The Workgroup on Computer-based Patient Records of the National Committee on Vital and Health Statistics held public hearings on October 14-15, 1999, at the Hubert H. Humphrey Building, Washington, D.C. Present:
Note: The transcript of this meeting is posted on the NCVHS website.
The NCVHS Workgroup on the Computer-based Patient Record (CPR) conducted a meeting and held hearings on October 14-15, 1999, in an ongoing process to study the issues and advise the Department of Health and Human Services on the adoption of uniform data standards for patient medical record information (PMRI) and the electronic exchange of that data. The Workgroup heard presentations from individuals on four panels. The Workgroup also reviewed its schedule and agenda for developing the briefing letter to the Secretary of Health and Human Services, and discussed the development of PMRI issues, assumptions, and general recommendations for message formats and medical terminology for its report.
John Kelly, Aetna U.S. Health Care
Mr. Kelly described how Aetna uses information and information technology (IT). Some data are collected in electronic form and stored in a single warehouse. Some data are distributed electronically, but the bulk of communication is conducted using paper. Using technology, Aetna places members into clinical/risk categories to help manage their care. Higher-risk members are placed in disease management programs, with results monitored. Patients in some programs have shown reduction in their use of acute care services, and improvements have been seen in medication compliance rates and patients’ self-assessed quality of life. The ability to communicate quickly as a result of early analysis of electronic data speeds up improvements in members’ care. Mr. Kelly asserted that any effort to accelerate the use of the Internet or access to the Internet in physicians’ offices would have positive effects.
Barbara Demster, Healtheon Corp.
Ms. Demster stated that the CPR permits transactions from disparate legacy systems to be presented transparently to authorized users with an Internet browser and proper security access identification. Consumers can access health care information online (a problem because of insufficient peer review), and some consumers keep their own health records.
Ms. Demster pointed to the interdependence of the personal health record, population health record, and care giver health record. Lack of standard formats and lexicons is a serious obstacle to conformity of terms and comparability, and thus to industry efforts to assess effectiveness and efficiency. Data standards are critical to assure data reliability and integrity; a unique health identifier will aid in matching data. An estimated 10+ percent of IT budgets is used to plug holes created by nonconformable data standards.
Government leadership can facilitate development and usage of data standards. Minimum data sets for responding to common recurring information needs within the delivery system would both address the confidentiality issue and improve data capture and quality. Current legislation does not encourage confidence in the privacy of health information. The public must be educated on how the Internet can be made safe for transmission of health data.
Mr. Kelly confirmed Dr. Fitzmaurice’s assumption that Aetna considers risks of high expense, hospitalization, death, and loss of health status. Aetna is also interested in quality factors vis-à-vis standards of practice for particular clinical categories. Increased incidence of mammograms occurs, for example, when members are contacted and urged to be involved in their own care.
Blackford Middleton, M.D., MedicaLogic, Inc.
Dr. Middleton defined the PMRI as any information in a paper- or computer-based medical record; he referred to NCQA’s frameworks for information space for PMRI and for adoption of IT within the clinical environment. The Internet must be added to the information-sharing equation. Dr. Middleton highlighted the necessity for comparable PMRI, for example, in efficient and effective health care delivery and more successful identification of best practices and management. Comparable PMRI is useful in data aggregation for individual patient care, information communication, and interoperability between disparate information systems. Data integration is the key ingredient to properly inform clinical decision making. Dr. Middleton suggested a framework for a national health care information infrastructure that would facilitate appropriate information exchange among all parties in the health care delivery process.
Dr. Middleton noted that over-coding produces a drop-off in usefulness of data, and stated that the industry should strive to model key information elements incrementally. MedicaLogic uses structured clinical lists derived from standard controlled terminologies, or standard codes, including ICD, CPT, ASTM, LOINC, SNOMED, and DSM, and some data is represented in proprietary format.
Dr. Middleton’s described several challenges, including, among others, balancing coded data requirements with usability of the project and finding a close matching structured data element from a standardized structured vocabulary with the user’s requested term. To address these and other problems, Dr. Middleton suggested a ranked, incremental approach by establishing: (1) national identifiers, (2) a minimal set of critical quality or outcome measures, (3) minimal required data sets to support derivation of the critical quality measures, (4) acceptable standard controlled vocabulary and codes for the minimum data set, and (5) acceptable methods to gather the minimum data set at the point of care. He urged the federal government to fund and coordinate public-private development of controlled vocabulary standards, which, when placed in the public domain, will trigger widespread adoption.
Dr. Middleton stated that more work is necessary to make the HL7 interfaces easier to operate and maintain. He identified three incrementally sophisticated levels of interoperability: basic, structured, and semantic. Coordination initiatives are not sufficient, but a ranked framework for structured interoperability standards, building upon current HL7 efforts, and new efforts arising in the electronic health care arena could lead to benefits within four years. Key to the process are identification of the critical quality measures to drive the business processes and impact on health care costs or quality, plus defining the minimal data sets and encouraging their adoption by major health care IT vendors. Dr. Middleton asserted that funding for academic research on a common reference model and terminology should be more tightly coupled to industry in order to run large-scale demonstration projects.
Capt. Mayes stated that unless one can demonstrate how specific coding can lead to better business outcomes, little motivation to adopt a system may be present. Some major health care companies have applied semantic management to standardized data representations. Mr. Kelly discussed the difficult cultural issue of identifying benefits and successful models, and then promoting them, particularly to the many physicians whose clinical practices are not yet automated.
Dr. Middleton and Capt. Mayes concurred that melding structured and semantic interoperability, and then working incrementally, are critical to success. Mr. Kelly discussed the importance of removing paper communication from the medical scene as an incentive to move to CPRs and to incorporate medical practice into the Internet system. Dr. Middleton acknowledged that what drives usable systems are whether the note supports the level of service billed and whether the note contains a problem list and a method to write a prescription on the formulary. Middleton finds he can gather data at the point of care in a way that supports the needs of both his clinical practice and the health care enterprise. Ms. Demster cited the problem of compliance regulations for fraud and abuse as an impediment.
Mr. Kelly endorsed national incentives for computerization, particularly in view of the key importance of the business case and of the cultural barriers that remain. Dr. Middleton noted that if a physician knows that his or her compliance in gathering certain data will be measured, that physician is more likely to collect those data. He also commented on the role of the grassroots consumer movement in motivating the medical industry to produce the data that will enable them to make informed decisions.
Dr. Zubeldia noted problems with the provision of free computers by Internet service providers and computer chains. Ms. Demster asserted that standardized data sets and outcomes research would positively impact quality of care as the bottom line. Mr. Kelly pointed out the value of simplification, focusing, and setting priorities, and Dr. Middleton concurred, emphasizing the necessity to work incrementally on relevant, attainable objectives. Dr. Middleton stated that ICD, CPT, DMS, and LOINC are sufficient for the present.
David Kibbe, M.D., Future HealthCare, Inc.
Dr. Kibbe explained that the health care industry is close to the point at which old ways of doing business are replaced by new methods as a critical mass of technological, cultural, and regulatory changes occur. He stated that “e-health” is both a product and a stimulus to new technology and its use.
Dr. Kibbe itemized the major emerging sectors of the e-health marketplace and described their roles. Applications providers serve as host and security guarantor for health data on the company’s servers; personal health information is easily accessible by patients, their families, and all authorized caregivers. All e-health companies interactively gather and exchange data from patients, physicians, hospitals, and family members.
Dr. Kibbe asserted that a set of standards that will make importing and exporting data from e-health sources inexpensively will help his business to become more successful and his clients more capable of providing better medical care. Universal adoption of HL7 and LOINC would be helpful. He urged the government to sponsor conferences of users and developers of e-health standards to define problems, issues, and research studies to show the value of uniformity of PMRI for business, patient management, and technology assessment uses. The government might purchase the rights to existing coding of vocabulary systems to make them available at low cost. Funding of standards can fill gaps identified at conferences and hearings, adopting PMRI standards for DHHS programs rather than inventing new codes, developing standards through government initiatives, promoting research in PMRI uses such as measures of quality and cost of care, chronic illness and emergency medicine registries, and establishing safe harbors for patient information accessible by patients and their families.
C. Peter Waegemann, Medical Record Institute
Mr. Waegemann focused on international PMRI standards and highlighted Sweden’s standard for medical record structure, France’s parallel efforts with the U.S. regarding legislation similar to HIPAA, and 7TC251 reorganization.
Workgroups on the ISO TC215 standard on health informatics are focusing on a generic domain model for health care, ownership and access rights, smart cards, and a general country identifier for patients, devices, and caregivers, among others. The first ISO standard is expected to be for public key infrastructure for health care.
Dr. Waegemann urged government agencies and the general medical community in informatics to be involved in the US TAG part of the issue. The U.S. is the only country that does not fund ISO travel and expenses, which presents a challenge to secure a consistent set of volunteers at worldwide meetings to represent the U.S. interests.
Dr. Waegemann noted that an ISO Workgroup is discussing reimbursement and the need for administrative interchange between countries. Dr. Kibbe stated that the government should remove cost barriers to using a proprietary coding system. Dr. Yasnoff’s suggested that the government might negotiate with a specific coding system.
Richard M. Peters, Jr., M.D., iTrust
Dr. Peters attributed a “frozen” health care IT industry to Y2K and no-confidence concerns expressed by the physician community about expensive and/or limited availability of practice-management solutions, plus radical changes in the health care market and in the reimbursement process; sharply reduced hospital revenues in the first half of 1999, and other factors. Near-term investment activity in new companies by venture funds, investment banks, and institutional investors—not traditional investors in health care—may double its current $500 million level in 2000. Investment targets have been connectivity vendors and consumer and provider health websites and portals.
Dr. Peters expects rapid consolidation of the online pharmacy market to take place; Internet practice management and the Internet electronic health record are other active areas. He also expects capital to flow in the near term into administrative outsourcing providers. Dr. Peters described the structure of the e-health market in detail and discussed both the negative impact of some market features on legacy companies in the marketplace and impediments to early adoption of the CPR.
Dr. Peters predicted that the primary driver in the next decade will be connectivity between the payers and providers and/or payers and patients, and listed several major players in this arena. Value can be achieved through connectivity regarding laboratory results and claims; connectivity players, with their additional capital, bring energy to an inexpensive infrastructure (i.e., the Internet).
Dr. Peters discussed how the current state of the market affects the standards issues: (1) e-health vendors are outrunning the standards efforts; (2) privacy and confidentiality are issues both in terms of security and in aggregation of data for inappropriate uses; and (3) new vendors have already defined a medical and health record, and they are already providing operability. Dr. Peters suggested solutions regarding privacy and confidentiality, defining and enforcing key standards quickly, inviting new e-health vendors into the standards process, and achieving national consensus on access rights and ethical data aggregation and use. Dr. Peters asserted that sale of aggregated data should not be permitted, although this is the stated business plan of some companies currently doing business. It is also important to push greater standardization of eligibility data. Dr. Peters endorsed certain terminologies and other possible standards, and suggested individuals to work on the process.
In summary, Dr. Peters asserted that investor money is driving a nontraditional view of the health care IT industry, driving new ways to do business. The standards process may be outstripped by individual efforts, and may lead to exchanging data in ways not contributory to patient care and detrimental to privacy and security.
Judy Ozbolt, Ph.D., Vanderbilt University
In June 1999, representatives of all nursing vocabularies, experts on language and standards, federal officials, officers of professional associations, health care providers, and representatives of the health informatics industry met at Vanderbilt University. They acknowledged that while different colloquial terminologies are needed for different specialties and sites, reference terminologies must contain all allowable terms to refer to the concepts in the practice, structured according to the concepts’ semantic relationships.
The UMLS may not be the long-term solution because of the independent evolution of the colloquial terminologies. One long-term solution could be to develop a reference terminology model that identifies and names the core concepts and their defining attributes, and depicts the semantic relationships.
Dr. Ozbolt urged government support through a contract to an institution and primary investigator, who would take responsibility for overall coordination and specific research and subcontracts to the institutions of all the collaborating investigators. Working conferences and dissemination of results must be supported. Dr. Ozbolt enumerated government organizations with an interest in this work.
The public good would include an ISO standard reference terminology model for nursing, plus strengthened collaboration and consensus in matters of nursing terminology, a reference terminology for nursing, a set of standard nursing terms, and data and information systems that are colloquial but mapped to the reference terminology. This will enable comparable nursing data across sites, systems, and time, which will justify nursing data and public health databases, and enable the mining of clinical data to assess the quality and effectiveness of nursing care. The long-term benefits would be new knowledge of how nursing contributes to health outcomes, and development of incentives to have data-based and knowledge-based initiatives to improve care at reasonable cost.
Dr. Peters urged that early steps be taken to define ethical practices in the area of data aggregation. He commented that large vendors do not perceive the need for standards, because they plan to dominate the market and establish de facto standards, and some people believe Internet standards are not critical, because with XML, data is easily translated and the format is inconsequential. Dr. Peters disagreed with both these views, asserting that if the standard is there, vendors have shown no resistance to abiding by it. Dr. Peters suggested that industry knowledge and contacts be used to draw into the discussions the smaller vendors.
Because applications service providers (ASP) that provide hospital information systems over the Internet are unregulated, Dr. Peters urged customers to do significant due diligence on the vendors and insist on contractual restrictions on data usage.
Dr. Peters suggested that the government and industry fuel the standards process with cash, particularly travel costs to meetings. Beyond funding, coordination is required. In deciding on standards, a simple majority should rule if consensus cannot be achieved. A public-private cooperative initiative is better than either public or private, but obstacles will be encountered in devising the mechanism.
Dr. Kibbe described an underlying cultural paradox, that of needing—but blocking—the master patient index effort, and of individuals and companies voluntarily giving up some of their confidentiality to gain information on a new service, for instance, to improve quality of care at an affordable price.
Mr. Waegemann stated that the U.S. does not accord priority or financial support to volunteers working in the interests of the U.S. on health care information standards. Dr. Ozbolt considers it possible to extend the nursing vocabulary development process to other vocabulary efforts. The U.S. effort has drawn into the process representatives from other countries.
Joan Kapusnik-Uner, Pharm.D., First Data Bank
Dr. Kapusnik-Uner noted that successful interoperability has been achieved to a significant degree within the pharmacy profession and among other interested parties. She credited NCPDP standards in this effort. Dr. Kapusnik-Uner described her company’s capabilities as a trusted source of drug-related information. She described the many purposes for which First Data Bank maintains databases, which map between the databases internally and to outside vocabularies. The company is involved in several standards development organizations.
Dr. Kapusnik-Uner described the functionality of a comparable CPR and its prerequisites. First Data Bank creates cross-reference data to enable its customers to interoperate with all the terminologies they use. She considers SNOMED-RT superior. She recommended that the government should help to define the CPR use cases for medical terminologies to set the bar at a minimal level, offer incentives for early adopters of new terminologies, support terminology-related outcomes research, particularly related to the NLM’s work, and support unique vocabulary initiatives.
Mappings must be done by experts, not volunteers. First Data Bank is converging to a single reference terminology model; doing so outside of the company may be achievable through HL7 though valid values for the HL7 model should be left between business partners, users and domain experts, not HL7 volunteerism. She expects UMLS 2000 to be a substantial improvement.
Floyd Eisenberg, M.D., SMS
Dr. Eisenberg discussed data quality, accountability, and integrity, primarily on vocabulary and measurement issues from the perspective of knowledge integration through a software vendor and ASP. SMS supports industry standard message formats and vocabulary. Dr. Eisenberg mentioned lingering problems, including costs associated with nonstandard interfaces. SMS will assimilate standards as they are defined.
PMRI is electronically maintained information about an individual’s lifetime health and health care that replaces the paper medical record as the primary source of information for meeting all clinical, legal, and administrative requirements. The role of vocabulary models is standardizing messages for a comparable patient-centered medical record. The primary barrier to a viable CPR is the lack of universal standards allowing data integration across diverse organizations. The keys to clinician use of IT is to make the computer the sole source of required information, the documentation process fast and simple, and value to the user easily demonstrable. A standard in bar coding and text parsing methodology would enable health systems with products from several vendors to manage the data-entry process more cost-effectively. Adoption of a standard reference terminology would facilitate translation and linkage between clinical and administrative terminologies. Appropriate encoding enhances the integrity of data, and sharing will engender use, which will engender quality improvements.
He urged the government to provide leadership by adopting a single vocabulary standard that integrates existing terminologies, placing the adopted standard in the public domain, publishing guidelines and protocols in standard format so they can be placed in operation as plug-ins, and providing incentives for accrediting and regulatory agencies and organizations to develop vocabulary and problem-based clinical quality measures.
Gary Dickinson, Per-Se Technologies, Inc.
Mr. Dickinson stated that for each individual patient or health plan member, PMRI forms an essential longitudinal chronicle of health status and interventions, describes the clinical and operational context and ascribes accountability, and serves as a foundation to the operations business record. He itemized the benefits of comparable PMRI, including ensuring trust, confidence, and validity of assessment, among others. Comparability, accuracy, precision, consistency, continuity, completeness, and persistence are essential characteristics of data integrity and quality. Mr. Dickinson focused significant attention on the many problems that exist under the current continued industry reliance on legacy systems, which has produced disparity in the current state of data quality, accountability, and integrity, and the ability to measure outcomes, quality, or performance, and other problems. He described a number of initiatives of the HL7 SIG on Accountability, Quality, and Performance, each focused on a key aspect of accountability and data quality and integrity.
Mr. Dickinson characterized private-sector progress as painfully slow and questioned whether sufficient incentives are in place to commit the necessary resources. He suggested that the government support existing initiatives with incentives for serious partners in the endeavor and that the government champion a coalition of public and private health industry organizations. He suggested a basic strategy to characterize the problems base, establish objectives and rationale, forge specific requirement statements, set measurable milestones and benchmarks for achievement, and develop implementable technical specifications as open consensus standards.
Mr. Dickinson noted that the government’s crucial role is to define HIPAA requirements by setting expectations for what both government agencies and the community at large might do and procure. A strategy is lacking in HL7 to determine needs, requirements, and objectives; the more industry input on setting priorities the better. Dr. Eisenberg considered the nursing terminology model to be a good one. Dr. Kapusnik-Uner stated that vocabularies at different levels of abstraction and granularity should be mapped in the UMLS and incorporated into the reference model. The UMLS is perceived to be a viable solution, especially with its promised enhancements, because of its view of the Internet and its capabilities.
Dr. Kapusnik-Uner asserted that First Data Bank significantly contributed to the HL7 RIM version 3.0. Currently code sets that would map to the HL7 RIM are proprietary.
To Capt. Mayes’s comment that a major barrier to UMLS is its unavailability for unrestricted public use, Dr. Kapusnik-Uner suggested that the government support licensing fees at some minimal data set level for a version of the UMLS for all vocabularies submitted to enable identification of the concepts in each vocabulary, the value set, which would be in the public domain.
Members discussed the impact on more rapid adoption of CPR if standards were developed and introduced, noted the price/cost/intellectual property obstacle imposed by knowledge vendors, and suggested a licensing fee regime. The conversation moved to complex information world of the future; Dr. Kapusnik-Uner described ways in which HL7 would appear in this world in terms of interoperability, mapping, and proprietary information. She suggested the value of convening a Vanderbilt-model conference to improve the model to define the message segments and elements. Dr. Kapusnik-Uner also described the relationship of J codes to NDC codes.
Workgroup members briefly discussed draft agendas for the December 9-10 and January 31-February 1 meetings. Mr. Blair noted that the full Committee restored the privacy, confidentiality, and security focus area to the topics that will be included in the report to the Secretary. The main focus of the meetings and interim conference calls will be to work iteratively and collaboratively toward defining and refining the issues, observations, assumptions, and recommendations that will form the report to the Secretary of Health and Human Services.
Members noted the importance of including a separate, high-level section on the rationale and driving forces behind the recommendations ultimately developed. Members agreed that this overarching discussion should not replace the discussion of the business case as a focus area. Dr. Zubeldia recommended including in the appendix some testimony before the Privacy Workgroup, plus the privacy, confidentiality, and security recommendations. Mr. Blair suggested also adding the proposal for the NHII.
Ms. Amataykul led the discussion of the convergence documents, explaining that she had compiled the most important issues in the areas of medical terminology and message format standards, based heavily on summaries previously prepared and on basic assumptions and issues. The current discussion paper included five basic assumptions relative to medical terminologies and two recommendations. Mr. Blair stated that by December 9, a list should be created of what the speakers have recommended during the hearings. Dr. Yasnoff suggested that the recommendations be operationally oriented.
Members identified issues with the meanings and focus of the assumptions, and resolved ambiguities. They expressed their wish to separate issues from assumptions, and discussed the following topics:
Dr. Fitzmaurice suggested framing subsequent discussions on each focus area in more specific components, such as issue, barrier, and general and specific recommendations.
Helene M. Guilfoy, ASTM
Ms. Guilfoy discussed standard E1384, content and structure of a CPR. The CPR defines all health-related information about a patient, requests practitioner orders, and includes patient identification information including demographics and legal parameters. Information in a CPR must be seamless and consistent over time, relevant, available in a common record structure, and commonly termed. She described the benefits of common terminology, some negative consequences of incomparability, and difficulties with existing terminologies. She also discussed companion ASTM standards to E1384.
Lee Min Lau, M.D., Ph.D., 3M Health Information Systems
Dr. Lau considers PMRI to be both clinical and nonclinical data that describes the patient’s information, including patient’s name, gender, race, religion, encounter-related information such as insurance payer information, where the patient is located in the health care facility, who are the caregivers, the patient’s discharge information, etc. Clinical information is also included, such as complaint, past history, family history, behavioral findings. PMRI must be linked to each episode of care and longitudinal. Dr. Lau described comparability, its uses, and its prerequisites, and then focused on elements of PMRI. Dr. Lau prefers definitions to be formal, such as that used by SNOMED-RT. She stated that if you know the meaning of each data item, the PMRI become comparable; systems need not be identical to be comparable.
The 3M dictionary is a working product that uses the existing terminologies SNOMED, SNOMED-RT, UMLS, LOINC, NDCs from First Data Bank, ICD-9-CM, DRGs, CPT4, and HCPCS. 3M has encountered the problems of some vocabularies being classification schemes, while others are controlled vocabularies; the distinction between interface terminologies and a reference terminology; incompleteness; errors; violation of principles; issues with updates and versions; and licensing and cost.
Dr. Lau described 3M’s data dictionary, which incorporates multiple source vocabularies and remains faithful to informatics principles. It maps between all source vocabularies. 3M adds, but does not eliminate, entries. 3M also submits concepts and terms to the standard developing organizations interactively.
Dr. Lau emphasized the following points: (1) Informatics principles are currently understood; (2) the market appears to be moving towards de facto standards; and (3) the market is also moving toward coordination, but the government should still lead by example through projects such as GCPR.
Dr. Lau observed that the end user should be the focus of the discussion. It is important to bridge the gap between informatics and the 100-bed community hospital that worries about making a profit or not losing too much money, or getting information to the clinicians or administrators. The greatest problems are the issues of cost and lack of understanding; the cost of using non-legacy terminologies is significant. She raised the question of how to get hospitals to use LOINC and noted that educating potential users should be a key component of the mission. Government funding for research development applications may have an important role in the process.
Lt. Col. Mark Rubertone, M.D., Chief, U.S. Army Medical Surveillance Activity
Lt. Col. Rubertone offered a functional overview of the Defense Medical Surveillance System, an electronic system for medical surveillance decision support. He described medical surveillance as the routine, systematic collection, analysis, interpretation, and reporting of standardized, population-based data for the purposes of characterizing and encountering medical threats to a population’s health, well being, and performance. He described the population covered in the Defense Medical Epidemiological Database (DMED), the 6.5 million individuals who have served on active duty or reserves since 1990, and the types of data sources. Lt. Col. Rubertone presented a live demonstration of the Defense Medical Surveillance System. He noted that because the Activity collects data from more than two dozen separate sources, some data must be mapped together using a common data dictionary. He also demonstrated the DMED by remote access via the Internet, stating that the goal is to receive a result from a detailed query of the database in 30 seconds or less.
Dr. Lau asserted that her company’s customers do not mind the licensing and cost for the dictionary, but they balk at the additional fee for SNOMED or First Data Bank. She stated that customers want to continue to use the terms and codes they want. Dr. Lau cautioned that until and unless all legacy systems are replaced, the gap between interface and reference terminologies will persist. Dr. Lau explained that some differences in coding systems can be resolved and mapped conceptually; it is not necessary to arrive at a one-to-one mapping.
Ms. Guilfoy stated her unfamiliarity with any vendor that has used E1384 for modeling. She noted that E1384 is a medical record itself, not a technology format. As a concept model, there is no expectation that it would map to HL7. The medical record could be transmitted with HL7. Because E1384 is conceptual, a vendor would not develop a database founded on the data elements within E1384.
Dr. Lau described the different roles that the concept unique identifier (CUI) and UMLS play versus SNOMED-RT. She explained that customers want to see their list of diagnosis of problems, each with an ICD-9 code, a UMLS code, and a SNOMED code. Lt. Col. Rubertone explained that although the two systems he described overlap, they have different objectives.
The Workgroup identified their key observations on the presentations of the current meeting. Among the issues discussed were:
Ms. Amataykul described her intention to identify for the Workgroup the key issues, the goals associated with them, and then proposed recommendations that were either clear or that remained to be defined. The ideas described in the working document would be linked to their individual proponents during testimony. Members suggested that the issues be clustered under the broad focus areas, and that the information be presented in terms of the issue, the causes, the goal, and the recommendation.
After thanking the participants for their contributions, together with Mr. Blair, Dr. Cohn adjourned the meeting.
Mr. Blair convened hearings of the Workgroup on Computer-Based Patient Records of the National Committee on Vital and Health Statistics on October 14, 1999. He explained that the purpose of the group, pursuant to the mandate of the Health Insurance Portability and Accountability Act (HIPAA), is to study the issues related to the adoption and exchange of uniform data standards for patient medical record information (PMRI). The Workgroup, whose efforts are currently in the data-gathering stage, will report its recommendations to the Secretary of Health and Human Services (DHHS) by August 2000. Workgroup members and other participants introduced themselves.
John Kelly, Aetna U.S. Health Care
Mr. Kelly noted that Aetna, the largest private managed care organization in the country, cares for more than 20 million Americans, contracts with more than 300,000 practicing physicians, and provides health benefits for more than 100,000 employers. He described the uses of information and information technology (IT) at Aetna. Using traditional provider, membership, medical services, claims, pharmacy, laboratory results, and other information, Aetna processes such functions as eligibility, claims payment, and utilization review, and also analyzes quality of care. The data are collected in electronic form and stored in a single warehouse. Some data is distributed electronically, but most communications are conducted using paper.
Using technology, Aetna places members into clinical/risk categories to help manage their care. Higher-risk members are placed in disease management programs and results monitored. Patients in the congestive heart failure management program, for example, have used fewer acute care services, and medication compliance rates and patients’ self-assessed quality of life have improved. A current study of asthmatics newly on steroids is showing decreased use (by 50 percent) of inpatient and emergency hospital services. Diabetes patients also have shown positive impacts.
Mr. Kelly asserted that access to accurate, reliable, and timely information that is shared with physicians and members can have significant positive impact on their care. The ability to communicate quickly as a result of early analysis of electronic data speeds up improvements in members’ care. Mr. Kelly asserted that any effort to accelerate the use of the Internet or access to the Internet in physicians’ offices would be positive.
Barbara Demster, Compliance Officer, Healtheon Corp.
Ms. Demster referred to the prepared questions of the Workgroup. She described the evolving nature of PMRI to a virtual record with inputs from multiple legacy systems that address specialized functions. The virtual record permits integrating transactions from disparate systems in various types of display and presenting them transparently at any time to any authorized user with an Internet browser and proper security access identification.
Ms. Demster cited a 1999 study that shows that a large proportion of consumers are self-educating by accessing health care information online, at a time when physician time with patients is declining. This access can be problematical because much of the material on the Web is not peer reviewed. However, many consumers are keeping their own health records, and electronic information exchange facilitates that practice.
Ms. Demster commended the Workgroup’s recognition of the overlap and interdependence of the personal health record, population health record, and care giver health record. These records’ lack of standard formats and lexicons is a serious obstacle to collecting data for quality assessment. Data standards are critical to assure data reliability and integrity. A unique health identifier will aid in matching data. A Baylor study has shown that most patients want access to their own medical records, to add information to them, and to access the records in plain English. She stated that an estimated 10+ percent of IT budgets is used to plug holes created by nonconformable data standards.
Strong government leadership, Ms. Demster asserted, will be required to facilitate the development and usage of data standards to bring conformity in data quality to health care databases. Computers’ ability to isolate data elements and act on them dovetails with the need to develop minimum data sets for responding to common recurring information needs within the delivery system; this would both address the confidentiality issue and improve data capture and quality. Current privacy legislation does not encourage confidence in the privacy of health information divulged to care givers, and withheld health information compromises patient care. Comparable data is necessary for quality population data; data standards and protections are necessary to allow collection of quality population data. The public must be educated on how the Internet can be made safe for transmission of health data.
Mr. Kelly confirmed Dr. Fitzmaurice’s assumption that Aetna considers the risks of high expense, hospitalization, death, and loss of health status. Aetna is also interested in quality factors vis-à-vis standards of practice for particular clinical categories, such as immunizations, mammograms, other screening techniques. Aetna compares the services that individuals are receiving and widely used recommendations, and then intervenes on the basis of the information. Mr. Kelly pointed to increased incidence of mammograms, for example, when members are contacted and urged to be involved in their own care.
Blackford Middleton, M.D., Senior Vice President for Clinical Information, MedicaLogic, Inc.
Dr. Middleton submitted written testimony and responded orally to the Workgroup’s prepared questions. He stated that the PMRI consists of any information in a paper- or computer-based medical record, and referred to NCQA’s frameworks for information space for PMRI and for adoption of IT within the clinical environment. The historical perspective must now be expanded to include use of the Internet for information sharing.
Dr. Middleton cited CPRI (1996) regarding the necessity for comparable PMRI; for example, efficient and effective health care delivery, more successful identification of best practices and management, and the use of guidelines and other support-decision tools depend on accurate, detailed, and comparable clinical information. Comparable PMRI is useful in managing health care delivery, both financial and clinical, across the continuum of care and throughout a person’s lifetime, including such functions as data aggregation for individual patient care, information communication, and interoperability between disparate information systems. Dr. Middleton discussed the need to exchange comparable data within a health care enterprise, between health care enterprises, and in inter-business transactions mediated across the Internet. Data integration is the key ingredient to properly inform clinical decision making. He suggested a framework for a national health care information infrastructure that would facilitate appropriate information exchange among all parties in the health care delivery process.
Dr. Middleton stated that as the degree of structured data rises, the information utility rises, but the burden of structured data gathering may eclipse the utility of the information; if we cannot get the data, we cannot do anything with it. Over-coding produces a drop-off in usefulness of data. The industry should strive to model key information elements incrementally. MedicaLogic uses structured clinical lists derived from standard controlled terminologies, or standard codes, including ICD, CPT, ASTM, LOINC, SNOMED, and DSM. Some data is represented in proprietary format.
Dr. Middleton’s greatest challenge is to balance coded data requirements with usability of the project. Other challenges are: (1) to find a close matching structured data element from a standardized structured vocabulary—given a potential lack of parity in definition, granularity, or intended use—with the user’s requested term; (2) that no widely used vocabulary includes a robust information model; and (3) that the user interface technology and information model may be necessarily different from the reference terminology and information model used within the system. To address these problems, Dr. Middleton suggested a ranked, incremental approach by establishing: (1) national identifiers, (2) a minimal set of critical quality or outcome measures, (3) minimal required data sets to support derivation of the critical quality measures, (4) acceptable standard controlled vocabulary and codes for the minimum data set, and (5) acceptable methods to gather the minimum data set at the point of care. He urged the federal government to fund and coordinate a public-private effort dedicated to rapid development of controlled vocabulary standards, which, when placed in the public domain, will trigger widespread adoption.
Dr. Middleton noted that by using HL7, vendors can interface their systems with remote systems successfully. He stated that more work is necessary to make the HL7 interfaces easier to operate and maintain, and identified three incrementally sophisticated levels of interoperability: basic, structured, and semantic. He commented that although progress will be made to a single reference information model, a modest commitment to coordination around structured interoperability is desirable, too. Coordination initiatives are not sufficient, but a ranked framework for structured interoperability standards, building upon current HL7 efforts, and new efforts arising in the electronic health care arena could lead to benefits within four years. Identification of the critical quality measures to drive the business processes and have immediate impact on health care costs or quality, plus defining the minimal data sets and encouraging their adoption by major health care IT vendors, are key to this process. Dr. Middleton asserted that ongoing funding for academic research to discover the best methods and approach toward a common reference model and terminology must continue, but research more tightly coupled to industry in large-scale demonstration projects would produce results within a 5- to 10-year time frame.
Capt. Mayes stated that although semantic management is the most difficult, it actually reflects the business process being automated; unless one can demonstrate how specific coding can lead to better business outcomes (which involves semantic management), little motivation to adopt the system may be present. Some major health care companies have applied semantic management to standardized data representations.
Mr. Kelly stated that Aetna’s information models are generalizable and that focusing on utility is critical. He commented on the gap between what is possible and what is happening in this arena, perhaps due to the difficult cultural issue of identifying the benefits and the successful models, and then promoting them, particularly to the many physicians whose clinical practices are not yet automated.
Dr. Middleton noted that melding the two philosophical orientations of structured versus semantic interoperability, and then working incrementally, are critical to success; Capt. Mayes concurred. Mr. Kelly suggested means to help the health care system move forward in computerization: electronic, not paper, communications with physicians and access to the Internet. Having a computer with an Internet connection now is the equivalent of having a phone—you cannot do business without it—and the discussion will not move further until all physicians are hooked into the Internet system. Dr. Middleton acknowledged that the fundamental drivers for usable systems are not semantic interoperability, but rather whether the note supports the level of service billed, contains a problem list, and offers a method to write a prescription on the formulary. Having abandoned dictation for CPR, Dr. Middleton finds he can gather data at the point of care in a facile manner that supports the needs of both his clinical practice and the health care enterprise. Ms. Demster cited the problem of compliance regulations for fraud and abuse as an impediment to communicating health care data electronically.
In noting that Aetna is considering incentives for computerization, Mr. Kelly stated that the company already offers incentives for certain other interactions. He cited the need for a national campaign for these efforts, particularly in view of the key importance of the business case and of the cultural barriers that remain. Dr. Middleton cited the example of the Providence Health System, which has a quality incentive program targeted to physicians; if a physician knows that his or her compliance in gathering certain data will be measured, that physician is more likely to collect those data. He also believes the grassroots consumer movement will drive the medical industry to produce data that will enable them to make informed decisions, for instance, in their choice of a surgeon based on CABG surgery outcome statistics, as is now possible in New York State.
Dr. Zubeldia noted problems with the provision of free computers by Internet service providers and computer chains. He questioned whether looking for perfection in standards is an impediment to doing anything at all. Ms. Demster asserted that although productive things can be done already, time and money are wasted on fixes for inexact standards. Standardized data sets and outcomes research would positively impact quality of care as the bottom line. Mr. Kelly pointed out the value of simplification, focusing, and setting priorities, and Dr. Middleton concurred, emphasizing the necessity to work incrementally on relevant, attainable objectives. Dr. Middleton stated that ICD, CPT, DMS, and LOINC are sufficient for the present.
David Kibbe, M.D., Chief Executive Officer, Future HealthCare, Inc.
Dr. Kibbe explained that his organization is a start-up company whose business case depends entirely on clinical information exchange over the Internet for case management and disease management purposes. Its clients are mostly community-based hospitals, integrated delivery systems, and large physician networks. The flagship product canopy is designed to help hospital-based case managers coordinate the care of complex illness patients and high-risk patients with multiple diseases, and to share clinical information in sites throughout a given community. The company targets populations most in need of disease management, and its products comply with HCFA standards for security and confidentiality of Internet transmission and maintenance of patient identifiable information on the Internet.
Dr. Kibbe believes the health care industry is close to the “strategic inflection point,” where old ways of doing business are replaced by new methods as a critical mass of technological, cultural, and regulatory changes occur. He stated that “e-health” is both a product and a stimulus to new technology and its use. Transforming factors are substitution (e.g., online retail pharmacies whereby technology is the lower-cost “middle man”) and innovation (technology permits conduct of valuable new tasks, such as a personal health website). These factors are related in a spiral fashion.
Dr. Kibbe itemized major emerging sectors of the e-health marketplace: content providers (electronic medical publications, disease-specific specialty sites [e.g., <www.alzs.org>]), portals or intermediaries (information products and services such as free e-mail accounts, expert health care advice [e.g., DrKoop.com], chatroom support groups), e-commerce companies (retail health care products such as pharmaceuticals [e.g., drugstore.com]), connectivity integrators (linkage of major parties to health care transactions, including payers, employers, patients, providers, and physicians, whose business plans involve making administrative transactions more efficient as well as utilizing databases of information for marketing, research, and client-support purposes [e.g., Healtheon, Quintiles]), and Internet software applications providers (ASPs) (new model of health care software sales and distribution offering users the ability to access software programs physically running on servers anywhere in the world via secure Internet connection [e.g., Future HealthCare]).
The ASPs serve as host and security guarantor for health data and information residing on the company’s servers; personal health information is easily accessible by patients, their families, and all authorized caregivers. Currently e-health companies all interactively gather and exchange data and information from patients, physicians, hospitals, and family members.
Dr. Kibbe asserted that a set of standards that will make importing and exporting data from e-health sources inexpensively will help his business to become more successful and his clients more capable of providing better medical care. He looks forward to making the current system universally accepted, easier to access, and less costly. Universal adoption of HL7 and LOINC would be helpful, despite any shortcomings.
He urged government to lead by working with the private sector to shape the e-health environment, particularly to sponsor conferences of users and developers of e-health standards to define problems, issues, and research studies to show the value of uniformity of PMRI for business, patient management, and technology assessment uses. The government might purchase the rights to existing coding of vocabulary systems to make them available at low cost. Funding of standards can fill gaps identified at conferences and hearings, adopting PMRI standards for DHHS programs rather than inventing new codes, developing standards through government initiatives, promoting research in PMRI uses such as measures of quality and cost of care, chronic illness and emergency medicine registries, and establishing safe harbors for patient information accessible by patients and their families.
C. Peter Waegemann, Medical Record Institute, American National Standards Institute’s Health Care Informatics Standards Board, and Chair of the U.S. Technical Assistance Group (US TAG) of ISO TC215 on Health Informatics
Mr. Waegemann focused on international PMRI standards, important for travelers, in border areas, and to reflect the presence of e-health care enterprises, such as telemedicine, telecare, and telepharmacy services. Referring to a hand-out that lists international standards efforts, Mr. Waegemann highlighted Sweden’s standard for medical record structure, France’s parallel efforts with the U.S. regarding legislation similar to HIPAA, and 7TC251 reorganization.
Workgroups on the ISO TC215 standard on health informatics are focusing on a generic domain model for health care; ownership and access rights; a general country identifier for patients, devices, and caregivers; functional requirements within an architecture for a CPR; messaging; concepts and presentation; and international and Internet security issues. The first ISO standard is expected to be for public key infrastructure for health care. A Workgroup is focused on smart health cards; soon 150 million smart cards worldwide will be used for health care purposes.
Dr. Waegemann urged increased participation by government agencies and the general medical community in informatics to be involved in the US TAG part of the issue. The U.S. is the only country that does not fund ISO travel and expenses, which presents a challenge to secure a consistent set of volunteers at frequent meetings throughout the world to represent the U.S. interests.
Dr. Waegemann noted that the focus of the ISO discussions is on clinical data. Although it is difficult to address the administrative area, a Workgroup is discussing reimbursement and the need for administrative interchange between countries.
Dr. Kibbe stated that where cost is a barrier to using proprietary coding systems, the government might remove that obstacle in order to move the agenda forward. He confirmed the reasonableness of Dr. Yasnoff’s suggestion that the government negotiate with a certain coding system, saying, “We would like the rights to distribute this free of charge in exchange for a grant of X indexed by inflation in perpetuity to maintain and expand the coding system.” Dr. Kibbe made it clear that his suggestions were as nonspecific as possible.
Richard M. Peters, Jr., M.D., iTrust
Dr. Peters described the state of the current health care information systems industry, lending support to the discussions of the first panel members. He stated that the health care IT industry is “frozen” because of both Y2K concerns and no-confidence concerns expressed by the physician community about expensive and/or limited availability of practice-management solutions. Consolidation in the health care IT industry has been seriously problematic, and nondelivery of contracted software has negatively impacted customers’ desire to take risks.
Some of the market-based drivers for IT problems include radical changes in the health care market and the reimbursement process; sharply reduced hospital revenues in the first half of 1999, drying up funds for IT; increased mismanagement in health care plans, which also impacts IT decisions; stalled capitation risk contracting, except with Medicaid plans and Medicare; and loss of investor confidence in the IT sector in health care.
Dr. Peters described the flat stock market performance of the major players in the industry, which he expects to continue to limit the availability of capital in the health care IT industry. He expects investment activity in new companies by venture funds, investment banks, and institutional investors possibly to double its current $500 million level in 2000. The targets have been connectivity vendors (such as Healtheon and Point Share, among others) and consumer and provider health websites and portals, predominantly on health (such as DrKoop.com and Medscape, respectively). Currently about 150 either health care consumer-specific or physician-specific portals are on the Net or currently seeking funding. In the realm of online claims and EDI, several companies are currently funded and moving into the market.
Of about 25 companies currently operating, 3 are major players in the online pharmacy market. Dr. Peters expects rapid consolidation of that market. Several companies either operate or plan to operate a prescriptions online operation using the Palm Pilot. Internet practice management is another active area, as is the Internet electronic health record. He noted that MedicaLogic is reconfiguring itself as MedicaLogic.com and has an offering in place whereby, for a fixed price, doctors can get computers, software, and an Internet-based CPR that is deliverable; currently, about 15 companies are “in that space,” Dr. Peters noted. He expected capital to flow in the near term into administrative outsourcing providers.
In terms of technology, 60 percent of the companies Dr. Peters described are intranet companies and 20 percent are pure Internet companies, all doing secure transactions; 20 percent are still client server, but are trying to move into the Internet space. Client architecture ActiveX is about 20 percent of vendors, HTML about 60 percent (a declining sector), and XML (a growing sector). Server architecture is divided as follows: NT, 60 percent; Unix, 40 percent; and main frame, 0. In terms of security, 60 percent of the sector uses passwords; public key infrastructure (PKI), 40 percent. PKI is the rising sector, particularly with the ongoing certificate authority work.
Regarding pricing and revenue models, 90 percent of new vendors market on a subscription basis and 10 percent by licensing. Internet development costs are markedly lower than traditional development costs, a situation that tends to price out legacy vendors. Dr. Peters discussed the negative impact of these market features on legacy companies in the marketplace, and described factors that have impeded early adoption of the CPR.
Dr. Peters predicted that the primary driver in the next decade will be connectivity between payers and providers and/or payers and patients. Companies moving in this direction include Healtheon, Point Share, Health Portals, and others. Connectivity in a health infrastructure is emerging; if enough transactions can be tied together to form a significant amount of data on patients, this will be of value to the payers and providers. Value can be achieved, for example, just through laboratory results and claims; connectivity players, with their additional capital, bring energy to an inexpensive infrastructure (i.e., the Internet).
Application services, the outsourcing of practice management, is also emerging. The future will bring outsourcing of hospital information systems and hospital registration with increased Internet use and declining client servers. The market of the future, asserted Dr. Peters, will be driven by cost to the payers, employers, providers (hospitals and physicians), and consumers, and by functionality.
Dr. Peters discussed how the current state of the market affects the standards issues. He stated that: (1) e-health vendors are outrunning the standards efforts; (2) privacy and confidentiality are issues in terms of both security and aggregation of data for inappropriate uses; and (3) new vendors have already defined a medical and health record, and they are already providing operability. iTrust, MedicaLogic, and Healtheon all have systems that talk to other vendor systems with interoperability. Standards will help that process move forward.
Dr. Peters also noted that investment dollars, and not revenue and profit, are driving change in health care IT now, because there is no way to make money at present. Although the current market is good for upstarts, it is bad for the industry as a whole, because legacy vendors already have a lot of data, have done a lot of work, and know a great deal about what needs to be done.
Solutions suggested by Dr. Peters include: (1) Privacy and confidentiality must be defined and enforced; (2) standards organizations of the federal government must target key standards quickly and get them in place; (3) new e-health vendors must be brought into the standards process; (4) new e-health vendors should receive feedback from long-time industry players to achieve interoperability; and (5) national consensus should be achieved on access rights and ethical data aggregation and use. Dr. Peters asserted that sale of aggregated data should not be permitted, although this is the stated business plan of some companies currently doing business. It is also important to push greater standardization of eligibility data. Dr. Peters endorsed Dr. Middleton’s proposal to create a nonprofit entity to push terminology forward and settle the issues. He endorsed using ASTM and HL7 to drive the XML side and UCEN and TC215 to back up; use Multim, First Data Bank, and Lexicom for pharmacy terminology, and SNOMED and MEDSEN and the National Library of Medicine for the actual wording. Dr. Peters proposed individuals such as Blackford Middleton, Paul Tang, Tom McNamara, Peter Elkin, David Kibbe, and a physician from his organization to help in the process.
In summary, Dr. Peters asserted that investor money is driving a nontraditional view of the health care IT industry, driving new ways to do business. The standards process may be outstripped by individual efforts and may lead to exchanging data in ways not contributory to patient care and detrimental to privacy and security.
Mr. Blair and Dr. Fitzmaurice noted that Drs. Peters and Kibbe had been invited to testify because of their significant overview of the system and their message that the world is changing, not necessarily because of the specific examples they mentioned.
Judy Ozbolt, Ph.D., Professor of Nursing and Biomedical Informatics, Vanderbilt University, and Member of the Board of Directors and the Executive Committee, American Medical Informatics Association (AMIA)
Dr. Ozbolt referred to her prepared testimony and discussed its highlights. In June 1999, Vanderbilt University hosted a working conference that included representatives of all nursing vocabularies, experts on language and standards, federal officials, officers of professional associations, health care providers, representatives of the health informatics industry. The informatics industry, National Library of Medicine (NLM), and HRSA provided financial sponsorship of the conference.
Participants learned that users employ colloquial interface terminologies to enter and retrieve data; these are what users see and are familiar with. The participants acknowledged that different colloquial terminologies are needed for different specialties and different sites. But reference terminologies must contain all allowable terms to refer to the concepts in the practice, structured according to the semantic relationships among the concepts.
Dr. Ozbolt cited the point of view of NLM’s Stuart Nelson. Its Unified Medical Language System (UMLS) maps from concept to concept for all terminologies, but it will not be the long-term solution because of the independent evolution of the colloquial terminologies. A long-term solution could be to develop, collaboratively and incrementally, a reference terminology model that identifies and names the core concepts and their defining attributes, and depicts the semantic relationships among them. The process would be to define and develop a draft model from a set of terms, to test the model against another set of terms, and then critique, revise, and rewrite until a robust model is developed. Conference participants volunteered to work on task forces and will next convene at the November AMIA meeting. Members wish to be known as the Nursing Vocabulary Summit Group, and they hope to meet next summer at Vanderbilt. The Summit Group jointly with AMIA drafted a proposal for the development of a reference terminology model to the US TAG; it was approved and will be balloted at ISO in November. Summit Group members have also begun to work with European colleagues active in terminology.
Dr. Ozbolt requested that the government support this collaborative effort through a contract, rather than competing research grants. The main contract would go to an institution and primary investigator, who would take responsibility for overall coordination and specific research and subcontracts to the institutions of all the collaborating investigators. Working conferences and dissemination of results must be supported. Government organizations that have an interest in seeing this work progress include the NLM, National Institute for Nursing Research, AHCPR, and the Division of Nursing at HRSA (for dissemination).
In the near term, the public good would include an ISO standard reference terminology model for nursing, plus strengthened collaboration and consensus in matters of nursing terminology. Midterm benefits include a reference terminology for nursing, a set of standard nursing terms, and data and information systems that are colloquial but mapped to the reference terminology. This will enable comparable nursing data across sites, systems, and time, which will justify nursing data and public health databases, and enable the mining of clinical data to assess the quality and effectiveness of nursing care. The long-term benefits would be new knowledge of how nursing contributes to health outcomes, and development of incentives to have data-based and knowledge-based initiatives to improve care at reasonable cost.
Dr. Peters discussed data aggregation, which is currently accomplished by payers and at the clearinghouse level, but not yet by such companies as Healtheon, for example, for which aggregation is part of their business plan. The government is under greater scrutiny and negative pressure because the public is aware that the government can maintain aggregated health files on its citizens. It is vital that early steps be taken to define ethical practices in this arena.
Dr. Peters stated that large vendors do not perceive the need for standards, because they plan to dominate the market and there will be de facto standards. He commented that they believe Internet standards are not critical, because with XML, data is easily translated and the format is inconsequential. Dr. Peters disagrees with both these views, asserting that if the standard is there, vendors have shown no resistance to abiding by it.
Dr. Peters suggested that industry knowledge and contacts be used to draw into the discussions the smaller vendors. Dissemination of information would be necessary, broadly and through a few key individuals. In the computer industry there is strong interest in, and understanding of benefits from, Internet standards, through the World Wide Web consortium and the IETF.
Because ASPs that provide hospital information systems over the Internet are unregulated, Dr. Peters commented, he hoped that their customers would do significant due diligence on the vendors and insist on contractual restrictions on data usage.
Dr. Peters suggested that the government and industry fuel the standards process with cash, particularly travel costs to meetings. Beyond funding, coordination is required. In deciding on standards, a simple majority should rule if consensus cannot be achieved. A public-private cooperative initiative is better than either public or private, but obstacles will be encountered in devising the mechanism.
Dr. Kibbe stated that ethical standards are essential to avoid the danger of companies misusing data. Dr. Kibbe described an underlying cultural paradox, that of needing—but blocking—the master patient index effort, and of individuals and companies voluntarily giving up some confidentiality to gain information on a new service, for instance, to improve quality of care at an affordable price.
Mr. Waegemann stated that individuals on ISO Workgroups are not necessarily employees of foreign governments, but they are funded well by those governments. The U.S. does not accord the same priority or financial support to the volunteers working in the interests of the U.S. on health care information standards.
Dr. Ozbolt considers it difficult, but possible, to extend the process that has been invoked to develop a nursing terminology to other vocabulary efforts. Thanks to the Internet and to the fact that the nursing informatics community is international in nature, the U.S. effort has drawn into the process representatives from other countries.
Joan Kapusnik-Uner, Pharm.D., First Data Bank
Dr. Kapusnik-Uner explained that drug databases and knowledge bases are the backbone of pharmaceutical care. Successful computerization and interoperability have been achieved to a significant degree within the pharmacy profession and among other interested parties. She credited creative vendors and NCPDP standards in this effort.
First Data Bank is considered a trusted, proven source of drug-related information, Dr. Kapusnik-Uner asserted. The company’s continuous internal maintenance function permits constant update; it customizes views as necessary; customers receive frequent incremental updates; and the vocabulary is based on a sustainable business model. First Data Bank uses many drug-model databases and many products derived from them. The company has many international drug database and employs native language speakers to help the company understand the uniqueness of how drugs are used in other countries.
Dr. Kapusnik-Uner stated that the term “drug” has different meanings to different data users. The CPR has enabled package drug codes, codes for inactive manufactured drugs, and higher levels of abstraction that are user friendly. First Data Bank maintains databases for the conditions for which drugs are administered, possible side effects, labeled and unlabeled indications, drug disease contraindications, and adverse drug side effects; they map between the databases internally and to outside vocabularies, ICD-9, ICD-10, and ICD-10-CMS. The company is involved in UMLS and in another metathesaurus initiatives, has partnered with Kaiser to beta test with SNOMED-RT for decision support, and participates in consensus building with NCPDP and HL7.
Dr. Kapusnik-Uner stated that a comparable CPR would allow unambiguous documentation and interpretation of patient data as well as physician support across systems, and would allow flexible data aggregation in the present and future. The degree of comparability depends on the task and its criticality; as part of a requirements analysis, her software programmers would lock decision makers in a room until they agreed on the degree and level of interoperability required, and then scope the interoperability issues. Various parties require different levels of granularity, and use cases would be ranked.
As a vocabulary developer, First Data Bank creates cross-reference data to enable its customers to interoperate with the terminologies they use. Dr. Kapusnik-Uner considers SNOMED-RT to be superior to ICD9-CM.
She suggested that a major obstacle is clinicians’ lack of time allocated to use the CPR properly; adequate time is essential for success. She recommended that the government should help to define the CPR use cases for medical terminologies to set the bar at a minimal level, offer incentives for early adopters of new terminologies, and support terminology-related outcomes research, particularly related to the NLM’s work, and support vocabulary initiatives such as the enterprise vocabulary server by Apolon.
Mappings must be done by experts, not volunteers. NCPDP has had success with a defined structure for common definitions and terms. Version 3.0 of the RIM should help HL7 initiatives, especially the drug parts of the RIM. First Data Bank is converging to a single reference terminology model; doing so outside of the company may be achievable through HL7though valid values for the model should be left between business partners, users and domain experts, not HL7 volunteerism. She stated the expectation that UMLS 2000 will be a substantial improvement and enhancement.
Floyd Eisenberg, M.D., SMS
Dr. Eisenberg’s organization has served the IT needs of participants in the health industry for almost 30 years. He focused his remarks on data quality, accountability, and integrity, primarily on vocabulary and measurement issues from the perspective of knowledge integration through a software vendor and ASP. He asserted that SMS supports industry standard message formats and vocabulary across the entire line of integrated solutions for the health care industry. Dr. Eisenberg itemized problems that still exist, including the inefficiencies and costs associated with nonstandard interfaces. He stated that SMS is committed to assimilating standards as they are defined.
SMS views PMRI as electronically maintained information about an individual’s lifetime health and health care that replaces the paper medical record as the primary source of information for meeting all clinical, legal, and administrative requirements. The role of vocabulary models is standardizing messages for a comparable patient-centered medical record. Information that is meaningful, interoperable, and sharable is the objective; it must be standard, and it is key to achieving the comparable PMRI. The primary barrier to achieving a viable CPR is the lack of universal standards allowing data integration across diverse organizations.
Dr. Eisenberg suggested that key to encouraging clinician use of IT is use of the computer as sole source of required information. The documentation process must be fast and simple, and value to the user easily demonstrable. A standard in bar coding and text parsing methodology would enable health systems with products from several vendors to manage the data-entry process more cost-effectively. Adoption of a standard reference terminology would facilitate translation and linkage between clinical and administrative terminologies.
Dr. Eisenberg noted that efforts to unify terminology and to provide crosswalks are progressing, but that reference terminologies must be consistently maintained. Appropriate encoding enhances the integrity of data, and that sharing will engender use, which will engender quality improvements.
The federal government’s role in standardizing vocabulary should include the following: (1) provide leadership by adopting a single vocabulary standard that integrates existing terminologies; (2) place the adopted standard in the public domain; (3) encourage nationally recognized sources of evidence-based clinical care recommendations to publish accepted guidelines and protocols in standard format so they can be placed in operation as plug-ins; and (4) provide incentives for accrediting and regulatory agencies and organizations to develop vocabulary and problem-based clinical quality measures.
Gary Dickinson, Manager, Health Care Standards, Per-Se Technologies, Inc.
Mr. Dickinson stated that by delivering integrated financial and clinical software solutions, comprehensive business management services, and Internet-enabled connectivity, his company helps health care provider organizations, integrated delivery systems, and physician practices simultaneously to optimize quality of care delivered and profitability of business operations. Per-Se Technologies processes more than 80 million health care claims annually. Mr. Dickinson described his leadership roles in the national and international development process for standards on health care informatics.
Mr. Dickinson presented a detailed description of PMRI. For each individual patient or health plan member, PMRI forms an essential longitudinal chronicle of health status and interventions, describes the clinical and operational context and ascribes accountability, and serves as a foundation to the operations business record. Comparable PMRI ensures trust, confidence, and validity of assessment; ensures value of health services to consumers and purchasers; and permits population measures and trends, measures of clinical operation, and clinical findings and trends to be validated and/or contrasted. Comparability, accuracy, precision, consistency, continuity, completeness, and persistence are essential characteristics of data integrity and quality; the greater the assurance of data integrity and quality, the greater the level of trust.
The disparity in the current state of data quality, accountability, and integrity, and the ability to measure outcomes, quality, or performance, according to Mr. Dickinson, is due to the industry’s continuing reliance on multiple, disparate systems of data and functional architectures, lowest-common-denominator data interchange standards, use of interface engines and hubs, data translation to a transmitted form, remapping to alternate data structures, data type conversion, data aggregation summarization, and derivation. Moreover, the designation of accountable health care parties, agencies, and actions is often inconsistent or incomplete. Persistent evidence of accountability for data amendment, verification, translation, and remapping is often difficult to track, and complete auditability of data content is often difficult or impossible. Problems also exist in the area of data definition, such as the lack of common and uniform data definitions, and uneven deployment of coding classification schemes.
Data translation is often inaccurate, inconsistent, or incomplete—lapses that breach data authentication and validation conventions and expose data content to repudiation of authorship. Data is often aggregated, summarized, or derived incorrectly or inconsistently. Data integrity quality edits and reviews are often lacking or inadequate. Data is often embedded in text strings, not as discrete data elements. Insufficient knowledge by users of the vocabularies creates problems, and medical judgment permits varied conclusions from the same data.
In response to questions on fixes for these problems and others, Mr. Dickinson described a number of initiatives of the HL7 SIG on Accountability, Quality, and Performance, each focused on a key aspect of accountability and data quality and integrity. These include enumeration of the top data-quality issues and focus on the points of authority and reference on which to base the SIG objectives and work plan; a white paper outlining key principles, objectives, and areas of focus; compilation of a roster of health care parties who compose the trust and accountability constituency for health care and information; privacy and confidentiality issues, including a conceptual diagram of trust and information flows; and work on persistent context templates, among others.
Mr. Dickinson characterized private-sector progress as painfully slow, and he suggested that the government support existing initiatives with incentives for serious partners in the endeavor; the private sector is unlikely to succeed without government involvement and substantial incentives. He urged that the government champion a coalition of health industry organizations, including government regulatory and accreditation agencies, standards developing organizations, providers, payers, vendors, consumers, employers, and others. Prime participants should include government regulatory agencies, U.S. accreditation agencies, informatics standards coordination and developing organizations (NCHSB, ASC-12N, DICOM, HL7, NCPDP, and others), and vocabulary coding classification developers (LOINC, SNOMED, Reed, ICD, CPT, and others). He suggested a basic strategy to characterize the problems base, establish objectives and rationale, forge specific requirement statements, set measurable milestones and benchmarks for achievement, and develop implementable technical specifications as open consensus standards.
Mr. Dickinson noted that HIPAA legislation mandated use of ANSI standards; it remains to be decided which standards meet particular objectives. The government’s crucial role is to define the requirements by setting expectations both for what government agencies might do and procure, and for the community at large. Although extensive work has been undertaken in HL7, a strategy is lacking to determine needs, requirements, and objectives. He commented that the more industry input on setting priorities, the better. Dr. Eisenberg stated that the nursing terminology model appears useful to bring people together to identify issues and gaps, and to determine best ways to work collaboratively.
Dr. Kapusnik-Uner stated that several vocabularies, such as NDC, COSTART, MEDRA, and others, which are at different levels of abstraction and granularity, should be mapped in the UMLS and incorporated into the reference model.
She stated that the UMLS is perceived to be a viable solution, especially with its promised enhancements, because of its view of the Internet and its capabilities; she mentioned a Stanford University project in which CUIs are used as a meta search engine strategy for the Internet for medical information. Dr. Kapusnik-Uner considers the UMLS as a viable broker of clinically viable, point-of-care vocabularies. She currently is sorting out the role of First Data Bank vis-à-vis HL7’s efforts.
Mr. Blair questioned whether First Data Bank has pulled back from its efforts with HL7’s vocabulary coordination efforts. Dr. Kapusnik-Uner asserted that First Data Bank significantly contributed to the HL7 RIM version 3.0 in identifying all the levels of granularity of what a drug is for all relevant viewpoints for the CPR; the model needs to be decided on and then code sets mapped to those message segments and elements to interoperability. HL7 must make the cases for what is sharable and what is not, and the government must set minimum standards and minimum levels of sharable information.
Capt. Mayes commented that a major barrier to UMLS is that it is not available for unrestricted public use. Dr. Kapusnik-Uner suggested that the government might support licensing fees at some minimal data set level for a version of the UMLS for all vocabularies submitted to enable identification of the concepts in each vocabulary, the value set; those would be in the public domain. Capt. Mayes concurred.
Dr. Eisenberg stated that standards would expedite the process of designing systems for individual customers; a starter set and available information would allow use of systems more quickly. The problem is devising a pricing structure for current knowledge vendors. He stated his personal view that a hierarchy of licensing fees might be one solution. Dr. Kapusnik-Uner highlighted the high level of well educated content experts needed to build and maintain a drug vocabulary. Dr. Cohn stated that the Workgroup and the Committee understand that the issue of equitability must be dealt with, but he continued to raise the issue of the increasingly complex world of information and what it will mean and look like. Dr. Kapusnik-Uner suggested that if HL7 messages are clear enough and the data elements are described within those messages, if a vendor wishes to plug in either a free set of codes or a set from another vendor, interoperability would be up to mappings that occur through HL7. She explained that mappings would be to the model, not vocabulary to vocabulary, so each vocabulary would have an HL7 mapping to the message segments. Dr. Kapusnik-Uner suggested that the best way to help this situation occur would be to have a think-tank consensus conference on the Vanderbilt model. She stated that the HL7 chairs also must agree on the short- and long-term goals of the committee. She drew a parallel between the objective of the Vanderbilt conference to what HL7’s Vocabulary SIG Technical Committee has done—to improve the model to define the message segments and elements.
Dr. Eisenberg clarified that his recommendation to encourage sources of evidence-based clinical care to publish guidelines in a standard, plug-in format could be accomplished using Arden syntax and/or the GLAF. The vocabulary must be sufficiently standard. He stated that clinical decision support at the point of care on recommendations is an important reason for creating a CPR.
Dr. Kapusnik-Uner described the relationship of J codes to NDC codes. J codes describe drugs administered in an outpatient setting—the package actually pulled from the shelf as distinct from the original order. Bar coding would constitute an easy fix. Dr. Zubeldia clarified that the NCPC codes cover the administration of the drug and the NDC code is the drug only. Dr. Kapusnik-Uner stated that First Data Bank mapping is done between them now. A comment from the audience indicated that other companies (such as Multim or MedicaLogic) also do the mapping.
Workgroup members briefly discussed draft agendas for the December 9-10 and January 31-February 1 meetings. Mr. Blair noted that the full Committee had decided on September 27 to restore the privacy, confidentiality, and security focus.
According to the draft schedule/agenda, at the January 31-February 1 meeting (in Washington, D.C.), the Workgroup will compile recommendations to present to the full Committee at its February 23-24 meeting. The current meeting ends the information-gathering stage. Ms. Amataykul and Mr. Blair have begun to converge issues, assumptions, and recommendations, particularly for message format standards and medical technologies. A conference call was tentatively scheduled for November to discuss updates to the draft documents presented to the Workgroup members.
The Workgroup will meet in Washington, D.C., on December 9-10 to discuss the issues identified and a more detailed outline of the final report. The members would review the first draft of the final report, including recommendations, in a January 10 conference call. The report would be updated, and members would develop a second draft at the January 31-February 1 meeting. This second draft would be critiqued and then approved by the Workgroup and then reviewed by the full Committee at the end of February. The Subcommittee on Security will also meet on December 10.
Dr. Cohn suggested that discussions about the business case focus on the compelling issues for which recommendations are required, not just the micro issues of funding conference attendance. Mr. Blair and Dr. Cohn noted that Dr. Middleton and others said the same thing in their testimony; Dr. Cohn urged that the views of these individuals be extracted to help key into the issues and actions on everything else. Ms. Amataykul suggested that she prepare a convergence document on business cases.
Dr. Cohn suggested adding to the report a discussion of the driving forces for standards just below the executive summary and prior to the presentation of the focus areas, in order to set the context. Dr. Fitzmaurice suggested that the discussion of these driving forces include the need for business profits, improved revenue, reduced costs, social needs such as improved efficiency in the health system, and incentives. The discussion also would encompass why it is important and to whom, and who will do something about it. The government would be one driving force. Dr. Yasnoff suggested using the title “Rationale” for that section. Dr. Cohn stated that the business case could include both tangible and intangible reasons. Dr. Yasnoff clarified that benefits to public health are not part of the business case, but nevertheless is a compelling reason for standards. Members agreed that this overarching discussion of incentives or rationales should not replace the discussion of the business case as a focus area.
Members discussed the content of the summary section, noting that most people will read the summary and ignore the main body of the report; on the contents page of the printed document, members agreed, the Executive Summary should be one line. Recommendations will be included in the summary. Dr. Zubeldia recommended including in the appendix some testimony before the Privacy Workgroup, plus the privacy, confidentiality, and security recommendations. Mr. Blair suggested also adding the proposal for the NHII.
Ms. Amataykul led the discussion of the convergence documents, explaining that she had compiled the most important issues in the areas of medical terminology and message format standards, based heavily on the excellent summaries previously prepared, and also basic assumptions and issues.
The current discussion paper included five basic assumptions relative to medical terminologies: (1) Convergence and interoperability among medical terminologies are necessary to be used as components of uniform data standards for PMRI; this will ultimately improve the quality of patient care; (2) key characteristics, attributes, and imperatives should be adopted by developers; (3) standard medical terminology should support collection of health care data once at the point of origin, and other uses of PMRI should be derived therefrom; (4) medical terminology standards should relate directly to health care message format standards; and (5) the role of government, if any, is to resolve the disparity of usage fees charged by different medical terminology developers. The two recommendations were: (1) The government should promote greater vendor and user acceptance of standard medical terminologies by officially adopting them within government agencies (e.g., the GCPR Project) through requiring their use and complying with minimum data sets, encouraging their adoption for use in performance measurement by accrediting agencies and other organizations; and (2) the government should develop policies and mechanisms to assist medical terminology developers develop, maintain, coordinate, and map national/international medical terminologies in an open process.
Mr. Blair stated that by December 9, a list should be created of what the speakers have recommended during the hearings. Dr. Yasnoff observed that the recommendations are not operationally oriented, but Ms. Amataykul reminded the members that this list is a high-level summary. Members identified issues with the meanings and focus of the assumptions, and resolved some ambiguities. Members expressed their wish to separate issues from assumptions.
Dr. Ferrans suggested as an assumption that medical terminologies need to be available at a reasonable cost. He suggested as an issue whether or not incentives are necessary to encourage users. Dr. Fitzmaurice suggested the issues of cost of developing, maintaining, and implementing terminologies, and the risk of their becoming obsolete. Mr. Blair suggested using the term “observations from testimony.” Dr. Ferrans suggested another observation: If people would use terminologies, there would be great benefit from it. Members added another: Vendors would adopt a standard that they feel is the right standard. Dr. Fitzmaurice noted that the government can provide an incentive by stabilizing a fluctuating field. Dr. Ferrans stated that the fact that everyone appears to want standards should fall under either conclusions or recommendations. Dr. Zubeldia suggested that the charge is not for NCVHS to create standards, but to propose parameters for the adoption of standards to be adopted. Mr. Blair emphasized the international nature of standards. Dr. Yasnoff added areas where government funding has been recommended, including travel, sponsoring research, sponsoring conferences. Dr. Ferrans suggested distinguishing among government facilitation, encouragement, and spending.
Dr. Fitzmaurice suggested framing subsequent discussions on each focus area in more specific components, such as issue, barrier, general recommendation, specific recommendation.
The meeting was recessed and reconvened on Friday, October 15, 1999. Mr. Blair again called the meeting to order and presented the context of the meeting for the benefit of the Internet audience.
Helene M. Guilfoy, ASTM
Ms. Guilfoy discussed the ASTM and particularly standard E1384, content and structure of a CPR. All other standards relate to this one and have attempted to incorporate consistent information and structure based on the E1384 structure. The CPR defines all health-related information about a patient, requests practitioner orders, includes patient identification information including demographics and legal parameters. Information in a CPR must be seamless and consistent over time, relevant, available in a common record structure, and commonly termed. Common terminology will help to bridge gaps between the experiences of practitioners in different disciplines without the need for translation and the occasion of misunderstood information. Ms. Guilfoy described the consequences of comparable PMRI using the examples of different pain scales in a hospital, whose use can result in improper administration of pain medication, and the crash of the Mars probe because of metric/English scale incompatibility.
Ms. Guilfoy cited difficulties with existing terminologies: inconsistent definitions and terms, such as patient status/classification/type of patient (but in X12.837, patient status is discharge information in a hospital information system); inconsistent values, such as marital status captured differently in four different established standards, which require translation when transferring data from one system to another.
Ms. Guilfoy discussed companion ASTM standards to E1384, noting that by using the data elements in 17.44, based on national guidelines for emergency care, it would be possible to treat the patient in the field, bring the patient into a hospital, admit the patient, and then discharge the patient to physical care or ambulatory service. The information would be in the same format consistently in the CPR. As ASTM was developing the standard for 17.44, gaps were discovered and then filled in E1384. ASTM is working on companion standards for pre- and post-natal views of the CPR that will permit a neonatal record that is consistent, longitudinal, and with a common format.
· Lee Min Lau, M.D., Ph.D., 3M Health Information Systems
Dr. Lau, whose work is mainly on the 3M Health Care Data Dictionary, focused her testimony on terminology. 3M considers PMRI to be both clinical and nonclinical data that describes the patient’s information, including patient’s name, gender, race, religion, encounter-related information such as insurance payer information, where the patient is located in the health care facility, who are the caregivers, the patient’s discharge information, etc. Clinical information is also included, such as complaint, past history, family history, and behavioral findings. PMRI must be linked to each episode of care and be longitudinal. PMRI must be comparable for a patient across all episodes of care; for an institution, comparable PMRI should be across all patients; for a country, across institutions, patients, locations, and time; internationally, across countries. Comparable PMRI is required for individual patient care and population-based functions such as outcomes research and cost assessment. To achieve comparability, data must be defined accurately and precisely; Dr. Lau differentiated between the data itself and the definition.
Dr. Lau discussed definitions of a piece of PMRI, which may include several levels of definition and should include synonyms and relationships among the data items. Granularity also should be taken into account. Meanings of each data item must also be addressed precisely and accurately. Dr. Lau prefers definitions to be formal, such as those used by SNOMED-RT. If one knows the meaning of each data item, the PMRI become comparable; systems need not be identical to be comparable.
The 3M dictionary is a working product that uses the existing terminologies SNOMED, SNOMED-RT, UMLS, LOINC, NDCs from First Data Bank, ICD-9-CM, DRGs, CPT4, and HCPCS. 3M encounters the problem of some vocabularies being classification schemes (which deal with discrete concepts), while others are controlled vocabularies (which deal with groups of concepts). Another issue is interface terminologies versus a reference terminology, and another is incompleteness (e.g., LOINC, which does not include all possible lab results; First Data Bank’s single-hierarchical presentation of aspirin). Other issues are errors, violation of principles, updates and versions, and licensing and cost.
3M incorporates multiple source vocabularies in the dictionary, remaining faithful to informatics principles and aiming at being practical and useful for its customers, by using a solution modeled after the UMLS—a map between each source vocabulary. 3M also submits concepts and terms to the standard developing organizations interactively. Dr. Lau illustrated her discussion using the example of interfacing a customer’s legacy lab system to the 3M CDR. The data dictionary is the point of entry; all the legacy system’s terms and concepts must be in the dictionary for the system to work. Coordination among terminologies requires crosswalks and mappings among terminologies, common definitions of concepts, and convergence (conceptual, not necessarily physical) to a single reference terminology or model.
Dr. Lau emphasized the following points: (1) Informatics principles are currently understood; (2) the market appears to be moving towards de facto standards, for instance, in lab results; (3) LOINC is well known and widely used; and (4) the market is also moving toward coordination (for example, LOINC and SNOMED), but the government should still lead by example through projects such as GCPR and CHCS2.
Dr. Lau observed that the end user should be the focus of the discussion. It is important to bridge the gap between informatics and the 100-bed community hospital that worries about making a profit or not losing too much money, or getting information to the clinicians or administrators. The greatest problems are the issues of cost and lack of understanding; the cost of using nonlegacy terminologies is significant. He raised the question of how to get hospitals to use LOINC and noted that educating potential users should be a key component of the mission. Government funding for research development applications may have an important role in the process.
Lt. Col. Mark Rubertone, M.D., Chief, U.S. Army Medical Surveillance Activity
Lt. Col. Rubertone spoke from the perspective of a developer of a system that capitalizes on pre-existing CPRs and standardization, and a developer of an integrated system that uses that information. He offered a functional overview of the Defense Medical Surveillance System, an electronic system for medical surveillance decision support.
The U.S. Army began to track HIV infections within the military in the 1980s, and in 1992, other diseases of military importance began to be tracked. Secure funding for the system is provided. The Defense Medical Epidemiology Database (DMED) was established in 1996 and later standardized across the services. In 1998, data sources expanded, and DMED version 2 was released in 1999.
Lt. Col. Rubertone described medical surveillance as the routine, systematic collection, analysis, interpretation, and reporting of standardized, population-based data for the purposes of characterizing and encountering medical threats to a population’s health, well being, and performance. He emphasized the importance of the routine and systematic collection of data, of getting data to decision and policy makers, and of the focus on a population, not just patients. The population in the DMED consists of the 6.5 million individuals who have served on active duty or reserves since 1990. Data is collected from pre-induction to post-discharge. Data sources include the pre-induction physical (including HIV test), Defense Manpower Data Center (for assignment and deployment information), hospitalizations, ambulatory care data, reportable diseases, health risk assessments, casualty, and environmental exposure data.
Lt. Col. Rubertone presented a live demonstration of the Defense Medical Surveillance System. He noted that because the Activity collects data from more than two dozen separate sources, some data must be mapped together using the Activity’s common data dictionary. He also demonstrated the DMED by remote access via the Internet, stating that the goal is to receive a result in 30 seconds or less.
Dr. Lau asserted that her company’s customers do not mind the licensing and cost for the dictionary, because they expect to contribute their share, but they balk at an additional fee for SNOMED or First Data Bank. She stated that the user community assumes that the informatics community will “do it right” regarding reference terminologies. Customers are more concerned that they will be able to continue to use the terms and codes they want; Dr. Lau stated that as reference terminologies mature, they will improve. She cautioned, however, that until and unless all legacy systems are replaced, the gap between interface and reference terminologies will persist. Dr. Lau stated that her company licensed the pharmaceutical database from First Data Bank, which has NDCs, which are not proprietary, but the links are. Dr. McDonald noted that CPT is licensed, but the fee is based on the number of people doing the coding.
Lt. Col. Rubertone explained that the DMED system’s fast response time reflects optimization in Oracle. The hardware is a super mini-computer by Hewlett-Packard.
Dr. Zubeldia noted the differences in viewpoint between arriving at a standard terminology versus accepting differences that can be resolved with interface terminologies linked to reference terminologies. Dr. Lau explained that some differences in coding systems can be resolved and mapped conceptually; it is not necessary to arrive at a one-to-one mapping.
Ms. Guilfoy stated her unfamiliarity with any vendor that has used E1384 for modeling. She used it in developing concepts for a clinical documentation model to enable four different hospitals to put their clinical documentation tool together. To actualizing the concepts within E1384, it would be possible to elicit from the CPR information on a specific lab test. She noted that E1384 is a medical record itself, not a technology format. As a concept model, there is no expectation that it would map to HL7. She stated that the medical record could be transmitted with HL7. Because E1384 is conceptual, a vendor would not develop a database founded on the data elements within E1384; its title is “concept and structure,” not record layout, which the vendor would need to implement a CPR.
Dr. Lau described the different roles that the concept unique identifier (CUI) and UMLS play versus SNOMED-RT. She explained that customers wanted to see their list of diagnosis of problems, each with an ICD-9 code, a UMLS code, and SNOMED. The HDD make a judgment call when the codes do not agree with each other.
Lt. Col. Rubertone explained that although the two systems he described overlap, they have different objectives.
Ms. Guilfoy stated that membership in ASTM costs $65 annually, which entitles a member to access to every standard within ASTM, including E1384, E16.33, and E17.44—she considered this to be close to nonproprietary. She would like to see vendors adopt a standardized data model in constructing their CPRs. Dr. Lau differentiated between an actual information model versus database schema, because not every vendor has both. There are no public domains, and the closest vendors get to sharing is to ensure that their information will support messaging standards such as HL7. 3M would be interested in supporting and exploring this kind of value because it contributes to comparability. Dr. Fitzmaurice noted that it would not be hard for the government to build a library and put it in the public domain. Capt. Mayes commented that progress would be made even if agreement were reached on common definitions of the entities. The relationships might be considered the value added within a particular system, so an entire model would not need to be exposed.
The Workgroup identified their key observations on the presentations of the current meeting. Dr. Cohn noted that not enough providers are plugged into the environment; some have billing systems, but some don’t even have that. Incentives might help get more providers plugged in. According to the HIPAA legislation, this discussion extends beyond standards to administrative simplification, decreasing costs, and increasing the quality of care. Promoting standards without a context of increased computerization does not fulfill HIPAA requirements. Dr. Fitzmaurice concurred, saying it is not just increased computerization, but increased communication. Aetna and MedicaLogic both said they want computers on doctors’ desks to communicate changes in guidelines, coverage, and benefits, and to enable doctors to receive decision support when they ordered drugs—all to improve quality of care and hold down costs of care. He stated that he did not hear discussion about increasing access.
Members clarified that the communication and sharing aspects of the information, rather than the underlying structure of the local system (e.g., CPR), is the focus of the issue.
Capt. Mayes pointed out the importance of recognizing, as Dr. Peters noted, that venture capitalists are driving creation of new systems, not the health care professions, as has been assumed in the discussions thus far. Dr. Ferrans expressed concern about the reselling of data and about privacy, security, confidentiality, and data integrity and quality issues. He agreed with Dr. Cohn that because enough people are not “plugged in,” it is relevant to the mission of the Workgroup to examine this issue in the context of electronic exchange of PMRI. Dr. Zubeldia suggested that increased urgency may be felt now in light of Dr. Peters’s information about the movement of venture capital into e-health market development. Capt. Mayes commented that while delivery of individual health care is local, much of the industry is internationalized (e.g., outsourcing of medical transcription), and recognizing the need for international standards and regulation is critical. Dr. Yasnoff noted that given the degraded financial situation of the health IT industry and health plans, it is important to remove unnecessary costs by fostering improved standardization. Dr. Fitzmaurice concurred with Capt. Mayes’s comment on international standards, adding that products (e.g., pharmaceuticals) are now flowing internationally, not just information. Good international standards will prevent a situation where standards are developed that are not compatible with U.S. medical practice.
Mr. Blair raised the issue of need for better coordination and interoperability for terminology among the drug knowledge base systems. Capt. Mayes noted that that situation was true for all terminologies.
Dr. McDonald cautioned against assuming that ISO and “international” are synonymous. He stated that U.S. interests are the most dominant in the world, and the ASTM, HL7, and DICOM are international. The ISO OSI versus Internet lesson is instructive, he commented.
Dr. Zubeldia raised the issue of standardization of administrative data. Vendors have developed proprietary systems, but HL7 communication standards are not tied to data content. Vendors might cooperate in arriving at a communication standard with data. Dr. McDonald agreed that the distinction of communication versus building the system as a standard is important, noting, however, that although no one tells companies how to build phones, all phones plug into any jack. He stated that medical vendors desperately want a standardized vocabulary and standardized messages; they would prefer not to expend extra time and labor. Dr. Lau concurred that, as a vendor, 3M would be happy to use standards; although the product is proprietary, the vocabularies and terminologies are not proprietary. 3M is neutral: If the source vocabularies are in the public domain, no one pays, but if they are not in the public domain, the end user (not the vendor) pays. With standards, vendors can support the customer’s need by working on the interface part and not the “soft” part. Dr. Cohn mentioned that nothing about standards requires that they be free; Dr. McDonald noted that HIPAA makes some provision about cost.
Ms. Guilfoy stated that vendors keep the record format proprietary, but the content should be standardized. If it could be determined what data is important to capture, those requirements could be submitted to vendors, who could fashion it as they saw fit, as long as it is available to health care providers. Dr. Ferrans noted that reduced licensing fees will accelerate more widespread adoption by end users and people capturing the proper mix of coded data.
Dr. McDonald confirmed Dr. Cohn’s understanding that the reference information model developed by HL7 specifies content. Dr. McDonald clarified distinctions between standards that are enabling (e.g., CD-ROMs on which any kind of music can be burned) and those that are prescriptive (e.g., ethnicity).
Capt. Mayes suggested the utility of analyzing a range of all the incentives that are potentially available through the federal government, from legislation to measures made possible by the discretion of a program manager. Dr. McDonald highlighted the ability and power of the government to adopt standards, which would serve as an incentive for the private sector to adopt those standards. He asserted that it probably will be more difficult to force standards in the broad medical area than for the administrative area. He also argued against selecting and supporting particular people to attend standards conferences, although he suggested an educational grant mechanism might work.
Mr. Blair explained his hope that by identifying the issues and observations in the current manner, it will be easier in the future to address these issues and make well-considered recommendations. Dr. Cohn concurred, and Mr. Blair suggested reviewing Dr. Middleton’s recommendations for the organization of the report.
Dr. McDonald identified two tensions: (1) System vendors prefer that vocabulary standards be free, and the merged vocabulary vendors have business plans to capitalize on their expensive investments, and (2) in the U.S., the market ultimately determines the price and no monopoly rules price or content. Dr. McDonald cautioned against moving toward a single standard with one solution that might lead to a monopoly and be disastrous in unanticipated ways—a uniform standard must be minimal or low cost. He noted also that regulated monopolies were guaranteed a profit, but not required to produce efficiently.
Mr. Blair suggested that the Workgroup focus on identifying the criteria for standards they would recommend be adopted, which would be based on function and the ability to meet market needs. If several involve user-fee issues, those problems should be addressed in a separate step. Dr. McDonald suggested that the process be done in parallel. He noted that with ACRI, the codes are free, but the connections and/or additional data are not.
Capt. Mayes urged the Workgroup to work toward creating a detailed legislative proposal. Other members suggested that they make specific recommendations that could be turned into law, and others noted that the process established by Mr. Blair to winnow down the issues and observations is a productive way to work. Dr. Cohn stated that the recommendations will be as specific as consensus among the Workgroup permits.
Dr. McDonald cautioned against considering the Internet to be a message format; discussions of message format relate to the specificity required in a communication. Mr. Blair stated his view that HL7 generally is successful in the marketplace, addressing most needs, and supported by most vendors and providers as users. He questioned whether communication via the Internet changed the needs for health care message format standards or HL7. Dr. McDonald summarized his response by saying, “Content is the content is the content, and HL7 does a pretty good job with the clinical content.” The current functionality does not answer all problems, such as denying access to John Smith’s CPR by his wife or employer, but these problems do not relate to the basic messages. Dr. Zubeldia added that for transport of the information, HL7 is a great vehicle, computer to computer and application to application; but consumers who wish to see only the information, use of HL7 is irrelevant. Dr. McDonald stated that it matters if “we start all over again.” Mr. Blair concurred with Dr. Zubeldia’s suggestion that the Internet be considered a pipeline. Dr. McDonald concurred with Mr. Blair’s assumption that the Internet is not changing how one looks at message format standards for PMRI; HL7 will continue to meet those needs within the pipeline of TC/IP.
Dr. Ferrans raised the issue that a voluntary message format standard that seems to meet the needs is already widely adopted, and he questioned the impact of these givens: Do issues remain regarding HL7 as a message format standard, and what (if any) role should the government play in facilitating or improving it? Dr. Zubeldia suggested that the role is to say officially that HL7 is the recommended standard, but he noted that some Europeans will suggest selecting EDIFACT. Dr. Cohn noted that it was not stated that, regarding other HIPAA standards, all X12 transactions were standards; the Workgroup could say that certain parts are a standard, or otherwise. Dr. Ferrans added that a message format standard can be viewed as an EDI standard; like X12, rulemaking processes have been established for EDI standards. Dr. Zubeldia commented that X12 as a standard has been in force for a long time. HIPAA has added the standard implementation guide, which, because the standard is insufficient on its own, will “make it happen with the standard data content” and requirements.
Dr. McDonald stated that some of the currently mandated administrative transactions could be described as a data set that indicates not only how to send it but what must be sent. He used the example of a serum potassium test. On the other hand, he suggested, if something is going to be sent, the vocabulary must be completely specified. The challenge is tougher to completely specify all vocabularies in health care, but good standard vocabularies will make messages more useful than without them. Dr. Ferrans added that even with HL7, significant mapping issues remain, and Dr. McDonald stated that the mapping issues represent a great problem.
Dr. Cohn stated that for a standard to be implemented nationally, it is helpful if a great deal of specificity is related to it. Dr. McDonald drew the distinction between prescriptive and clear standards. Dr. McDonald stated he sees a need for both HL7’s reference information model and ASTM E1384. The E1384 summarizes all the things you can enumerate that might go into a medical record and how the things might be grouped. The data model is completely technical, and it would ensure that the message would always be coherent. Dr. McDonald asserted that HL7 is “coming close to happiness.”
Dr. Zubeldia noted HCFA’s experience with an implementation guide for Medicaid Part B developed by experts. The private sector implemented the standard for government use only, and then the private sector used the government-developed implementation guide as the model for the commercial part of the 837. He suggested that this model be used with government incentives for HL7. Capt. Mayes seconded the utility of this approach, stating that CDC is working with HL7 to define a message standard for immunization, and GCPR has decided to use HL7 as the basis for its activities, with the explicit idea that it will be made publicly available. For message standards, he added, the general consensus is that in the clinical arena, HL7 is a dominant player both domestically and internationally. This is a way to say that the Workgroup recommends that government projects that require public reporting of clinical information adopt HL7 as the message format, and that this is done by building implementation guides through the HL7 process. The recommendation should be to expand on what is already being done successfully by CDC and GCPR. Dr. Zubeldia added that the government can say we will only medical systems that are compliant with implementation guides developed for the government.” Dr. McDonald asserted that that would be the most powerful thing to do. Capt. Mayes added the option of interfacing between different systems. Dr. Yasnoff discussed the need for a certification mechanism that says, “We have taken messages from this system, and we certify that these are really the HL7 messages” that do not have extraneous things in them. Dr. Ferrans concurred, noting the example of multiple, flawed interpretations of DICOM that have hampered true interoperability. Dr. Zubeldia cautioned that if the government adopts a standard for others to use, a backlash may result; leading by example, whereby the government says, “We’re adopting a standard for us to use, and if you want to interface with us, you must use that standard,” is preferable. Dr. Cohn summarized the discussion, identifying the important role of implementation guides in the process and the issue of compliance with the standard.
Regarding the role of federal funding, Dr. McDonald stated that a little money goes a long way. He noted that those who are dedicated to the idea of standards currently are involved in the arena. Development of the Internet is a good model for development of these standards. Capt. Mayes stated that he supports a grant/contract approach. Dr. McDonald stated that a funding mechanism would be welcome for maintaining a total vocabulary, such as SNOMED, which would solve the availability problem. Capt. Mayes added that funding for certification must be available as well; participants suggested as models NIST or ENAC.
Mr. Blair opened a discussion about data quality. He noted that suggestions were made to integrate data quality mechanisms into the message development format in HL7. The Joint Commission and NCQA are expected to provide incentives for improved data quality to enable them to do performance measures. Capt. Mayes suggested the value of acknowledging the complexity of the data quality concept and of developing a more formalized view of data quality, especially given its relationship to security and to performance measurement. Dr. Cohn mentioned other aspects of data quality, precision of the definition and accuracy of the data.
Ms. Amataykul described her intention to identify for the Workgroup the key issues, the goals associated with them, and then proposed recommendations that were either clear or that remained to be defined. The ideas described in the working document would be linked to their individual proponents during testimony. Members suggested that the issues be clustered under the broad focus areas, and that the information be presented in terms of the issue, cause, goal, and recommendation. Mr. Blair reminded participants that this step is the beginning of an iterative process. He stated that it would not be necessary for Ms. Amataykul to focus to a great extent on the recommendations, since those were likely to emerge through discussions.
Dr. Cohn raised again the issue of guidance for available government incentives. After thanking the participants for their contributions, together with Mr. Blair, Dr. Cohn adjourned the meeting at 12:35 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 5/30/00
____________________________________________________________