standards for patient medical record information (PMRI) and the electronic exchange of such information. During the two days, the Work Group heard presentations from and talked with seven panels populated mainly by developers of medical terminologies. Participants had been sent questions to address concerning medical terminology, code development, the government's role, and related issues.
Drs. Cimino and Chute introduced the issues relating to the development of comprehensive and comparable clinical vocabularies. Dr. Cimino described a model of patient data collection and recording whose problems have not been resolved after more than two years of attention. To help break the logjam, Dr. Cimino presented a list of "do's and don'ts" of development. Dr. Chute described positive aspects of the collaborative work on terminologies led by CPRI and other groups, and identified several areas in which tasks still must be accomplished. Drs. Fitzmaurice, Cimino, and Chute discussed starting points for new, useful measures of quality of care.
The panelists described aspects of the business case for standardized terminologies and urged the government to take steps to promote their development. Dr. Campbell described his enterprise's efforts to develop nationally comparable data across the organization. He identified "strategic imperatives" that should shape terminology development, but asserted that systems do not achieve these imperatives because of failure to invest in robust infrastructure in a collaborative environment. Using the engaging illustration of U.S. funding of air mail delivery in the early days of air travel, Mr. Tuttle suggested government funding of EMR development "on the margin," possibly by offering incentives for encoding reimbursement claims with clinical descriptions fitting standard, beneficial criteria. Mr. Mayes pointed out the necessity to identify appropriate types of government activities by individual agencies, then approaching margins specifically.
Ms. Giannini, Dr. Pollock, and Dr. Lapp described details of terminology standardization efforts for alternative practitioners and nurses, emergency medicine practitioners, and dental clinicians, respectively. Ms. Prophet described the features of ICD- 10-PCS, its relationship to earlier versions, and the work remaining to be undertaken; Dr. Berglund spoke about ICD-9-CM and ICD-10-CM in the context of international classification of diseases. Dr. Gordy described the AMA's Current Procedural Terminology and the AMA's efforts in the standard development arena.
Dr. Berglund noted the value in government specification of standard ways to access and express PMRI. The AMA believes that the federal government should limit its focus to broad recommendations for medical terminology frameworks with code sets maintained by the private sector. The federal government could help identify gaps in standards and help the private sector to facilitate and set priorities in standards development. The AMA does not support federally mandated PMRI standards. The ADA endorses designation of terminology sources, rather than standardization of a specific terminology version, and the minimization of duplicate effort. Continued government participation in the standards process will promote comparability.
The panelists discussed characteristics of the current system that result in reluctance by practitioners to code any information that will not result in reimbursement, commenting on the negative impact of this state of affairs on complete PMRI. They asserted the value of data for data's sake.
Dr. Bidgood introduced the Digital Imaging and Communications in Medicine (DICOM) structured reporting model for nonlinguistic evidence, such as X-ray and other images and wave forms. Dr. Spackman described efforts to advance the role of SNOMED as a national reference terminology for patient health information, as well as collaborative efforts with the UK's National Health Service (NHS) to merge SNOMED with UK Read Codes. Messrs. Goltra and Lareau described MEDCIN, a clinical engine for point-of-care documentation in the diagnosis and treatment of a patient, whose intelligent prompting is a unique feature of its user interface terminology. Dr. Huff focused on LOINC and intercomputer communications. Ms. Martin described the features of the Omaha System, initially used by administrators and multidisciplinary clinicians employed in community- focused community settings. Dr. Saba described the Home Health Care Classification System (HHCC), which helps to code care provided during and between home visits; because reimbursement is based on a visit, what transpires during that visit is not coded or reimbursed, and measurement of care outcomes is not possible.
Dr. Bidgood suggested that the government support implementation of the DICOM model by funding R&D for the user interface and development of robust clinical coding schemes. Dr. Spackman urged the government to take a role in testing, improving medical terminologies, fostering collaboration among developers, adopting and requiring standardization of terminologies, and providing incentives to submit data electronically. Mr. Lareau urged the government to encourage vendor cooperation and mapping between terminologies, promote low-cost basic data elements, and encourage vendors to participate in standards bodies. Dr. Huff recommended that the government use HL7 and support medical terminologies with public funds, and urged that the best, most cost-effective terminologies be chosen, including among them LOINC, SNOMED, and Read, possibly MEDCIN, improved NDC codes, ICD-9-CM, ICD-10-CM, and procedure coding systems. Ms. Martin stated that the government should help to coordinate and disseminate terminology information, coordinate efforts, review and update terminology standards and guidelines, and address security issues. Dr. Saba urged the government to narrow the gap in patient care data; she suggested conducting a study that compares traditional HCFA codes for the reimbursement of home visits with Oasis's quality indicators and HHCC terminologies.
The panel discussed how a future shift to a prevention model might impact the classification system and how such a shift might be incorporated. Ms. Martin asserted that the Omaha System has built-in prevention features to address prevention of health-related problems. Work Group members and panelists discussed differing views of the future role of government. Participants discussed the difficulties inherent in coming to consensus on terminology issues; even the concept of framework is not universally accepted.
Mr. Kennelly, representing the new Medical Device Communications Industry Group associated with the IEEE to accelerate the standards development and implementation process, supports the IEEE 1073 standard designed specifically for patient-connected devices such as ventilators; he described the terminology portion of the standard is an object-oriented representation of device parameters and their attributes. The emergence of this industry group is evidence of the market's interest in device data. Dr. Gabrielli described the "first medically intelligent computer system that can analyze and digitize free narrative medical text." Mr. Jordan presented information on automated pharmacy information systems and industry standards.
Mr. Kennelly foresaw no government role in development of the terminology, just in its maintenance and in the encouragement of use of the standards through policy setting. The role Dr. Gabrielli envisioned for the government is to help orchestrate a large-scale implementation of a formal dictionary.
Participants discussed the value of a system to share pharmacy data with clinicians; Dr. Gabrielli cautioned about a gap between the patient's clinical record and the pharmacist's, based on issues of confidentiality, divergent understandings of clinical terminology, and control over communication with patients. Dr. Fitzmaurice observed the various disconnects among segments in medical care and raised the issue of where the patient's record ought to be located. The participants asserted that the more important issue is that wherever the record is located, it must be available wherever and whenever it is needed.
Panelists discussed existing nursing data systems and concurred that currently, the visibility of the nursing contribution to patient care is minimal. Dr. Jones explained the workings of NANDA's nursing diagnosis system and asserted that the PMRI should reflect a multidisciplinary approach to documenting care. Prof. McCloskey described the Nursing Interventions Classification, a system that classifies what nurses do in multiple settings. Ms. Moorhead spoke on the emerging Nursing Outcomes Classification. Dr. Ozbolt described the impact of the lack of standardized nursing terminology, including the inability to research the effectiveness and cost-effectiveness of nursing services. She offered to submit to the Work Group a report of an upcoming conference of nursing vocabulary developers and others; additional research is necessary to reconcile the nursing systems and to evaluate their contribution to patient care.
Dr. Rothwell described the Structured Health Markup Language. The mission of the project is to define a granular, unambiguous representation of terms and phrases in an easily understood architecture; use easy, inexpensive, widely supported tools for authoring, maintenance, and use; and provide the product as a nonproprietary standard.
Dr. Neal described MedDRA, an international, multi-axial medical terminology developed for drug regulatory purposes. Capt. Hess discussed the FDA's National Drug Code. Ms. Coates described the Universal Medical Device Nomenclature System, which is used worldwide to classify medical devices for indexing, storing, and retrieving device-related information. She urged the government to encourage use of standard terminologies, to provide funding for developers and for maintenance, to coordinate international efforts, and to sponsor a conference on standards, data elements, and policy issues related to PMRI. Ms. Meredith urged the government to support the development of a drug reference terminology.
The panelists discussed the different purposes and uses of drug-related information systems, and noted the fact that communication between systems is limited in many cases. Ms. Bickford urged that an occupational health perspective be included in development of standard terminologies.
Dr. Fitzmaurice summarized the major points of many presentations. Dr. Cohn presented brief synopses of the hearings' major themes. Among the themes were the importance of the business case, the state of the market for terminologies, the need for funding of consensus development and maintenance, the role of the government in promoting and developing standards and infrastructure, the role of various existing classification systems, and the impact of not following official coding guidelines. Participants further discussed views of government funding.
Looking forward, participants suggested gaps in the knowledge base that might be filled at future hearings, such as the impact of the wellness movement, environmental hazards, and genetics. Several participants commented on the need to "get the principles right," including flexibility and adaptability. Sufficient discussion has also not taken place on the user interface of technologies.
Participants discussed possible agenda items for the June meeting, including such topics as MedDRA, different viewpoints on terminologies, and the testimony of Claudia Teshiet, John Madison, Betsy Humphreys, and Dr. Greberman. Dr. Cohn suggested reserving time to discuss what has been learned, what must be changed in the work plan, and next steps. An upcoming letter to the Secretary should also be placed on the agenda. The issue of confidentiality and privacy vis-a-vis states was raised. The Work Group set in motion a process of preparing and circulating short working drafts on major issues, including PMRI.
Mr. Blair convened this hearing of the Work Group on Computer-based Patient Records (CPR) of the National Committee on Vital and Health Statistics (NCVHS) pursuant to the mandate of the Health Insurance Portability and Accountability Act of 1996 (HIPAA) for the NCVHS to study the issues related to the adoption of uniform data standards for patient medical record information (PMRI) and the electronic exchange of such information. Following introductions of the participants, Mr. Blair itemized the focus areas defined by the Work Group: message format standards, medical terminologies, quality and accountability of data, requirement to address diversity of state laws and regulations related to electronic health records, some infrastructure issues, and some cost/benefit issues related to the subject areas. He then introduced the members of the first two panels.
Dr. Cimino discussed the basics of a model of patient data collection and recording whose problems have survived for at least the past two years. Inaccuracies in recording data and in translating code have not yet been solved. In some institutions, the same patient data is collected multiple times for different purposes, exposing the data to multiple errors. Dr. Cimino recommended collecting patient data once, putting it into a clinical repository, and, based on the terminology used to code the data, generating financial and research data from that repository.
Dr. Cimino has identified 12 desiderata for controlled terminology, several of which he described. The most important is concept orientation, the notion that each term is backed by a discrete concept or meaning. Each code is linked to only one concept to avoid ambiguity in the coding system. Clear semantics must be used and codes must be linked to meanings that do not change, to enable longitudinal examination of a patient record, and concepts must be permanent (although they can be retired, or "emeritus").
Among Dr. Cimino's recommendations are the following: (1) Don't limit the depth or breadth of a hierarchy in a terminology; (2) don't limit to a strict hierarchy, unless conditions are mutually exclusive. Multiple hierarchies are critical to medicine. An organism is not a disease, although they can have the same name; (3) don't put meanings into codes; (4) eliminate the use of NEC (not elsewhere classified) to pigeonhole terms; instead use modifiers to convey that an “other” rather than “unspecified” diagnosis is present; and (5) strive for content that matches the granularity of the primary creators of data, not content convenient just for financial purposes. The feedback mechanism must include all data users; and (6) develop a timely, concept-oriented update process to reflect changes in clinical medicine. Reasons for changes in concept names must be documented to enable mapping to old (or new) terms.
Dr. Cimino also discussed lessons learned from ICD, which is a strict hierarchy, a number coding system, with an update process which, in his opinion, was changing the codes inappropriately (Dr. Cimino does not favor ICD). SNOMED-RT offers many positive lessons to be learned. LOINC offers two unique lessons: (1) They convened people who created the technologies in the same room to create a terminology that covers everything in the laboratory systems; and (2) they created a knowledge model. As an example, in the nomenclature, the first part is the thing the test measures; the second is the thing that measures it; and the third is a period of time. This process and information permitted the creation of unique identifiers.
Dr. Chute described CPRI's focus over the past few years, stating that "those with more detailed, reliable, and comparable information for cost and outcome studies . . . are going to win in the marketplace." He considers this activity crucially important to the business of health care. He presented a graphic depicting the circular relationship of better patient care yielding more knowledge and increased knowledge yielding better patient care, all held together by a hub of terminology. He also stated the importance of standard ways of representing and measuring health information.
Collaborative work on terminologies has significant history; information generated by CPRI activities, including comprehensive meetings summaries, appears on the CPRI website <www.cpri.org/terminology>. At the first CPRI meeting, tentative agreement was reached on the definition of clinical terminology; it describes standardized terms and their synonyms with a clinical granularity, and terms can map to a higher-level classification system. Dr. Chute pointed out that the desiderata that pertain to a higher-level classification may not be identical to the desiderata that pertain to a detailed nomenclature. Another concept recognized at the first CPRI meeting is the idea of an entry-level terminology that can in turn be an entry point to an underlying reference terminology that can be mapped into an "administrative system du jour," such as ICD-9-CM, CPT, or ICD-10-CM, etc. Another concept is that with an underpinning clinical representation of patient events, one should be able to derive a higher- level classification system that characterizes patient data in the way that the clinical classification system is intended. Rather than referring to awkward tables that provide incongruous mapping, it is possible to go from highly evolved, differentiated classification systems to their clinical root and remap at the level of the underlying clinical representations to find an analogous representation in a classification of a different color. The final point that emerged from the CPRI meeting was that the idea of free and available terminologies is impractical, given no alternative funding mechanism or corporate resources for the purpose of creating and maintaining them.
The April 1999 CPRI conference examined case studies with bilateral coordination between a classification and a coding system for nomenclature. The hope was to generalize the lessons learned. Dr. Chute identified several of the coordinations: NLM facilitation of bilateral mapping between and among CPT, ICD, and SNOMED; SNOMED's use of existing LOINC laboratory codes; letter of intent for Lexical Technology and Ontext to collaborate to develop terminologies; and the merger between the College of American Pathologists' (CAP) SNOMED and the UK Clinical Terms Project (formerly known as Read Codes) to unify the major Western nomenclatures that address clinical patient information.
Other efforts have also borne fruit relatively recently. The ISO Working Group on Terminologies has emerged to address metastandards around terminology development. The HL7 Vocabulary Working Group, which has become the predominant forum for practical consideration of terminology issues, has announced its intent to register terminology systems. NCVHS provided a forum for detailed updates on many major systems, from which four major points emerged: (1) Recognition is increasing of the business relevance of terminology and health care; terminology is recognized as enablers to quality improvement and outcomes, and can enhance clinical efficacy and provide reliable linkages for decision support. (2) A spirit of cooperation and coordination is evident in (a) the case examples and (b) positive consolidation in the field. (3) Developers have embraced the continuum from nomenclatures to classification in a synergistic, cooperative, collaborative way. (4) Public funding has been provided as an experiment to support LOINC in its initial development of an infrastructure for the public good, with an aim toward reducing the end-user costs.
Tasks still necessary to accomplish include: (1) engaging payers and providers, (2) focusing both on where a structure and representation of information, as characterized within a terminology, meet the reference information model, and on how to reconcile distinctions between those views of information, and (3) acceptance of terminology as a crucial infrastructure for "effective and quality health care to be delivered efficiently."
Drs. Chute and Cimino stated that their positions are fundamentally the same, that for many desiderata, "one size fits all," but examination on a point-by-point basis is in order. Dr. Chute noted that in a perfect future, one would recast existing classifications to be premised on an explicit combination of reference terminology elements. He pointed out that the CPRI summaries have glossaries to address confusion in terms. Dr. Cohn noted that the CPR Work Group has helped to plan both the Terminology I and Terminology II conferences.
Dr. Chute stated that additional elements are necessary in developing the continuum from classifications to nomenclatures. An important objective is "sliding that slider" closer to the nomenclature end of the continuum. He acknowledged the endorsement and recognition that detailed, specific, granular representation of medical events, diagnosis, and concepts are as important as (if not more than) the larger categories that appear on hospital bills.
Dr. Cimino stated that the dual approaches of developing an appropriate framework and permitting the pieces to evolve, and of standardizing specific areas and helping develop specific pieces of the puzzle, should both be taken. A high level of organization is necessary to determine where all the pieces should fit, but they can develop together.
Ms. Giannini suggested when to map to the code sets--the life of a claim on the administrative side and the use of the claim on the clinical side. She observed that the life cycle of a claim for administrative purposes between providers and payers is often overlooked. Dr. Cimino cautioned against using claims data for clinical purposes; since the most important goal is to do patient care, rather than to pay for it, it is appropriate instead to collect clinical data and then generate claims data from that. Ms. Giannini and Dr. Cimino agreed that information must be attached to the patient, and that it is important to know all the individuals who will use the information and the purposes for which it is to be used.
Drs. Chute and Cimino agreed that because of continued product variability, it is difficult to predict the future of vocabularies and how they will map to each other. Dr. Cimino stated that the National Library of Medicine (NLM) does not see itself as expert in the area of representation of clinical data, but that NLM has a crucial coordination role in dissemination.
Regarding maintenance of vocabularies, Dr. Cimino stated that most of the public funding of the LOINC effort pays for travel to convene the volunteers who do the work. He recommended paying content developers or imposing license fees. Dr. Chute suggested that terminology development is an ideal target for appropriate government support.
Drs. Fitzmaurice, Cimino, and Chute discussed starting points for new, useful measures of quality of care. Dr. Chute pointed to the desirability of harmonious synthesis of information models about patient record with information structures of terminology, an activity that has not yet taken place. Dr. Cimino observed that the patient record would include the details of care, and an ad hoc study would determine which data items to examine. Dr. Fitzmaurice summarized the viewpoints, stating that it would be wise to create fundamental terminology to be able to derive quality measures. Dr. Chute added the necessity to distinguish what belongs in the terminology and what belongs as other attributes of recorded information about the patient. As an example, he stated that the concept of trimester of pregnancy should not derive from the terminology, but should derive from information in the record.
Dr. Campbell, in his presentation of an enterprise view of terminology, described Kaiser's efforts to standardize terminologies in order to develop nationally comparable data across the entire organization. Business-requirements people are key players in Kaiser's development of clinical information systems; their contributions help to counter significant challenges to budget requests to implement projects of an esoteric nature, particularly when esoteric projects map to business requirements.
Dr. Campbell stated that Kaiser Permanente's new vision statement includes the notion that use of the EMR and other clinical information system tools are "critical enablers on the path to significant improvements in health care outcomes and quality service." Three elements necessary to achieve the vision are: (1) nationally interoperable information systems that integrate applications from multiple vendors and from multiple application areas, (2) interoperable information systems founded on robust terminologies, and (3) robust terminologies more important than a single application implemented across the enterprise.
In the early 1990s, Kaiser made a strategic investment in terminologies that fit within strategic CIS needs. The terminologies were to be EMR-vendor neutral, with the interface portion of the terminology separate from the reference terminology; to have scientific validity (as does the LOINC effort); to be well maintained; to be self-sustaining, even in the absence of the developer; and to have a scalable infrastructure and an efficient, centralized process control to keep the terminologies safe from semantic violations. Few terminology systems currently meet Kaiser's strategic imperatives, Dr. Campbell stated, mainly because of failure to invest in robust infrastructure. The support for infrastructure, he asserted, "must be part of a self- sustaining revenue model for the terminology, and a scalable infrastructure requires significant resources." Developing an infrastructure that requires distributed systems, dedicated database administrators, replication of databases across multiple sites, and the ability to handle large transaction loads is "nontrivial."
In forging collaborations, Campbell asserted, process and scalability are more important than the starting point; content is less important than the willingness to work together. A group with better resources, infrastructure, and scalability, whose terminology system could be sustained in a way to meet Kaiser's enterprise needs, he said, could quickly eclipse a system that would make "today's terminology solution work for five years." Another major challenge is securing sustained organizational commitment for the process.
Dr. Campbell discussed two examples of systems that could benefit from the infrastructure described above with regard to distributed scalable terminology solutions: national drug codes (NDC) and LOINC codes. NDC codes are EMR-vendor neutral, have no intrinsic proprietary advantage, are scientifically valid (for the purpose for which they were originally intended, but subject to improvement) and self-sustaining (they are part of a regulatory mandate), but they are not well maintained; no organization accepts accountability for publishing on a daily basis a quality-reviewed version of the NDC codes as a universal standard to anyone.
The system of NDC codes has no scalable infrastructure or process control, but the system is amenable to improvement at moderate cost and low risk. (Improving an existing system is less risky than developing and validating a new one.) The goal here would be to refine existing organizational accountability (as opposed to creating new organizational accountability) and to add new functionality incrementally as part of the process to improve scientific validity and solve other imperatives. Dr. Campbell's view of an improved NDC process: With a distributed development process with robust infrastructure, one could start with an NDC version with all codes to date. Manufacturers and repackers would submit electronically their version of the code to a central site, which would be filtered through a quality control process. If the manufacturer/repacker made a mistake in submission, a correction would be requested within 24 hours. Upon passage through the quality control system, it would be published as the next version of the NDC codes. If this process were conducted in a scalable way to meet all the needs of all U.S. manufacturers and repackers, this credible, scalable solution would eliminate, for example, the triplication of effort now undertaken by the Food and Drug Administration (FDA), the Department of Veterans Affairs (VA), and the Health Care Financing Administration (HCFA). But infrastructure must be first implemented that does not yet exist.
Dr. Campbell described the attributes of LOINC codes: EMR-vendor neutral, high scientific validity, fairly well maintained (but amenable to increases in scale), but not self- sustaining (government funds would help) and lacking scalable infrastructure and process control. An improved LOINC process would allow distributed development with robust infrastructure. A diagnostics vendor, for example, could propose a new test name, which would go through a quality control process or review for scientific validity. Successful passage would result in publication in the next version.
Dr. Campbell observed that this type of infrastructure depends on facilitating collaborative development. He described the CMT project, of which Kaiser was an original partner with the Mayo Clinic, the CAP, and Stanford University, and whose goals were to develop and evaluate distributed development methodologies that were organizationally scalable and developed on a distributed and scalable computer architecture that could grow with the project. The foundation of the original work was SNOMED. The process used by the CMT project would be appropriate for either LOINC or NDC codes. Dr. Campbell submitted a document that presents a comprehensive view of the project.
Dr. Campbell asserted that multiple developers participating in the process may not be the most efficient way to develop a terminology, but he claimed it to be the most appropriate way to develop a terminology of "general purpose." LOINC's success was due in part to the small size of the domain in terms of concepts.
In summary, Dr. Campbell recommended that the government should do the following: (1) work to facilitate collaborations; (2) collaborate with itself and with industry; (3) share infrastructure (for example, among HCFA, VA, and FDA); and (4) invest in infrastructure. In addition, terminology efforts of other governments should meet uniform standards of data representation, configuration management, and reusable tools and processes.
Mr. Tuttle urged the Work Group members to think about the way that airmail influenced air travel as a way in which terminology might influence the EMR. In the 1920s, although air travel was "primitive, disorganized, and chaotic," the government paid an airline contractor whether 1 letter or 50 letters were to be flown. The government did not pay all airlines for mail service. Contracts were awarded that typically enabled small, local airlines to stay in business. Mr. Tuttle proposed that the government pay for EMR results in the same manner.
Most EMRs that have been built have failed, Mr. Tuttle asserted, but some electronic filing cabinets flourish. Using electronic filing cabinets, physicians can access chronological medical information by using a patient identifier. Electronic filing cabinets are always available, accessible from anywhere, and embraced by users. But data is not comparable from system to system, and there are few opportunities for economies of scale.
Speaking again of the airmail industry in the 1920s, Mr. Tuttle noted that the post office gave contracts to various local mom-and-pop airlines. The government did not pay for the planes, airport, pilot training, or passengers, and competition, standards, and regulation were incremental and reactive. At the same time that the post office paid airlines on the margin, they minimized their risk and did not cancel railroad contracts to carry the mail.
Mr. Tuttle suggested that for EMRs, when claims are now submitted for payment, the government pays X dollars in reimbursement. He proposed that the government pay in addition a very small fraction of X for, for instance, clinical descriptions that would fit the beneficial criteria described in the earlier panel. This represents financing infrastructure development on the margin, whereby the government provides a little incentive for meeting certain criteria. Mr. Tuttle suggested that whereas railroads provided predictable and reliable service as a standard, airlines provided optional, and more risky, service. More risks can be taken with airlines if the railroads are still running. "When the weather was bad, Lindburgh got on the train."
Mr. Tuttle urged paying on the margin for comparable results. The hypothesis is that comparable clinical descriptions will be "worth the public's while to pay for." Mr. Tuttle noted that the government should help in developing terminology in the following ways: (1) Marginal subsidies would level the playing field for everyone who is a resource (e.g., academia, government, private sector entities); (2) since more comparability would benefit the public, marginal dollars should be spread among the terminology providers; and (3) comparable prescription suppliers should be subsidized. To qualify for government subsidies, terminology providers should put up a terminology server that is optionally mirrored for the government to get comparable descriptions (i.e., what it is paying for). The server should: (1) be the authoritative archive of all versions of the terminology since the start of the subsidy, (2) be on the Web for all licensees to keep user costs low; (3) support class-based queries (e.g., this group of diseases or lab tests); (4) supply changes to the terminology according to accepted standards; (5) support a longitudinal query; and (6) have no patient information in it.
Mr. Tuttle's plan to "unstick" the problem and set up a viable system has the following elements: (1) Define the early objectives; (2) budget for the seed dollars (i.e., pre-pays to level the playing field); (3) define competition among players; (4) subsidize winners on the margin by the government; (5) leadership; (6) budget for reliable servers; (7) award subsidies on merit; (8) sustain projects on the margin, adjusted in value; and (9) apply a sunset provision.
Dr. Chute referred to the ISO Working Group III, which is promoting a metaterminology on terminology. Dr. Campbell explained that the challenge is not only to agree to a standard, but also to use the terminology agreed to. Dr. Cohn referred to the framework document, which posits one person's consensus view of these definitions.
Mr. Tuttle discussed the utility of the metathesaurus, claiming that the only way to measure its usefulness is to conduct a project on the scale he proposed in his testimony. The metathesaurus does not yet explicitly represent the meaning of codes, although it is useful in helping people navigate among different terminologies. Mr. Tuttle stated that the real test of the proposed project is to see whether computer programs can do what is necessary to achieve the project goals.
Mr. Mayes concurred with Mr. Tuttle's idea of marginality, but pointed out the political futility of seeking to fund infrastructure for infrastructure's sake. He suggested that more effort must be invested in identifying appropriate types of government activities by specific agencies, and then approaching those margins specifically.
Dr. McDonald endorsed the idea of approaching the original sources and vendors of items related to LOINC and NDC, and to encourage them to volunteer their services in this endeavor in a self-interested manner. He also suggested an increased focus on human artifacts as the dominant source of information. Dr. Campbell noted that he seeks reusable infrastructure that could benefit more than one particular terminology effort.
Mr. Cassidy questioned the appropriateness of requesting government funding for a project to provide interoperability and the ability to recognize the precise meanings of definitions of terms. Dr. Campbell responded that agreeing on high-level standards for the meaning of concepts used in the medical record is of paramount use. He disagreed with Mr. Cassidy that viable efforts are not in evidence, pointing to Kaiser Permanente's extensive work with SNOMED and other efforts, which, Dr. Campbell noted, could be better funded.
Ms. Giannini discussed an alphabetical code set to describe alternative practitioners, including nurses and midwives. The code fits into the HCFA 1500 form and the UB-92 in the procedure code space, and is meant to be a procedure code, as are HCHCs. The code set describes, state by state, which provider types can use specific codes (medical doctors, naturopathic doctors, nurse midwives, registered midwives); this feature makes this code set different from others. Some codes have strong CPT crosswalks; for others that have none, the provider must submit a written report.
Ms. Giannini demonstrated how the system permits attachments, on a code level, of decision support as used by payers when they adjudicate claims, which saves time and money. The system also permits attaching the national training standard on the code level to refer to from a claims adjudication level.
The system supports the provider, claims adjudicator, and medical decision staff with information that permits informed claims decisions on alternative medicine. Providers included in the code set have national training standards, are licensed by at least one state, and can secure malpractice insurance.
An objective is to have universal Internet access that will permit real-time support and decision making to support the coding person and the paying agency, so that the knowledge level rises. Ms. Giannini also presented to the Work Group a written response to the prepared questions.
Dr. Pollock stated that CDC's Injury Prevention and Control Program is working to improve the accuracy, completeness, timeliness, and accessibility of emergency department (ED) data and public health surveillance of injuries; to be able to monitor the incidence, causes, and effects of injuries; and to evaluate the effectiveness of countermeasures. Dr. Pollock noted that variations in data entry into ED record systems impede collection, communication, and reuse of ED data for secondary purposes.
With support from the emergency medical and nursing professional associations, CDC began in 1994 to coordinate a public/private partnership that has developed Data Elements for ED Systems (DEEDS). Select specifications from DEEDS have been incorporated into the LOINC database, the HL7 Implementation Guide for Claims Attachments, and the forthcoming HCFA Notice of Proposed Rulemaking (NPRM) for HIPAA-mandated claims attachment standards. The ED attachment proposed by HCFA is essentially a subset of DEEDS data elements. DEEDS also serves as source material for statewide ED data standardization efforts in North Carolina and Massachusetts, and DEEDS specifications are in use in CDC-funded injury surveillance projects in several states.
Dr. Pollock stated that the vendor community's response to DEEDS has been uniformly positive, and several vendors report DEEDS compliance.
In the DEEDS development effort, an ED record system's main function is to store clinical data and to facilitate its retrieval during direct patient care. DEEDS' scope of coverage focuses on data elements in current clinical use; its data elements are organized in the approximate temporal sequence of data acquisition during an ED encounter. A structured format is used to document each data element, including a concise definition, specification of data type and field length, description of when data element repetition may occur, coding specifications for coded elements, and reference to data standards for guidelines that were used to define the data element and its field values. As much as possible, specifications for DEEDS data elements incorporate national standards for health care data, particularly those applicable to EMR systems. Data types and other relevant specifications conform to HL7 version 2.3, and an appendix maps DEEDS data elements to HL7 fields and segments. Other standards include the Census Bureau's industry and occupation codes, Office of Management and Budget standards for classifying race and ethnicity, the X12 health care provider taxonomy, LOINC codes for laboratory result types, and ICD-9-CM external cause injury and condition codes.
In some instances, when a standard terminology or code set was unavailable for use in DEEDS, a recommended set of terms and codes was developed; DEEDS includes its own code sets for mode of transport to the ED and for patient acuity. In other instances, additional research and development (R&D) are needed to design terminology or coding specifications to select a set of terms and codes from available candidates. Work is also needed on chief complaint medication identifiers, patient outcomes, and several other coded data elements. Chief complaint is particularly important in triage decision making and in decision support, and a key determinant of resource use and service intensity; the aggregate provides a crucial unit of analysis for evaluating episodes of care. Dr. Pollack asserted that with adequate support, a field-ready set of chief complaint terms and codes can be identified or developed in two to four years. Terminology could be selected from existing vocabularies, as long as they allow representation of undifferentiated complaints (health problems for which etiologic attribution is premature), an important issue in emergency medicine practice. Also important is ease of use and reproducibility of a chief complaint system that functions effectively in a high- volume clinical environment, with multiple users working at various levels of clinical experience.
Further work is needed on DEEDS to expand the scope of coverage to all types of data entered in ED records. The public/private partnership used to develop DEEDS and its reliance on existing health data standards set a precedent for future revision and expansion. Dr. Pollack also presented written testimony to the Work Group.
Ms. Prophet discussed ICD-10-PCS, whose goals were to replace the 20-year-old Volume 3 of ICD-9-CM. In 1992 HCFA funded a project to produce a preliminary design for a replacement of Volume 3, and then in 1995 awarded 3M HIS a three-year contract to complete the development of a replacement system with four objectives--completeness, expandability, multi-axial (each code character would have the same meaning within and across sections wherever possible), and standardized terminology with discrete meanings for each term.
During the first phase of testing, 5,000 mainly surgical records were coded using ICD- 10-PCS. During review, testers identified several needed revisions. Following the review, the testers recoded 100 records using both ICD-9-CM and ICD-10-PCS and compared them. Among the issues identified was the lack of a "not otherwise specified" code option, which was deliberately omitted to force specific detail. However, because of concerns expressed by coding personnel about insufficient information, a compromise was reached that established a default character in some areas and also reduced the number of approaches. The dual numbering issue has not yet been completely addressed; the draft ICD-10-PCS shows what the code numbers would be with either embedded or not embedded meaning. Training manuals were revised to address some concerns, and some root operations were redefined.
The test process showed that ICD-10-PCS has more specificity and detail than ICD-9- CM; ICD-10-PCS is easier to expand the system further; its hierarchical structure has a specific, defined meaning; it is easy to learn and understand; and testers received the standardized terminology well. Testing showed that use of ICD-10-PCS should lead to improved accuracy and efficiency of coding. Some nonsurgical sections have not yet been tested significantly.
The electronic version of ICD-10-PCS is available on HCFA's home page. ICD-10- PCS will map to ICD-9-CM, and the training manual is also available. Findings were presented at the May ICD-9-CM Coordination and Maintenance Committee meeting. Some nonsurgical areas needed more training and some brand-name pharmaceuticals used in the medical records were not adequately explained within the system of PCS, which caused difficulty in coding pharmaceuticals. Attendees at the May ICD-9 committee meeting suggested that non-Medicare records must be tested more with the system.
Dr. Berglund spoke about the International Classification of Diseases as it relates to PMRI from the perspectives of both ICD-10-CM and ICD-9-CM. ICD-10-CM is based on ICD-10 from the World Health Organization, and it also draws from ICD-9-CM. ICD-10 was implemented for mortality reporting in the U.S. in January 1999. It has the broadest scope of any ICD revision to date, includes the following changes: Alphanumeric codes, restructuring certain chapters/categories, addition of new features, and expansion of detail.
Dr. Berglund stated that the ICD is not intended to capture all the information in the medical record--and it doesn't--and that it was designed to collect aggregate data for both national and international comparisons. He asserted that for an aggregate classification, such concepts as "not elsewhere classified" are necessary. Although the ICD has a strict hierarchy, the use of "excludes" notes facilitates finding conditions from multiple hierarchies.
Dr. Berglund described characteristics of ICD-9-CM. Its current uses include public health tracking, research, and billing. Its domain is primarily diagnoses and reasons for encounters with the health care system. Its scope includes secondary areas, such as signs and symptoms, as well as some medical test results and nursing diagnoses. Its settings include inpatient, outpatient, long-term, and home-care, plus clinical and epidemiological research studies, and public health reporting.
ICD-10-CM will be a major expansion with increased specificity and revisions in many areas. Major changes are not expected to ICD-9-CM itself, although it has yearly updates in an open process. Detail is added to ICD-9-CM by making new and more specific codes. The aggregate codes then will no longer include those conditions for which the new codes were created. Dr. Berglund stated that he did not consider this a change of meaning, in disagreement with Dr. Cimino, although he conceded that this can be an important issue for data. However, a detailed conversion table enables longitudinal tracking.
Dr. Berglund stated that changes in ICD-10-CM must conform to WHO conventions; ICD-10 categories were evaluated to provide codes for ambulatory and managed-care encounters, decision making, and outcomes research, and certain aggregate codes were expanded. Major modifications to ICD-10-CM include addition of a sixth character for more detail, creation of full code titles, addition of laterality, and creation of certain combination codes.
Dr. Berglund stated that ICD looks at aggregate data at a coarser level of granularity than do many other terminologies and nomenclatures, and that other systems often map to the ICD, such as in the Unified Medical Language System (UMLS) metathesaurus in the NLM. ICD-9-CM is referenced by both HL7 and X12, and is the de facto standard method of representing diagnoses in both HL7 and X12.
Regarding issues deserving government attention, Dr. Berglund pointed out the importance of maintaining ICD-9-CM until ICD-10-CM implementation can begin, but ICD- 10-CM should begin to record diagnoses as soon as is reasonable. He asserted that it is important both to facilitate coordination with the WHO in relation to international terminology use and statistical reporting, and to continue to facilitate standards for electronic interchange of detailed medical information and mappings between systems.
Dr. Berglund also discussed issues related to comparability of PMRI. Privacy concerns may point to systems with less (rather than more) detail, but security needs are not diminished. It would be beneficial to mandate under HIPAA the use of official coding guidelines for systems in use. Government specification of standard ways to access and express PMRI could encourage coordination both among terminology developers to facilitate mapping between systems, and between terminology developers and message standard developers.
Dr. Berglund noted that the ICD alone will not meet all the needs of research or of public health, but to capture diagnoses, ICD-9-CM currently meets most needs.
ICD-10-CM has gone through an open comment period and recommendations have been reviewed. Tabular list revisions are being finalized, and future work will include revising the index, crosswalks, and guidelines; developing training materials; and formal testing in a comparability study with ICD-9-CM. In addition, a database or tagged version that could facilitate mapping must be developed. The National Center for Health Statistics' (NCHS) web page may be accessed at <http://www.cdc.gov/nchswww/about/otheract/icd9hp2.htm>.
Dr. Gordy presented the AMA's view of uniform data standards-based patient records. Dr. Gordy stated that a patient's medical record should include sufficient information for physicians and other health care professionals to assess previous treatment, ensure continuity of care, decide on future and further treatments and clinical activities, and avoid unnecessary therapies or tests. Comparable, accurate PMRI will help realize the clinical and research utility of this information, but comparability enhancements must not detract from the needs of patients and health care professionals.
The AMA's Current Procedural Terminology (CPT) is used primarily as a working clinical nomenclature to describe medical procedures and services. CPT describes medical, surgical, and diagnostic services, and addresses medical services, plus physical therapy, occupational therapy, speech, language, and hearing, optometry, some chiropractic, dietary, psychological services, and other health professional services. It describes clinical practice in a manageable number of codes; pre-coordination of terms meets the needs of health care professionals and payers.
CPT has been adopted extensively by payment and administrative systems, and by a broad spectrum of practitioners. Market acceptance of CPT is evident in its use by many health care professionals to report health care claims. All health care institutions and third-party payers use CPT. Many countries and many U.S. government agencies use CPT; CPT is used as the basis for the resource-based relative value system for Medicare.
Dr. Gordy stated that the AMA is currently developing CPT-5 to meet HIPAA criteria for longer-term clinical procedure code set for administrative and financial transactions, and to be an integral element of the code sets for CPRs. CPT-5 aims to: (1) reduce ambiguity and enhance consistency and accommodate nonphysician needs; (2) ensure that managed-care plans can use CPT and make changes to accommodate practice guidelines and report cards; (3) make additional changes to prepare for computerized records, such as increased granularity; (4) incorporate and enhance functionality; (5) address interest in a more hierarchical approach to CPT and associated specialized queries; and (6) refine the editorial process to allow for open exchange of information through broader participation.
CPT's users include clinicians, payers, managed-care plans, and researchers. CPT-5 will include comprehensive descriptions of clinically recognized and generally accepted services provided by health care professionals to patients and populations. It will be designed to communicate information about clinical services performed and to address the needs of health care professionals, patients, and payers, for administrative, financial, and analytical purposes. Dr. Gordy described in detail the components of CPT-5 and its structure.
Dr. Gordy explained that the AMA, NLM, and the CAP have taken initial steps to map CPT to SNOMED. In addition, CPT contains many concepts similar to those in ICD-9-CM, Volume 3. Former relationships with other terminologies have been modeled in the NLM's Metathesaurus. CPT is used in ANSI ASC X12 and HL7 standards as required data elements.
Dr. Gordy asserted that the issue of the adoption of uniform data standards and the electronic exchange of this information is fundamentally different from HIPAA administrative transactions. No federal legislation requires implementation of PMRI standards or code sets, and it is not clear if the HIPAA model for administrative standards that is intended to address inter-enterprise communication is fully applicable to PMRI standards that involve a large intra- enterprise focus. The AMA believes that the federal government should limit its focus to broad recommendations for medical terminology frameworks with code sets maintained by the private sector. The federal government could help identify gaps in standards and help the private sector to facilitate and set priorities in standards development. The AMA does not believe that federally mandated PMRI standards are likely to be appropriate or productive.
There is no single solution to the vocabulary needs for electronic PMRI. Clinical interface terminologies, reference terminologies, and administrative terminologies, despite some overlap, each play an important role. SNOMED has a major and promising role as a clinical reference terminology for CPR and other purposes. In addition, however, CPT is a clinically based, working terminology that reflects the preferred terms and communication needs of physicians and other health care professionals with carefully designed pre- coordination of terms.
Dr. Gordy concluded that coordination among code set developers and message standard developers follows a bilateral and limited multilateral model driven by the needs of each code set and developer. This model is exemplified by coordination among NLM, AMA, and the CAP, as well as the work of the CAP and LOINC. Both X12 and HL7 have incorporated SNOMED, CDT, and CPT based on user needs.
Dr. Lapp responded to the questions prepared by the Work Group. The American Dental Association (ADA), which favors a health record, believes that for optimal patient benefit with assurance of confidentiality, appropriate health information should be available at the time and place of care to practitioners authorized by the patient, through the development of the CPR.
The ADA has identified fundamental criteria essential for information to contribute to quality outcomes and efficient, economical delivery: quality, utility, proximity, accessibility, and confidentiality. Availability of comparable health information at the point of decision enables the patient, care giver, and manager to use automation efficiently to enhance their best judgments; lack of comparability reduces quality of care and increases cost.
The ADA seeks to achieve uniformity, consistency, specificity, and accurate reporting of dental treatment; the ADA-developed dental code is published in the ADA's current dental terminology. The dental code is the accepted standard for dental procedures. The dental code, used by all practicing U.S. dentists in all settings, has been proposed as the HIPAA standard dental code. The third edition of the code will be released this summer. The ADA's Systematized Nomenclature of Dentistry (SNODENT) will offer a comprehensive terminology to establish and define dental oral disease classifications and co-morbidities. SNODENT will be an integral part of the CPR, and dental practice management system vendors are expected to incorporate SNODENT in their systems. Dr. Lapp noted that SNODENT is a microglossary of SNOMED for which the ADA has a licensing agreement with CAP.
The ADA is a designated code source; the dental code is used in the X12 message format. Although both the ADA's dental procedure codes and SNODENT could be used with HL7 and LOINC, the formats are not currently used in dentistry. ADA advocates acceptance of both SNODENT and the dental procedure codes as the required terminology for dentistry. ADA is also working internationally to promote adoption of standards for comparable health record information.
In terms of the government's role, ADA endorses designation of terminology sources rather than standardization of a specific terminology version, and the minimization of duplicate effort. Continued government participation in the standards process will promote comparability. The government should support ISO TC 215 Work Group I, which is working on an international model of the health record.
Dr. Berglund related the nonuse of official coding guidelines by payers to reluctance to reimburse for certain sets of codes. He considered the central issue to be more a money issue than one of data quality, but when unique coding guidelines are used, quality is affected and comparability is lost. Ms. Prophet concurred, stating that some payers refuse to reimburse unless a claim is described in an inappropriate way. Dr. Gordy noted that contractual arrangements play a role, such as Medicare's general policy to pay for one service per day.
Dr. Huff noted that the codes described in the presentations on CPT and ICD-9 as being used and approved for use in HL7 were for billing diagnosis and billing procedure codes, and in both cases, those nomenclatures have been found insufficient to describe clinical detail. Dr. Fitzmaurice commended the AMA on incorporating more uses than just reimbursement, and particularly the potential for incorporating quality measures, in its upcoming CPT-5 version. Dr. Gordy stated that the specific needs of practicing physicians in ambulatory care settings will be addressed in CPT-5 in the future. Dr. Gordy clarified that ICD-9 is used to report in-facility procedures, and CPT is used to report outpatient care. Ms. Prophet stated that ICD-10-PCS is a complete procedural system that covers all types of procedures, except evaluation and management, found in CPT. In CPT-5, a work group has been charged with developing a code set for those nonphysician providers who currently cannot use CPT codes. Ms. Giannini stated that her group has been working on codes for billable procedures in the Nursing Intervention Classification (NIC) system. Ms. Prophet noted that nursing services are not completely covered from the PCS point of view. Dr. Berglund stated his uncertainty about the usefulness of ICD-9-CM nursing codes in billing situations.
Dr. Ferrans applauded AMA's focus on the CPR aspect of the human/computer interaction on the physician data input step and working with software vendors on the design of interfaces. Dr. Gordy reiterated the difficulty that the AMA observes in meeting inter- enterprise needs, stating that the organization would cooperate in studying the issue.
Dr. Saba and Dr. Cohn discussed the different purposes of health care terminologies and the difficulty in justifying entering nonreimbursable codes in a health record. Ms. Giannini noted that terminologies have not been measured for the purpose of payment, and that having a relative value unit associated with a procedure is how the industry wishes to do business. Dr. Berglund stated that data for data's sake is important in the public health environment. Dr. Gordy noted that in an intra-enterprise setting, certain HEDIS requirements might be fulfilled with code sets, but a surgical procedure in an ophthalmology practice, for example, may never capture all the procedures used.
Dr. McDonald raised the issue of the difficulty in coding for chief complaint, such as depression, of which there are many variations. Dr. Pollock noted the wish in emergency medicine and the nursing community to develop a standard way to express a chief complaint and to be able to teach it. Dr. Berglund identified a tension between a clinician identifying depression as a chief complaint and a psychiatrist wanting additional detail. Ms. Prophet stated that the not otherwise specified (NOS) option was added to PCS to address the need for a broader description.
Dr. Bidgood represented the Digital Imaging and Communications in Medicine (DICOM) Standards Committee. Stating that his written testimony answers the Work Group's prepared questions, he discussed the importance of the standard open-structured representation of nonlinguistic evidence, such as images and wave forms, in patient medical records--a larger factor in comparability than the coded description per se. The challenge is for codes to maintain sufficient breadth and depth in their representations.
Dr. Bidgood noted that nonlinguistic evidence is relatively expensive to produce and document, and also relatively inaccessible. Its value is proportional to the quality of its diagnostic descriptions. The DICOM structured reporting (SR) model offers a standard interchange medium for attaching pretext, numeric measurements (as with LOINC coding), codes as with SNOMED, LOINC, or other standard coding systems to features, and permits an audit trail back to the root. The goal is open representation of image content in which image features are stored as objects, whereby specific evidence is documented that leads to conclusions.
The ability to share nonlinguistic data in commercially available equipment is new. Dr. Bidgood stated that with the application of robust, granular, context-specific terminology from SNOMED or LOINC, for example, to the structured nonlinguistic evidence model achieves closure of the loop from practice guidelines to clinical data sets to performance measures upon explicitly marked data, and back to feedback to the practice guidelines.
Dr. Bidgood recommended that the government offer an incentive or promote implementation of the DICOM model. R&D is needed for the user interface, and endorsement of robust clinical coding schemes would also be helpful. Dr. Bidgood supplied the Work Group with references and a copy of the SR documentation model, the model behind DICOM SR.
Dr. Spackman represented the CAP in his role as chair of the SNOMED Editorial Board. He submitted answers to prepared questions in written testimony, highlighting for the Work Group the usefulness of secondary uses of detailed clinical information, the comprehensiveness and compositional nature of SNOMED, and its domestic and international acceptance for a variety of uses. SNOMED's licensees also include a broad spectrum of categories, including information system developers, payer-provider systems, insurance companies, research entities, and government agencies.
Dr. Spackman described four CAP initiatives to advance the role of SNOMED as a national reference terminology for patient health information, including the expansion of its functionality, merging SNOMED codes with the United Kingdom's (UK) Read Codes, increasing the breadth of stakeholder participation, and initiating the process for SNOMED to become recognized as an ANSI-accredited standard.
SNOMED RT, currently undergoing broad beta testing, combines the granularity and comprehensiveness of SNOMED with improved clarity of meaning and analytical power, vendor neutrality, multiplicity of hierarchies, and explicit semantic definitions.
The collaboration between CAP and the UK's NHS takes a new approach to computerizing scientific terms that physicians, nurses, and allied health professionals use to manage patient records and medical communication. The strength of SNOMED RT in specialty medicine will combine with the richness of the Read Codes Version 3 in primary care. The merger will decrease duplication of effort of SNOMED and NHS, create the most comprehensive language of health, and provide an essential building block for a common computerized medical language for use worldwide.
SNOMED's Editorial Board recently added representatives from NHS, VA's Health Administration, and the Department of Health and Human Services. The SNOMED Authority now includes representatives with experience in enterprise-wide implementation of clinical systems, needs of the global market, and international standards-setting activities. SNOMED International will maintain its not-for-profit status, but will eventually become self-sufficient; the CAP is committed to widespread availability and access to SNOMED with minimal cost or restrictions. The CAP is currently submitting SNOMED for ANSI accreditation.
SNOMED is currently mapped to ICD-9-CM codes; the CAP intends to enhance that mapping, in together with NLM and the NCHS, to include rules and code-to-code linkages. A mapping of SNOMED terms to CPT-4 also has begun with the AMA and NLM.
Dr. Spackman's recommendations for government action include a role in testing and improving medical terminologies, and fostering collaboration among developers, adopting and requiring standardization for clinical decision making and support systems in federally funded health care initiatives, providing financial incentives to providers for submitting government- requested public health data in electronic form, and facilitating comparable PMRI by adopting standardized clinical reference terminologies for the CPR.
Dr. Spackman noted that SNOMED conforms with Dr. Campbell's five criteria for a useful clinical technology. Dr. Spackman asserted that SNOMED is the most complete reference terminology currently in use that captures all aspects of the patient medical record in computer-readable format.
Mr. Goltra asserted that different types of terminologies are appropriate for different uses, for example, clinical and billing. Within billing, a code that can be split to multiple separate concepts can be a problem in clinical practice, but for billing purposes, it can be an asset in grouping together like kinds of services for reimbursement.
Mr. Lareau answered some of the Work Group's prepared questions and then described MEDCIN. Medicomp's view of PMRI is as useful primarily to diagnose and manage a patient; other uses include reimbursement, research, and patient care. The need for comparability between data elements varies by the use of those elements. MEDCIN is the basic clinical engine for point-of-care documentation in the diagnosis and treatment of a patient. Whereas some want to split information down into more atomic levels for research, others might want to group information for billing. Links must be forged to other terminologies. Eight vendors at ambulatory sites with about 165,000 users are now MEDCIN users.
Slowness and confusion inherent in asking doctors to capture information in historical files and then map the information mitigates against this process. In many cases, a template approach may be inappropriate in that it does not recognize providers' reductive decision- making process. Medicomp uses intelligent prompting--a form of dynamic templating that adjusts to the clinical situation during the encounter and produces findings for documentation, diagnosis, and therapy. MEDCIN provides a relevant list of items for documentation regardless of whether the patient is approached diagnostically or symptomatically.
MEDCIN is used primarily as a user interface terminology to enable vendors to provide clinically relevant information with a coded structure beneath it. Medicomp is a member of HL7 and ASTM. Mr. Lareau recommended that the government should encourage vendor cooperation and mapping between terminologies, promote low cost of basic data elements, and encourage all vendors in any government-sponsored project to participate in standards bodies.
Dr. Huff serves as co-chair of LOINC with Dr. McDonald, is chair-elect and co- chair of the Vocabulary Committee of HL7, and advisor to the SNOMED Editorial Board. He focused his presentation on LOINC--which contains 20,000 codes, is available and free for download from the Web--and the communication of information between computer systems. As background, Dr. Huff noted that in intra-enterprise activities, the goal is to reduce redundant data entry and provide better clinical care; inter-enterprise activities can include public health reporting, outcome studies for clinical trials, and patient administration. Interface standards are relevant to decreasing the cost and increase the ease of communicating. The object is to communicate information from one independent system to any another; in that context, HL7 defines message structures.
Dr. Huff asserted that the record structure is as important as the terminology; the combination of those two things leads to semantic understanding of what is transmitted. He acknowledged that people are not using the data in HL7 consistently throughout the U.S. and the world, a condition that increases the time necessary to map between two implementations of an HL7 interface. The key is to use a consistent vocabulary when communicating externally; Dr. Huff used the universal language Esperanto as an example.
LOINC's goal is to make a universal coding scheme for observations within messages. Multiple coding schemes can be used, but the strategy within HL7 would require knowledge of the synonymy between/among systems.
Dr. Huff then explained an "inexplicable association between vocabularies and the structure." An important question is: How does the record structure correlate with the use of the vocabulary, and how do I use that combined structured terminology to communicate semantically from one system to the other so the other systems understand exactly what I mean?
LOINC is proposed for use in claims attachments as part of the HIPAA initiative, and it is part of the DEEDS database, among other applications. Dr. Huff recommended that the government use HL7 and support free-for-use medical terminologies with public money, possibly in a public/private partnership. He acknowledged the role for the NLM and the UMLS metathesaurus. He urged that the best, most cost-effective terminologies be chosen, including LOINC, SNOMED, and Read, and possibly MEDCIN, and NDC codes with improvements in clinical drugs that have already been worked on with the UMLS Metathesaurus. There is also a role for ICD-9-CM, ICD-10-CM, and procedure coding systems.
Ms. Martin, representing the Omaha System and speaking from the perspective of multidisciplinary administrators and clinicians employed in community-focused settings, submitted to the Work Group her formal answers to the prepared questions. Nurses, physical therapists, registered dietitians, occupational therapists, social workers, speech-language pathologists, and home health aides, Ms. Martin asserted, need a comparable record to share details about their clients' needs, the services they provide, and the outcomes of those services; the administrators need a comparable record to extract essential clinical, staffing, program, and cost data. She noted that a structured record that uses health care terminologies is essential for an effective, automated, integrated, community-focused information system. Complete, accurate record information is a requisite for successful, continuous quality improvement programs, reimbursement, accreditation visits, and audits.
The Omaha System provides a simple, comprehensive, research-based structure to describe and quantify the concerns of clients and the practice of nurses and other health care professionals. Omaha System use has expanded dramatically to include nurses, other clinicians, managed care center staff, hospital-based and managed care case managers, nursing educators and students, acute care setting staff, nurse practitioners, and the international community. The American Nursing Association (ANA) has recognized the Omaha System, which is incorporated into the International Council of Nurses' International Classification of Nursing Practice and included in the JCAHO and NLN accreditation standards.
Ms. Martin described the taxonomic framework of the Omaha system’s three parts: problem classification scheme (comprised of nursing diagnoses), intervention scheme (comprised of activities provided by clinicians), and problem rating scale for outcomes (an evaluation tool for measuring client progress from admission through dismissal). The Omaha system is based on actual client data collected and tested in the practice setting. Software developers who include the Omaha System in their products control the selection of platforms and message format standards. Ms. Martin asserted that government agencies should help to coordinate and disseminate terminology information, coordinate efforts between the clinical work of the professional organizations and the reimbursement issues of third-party payers, review and update terminology standards and guidelines, and address security issues. She stated that a government-sponsored research project is underway that will help improve identification of problems, selection of interventions, and measurement of outcomes.
Dr. Saba submitted written testimony to the Work Group that answered the prepared questions and in her remarks focused on the Home Health Care Classification System (HHCC). She asserted that progress notes in the paper record reflect the care in events that are coded, but the "between part" is not coded; the CPR lacks patient care data that reflects quality and measures outcomes. If HIPAA is to reflect the patient record, care data must be included in the CRP.
The HHCC relates to care provided during and between home visits. Home health care traditionally uses a medical model, although physicians are not providing care. Reimbursement is based on a visit, but what transpires during that visit is not coded or reimbursed; measurement of care outcomes is not possible. Dr. Saba urged the government to examine how to narrow the gap in patient care data.
HHCC's design is similar to ICD-9 and coded like ICD-10. Twenty home health care components represent the framework between the vocabularies of HHCC and ICD-9. HHCC can be linked to medical diagnosis, the entry to the health care system for reimbursement; Medicare funding; and third-party payers. The unique HHCC vocabulary includes psychological, physiological, functional, and behavioral care components. The hierarchical HHCC is enhanced by the use of three modifiers, and the interventions are expanded by using four types of action. Using wound care as an example, Dr. Saba stated that one can assess wound care, teach about it, manage it, and refer the care to another care giver; each type of care would cost differently and would have different implications for the quality and outcome of the care.
The advantages of HHCC include its ability to track care over time and to relate to many other existing terminologies (framed like ICD-10, uses HCFA 486, complementary to Oasis). HHCC is approved by the ANA, is in the UMLS, is coded in the Cumulative Index of Nursing and Allied Health Literature, and is incorporated into the International Classification of Clinical Nursing Practice under development.
Dr. Saba is currently testing the HHCC in the U.S. military forces, but funds are also available for testing in the civilian environment. Dr. Saba recommended a government study that compares the traditional HCFA codes for the reimbursement of home visits, Oasis with the quality indicators, and the HHCC terminologies.
Dr. Saba clarified that the six ANA-recognized vocabularies are Omaha, NANDA, NIC, HHCC, Patient Care Data Set, and Nursing Outcomes Classifications (NOC). The HHCC is on Georgetown University's web page, downloadable free of charge. Dr. Saba and Ms. Martin noted that Alternate Link is using portions of the HHCC. The HHCC was designed purely for the home health environment. Omaha System addresses the entire spectrum of care.
Dr. McDonald stated that the term vendor neutral is not a "sufficiently worthy goal," and he discussed the relative value of having a soft monopoly and of patenting standards. Dr. Spackman described general principles used in HL7 development to promote economic fairness. Mr. Goltra stated that the cost of MEDCIN's nomenclature is $25 for an unlimited number of users per site; it is available for HL7 messaging at no cost.
Dr. Spackman explained that because the United Kingdom has provided intellectual property rights to the CAP in the mutual work that will be created, there are no problems with recoding.
Ms. Bickford stated that the ANA now recognizes seven data sets; the new one is the Peri-operative Nursing Data Set, and she offered to provide contact information to the Work Group.
Dr. Saba discussed a model developed by Mitretech Corporation to determine whether a home health care professional should make a home visit and be reimbursed, or whether the care can be provided through two-way teleconferencing for reimbursement or for providing the equipment. The model requires testing, especially with a decision-support model. Ms. Martin added that in addition to telehealth, community-focused services are not site specific.
The panelists discussed how a future shift to a prevention model might impact the classification system and how such a shift might be incorporated. Dr. Spackman pointed out that SNOMED and Read do not exclude prevention procedures, but acknowledged that the terminologies will evolve over time; proactive maintenance of vocabularies is extremely important. Mr. Goltra stated that a patient's history will also be affected, for example, by their genetic proclivities. He foresees massive reclassification based on changing definitions of diagnosis. Ms. Martin stated that the Omaha System has built-in features to address prevention. Dr. Saba observed that a lifelong CPR has been developed by the military to collect the history of U.S. servicemen. Dr. Bidgood noted that the marginal benefit of new genetic information imaging as a screening for serious disease will be evident; DICOM's structured reporting is currently working on initiatives involving screening for breast carcinoma, diabetic retinopathy, and malignant melanoma.
Dr. Bidgood stated that DICOM's advantage is that each work station that supports a DICOM image also supports the semantics of the coordinate system of the image, which enables representation of any region of a DICOM image concisely and unambiguously. DICOM can link information along any point in the spectrum of expression. The system works for radiological images and wave forms. Dr. Ferrans observed that the ability to leverage X- ray information may prompt significant discoveries in patterns of disease.
Dr. Spackman asserted that in the joint CAP-NHS project, a well-defined structure and process and distribution of effort might lead to reduced maintenance costs in the future.
Mr. Blair stated that as long as the private marketplace is making progress, the government may stay out of the funding business; the prepared questions regarding a government role referred to dysfunctions in the marketplace. Dr. Cohn stated that no decisions have yet been made about government funding.
Dr. Saba noted the desirability of developing a uniform, freestanding framework that links all vocabularies. The government could mandate everyone to code their data sets in a way that maps to the framework. Dr. Spackman, in stating that even the concept of framework is not universally accepted, noted that the government should not impose standards or regulations that would stifle development and collaboration. He observed that the picture of the PMRI incorporates more than reference terminology, and that all other pieces of the puzzle are not yet in place. Mr. Goltra echoed Dr. Spackman's view, pointing to new awareness in the marketplace of the need for changes in frameworks. Dr. Bidgood noted the difficulty inherent in trying to push the diverse existing coding standards into a common framework.
Dr. Spackman described his view of the appropriate role of government: to encourage a process whereby vendors of clinical information systems become more eager to implement detailed clinical terminologies in their systems--that the marketplace be fostered and expanded--with less uncertainty about some decisions regarding clinical technologies. The government should avoid mandating rigid requirements, but should lead by example, specifying the approaches it will take in data collection.
Dr. Bidgood suggested that government interactions may provide lessons learned, particularly about the cost-effectiveness of a particular approach. He stated his wish to keep the payment issue independent (but linked); mandates for what needs to be transmitted should have a different set of requirements from the rest of the PMRI. The parts of HIPAA mandated in the law would be standardized. Mr. Goltra noted the power of government in that HCFA produced a virtual nomenclature overnight vis-a-vis the EMN codes.
Participants discussed inducing the field to use code and classification systems without a reimbursement carrot. Dr. Spackman suggested that pathologists, for example, should not code, but rather use structured data entry or auto encoders to get the meaning of their pathology reports into a form that can be used for all secondary uses of data. These secondary uses must be made easy for the primary care physician, etc., to get data into that form. Mr. Goltra noted that Medicomp's intelligent prompting may help to minimize resistance. Mr. Lareau described Medicomp's effort to offer incentives to the provider to use the codes. Ms. Martin noted that the power of client data, especially data generated by automated information systems, motivates staff to use standardized vocabularies. Client data are valuable for clinicians’ practice and for management and reimbursement purposes. This new attitude of valuing data is also evident in the academic community. Dr. Saba pointed out that this process will be delayed until health care professionals are taught to use the technology in the care of patients.
The hearing was recessed at 5:40 p.m. until the following day, May 18, 1999, at 8:45 a.m.
Mr. Blair reconvened the hearing at 8:51 a.m. on May 18, 1999; presented an overview of the previous day's proceedings; and initiated introductions of the participants. The first panel then made its presentation.
Mr. Kennelly represented a new group of the Institute of Electrical and Electronics Engineers (IEEE) related to industry standards and technology. Five companies already have joined the group, whose goal is to accelerate the standards development and implementation processes. The group supports the IEEE 1073 standard, designed specifically for patient- connected devices such as infusion pumps and ventilators. Mr. Kennelly submitted a written response to the prepared questions.
Two types of data are relevant to PMRI for medical devices, historical summary data and comprehensive data on current condition. Mr. Kennelly asserted that the many sources of PMRI are vitally important, but their capture methods are different for various sources. One way to incorporate nursing or other clinical notes into the technology might be by a dictation device.
Mr. Kennelly stated that the terminology portion of IEEE 1073 is called the Medical Device Data Language (MDDL), an object-oriented (OO) representation of device parameters and their attributes in a high-level, OO method developed in the U.S. and Europe. The terms include physiological concepts and device terminology. Similar to DICOM, the IEEE 1073 standard is a seven-layer communications standard that specifies all connectors, cables, data rates, network, and transport up to the data language for representations and formats of data.
The market is now turning to the need for device data, and people want to do real-time inventories and pay for use. Many applications are now being introduced into the market that require robust, reliable data communications with medical devices. A key sign of this market acceptance is the formation of the Medical Device Communications Industry Group, which requires an annual fee of $25,000 to advance the IEEE 1073 standard, and the fact that ten additional companies are contemplating joining. Another sign of acceptance is the adoption of the nomenclature in Europe of the CEN TC251 prestandard, which is 95 percent compatible with the IEEE 1073 standard.
Development of the terminology will expand to cover new devices and new parameters for current devices. The relation to other terminologies is through bilateral mapping to permit new device uses; the entire terminology can be used by any other application through the ISO object identifier. Mr. Kennelly sees no government role in development, but rather in maintenance. The government should also encourage the use of standards through policy setting, particularly through purchasing rules and FDA rules.
Dr. Gabrielli stated that his company has completed and made operational the first medically intelligent computer system that can analyze and digitize free narrative medical text. He described the major problems solved to reach the goal; he also submitted written responses to the Work Group's prepared questions.
Medical text analysis required the input of a multidisciplinary team to accomplish such tasks as grammatical analysis and semantics, and development of computer software. Over ten years, the researchers have developed a hierarchical lexicon of almost one million entries, with each term linked to its supra- and sub-terms and to its synonyms. Development has also required finding structures in the text that carry clinical information, called "clinical facts," characterizing them, and establishing rules about them. The narrative generally is composed of clinical facts (25-30 percent) plus other relevant information. Other issues to be resolved were ambiguity in the English language and grammatical ambiguity.
The lexicon currently contains 400,000 medical terms and is projected to have around 450,000; Dr. Gabrielli contrasted this with the Read Codes, which he noted were barely at halfway for text processing. He stated that a single umbrella lexicon ultimately will be required that will cover SNOMED, Read Codes, and psychiatric codes, and that will cross-reference ICD, CPT, LOINC, FDA's COSTAR, and all other terminologies in use.
Dr. Gabrielli asserted that if the patient record is fully coded, one can perform automated reimbursement coding, which would save about $45 billion annually. One issue to be resolved is proprietary term listings; Dr. Gabrielli suggested that the government might help orchestrate a large-scale implementation of a formal dictionary. This situation might eliminate the need for the professional organizations, because an operator could apply algorithms for monitoring all medical cases with some clinical condition. Under these conditions, maintaining the lexicon would be an issue. Development would be necessary to create computer-oriented factual guidelines designed to be processed by computers. An anonymous clinical database for millions of clinical case histories would be necessary. Dr. Gabrielli also called for the rapid development of speech recognition.
Mr. Jordan focused on the questions provided by the Work Group and noted he will submit further written testimony. Mr. Jordan stated that PMRI should describe past and current experiences, status, and results of an individual patient's medical therapy encounters. All relevant information should be included from each of the patient's providers. PMRI should include information on prescription and over-the-counter medications, including interactions, compliance, and achievement of expected outcomes. Mr. Jordan asserted that when pharmacists communicate with health care providers, the outcomes of therapy are often improved.
Mr. Jordan explained that the National Council for Prescription Drug Program Standards (NCPDP) provides clinical support tools within the standard using the Professional Pharmacy Services (PPS) codes, which support the conduct of online real-time drug utilization review (ORDUR). These standards support coordination among pharmacists and could be used similarly with other health care providers. PPS codes facilitate documentation of electronic transmission of information related to professional pharmacy services and to support ORDUR activities administered by third-party claims processors and within pharmacists' own practice management systems; the objective of the code set is to improve quality and continuity of care, standardization of electronic documentation, and billing infrastructure. The codes help identify and prevent inappropriate prescription drug therapy at lowest possible cost. The PPS codes are implemented by virtually every vendor active in the prescription drug benefit industry and pharmacy computer vendors. More than 95 percent of all U.S. community pharmacies in the U.S., a number of foreign countries, and 47 states use NCPDP's telecommunication standards.
Mr. Jordan noted that some concern has been expressed about the codes providing too much information, particularly regarding ORDUR. This problem is under examination by the American Pharmaceutical Association (APhA) and other organizations, with the goal to improve the terminology and its use.
APhA and seven other pharmacy practice organizations have developed a Pharmacy Practice Activity Classification (PPAC) system, currently being incorporated into NCPDP standards and improved through consensus. PPAC has been accepted by NLM as a valid vocabulary and classification system. Mr. Jordan asserted that as the need for drug therapy management services expands, so will the need for PPS and PPAC codes. He noted that coordination of standards development organizations and data dictionaries with assigned secretariats for health sector domains will help ensure success in coordinated, useful PMRI.
Mr. Jordan stated that the pharmacy code sets currently are unique to pharmacy, developed to accommodate ONDUR but enhanced to optimize health care through the screening system. The professional service intervention codes record the pharmacist's intervention. APhA believes that other medical professionals who look at EMR may find this data valuable. NCPDP has a standard, called Script, that physicians use to prescribe and send an electronic prescription to a pharmacy. A system is currently running that includes the payer industry, pharmacy chains, and physician practice management systems with electronic communications. Physicians are now seeing the same interaction warnings that pharmacists see.
The information sent to pharmacy benefit managers includes NDC codes and sometimes GPI, which therapeutically classifies drugs. Adverse reaction codes are part of the system, but Mr. Jordan stated that if they were not mapped to other codes in other classification systems, such mapping would be appropriate.
Mr. Jordan stated that the pharmacy industry has been trying to move forward the standards-setting coordination process, taking an active part in HL7, X12, and other organizations. He noted that NCPDP is the expert body on standards for the pharmacy business, and that the IEEE standard is an interesting model.
Mr. Mayes commented that the pharmacy industry has done well in automating its business processes and introducing automation into patient management practices. He encouraged the pharmacy world to participate fully in the standards development process to leverage this type of PMRI to meet business needs in health care. Mr. Jordan noted that the PPAC system has been recognized by the NLM, and NCPDP has accepted it as a data element within its standards.
Dr. Gabrielli identified a gap between the patient's clinical record and the pharmacist based on issues of confidentiality, pharmacist's understanding of clinical terminology, and control over pharmacist/patient communication. He commented that other operating systems, such as CORPUS, may be able to do similar things. The solution involves more exposure of the developers and business people to these issues. He cited the case of an HMO with one million annual visits that had been computerized; his company found that 48 percent of hypertensives were untreated; the quality control chief did not return Dr. Gabrielli's telephone calls because of embarrassment.
Dr. Fitzmaurice observed that the pharmacy has information the doctor does not, the computerization company has information the health plan doesn't have, and there is information that nobody has, but a device could pick it up, and he questioned where should be the "repository for me." Mr. Kennelly said it is important only for the records to be accessible, whether by Web or card; it needs to be wherever you need it. Dr. Gabrielli added that there will be at least two large partitions of clinical information: the identified clinical record, which must be protected and controlled (the patient controls his own record); the other is the unambiguous electronic database. Mr. Jordan concurred that the record needs to be where it is needed; for the foreseeable future, the record will be disaggregated information in various medical settings. He expressed hope that a master patient index could be developed and a portal established to exchange information, giving adequate consideration to confidentiality, control, etc.
Dr. Ferrans observed the hidden clinical benefit in capturing automated data from medical devices to free nurses from hand recording the information and commented on the value of physician access to pharmacy decision-support information. Mr. Jordan stated that although inpatient pharmacy has long been automated, use of the community-based decision- support mechanism is not broadly incorporated into patient care. Dr. Ferrans cited a recent meta-analysis published in JAMA that found an overall incidence of 6.7 percent for serious adverse drug reactions in hospitals; drug-related morbidity and mortality costs $136 billion annually in the U.S. He asserted that inpatient clinical decision-support systems for inpatient pharmacists and physicians would be integral to the medical record.
Dr. Ferrans cited the benefits to taking a thorough patient history and matching terms with a lexicon on the "back end." Dr. Gabrielli stated that his company's model takes data from the dictation back to the physician, much like a hotel takes shined shoes back to a guest's room overnight. He discussed the benefit of longitudinal records and the need in today's medicine to quantify the impact of a therapy, and asserted that his terminology is far ahead of others.
Dr. Jones described the nursing diagnosis--which defines a clinical judgment about an individual, of which there are three components: label, definition, and the environmental, physiologic, or genetic chemical elements that increase the vulnerability of an individual--and stated that the mission of NANDA is to increase the visibility of the nursing contribution to patient care. NANDA's work focuses on the availability of a centralized patient-family-community database to describe health and wellness, and a national data set. She asserted that nursing language should be included in any standardized database or patient medical record, to articulate both a disciplinary focus of care and a perspective within the care team, and to satisfy legal requirements for nursing care documentation.
Dr. Jones described the focus of Taxonomy 1, the current status of nursing diagnosis. Approximately 140 nursing diagnoses are currently in use, nearly 60 of them recently added or updated. In addition to her prepared remarks, she presented to the Work Group Nursing Diagnosis: Definitions and Classification, a book translated into ten languages and used internationally, although some translation issues have arisen.
The Taxonomy Committee stages and classifies nursing diagnoses; the structure identifies the things that nurses locate in patients. The goal is to make terminology computer accessible. Dr. Jones described in some detail both the classification system in use and the proposed new system, Taxonomy 2, which would institute more consistency within diagnostic statements, greater conceptual focus, better articulation of knowledge within the discipline, and more clarity in terms of practitioner usefulness. Dr. Jones presented an example of the operation of the proposed taxonomy, which includes 12 domains (e.g., health perception, values, beliefs, comfort, but still under revision) and six axes (e.g., diagnostic concept, level of acuity). She stated that the ultimate PMRI contains data that captures the patient health/illness experience within the existing framework of medical practice. Precision and accuracy are essential. At present the PMRI "does not contain vocabulary that allows the nurse to document data relevant to nursing practice within the domain of the discipline."
The nursing diagnosis in clinical practice, research, education, and administration, asserted Dr. Jones, helps to focus the work of nursing as a "thinking discipline," not just a "doing discipline." The nursing diagnosis is currently included in the UMLS and also works in concert with other nursing classification systems (see next two presentations). The NANDA taxonomy describes nursing phenomena of concern, creates a structure to document care, develops a database that will identify high-incidence nursing problems, helps to define complexity, and describes inpatient and outpatient care needs. It also helps inform administrative decisions and promote an understanding of the human experience across populations and settings. Dr. Jones observed that PMRI should reflect a multidisciplinary approach to documenting care.
Prof. McCloskey described the Nursing Interventions Classification (NIC). As background, she pointed out that although nurses are the largest group of health care providers and spend the most time with patients, the nature and impact of nursing services are virtually invisible. She asserted that nursing data must be included in the patient's health care record to study costs and effectiveness of nursing care, and to determine the relationship of specific nursing interventions to the interventions of other health providers.
The NIC, a comprehensive, standardized classification of interventions that nurses perform, has multiple uses in health care settings. The classification includes all interventions that nurses do on behalf of patients and can be used in all settings. Prof. McCloskey referred to her handout in describing the system. NIC interventions have been linked with NANDA Nursing Diagnoses, Omaha System problems, and NOC nurse-sensitive outcomes. The classification is continually updated with an ongoing process for feedback and review. The National Institutes of Health helped to fund the ongoing research to develop NIC, which is recognized by the ANA, and is included in NLM's Metathesaurus and the nursing indexes of the Cumulative Index to Nursing Literature, (CINAHL), and Silver Platter. NIC is included in the Joint Commission on Accreditation for Health Care Organizations Manual as one language to meet the standard on uniform data collection. The National League for Nursing produced a videotape about NIC to facilitate its teaching to nursing students and practicing nurses, and Alternative Link has included NIC in its ABC codes for reimbursement. The Center for Nursing Classification estimates that more than 300 clinical agencies now use NIC, and 150 schools of nursing use NIC in their curricula. Several translations of NIC are in print or forthcoming.
Ms. Moorhead spoke on the relatively new Nursing Outcomes Classification (NOC), which captures the range of nursing specialties in different settings. The NOC is a comprehensive, standardized classification of patient outcomes developed to evaluate the effects of nursing interventions on treatment. An outcome refers to an individual, family, or community condition measurable along a continuum and responsive to nursing interventions. Ms. Moorhead referred to her handout to describe an example of the use of NOC. Each outcome has a definition, list of indicators, and scale to measure patient status. The NOC team believes that a taxonomy blended with NANDA may be developed in the future.
Ms. Moorhead stated that the research has progressed to the second phase of measurement-scale testing. Several tools can assist in the implementation of NOC, and outcomes have been linked to NANDA, Omaha System, Resident Admission Protocols (used in nursing homes), and NIC. NOC is recognized by the ANA as a standardized language and is included in UMLS and CINAHL's index to nursing literature. NOC is used in practice, nursing education, and research, and is being translated into other languages for international use.
Dr. Ozbolt addressed the prepared questions of the Work Group. She stated that PMRI includes all data recorded during care by providers of all disciplines--usually assessment, diagnoses, orders, and documentation of care, but rarely goals and clinical outcomes (including physiological, cognitive, affective, behavioral, or functional). Records are important to ensure continuity of care and as a source of data for payers, managers, and researchers. Clinicians and other data users may require different concept representations to get at the semantic meaning of what they need to do their work.
Because the services of nurses and allied health personnel are not recorded in a retrievable fashion, effectiveness and cost-effectiveness of services that represent one third of a hospital's typical operating budget cannot be researched. Dr. Ozbolt asserted that "nursing has essentially been treated as part of the furniture," rolled into the room rate.
Nursing has no standard vocabulary with atomic-level elements with combinatorial grammars. Such vocabularies, Dr. Ozbolt stated, could: (1) be used as building blocks for care plans (e.g., on particular hospital units to project care requirements and staffing needs to preclude the need for redundant data collection from an expensive nursing classification or acuity system); (2) provide decision-support via hyperlinks to atomic-level elements; (3) create databases for quality and research; and (4) have sensitive measures for quality and effectiveness.
Dr. Ozbolt recommended that the government require that reports be based on clinically validated terminologies at appropriate levels of granularity and that terminologies meet emerging criteria for clinical terminologies, and support research and conferences to develop and test terminologies in nursing and allied health. Comparability of PMRI, which does not exist in most places, is necessary to reimburse, study, and improve patient care services. The ANA recognizes the Patient Care Data Set.
An upcoming conference funded by public and private sources will convene all vocabulary developers recognized and under consideration by the ANA, plus language and standards experts, representatives of federal agencies and professional organizations, and others to develop guidelines and desiderata for nursing terminologies. Dr. Ozbolt will provide the Work Group with a copy of the conference report.
Dr. Ozbolt stated that nursing acuity systems are commercial products designed to predict resource consumption for patient care. With prospective systems, it is not possible to know whether the patients receive the care, and retrospective systems cannot evaluate whether what was delivered was adequate; a 1988 study confirmed a study conducted a decade earlier that showed that the systems possessed neither validity nor reliability. Prof. McCloskey added that these systems are too customized to be comparable, and that research is necessary to see if an index of interventions could be developed. Dr. Ozbolt stated her goal to develop a way to project from a networked pathway system the care required in an upcoming shift on a unit. Dr. Jones suggested that the definition of acuity and the way it is operationalized are part of the dilemma.
Dr. Ozbolt stated that research is ongoing, to the extent that data is available, regarding the contribution of the nursing code sets and classifications to improved quality of care. Prof. McCloskey noted that research depends on people using the classifications with information systems. The vendor community is extremely broad and vendors are relatively small; she urged the Work Group to encourage vendors to use standardized language, which would produce the data to do effectiveness research and improve clinical pathways. She commented that only recently have critical paths begun incorporating nursing standardized data, but not based on a database of research. Dr. Ozbolt stated that the goal at Vanderbilt for standardized language built into critical pathways is the ability to customize pathways on an ad hoc basis.
Dr. Rothwell described his organization's Structured Health Markup Language (SHML). He identified the major issues affecting PMRI, for example, the shift to a disease prevention/management paradigm and the rapid rise of Internet standards. The Health Language Center focuses on adherence to guidelines and eligibility; when these processes work, the areas of care and secondary uses can be addressed.
The Center's task is to extract both text and structured data from the PMRI and to integrate the information from both sources to derive the clinically relevant content unit (CRCU), which is a finding. Natural language processing (NLP) is an extended XML formalism and a description of a set of tags using the XML formalism to characterize the information in the record in a new way. Dr. Rothwell described the function of NLP, whose formats include, for example, demographic data, verbs, patient state data on diagnosis, changes, and others. SHML is piggybacked on XML, whose structure and function Dr. Rothwell described in some detail. XML is concise, precise, readable, and transportable to databases (relational or object); SHML is flexible in how it presents data using descriptive tags.
Dr. Rothwell demonstrated a mapping of SHML, showing both language class and medical knowledge class. A traditional tag set is applied, in addition to such new tag sets as verbs, prepositions, dietary/food, time, location, alternative care, and others. Dr. Rothwell noted that processing speeds can be enhanced if the viability of the approach can be demonstrated. The SHML approach has the ability to resolve ambiguity and to deal with multiple hierarchies. In summary, Dr. Rothwell suggested the following: (1) adopt rules and notation in place for SGML/XML for the PMRI; (2) structure the EMR; and (3) use XML rules, notation for content. He stated that SHML/XML works with language but does not reinvent it, that XML provides structure and contextual meaning to a document, and that XML is a self-describing data structure. He stated that the mission of this project is to: (1) define a granular representation of terms and phrases that, within a given language (domain), unambiguously define clinical concepts; (2) provide an adequate representation of the terms and concepts in an easily understood architecture; (3) provide for discrete mapping to other nomenclatures and code sets; (4) use easy, available, inexpensive, and widely supported tools for authoring, maintenance, and use; and (5) provide this as a nonproprietary standard under the auspices of a private, non-profit entity--the Health Language Center.
Dr. Neal provided the Work Group with a written response to its prepared questions, a history of the development of the Medical Dictionary for Regulatory Activities (MedDRA) Terminology, and a Federal Register ANPRM. She presented an overview and update of MedDRA, an international medical terminology developed for regulatory purposes. MedDRA is a multi-axial terminology built upon a five-level hierarchical structure.
Dr. Neal explained that the need for a standard terminology is great in the regulatory decision-making process to: (1) facilitate consistent coding to avoid translation errors, (2) decrease time spent by reviewers in normalizing data and making comparisons, (3) decrease submission time for new drug applications and new investigational drug applications, (4) improve product labeling, (5) facilitate the shift to electronic submissions, both pre-market and post-market, and (6) improve the ability to perform epidemiological studies on a global scale.
In order to foster international usage, the terminologies included in MedDRA 2.1 are the MCAs, medical terminology, COSTART version 5, WHO-ART, J-ART, ICD-9 and the clinical modification, and HARTS. Topics such as drug and product terminology, study design, patient demographic terms, population qualifiers, and descriptors of severity were initially tabled and my possibly be addressed at a later date. The terminology was designed for use throughout the regulatory cycle and prompted by the rapidly changing environment in the FDA. Since November 1997, all adverse event reports received for drugs or therapeutic biological products are being entered into the new Adverse Events Reporting system (AERS) database and encoded in MedDRA.
A November 1998 ANPRM announced the FDA's plans to require use of certain electronic standards, including MedDRA (in the ANPRM, the term M1 refers to MedDRA). In view of the FDA's goal to achieve a paperless environment by 2002, comments were solicited regarding: (1) whether and which exemptions should be granted to the rule, (2) how users see the cost benefits versus cost burden, and (3) what time frame for implementation would be reasonable.
FDA's next steps include continuing post-market implementation of MedDRA, preparing a NPRM for electronic ADR reporting, and applying the post-market experience to pre-market activities. Dr. Neal introduced several informational web sites.
Capt. Hess presented a written response to the Work Group's prepared questions and focused his oral presentation on the National Drug Code (NDC). He described the NDC's three configurations.
The Drug Registration and Listing system consists of a relational Oracle database that contains information on over 124,000 (>48,000 prescription drugs, >65,000 over-the-counter drugs, and >11,000 bulk drugs). This system is used to produce the NDC Directory, which contains a subset of 124,000 drugs that does not include the bulk drugs or most of the over-the counter drugs. Inclusion in the directory does not denote approval by the FDA.
Most products have an NDC assigned to them prior to approval because the NDS is required before a product can be marketed. Files are available on the Internet. The current audience is third-party providers worldwide. No enhancements to the NDC are planned at present. Additional integers would change the three different configurations, and numbers will run out in a few years; the different parts of the NDC are sometimes recycled when specific criteria are met. For instance, the labeler code can be recycled after a firm as been out of business for at least five years. Also, the product code and the package code can be recycled after a product has been off the market for at least five years. This does not happen very often. The NDC is indirectly mapped to the UMLS through Micromedex; there is some overlap with other files, such as the FDA's Drug Product Reference File, which produces the "orange book," Approved Drug Products with Therapeutic Equivalence Evaluations.
Ms. Coates stated that ECRI--a nonprofit, nongovernmental health services research agency and a collaborating center of the WHO on information transfer on medical devices has developed, maintained, and continuously updated a standard controlled terminology and coding system for medical devices. Ms. Coates responded to the Work Group's prepared questions, stating that the PMRI must be comprehensive, comparable, accurate, confidential, accessible, and retrievable to support analysis from many perspectives. High-quality, sufficiently granular data must be available to support clinical decision making, outcomes research, and epidemiological studies. Comparable PMRI supports data retrieval for such activities as health services research, outcome studies, and technology assessment. Comparable data must be available, accurate, and precise. Broadly used coding systems, such as CPT and HCPCS, do not uniformly identify devices explicitly, a growing problem as the number of devices increases.
ECRI's terminology, Universal Medical Device Nomenclature System (UMDNS), classifies medical devices for indexing, storing, and retrieving device-related information. UMDNS is widely used by government agencies, health care systems, and facilities worldwide. The terminology covers almost any nondrug item used in patient care. The terminology, translated into ten languages, is used internationally in thousands of institutions. Its applications include regulatory databases on medical device adverse incidents and software for inventory control. It is the interim standard for electronic communications and medical product registration for the European Union, and serves as the primary source vocabulary for the final European standard, Global Medical Device Nomenclature. ECRI currently assists the FDA to harmonize its present nomenclature system with UMDNS and ultimately with the final global standard. ECRI is expanding the terminology to incorporate European devices not used in the U.S.
UMDNS has been incorporated into UMLS; the U.S. National Library of Medicine (NLM) contracts with ECRI to map medical device concepts to other nomenclatures in UMLS. UMDNS is linked to MESH, ICD, CPT, SNOMED, and others, and its terminology is incorporated by reference in SNOMED and SHML. ECRI has also mapped its device terminology to the FDA's product categories and to terminology developed by Japanese health authorities.
Ms. Coates urged the government to encourage the adoption of standard terminologies for PMRI, to make funding available to terminology developers to enhance and maintain high- quality terminologies, to coordinate its activities with ongoing international efforts better, and to sponsor a conference on standards, data elements, and policy issues for the PMRI and publish the proceedings.
Ms. Meredith explained that Cerner Corporation, a leading supplier of clinical and management information and knowledge systems for health care organizations in the U.S. and abroad, services 1,000+ clients; Multum focuses on the creation, maintenance, and distribution of drug information. More than 2,000 organizations license drug and drug product information from Multum, whose drug terminology is a reference vocabulary in standards produced by the NCPDP and is being incorporated into the UMLS.
Ms. Meredith explained that generic drug names in the U.S. have not been codified to promote interoperability and are not sufficient to describe a drug to be ordered, prescribed, or included in an EMR. The proliferation of "numerous disparate and incomplete pseudo- standards" has been a serious problem in achieving interoperability. Serious shortcomings exist in basing information on NDCs--for example, the absence of a common meaning and standard format and of regular updates to the public list--which are not an appropriate nomenclature for clinical practice because of lack of interoperability.
She asserted that the health care informatics community needs a reference terminology for drugs that contains generic name information, plus modifiers for strength, dose form, and route of administration, which would promote interoperability. Ms. Meredith urged the federal government to support strongly the development of a drug reference terminology.
Ms. Meredith explained that the Multum terminology is similar to the NDC, but no commonalities exist among the for-profit companies. Mr. Mayes stated that HCFA licenses for public use for free Multum's drug list; NDC was never meant to be a clinical vocabulary. Ms. Meredith stated the need for a bridge among terminologies, and noted that the drug list would continue to be provided free of charge.
Capt. Hess noted that despite problems with NDC, its use is mandated by the Food, Drug, and Cosmetic Act. Databases currently include all drugs on and off the market, coded with standardized nomenclature for the drug's proprietary and generic names, dosage form, and route; however, NDCs were designed only to identify a product. Mr. Hess stated that the NDC is updated daily and files are available on the Internet in downloadable form, although he did not know whether the file available on the Web is the most up-to-date version. A current copy may be requested from the FDA.
Dr. Rothwell stated that foreign language represents difficulties for SHML. He noted that information on acupuncture, for example, is defined in SHML dictionaries and assigned tags; the intent is to be inclusive. The work on SHML is complementary to the record architecture activities of HL7, in that SHML addresses the content issues within the sections outlined by HL7's architecture.
Dr. Neal stated that MedDRAis built upon a 5-level hierarchical structure, and more information is available on the MedDRA Maintenance and Support Services Organization website. She stated that there is no mapping between SNOMED and MedDRA--as Dr. Fitzmaurice suggested would be helpful to the clinician to avoid having to learn two codes sets.
In the case of adverse drug events, Dr. Neal stated that physicians report either directly to the FDA on a MedWatch form, which is coded by contractors into the MedDRA codes, or to the product manufacturer, which then sends the form to the FDA, where it is coded. The ultimate intent is to have the drug company send precoded electronic reports to support the Agency’s move towards a “paperless” environment. MedDRA was not designed to be implemented with an EMR that a clinician would use directly, although MedDRA will be used if clinical trials fit into primary care practice.
Ms. Bickford urged that nonregulated drug compounds and environmental chemicals, toxins, and exposures (from an occupational health perspective) should be considered for inclusion as part of PMRI. Mr. Mayes stated that the Environmental Protection Agency is working formally on the latter issue.
Dr. Neal confirmed that although MedDRA was developed for regulatory uses, it covers a vast expanse of procedures, symptoms and signs, and diseases and diagnoses. She stated that TRW is leading an industry consortium to manage the distribution of MedDRA. The terminology is available at an affordable rate to institutions.
Capt. Hess stated that all pharmacies and pharmacy vendors use the NDC for ordering and filling prescriptions. If the database were brought into providers' offices, it would be possible for them to use it, but accessing the code itself is not valuable; a code associated with additional descriptors would be essential. Ms. Meredith pointed out the difficulty in choosing among multiple identical products. On the device side, Ms. Coates stated, the ECRI terminology was developed to organize information about medical devices to support many different applications, including procurement in hospitals. Both a Universal Product Number and a higher level classification system such as UMDNS are necessary to organize device information and identify it appropriately.
Dr. Fitzmaurice summarized the major points of the presentations. Dr. Cimino provided guidance on the qualities of a classification or terminology. Dr. Chute stressed the considerable ongoing collaboration on terminology development. Dr. Campbell's lessons learned regarding terminologies were valuable, considerably from the big picture standpoint. Mr. Tuttle sparked thinking about government subsidies at the margin, using a pilot program to elicit quality improvements.
The presentations on specific code sets indicated development progress. Many codes are not reported because they are not reimbursed, resulting in incomplete PMRI. Advice to the government included supporting government code systems and terminologies, supporting and making freely available some codes and terminologies, making cost-effective choices, not over- studying.
The market is becoming more interested in encoding. The issue of funding for standardized PMRI was raised. People recognize the value of data for care, but payment cost- effectiveness requirements currently drive medical information distribution. Standards are under development for medical-device data. Dr. Gabrielli discussed parsing down information to the basics, as opposed to specialty-by-specialty work.
Many nursing code systems exist, and more research is necessary to reconcile them and to evaluate their contribution to patient care. Dr. Rothwell is merging XML with structured language, but questions remain.
In the arena of drug codes, the puzzle is to capture information on what a doctor prescribes, what the pharmacist fulfills for the patient, and how the patient takes the drug--and get the information to the emergency room doctor to deal with an adverse reaction.
The Work Group note takers planned to submit their notes to Dr. Fitzmaurice for synthesis.
Dr. Cohn summarized the hearings in terms of major issues on which the Work Group may need to take action. Dr. Chute discussed the business case and issues related to cost of development and selling the terminologies, versus government subsidies; the bottom line is that either private or public funds will support development. Discussions revealed that although terminologies are getting better, the market is not ripe for them. One speaker described the dual need for a top-down and a bottom-up approach. Several participants mentioned the needs to convene conferences for consensus development and to fund maintenance. It is also important to consider vendor neutrality. Dr. Campbell's construct of strategic imperatives should be referenced in the discussion on principles.
Among the roles discussed for government were: (1) to adopt and use standards, and (2) to fund "last-minute infrastructure." It is important to consider ICD-9, ICD-10, and CPT I; ICD and CPT have significant clinical uses that merit discussion. An impediment to data quality occurs when payers do not follow official coding guidelines. The Work Group looks forward to receiving the report from the upcoming nursing terminology summit.
Dr. Cohn remarked on redundancies in recommendations, particularly in that Drs. Gabrielli and Rothwell are doing similar things, from the perspective that there are many ways to get data into a PMRI. Questions remain about where drug terminologies fit into an overall framework, particularly in view of the deficiencies of NDC. Mr. Blair noted Multum's frustration with the coordination process and its hopes that the government will serve as catalyst.
Mr. Mayes noted the academic tone of requests for the government to award grants, which are used for development; contracts are the mechanism for ongoing programs. Successful companies within their own markets do not care about government involvement. He noted that Mr. Tuttle's suggestion for government funding on the margin was the most intriguing, given the government's need to focus on the business case. He observed that the tighter the recommendations are linked to the business case and to missions of specific agencies, the more likely they are to be implemented.
Mr. Blair added that the majority of views expressed in this round of hearings were from developers, and that the voices of vendors and other users were not present. Mr. Kun suggested that wellness and other concerns, such as environmental hazards and updates in genetic knowledge, must be addressed in PMRI in order to avoid obsolescence of terminologies. Several members discussed the necessity to get the process and the principles-- such as flexibility and adaptability--right. Dr. Kolodner emphasized that growing complexity should impact on the decisions the Committee makes on specific recommendations; he introduced the analogy of an ecosystem of a forest in which many different entities must interact. Dr. Ferrans suggested that Paul Tang's concept of alignment of incentives should be a guiding principle, and because there is such a variety of implementation possibilities, the recommendations are either going to be very high level, broad, and principle based, or complex in response to the different components and segments--or a combination of both. Dr. Ferrans also noted the lack of discussion about the user interface of terminologies.
Mr. Garvie mentioned the need for a synthesis methodology to capture all the information that has been presented to the Work Group. Dr. Greberman suggested using specific examples to show possible solutions to the issues identified, such that various users will have access to the information they need and to find ways to work together through mapping, for example. He also suggested investigating the need to discuss MedDRA at the next meeting.
The Work Group members next discussed the agenda for the June meeting. Dr. Cohn suggested the need for reflection on how to proceed most productively with the remaining hearings and how to synthesize the information already collected. He also noted the need to hear differing viewpoints on terminologies. Claudia Teshiet and John Madison also will testify. Dr. Cohn also noted the need to have focus areas critiqued by individuals working on the GCPR and to discuss opportunities for collaboration and synergy. Betsy Humphreys will be asked to comment on the testimonies heard by the Work Group, as will Dr. Greberman. Dr. Cohn stated that most of the afternoon should be spent discussing what has been learned, where in the work plan the Work Group stands, what must be changed in the work plan, and next steps. He also suggested discussing the substance of a letter to be sent to the Secretary in September concerning the Work Group's efforts. An FDA testifier on device codes may also be invited.
Dr. Ferrans questioned whether attention should be focused on terminologies not yet discussed, but Dr. Cohn suggested that the next effort must be defining PMRI to determine the boundaries. Mr. Blair stated that testimony on the "other terminologies" should be presented in person.
Mr. Mayes raised the importance of the issue of confidentiality and privacy, and noted that if state laws are to be examined, it should be done reasonably soon; the letter to the Secretary should also reflect discussions of the Privacy Work Group. Dr. Ferrans suggested sources of testimony for states' issues and medical technology.
Mr. Heffernan discussed the possibility of circulating a summary of the issues--for example, convergence on concepts--for comment. Dr. Fitzmaurice suggested initiating a process of brief working drafts, for example, first on the PMRI and second on the functions of PMRI. Another focus would be the business case for electronic PMRI and its many components. The Work Group began to set this process in motion.
Mr. Blair and Dr. Cohn thanked the members for their participation, and the hearings were adjourned at 5:00 p.m.
I hereby certify that, to the best of my knowledge, the foregoing summary of minutes is accurate and complete.
/s/Jeffrey S. Blair February 2000
_____________________________________________-
Chair Date