The Work Group on Computer-Based Patient Records of the National Committee on Vital and Health Statistics met on March 29-30, 1999, at the Hubert H. Humphrey Building, Washington, D.C. Present:
The Work Group on Computer-Based Patient Records held a two-day hearing on message format standards and data quality, accountability and integrity. They heard from a total of four panels, three of them on the first topic.
The members of panels 1-3 were sent a set of questions about the following topics, to guide their presentations:
Panel 4 was asked about:
All panels were asked to comment on:
Dr. Beeler described the genesis and development of HL7 and its current work on Version 3, and he outlined the merits of this new version. He was the first of several panelists to recommend that the government "be supportive and not directive." He also was one of many to encourage more federal funding for standards development efforts. In addition, he stressed that semantic interoperability should be technology-neutral.
Mr. Shakir specializes in modeling, and he stressed its potential as a collaborative effort. He described the reference information model (RIM) developed by HL7 and noted that proprietary attitudes are an impediment to model development. He recommended that there be a single reference model to which others can be mapped.
The group discussed the idea of government-funded extramural research on harmonization, a proposition that was not enthusiastically received. The panelists recommended that the government simply provide further support for the work that is already underway.
Dr. Dolin's topic was XML technology in HL7's PMRI exchanges. XML embeds tags in documents that tell a computer how to process the document. The declaration of this arrangement is placed in a document-type declaration (DTD). He noted that XML has the same usefulness for medical publishing as SGML, and it can be rapidly deployed over the Internet through conversion to HTML. It has been accepted as the transfer syntax for HL7 since an interoperability demonstration in early 1999, and it will be used for Version 3.
XML provides a syntax for HL7 semantics, tying to the shared information model that is central to the patient record architecture (PRA). Dr. Dolin discussed some of the mapping issues, both between DTDs and more generally between information models. The difficulties of mapping various information models make it preferable for all groups creating DTDs to derive them from PRA, to map constructs to the HL7 model, or to get involved in the harmonization process with the RIM.
Mr. Rishel reviewed some of the strengths of HL7, notably its adaptability to different information system environments and capabilities. When data are not completely comparable, accommodations are made so that patients can be treated. In this environment, the government's role should involve a series of steps to improve quality and reduce bilateral negotiations for information exchange. The major obstacle to such exchange is misunderstandings about vocabularies. He discussed the multiple components of messages and echoed fellow panelists in stressing the importance of information models. In developing clinical templates, which are essential, HL7 will reach out to professional societies for specific templates and publicize the work of the specialty committees. HL7 can use help with the costs of operating the methodology and with outreach to professional societies to achieve the templates. He urged the government also to negotiate public rates for a few vocabularies.
The group discussed the possible need for a "super-harmonization process" above the HL7 level because of the multiplicity of information models. Some panelists felt that support for existing efforts and a way of cross-walking among existing models is what is most needed. Others urged that one model be used to which others would map.
Work Group members also raised questions about the respective business and clinical cases for modeling, clinical vocabularies, and message exchange standards; international coordination; the role of the federal government; and the relevance of HL7 mechanisms to the data sets in which NCVHS has a strong interest.
Mr. Beatty, who is affiliated with X12, stressed the need for coordination among the business and data models of the various SDOs. This need, plus the need for a single reference model, will grow as the computerized patient medical record develops. X12 hopes to see the SDOs work together further to build the reference model.
Mr. Solbrig pointed out that a specific enterprise viewpoint underlies the information model of every organization and should be understood before any information model is embraced. He recommended that the federal government clarify the enterprise model underlying its own information needs relating to PMRI. He does not favor moving to facilitate the development of medical information message format standards until objectives have been clarified and models developed.
After discussing the features of document records to which electronic records must map, Ms. Sokolowski outlined the merits of XML in meeting these criteria. She explained specific characteristics and applications of XML and the function of document-type definitions (DTDs). For a federal role, she recommended facilitation of communication among SDOs and financial support for standards efforts.
In the discussion session, Dr. Fitzmaurice stressed that caregivers need to verify and validate information, a capacity not currently part of XML. Dr. Beeler speculated that there is unanalyzed overlap between HL7 work and the ASTM activity on templates. The group discussed the merits of pulling together a single national data dictionary, with the sentiments of the panel being that it is not worth the effort. There was brief discussion of object-oriented technology.
Mr. Harrington said that in an environment of heterogeneity, interoperability is a top priority. Message format standards are one important paradigm for communicating information in a heterogeneous environment. They can be accelerated through precisely defined profiles for standards and by middleware to support implementation. He stressed the private sector's demonstrated willingness to work together, and recommended that the federal government remove legislative barriers to the HII and require federal agencies to use standards.
Mr. Shafarman was one of several panelists to recommend taht the government let the market decide which syntaxes and technologies are appropriate. He also argued for the principle of technological neutrality, an HL7 goal. In addition, the government hsould support the development of "base-level" clinical vocabularies and of clinical templates that can be phased in one clinical domain at a time.
Mr. Pratt pointed out that PMRI permits a reengineering of the health care delivery process. He noted that all of the day's panelists emphasize vocabulary as the essential model. Observing that HL7 Version 3 is expected to reduce negotiations about interface to the resolution of minor site-specific requirements, he called for federal funding to accelerate progress to this end. The government also should help industry with the "investment spike" involved in HIPAA compliance, e.g., through targeted tax investment credits. His recommendation that the government designate an independent agency/ies to design and execute conformance testing generated considerable discussion later in the meeting.
Mr. Meyer said that no single vocabulary at present fully meets the needs of the PMRI; but the work of the U.K.'s National Health Service is encouraging. In the U.S., HIPAA may provide the impetus for a similar initiative, with the NCVHS recommendations on vocabularies setting the stage for such an effort. A document architecture, such as the one the SGML/XML SIG is developing, is the most efficient approach to exchange and processing. The PRA, however, has engendered competition oveer who will be the arbiter of the resulting DTDs, a divisiveness that is counterproductive. This arena badly needs coordination, something that might be provided by ANSI HISB. Regarding the choice of a standard vocabulary, he recommended that the government conduct research and recommend a process for closure.
In the discussion period, Mr. Harrington outlined the Andover Working Group's actions to improve interoperability for clinical messages. The point was made that interoperable standards require more attention to detail, which initially may take more time but ultimately will result in quicker hook-ups and diffusion of the standard.
The group discussed the idea of an independent agency to design and execute conformance testing. Mr. Pratt noted that the National Institute for Standards and Testing fits the bill perfectly. Concerns were explored about the "apparent cross-purposes of the ASTM effort to define DTDs external to HL7's patient record architecture." Conformance testing at the Version 2 level was suggested, and HL7 was encouraged to specify the criteria needed to claim conformance. The group discussed whether HL7 will meet the HIPAA intent of improving the efficiency and effectiveness of the health care system.
It was noted that a central question is, What can we do to to make standards an easier process within business? The business case is not always obvious to people, even though it exists. What incentives are appropriate to encourage standards participation among vendors and providers? Finally, the group discussed the communication resources now available for sole or small-group practitioners. Panelists pointed to the Internet as the best medium for this purpose.
Dr. Bormel challenged the industry to recognize that automating the complexities of patient information will take considerable work and time, and then to figure out how to minimize these factors and properly incentivize the desired behavior. He outlined a schematic for understanding the health care world, with the variables of acuity, cost, time, and population. The issues for PMRI differ as a function of these variables. Drawing a distinction between "patient" and "person" information, he stressed that trying to integrate personal, patient and population types of CHR would result in huge costs and a failure to operationalize basic business and clinical objectives.
Dr. Sutherland discussed the potential role of automation in reducing adverse events, reducing costs, and improving quality of care. He then reviewed the issues in getting providers to automate, stressing the need for benefits and incentives to bring about this behavior change. He focused on the Internet as the chief driver of change, outlining the steps needed to realize its potential. Noting that epidemiological clinical trials have taught many lessons about data collection that could be applied to the health care environment, he recommended collaboration in this area between NIH and NLM. The best role for government is in supporting development of standards and providing incentives for quality.
The discussion with the panelists focused on the need for coordination of common vocabulary development and on the early steps needed to establish access to patient records for all relevant providers. Dr. Cohn proposed that the Work Group organize a briefing on the recent literature on the costs and opportunities related to quality, i.e., the links among quality, standards, and PMRI.
The group listed the main points they gleaned from the foregoing panels, and identified topics on which they want to follow up. See the detailed meeting summary below for the list of topics.
Members reviewed a "first cut" of questions for the May hearing panelists, which will be revised based on this discussion. The group also discussed the objectives for the May meeting, the types of presenters and some suggested names, and possible meeting formats.
Members agreed on the desirability of organizing a joint meeting on data quality with the Work Group on that subject, and of considering holding one hearing outside Washington, D.C. They will revisit these decisions in May or June.
Members agreed to review the guiding principles for selecting a standard at the Work Group's June meeting.
In response to a proposal from Mr. Blair based on concerns that the Work Group's current name lends itself to misinterpretations of its mission, members and staff agreed informally (with one abstention) to propose the following name to the Executive Subcommittee: Work Group on Patient Medical Records Information Standards.
Dr. Cohn began with introductions. He noted the Work Group and this activity originated in HIPAA mandates to the Secretary and NCVHS. The Work Group's specific focus is the issues related to adoption of uniform data standards for patient medical record information and its electronic exchange. It will prepare the NCVHS report due in August 2000, based on the findings in a series of hearings planned for this year. Today's hearing will focus on message format standards; tomorrow's will focus on data quality, accountability and integrity.
Mr. Blair added that the agenda is crafted to educate and inform Work Group members-- e.g., about syntaxes and modeling as a tool.
This panel represents providers and consultants; vendors, who are also HL7 constituents, are on panel three. Dr. Beeler noted that the answers given to the questions posed by the Work Group for this meeting would be informal, and also can be inferred from HL7's activities.
He characterized patient medical record information (PMRI) as "data about the events and findings that characterize an individual's health status and health care." The amount of detail depends on what is needed. He reviewed a range of uses, noting that use of the record by a single integrated health care system is becoming increasingly important. The fundamental use is for clinical decision making; other uses are management and assessment.
Regarding comparability of the data, he asserted that correct decision-making depends on comparable data; the less comparable the data, the greater the difficulty in making good decisions. A key factor in comparability is having a common semantic understanding of the information. Data interchange standards are important because decision making requires the communication of data between disparate computer systems.
Thus, the core issue is interoperability-- "the ability of two or more systems or components to exchange information and to use the information that has been exchanged" (IEEE). These two facets Dr. Beeler termed functional and semantic interoperability, respectively, and he stressed that the second is as essential as the first. Given the variety of software designs in use by various health care systems, there must be a common understanding of these core principles: 1) the data model, 2) the data representation or vocabulary, 3) the notion of when to communicate, and 4) security. Citing the ISO TR9007 standard statement about the necessity of having both semantic and syntactic agreement, he described ISO's model of layers for communications -- from the bottom up, a channel, services, syntax, and semantics (including "nouns," "verbs," and "adjectives"). HL7 specifications have focused on semantics and syntax and, recently, also on services.
Dr. Beeler gave some background on HL7, which was founded in 1987 and now has 450 organizational members and has affiliates in seven countries. It meets three times a year. ANSI accredited HL7 in 1994 as a standards development organization (SDO). HL7 is currently working on Version 3, which will be based on a consensus information model and agreed-upon vocabularies, and built on a set of industry-based technologies and capabilities. This will provide standards not only for messaging but also for document, component, and medical logic representation. It separates information content definition from transmission formats.
The approach to developing these Version 3 standards is different from previous versions in that it is model- and repository-based. There are four design phases, which Dr. Beeler described. He then discussed the importance of semantics as the core issue, noting as examples the various possible interpretations of "attending physician," "identifier," and "current episode." HL7 builds and applies common semantic models on domain expertise. The experts, a critical factor, are willing to help, although logistics are a problem and funding must be found for getting them to meetings and supporting their collaborative work.
The role for government is to "be supportive and not directive." It should avoid the temptation to focus on "the current 'hot'technology." The foundation for semantic interoperability must be durable and independent of technology change as it anticipates "the world as it will be."
In HL7 and the Huntington Group, Mr. Shakir focuses on modeling activity and methodology, and also on relationships between HL7 and other SDOs. He noted that modeling is, or can be, a collaborative process. Models can help improve communication and provide a basis for reconciling differences and identifying gaps in understanding.
HL7 has developed a reference information model (RIM). Through a collaborative harmonization process, which Mr. Shakir described, HL7 technical committees reconcile the differences in their understanding, resulting in a coherent shared view represented in the information model. This process involves the international affiliates as well. External contributions, such as from X12, are recognized.
A variety of models exist, for business and for information. HL7 and X12 have been communicating with each other about their respective modeling activities. Canada and Australia have such models, as do many large health care concerns and the Department of Defense. The most important role of business/information models in this context is to permit organizations to collaborate on the patient medical record.
Impediments to the process include proprietary attitudes on the part of some organizations. The process is also dependent on linkages to vocabularies and behavior models. There must be adjustments for local variations. Only a few vendors and fewer SDOs are using a model-driven approach to product development. In contrast, modeling is integral to the development of HL7's Version 3. It is important for this to happen with other products, as well, as it is very costly to maintain cross-references between various reference models.
To overcome the barrier of proprietary models, HL7 developed a process that allows organizations to contribute their models as seeds while protecting anonymity and intellectual property. In another area, HL7 is integrating modeling with vocabulary development, and it has built model management tools to support harmonization. It needs additional funding to sustain these efforts.
The federal government can participate in model harmonization efforts sponsored by SDOs, thereby encouraging others to participate. It also can provide funding to SDOs for infrastructure costs to maintain the model. Mr. Shakir joined Dr. Beeler in urging the government not to legislate the adoption of a single business or information model; rather, the model should be allowed to evolve. However, it is useful to have a single reference model to which others can be mapped. The government can also establish guidelines for model development.
Dr. Yasnoff asked about the need for government-sponsored research on the harmonization of models in respect to vocabulary, and Mr. Shakir said the expertise is available in the private sector but funding is needed. Mr. Rishel commented that the greatest need in this area now is for practical applications rather than fundamental science. Dr. Beeler said the tasks for information modeling are far less complex and voluminous than the coding linkages required for the UMLS. He noted the need to specify a respository of model harmonizations.
Mr. Blair noted that the Work Group would spend two days on medical terminologies and their coordination in a May hearing, providing additional opportunities to pursue this important subject.
Dr. Dolin pointed out the distinctions between the UMLS and the RIM harmonization process: the former is mapping terminologies but not creating them; the latter is creating a model that harmonizes all the others. He recommended that the government support the work within HL7 rather than creating a parallel process.
Dr. Yasnoff explained that what he was getting at in his question was the possible need for an extramural program like UMLS. Another question is where the repository of the models will be and how they will be maintained. These questions will be revisited.
Dr. Dolin's topic is XML technology in HL7's PMRI exchanges. These activities are in a draft or proposal stage and have not yet been subjected to HL7's formal ballot process. XML embeds tags in documents that tell a computer how to process the document. Dr. Dolin explained the differences among HTML, SGML, and XML. The latter two are user-defined, with XML being simpler than SGML. (For details of this and all other presentations, see transcript and slides on NCVHS Web site.) XML can be used to define the tags and their legal arrangement within documents. The declaration of this arrangement is placed in a document-type declaration (DTD). The net result is "a chunk of computer code" that can be validated.
HL7 created the SGML Special Interest Group in 1996, renaming it the SGML/XML SIG when XML was approved by the WWW consortium in February 1998. XML will carry on and expand the well-established tradition of SGML use in medical publishing. It is useful for rapid widespread deployment partly because it is readly converted into HTML for deployment over the Internet. It also permits consistency in content and presentation and enhanced search performance including context-specific [<sic. -->content-specific?] searching capabilities. XML also is an application-independent persistent format. Dr. Dolin showed an example of the use of XML with Kaiser's internal clinical practice guidelines, described in the March JAMIA.
In messaging within HL7, XML serves as the transfer syntax -- i.e., the explicit expression of the sender's semantics in a format understandable by the receiver. Dr. Dolin cited the report of the European Committee for Standardization (CEN) on syntaxes for existing interchanges, published in 1993. It was extended in 1997 to include SGML, which was found to compare favorably to other syntaxes. Mr. Rishel coordinated a 10-vendor HL7-XML interoperability demonstration in February 1999, and all participants rated it a success, thereby entrenching XML as a transfer syntax for HL7.
Dr. Dolin discussed various ways to map HL7 into XML syntax. He stressed that all relevant metrics must be factored in, whether or not they are easily quantifiable; the optimal message representation balances functional, technical, and practical requirements. He showed the findings to date on using XML as a transfer syntax for HL7 messages. Based on these findings, HL7 plans to use XML as a transfer syntax for Version 3 messages.
The benefits of XML for medical publishing also apply to clinical documents within HL7, making it possible to capture the semantics of clinical narratives. XML provides a syntax for HL7 semantics, which ties to the shared information model that is central to the patient record architecture (PRA). The architecture provides a common mechanism for referencing standard terminologies. It also has levels that correspond to conformance with various types of reporting.
Dr. Dolin discussed some of the mapping issues, both between DTDs and more generally between information models. He said the onus is on the developers of the PRA to make sure they have captured the semantics necessary to facilitate mapping, although there is still the potential for difficulties. He then acknowledged the PRA editorial group, all of whom were present at this meeting.
In summary, he stated that HL7 is defining standards for shared semantics of health care information that can be readily expressed and transferred in XML, which is a sound and valuable technology that will facilitate HL7 implementations.
Regarding the potential government role, he observed that the evolution of the medical record is an ongoing process that requires a balance between the need for implementaiton and the need to deepen shared understanding of clinical information. He encouraged NCVHS to identify HL7 and the PRA as the organization and process that will continue to develop the models. The difficulties of mapping various information models mean that it would be preferable for all groups creating DTDs to derive them from PRA, map constructs to the HL7 model, or get involved in the harmonization process with the RIM. Finally, grants to accelerate standards activities are a major need.
Asked by Dr. Cohn how XML would help with issues of comparable patient medical record information, Dr. Dolin referred to its role in validating a subset of HL7 business rules. He agreed with Dr. Cohn's characterization that the XML processors make it possible to enumerate a list, which becomes a way to enforce a vocabulary standard within HL7. In response to a question from Dr. McDonald, he commented on HL7's role in the "elements vs. attributes" debate.
Mr. Rishel said a key point for him is that "standards, like anything else, is an experience cycle" moving progressively through stages of judgement and experience. He reviewed some of the strengths of HL7, notably its adaptability to different information system environments and capabilities. A problem has been the bilateral agreements that have arisen during implementation. Standardized interfaces enable organizations to be more efficient.
Comparability is achieved by addressing vocabulary, the reference information model, and clinical templates. When data are not completely comparable, accommodations are made in order to treat the patient. In this environment, the government's role should involve a series of steps to improve quality and reduce bilateral negotiations for information exchange. The major obstacle to such exchange is misunderstandings about vocabularies. There is a cost barrier to the full exchange of comparable, fine-grained information, however.
In the area of the information model, HL7 has "a semi-definitive approach to vocabulary" in that it does not define codes or direct industry to use a particular vocabulary. It is seeking the most appropriate technology for exchange. Good technological support for information interchange is rapidly expanding, and SDOs should take advantage of all of them, in Mr. Rishel's view, rather than advocating one or another. He acknowledged that proprietary vocabularies require licenses from payers.
He discussed the multiple components of messages and echoed fellow panelists in stressing the importance of information models. Clinical templates provide a way to develop fine-grained rules so that things can be expressed comparably -- to make it possible to define specialized information structures using a common methodology. HL7 is currently working out how to accomplish this. It wants to use its standard process, which has been proven to be "fair and workable," to establish templates. Considerable subject expertise is needed, and HL7 will reach out to professional societies for specific templates and publicize the work of the specialty committees. Mr. Rishel noted the relevance of clinical templates to claims attachments. An approach has been balloted and published by HL7 and is now being reviewed by HCFA.
Regarding the government's role, he cited the legislative influence on the automobile industry as a good example: for pollution reduction and better gas mileage, Congress "created a series of goals over time that kept raising the bar." A balance must be maintained between improved functionality in the standard and the costs for vendors to improve their information systems. The RIM should be a reference point for future design. The typical release cycle for vendors is 18 months to two years, so it takes industry at least that long to react. Mr. Rishel advised government to be selective in what it wants in two years, and not to let perfectionism impede improvement. He noted that clinical templates are necessary in the absence of trading partner agreements. HL7 can use help with the costs of operating the methodology and with outreach to professional societies to achieve the templates. He urged the government also to negotiate public rates for a few vocabularies.
Dr. Cohn thanked the panelists for this informative session. He asked for comments on the need for a "super-harmonization process" above the HL7 level because of the multiplicity of information models. Mr. Shakir said what is needed is a way to cross-walk among models, not a super-model. Dr. Beeler observed that every model has a specific purpose, and one question would be what purpose a super-model would be intended to serve. He sees value in a more abstract (simplified) model of health care to which all SDOs could map -- an activity that has been proposed by several different entities including Great Britain's National Health Service.
Col. Ray asked how models are implemented and integrated -- e.g., how the GCPR can harmonize the HL7 RIM and its own functional area model for data (FAMD), especially as changes come over time. Mr. Rishel recommended a process of tracking back and forth between the models rather than "looking for the one true model." He also referred to HL7's meta-model and its message development framework document.
Dr. Fitzmaurice asked about the business case and the clinical case for modeling, clinical vocabularies, and message exchange standards. In response, Dr. Beeler cited increases in the effectiveness of information --a better path through the data-- and in the efficiency of patient care. He added, however, that these are "still premises" that are yet to be proven. A third benefit, one he is sure of, is process improvement.
Dr. Dolin asserted that "the information needs of office providers are not being met" with paper systems and he noted the need for decision support applications.
Mr. Shakir related these facets to modeling, which by establishing interoperability can make it possible to capture the multiple communications among all the players, both individuals and organizations.
Dr. Fitzmaurice then asked if the case extends to international coordination in modeling. Dr. Beeler observed that the health care computer systems business is global, and borders mean less and less. The commercial argument for international coordination is stronger than that based on patient movement. Mr. Shakir, whose family lives in Ghana, pointed out that other nations' public health systems can benefit from the clinical studies done in the U.S. if their information models are consistent. Dr. Fitzmaurice cited clinical drug trials as another reason.
Dr. Cohn asked for further responses to the Work Group's question about what role the federal government should play that might bring about results in one to four years. These areas were mentioned:
Mr. Shakir commented that the panel has presented a three-layer cake: semantics (including the information model), syntax, and implementing technology. The most collaboration is needed for the information model, the vocabulary, and clinical templates. The SDOs have demonstrated the will to work together; the problem is that they share members, who can't go to all the meetings. The government could help by encouraging collaboration among the SDOs and underwriting the costs. It should stay away from dictating the syntax, letting the market decide which is best.
Dr. McDonald observed that NCVHS has a strong interest in data sets, and asked how the HL7 mechanisms relate to that interest. Mr. Rishel called data sets "aggregators" for specific purposes, and said the greatest problem they face is variability in the source systems. The templates being developed have the potential to generate comparable data once they are introduced into the source systems. In the short run, he asserted that more use should be made of the public Internet, which can be made secure "by a few programmers in a few months."
(Note: Claudia Tessier, Chair of ASTM E31, was unable to attend this meeting. Her testimony will be posted on the Web and she will be invited to testify at another hearing.]
Mr. Beatty represents Mayo on X12 and chairs the Insurance Subcommittee (X12N), which develops EDI standards for the health care industry, property, casualty, life insurance and reinsurance industries. These standards facilitate the nine transactions required under HIPAA. X12 and HL7 jointly developed an implementation guide for the 275 patient information transaction, merging the strengths of both organizations.
The Insurance Subcommittee has done business and data modeling to assure consistent transactions. The task is facilitated by the X12 Business and Information Modeling Task Group, which has produced a data dictionary, a data model, and a business model, all available on the X12 Web site. The models, which continue to evolve, are being incorporated into other development processes. X12, HL7 and NCPDP all have business and data models. The need for coordination among them, and for a single reference model, will grow as the development of a computerized patient medical record proceeds. X12 hopes to see the SDOs work together further to build the reference model.
Mr. Solbrig is Senior Product Development Specialist with 3M, which he has represented on several national and international SDOs, notably the CORBAmed task force. He organized his talk around the Work Group's questions after first pointing out that information exchange within an enterprise involves many implicit shared understandings that must be accommodated in another way with large-scale information exchange. He defined semantics, pragmatics, and syntax as elements of communication and asserted that the choice of syntax should be left to network architects and engineers. Healthcare-specific syntax specifications are no longer necessary; so any federal involvement in this area should be very general and might even just complicate things.
Regarding models, he referred to the ISO Standard 10746, the reference model for open distributed processing, which identifies five "viewpoints" of models, including enterprise, information, and computation. Noting that today's discussion is focusing on the second viewpoint, he cautioned that the enterprise viewpoint underlies all information models and it should be understood before an information model is embraced. Each business has its own perspective. For example, the NHS (England) and HL7 information models don't always line up because of different underlying business models. CORBAmed is currently developing its own enterprise model to serve as a roadmap.
As to the federal role in modeling, he suggested that it fund an effort to specify the enterprise model for PMRI from the federal perspective. After developing this model specifying the intended uses of information, appropriate information and computational models can be developed.
The useful emerging standards or technologies for the collection, storage and communication of comparable medical information are "a distributed system on a national level." All the tools and techniques used to implement such systems locally are applicable and essential. One tool to help manage complexity is a meta-data "repository of information about information" being developed by NCI TSL8, with participation by several U.S. federal agencies as well as Australian Health Care, et al. The federal government, Mr. Solbrig said, needs to recognize this as a national project and develop national-level specifications for it, and also provide some funding.
He does not favor moving now to facilitate the development of medical information message format standards. First, specific objectives should be clarified, followed by good informaiton and computational models; then come the technical decisions.
Ms. Sokolowski co-chairs the HL7 SGML/XML SIG and also the ASTM E31.25 (focusing on XML and XML DTDs). She started by talking about the nature of health care information and observing that the components, organization and function of electronic records must map to today's paper-based documents. Currently, there are multiple formats and processes and technologies for electronic conversion. The criteria to be satisfied include having standardized documents, context for narrative text, consistent place for diagnosis and for symptoms. The technology should be durable and allow timely access and the ability to search for specific information, viewable in different ways.
She discussed the ways that XML (Extensible Markup Language) meets these criteria, noting that its most important attribute is that it communicates information in such a way that both humans and computers can understand it. Its durability despite technological change stems from the fact that it is a subset of SGML, which is already 13 years old. Ms. Sokolowski explained specific characteristics and applications of XML and the function of document-type definitions (DTDs). XML DTDs provide context for information, unlike HTML.
She asserted that to use XML in health care and the patient medical record, some standardization of the clinical documents types expressed in XML DTDs is needed, one that can be transferred to a new syntax currently under development. This process should be accelerated. In addition, government forms such as the HCFA 1500 should be standardized in XML. She noted the need for someone to facilitatate communication among the SDOs, especially in relation to XML activities. This is as another possible government role, in addition to providing resources to accelerate the ASTM E31.25 effort and the W3C work on DTD syntax.
Dr. Fitzmaurice stressed that caregivers need to verify and validate information, a capacity not currently part of XML.
In response to a question from Dr. McDonald about the level of abstraction of XML, Ms. Sokolowski said information on the type of mammogram, for example, would be derived through document analysis.
Ms. Frawley described a joint project of the ASTM work group on medical record documents, Ms. Sokolowski, and the American Association for Medical Transcription, looking at documents being transcribed in terms of formats, the degree of commonality among data elements, and the types of reports. The study highlights the lack of standardization and commonality among paper records today.
Asked about his views on the need for a "super-harmonizing" process, Mr. Beatty referred to his earlier comments about the differing business perspectives of different organizations. X12 has developed ways of communicating information between organizations that have different business and data models. Common reference models would permit interorganizational communication about such things as eligibility.
At Mr. Blair's request, Ms. Sokolowski described the work of the HL7 SGML/XML Special Interest Group, which she co-chairs. It has a different focus from the ASTM group, which was formed in November 1998. The two need to cooperate. Dr. Beeler speculated that there is unanalyzed overlap in what the ASTM group is doing on templates.
Asked by Dr. Fitzmaurice about the merits of pulling together a single national data dictionary, Mr. Beatty said he did not think this is necessary if there is a single reference model. Dr. Sutherland of IDX (see Panel Four) commented on the differences (not irreconcilable) between his views and those of Mr. Beatty. He described the efforts of the Joint Working Gorup for a Common Data Model, which he chaired. Differences between the models will inevitably arise, and it is necessary to retain control over the direction of the RIM. Having a common reference information model may not be worth the energy it would take to build it; he favors working on collaboration and mapping.
There was brief discussion of object-oriented technology, which Mr. Rishel characterized as based on a powerful idea that combines data and algorithms and permits better programming. He offered some further technical comments on its capacities. Mr. Solbrig observed that looking at the business models of "the two organizations" [not specified] explains previous conflicts.
Mr. Harrington spoke on behalf of Hewlett-Packard and the Andover Working Group. The latter, founded by Hewlett-Packard in 1996, has more than 300 member organizations and works to accelerate and broadly deliver interoperable standards-based solutions for health care. Both organizations believe that because heterogeneity is "a fact of life" in the health care information infrastructure, interoperability is a top priority. Further, message format standards are one important paradigm for communicating information in a heterogeneous environment. They can be accelerated through precisely defined profiles for standards and by middleware to support implementation.
On the Work Group's question about definitions, Mr. Harrington cited the IOM definition of the patient record and AHIMA's delineation of the information contained therein. He noted that the multiple uses and users of PMRI necessitate that they be comparable, i.e., "semantically equivalent."
On syntax, he cited the work of IEEE P1157, CEN TC 251 and HL7 in demonstrating the possibility of mapping a common information model to and between various interchange formats without semantic loss. Expressiveness, efficiency and extensibility are factors in choosing a syntax. The Andover Working Group has worked on mapping information between IEEE 1073 and HL7. Work on XML, as well as on HL7 Version 3, shows the private sector's ability to work together and make progress.
In addition to message format standards, other approaches to interoperability are also important and all will benefit from development of a common information model.
Regarding emerging standards and technologies, the Andover Working Group has concluded that the availability of middleware to support implementation of standards is critical. It has done some promising work with other groups on component-based support for implementation. Mr. Harrington also noted the important work on SGML/XML.
On a federal role, Hewlett-Packard sees the greatest impact from 1) removing legislative barriers to the implementation of the HII; 2) participating with accredited SDOs and their consortia; and 3) requiring the use of standards by government agencies.
Mr. Shafarman co-chairs the HL7 Control Query Committee and is on the board of its international committee. He began by calling attention to several current paradigm shifts. One is in standards work, from just looking at messages to looking at exchange technologies. Another is from viewing vocabulary content models as separate from information models, to an understanding that context gives meaning to vocabulary. Other important shifts: toward greater interoperability, and to seeing documents as collections of atomic, reusable information units. OACIS believes messaging and object interfaces must have vocabulary and information models in common. The thinking is in terms of object models, in which the object has behavior and attributes and different specializations can be created of an object. The ability to add specializations is a key developmental step with the HL7 RIM.
OACIS sees the definition of the PMRI as broader than basic clinical information but restricted to information that contributes to patient care, e.g., excluding financial and administrative data. (Other panelists did not mention this distinction.) The organization is evaluating vocabulary and information models in terms of how much interoperability they yield. This opens the way to interfaces with clinical decision support and research, among other things. A characteristic is that model-based paradigms are continuously extensible via specialization for storing new knowledge.
Regarding syntax, Mr. Shafarman pointed out that a syntax does not by itself support information standards; doing so requires particular uses and forms. Thus, the syntax on its own does not result in interoperability. OACIS supports HL7's goal in this area, (i.e., technological neutrality). He suggested that NCVHS take taht approach and let the market decide which syntaxes and technologies are appropriate. The focus, instead, should be on information modeling and vocabulary standards.
The federal government should investigate how to help with the extension and evoluation of the HL7 RIM and support the development of "base-level" clinical vocabularies. Federal support is also welcome for develomeht of clinical templates. These can be phased in one clinical domain at a time rather than all at once, starting with the most critical. Eventually, the federal government should encourage standards for rules in data warehousing.
Mr. Shafarman pointed out that work has already begun in many areas. He cited CDC's electronic lab reporting project and the HIPAA-related work on claims attachments. He encouraged the federal government to do all specifications for new projects requiring clinical information and information interfaces in a technology-neutral model that can evolve along with the technology. He joined Mr. Rishel in urging government not to let the perfect be the enemy of the good in making progress in this area.
Mr. Pratt co-chairs HL7's SIG for Object Brokering Technologies. SMS is committed to supporting industry standard message formats and common vocabulary and has supported their development. He reviewed the chief merits of standards. Regarding PMRI, he noted that it does not just recreate the paper medical record but permits a reengineering of the health care delivery process.
He noted that all panelists today emphasize vocabulary as the essential model, and that the primary barrier to achieving a viable CPR is the lack of universal standards that allow data integration.
He reviewed the advances achieved by HL7 Version 2 and the further progress expected from Version 3 toward plug-and-play compatibility. It is expected to reduce negotiations about interface to resolving minor site-specific requirements. Progress to this end can be accelerated through further funding. In addition to the competing demands of product development and standards development, the industry is challenged by declining revenue and increased regulatory initiatives.
In this climate, the federal government should help industry with the "investment spike" involved in HIPAA compliance. Mr. Pratt recommended targeted tax investment credits as an effective incentive. Second, the federal government should help HISB develop and execute a stronger charter that provides real leadership in health care informatics. Third, it should designate an independent agency/ies to design and execute conformance testing.
As his organization's informatics standards liaison, Mr. Meyer is active in several SDOs and monitors other organizations promoting standards. He sits on ANSI's HISB and the US TAG to ISO TC215. His organization also supports standardization of the industry and is working hard to promote interoperability.
He noted that the goals of improving patient care and reducing costs are achieved by comparing many kinds of information and combinations of data. Comparison depends on true comparability, apples to apples, as well as on precision. This in turn depends on a standard terminology geared toward evidence-based medicine. No single vocabulary today fully meets the needs of the PMRI; but the work of the U.K.'s National Health Service is encouraging. In the U.S., HIPAA may provide the impetus for a similar initiative, with the NCVHS recommendations on vocabularies setting the stage for such an effort.
Mr. Meyer reviewed some of the syntaxes and technologies available, noting that they have little applicability to many legacy systems. While the HL7 data dictionary may be "the glue that cements the process of PMRI exchange" and is already a de facto standard, Version 3 will be a quantum leap forward. Its primary syntax is XML, which allows the inclusion of meta-data in the context of the message content. The RIM provides a graphic representation of the enterprise. Mr. Meyer recommended a review of the HL7 message development framework, to understand how modeling drives the development of message format standards.
The impediment to extending the scope and applicability of the RIM is continuing harmonization of existing models. Government can support this process through participation; by making the DOD, VA and public health models available; and by promoting harmonization of the RIM, the HIPAA data dictionary, and the ASC X12 data model. A document architecture, such as the one the SGML/XML SIG is developing, is the most efficient approach to exchange and processing. The PRA, however, has engendered a competition to control or be the arbiter of the resulting DTDs, a divisiveness that is counterproductive. The arena badly needs coordination, and Mr. Meyer recommended that ANSI HISB provide mediate a closure. NCVHS, through the CPR Work Group, can provide leadership by supporting a document architecture once an appropriate candidate has been chosen.
Also in need of coordination is the issue of vocabulary terminology and code sets. He applauded the efforts to this end by Drs. Chute and Cohn in a politically charged environment where the benefits for patient care and administration are vying with entrepreneurial forces. Mr. Meyer raised the question of how a standard vocabulary should be chosen, and recommended that the government conduct research and recommend a process for closure, if not actually declaring a standard.
At Mr. Blair's request, Mr. Harrington outlined the Andover Working Group's actions to improve interoperability for clinical messages. They convened a group of leading vendors and users and looked at the process model for a multi-vendor profile, based largely on the HL7 Version 3. By getting to a definition that eliminated optionality, and other aspects, they demonstrated the ability to reduce burden and achieve plug and play. Participants commented on the rapidity of the integration. Dr. McDonald acknowledged Mr. Harrington's seminal role in object modeling.
Mr. Shafarman agreed with Mr. Harrington that Version 3, especially the RIM, can help generate standards more rapidly. Mr. Harrington, in turn, agreed with the observation that doing more interoperable standards requires more attention to detail, which initially may take more time, but will ultimately result in quicker hook-ups and diffusion of the standard. The hope now is that vendors will sign up to support various profiles. He noted the "pieces of the value chain" that need to come together in this process, which will be driven by the market. The ability to do a registration of a process and enforceable contract is a key factor; another is middleware.
Mr. Pratt described the efforts of the SIG on Object Brokering Technology to achieve interoperability between the application and the middleware.
Dr. Fitzmaurice followed up on Mr. Pratt's recommendations about an independent agency to design and execute conformance testing. Mr. Pratt said that HL7 Version 3 will offer the possibility for the first time of having a single HL7 interface, which is not the case now. Verifying that a vendor's interface passes the test for HL7 Version 3 should not be done by the SDOs, which are dominated by vendors who would rather invest their time in developing and enhancing the standards. He noted that the National Institute for Standards and Testing fits the bill perfectly.
Asked by Dr. Fitzmaurice to expand on his comments about DTDs, Mr. Meyer noted the concern about the "apparent cross-purposes of the ASTM effort to define DTDs external to HL7's patient record architecture." If a single such architecture is needed, the HL7 RIM is a good candidate for the national standard. He explained that the X12.N is very focused on electronic commerce while the RIM is much more comprehensive and offers a graphic visulalization of the health care enterprise.
Returning to the issue of conformance, Dr. McDonald commented that the failure to conform often stems not from variability but from users' never having "read the book." He suggested conformance testing at the current generation level. Mr. Meyer described a gap analysis tool that weighs products against other interfaces to ensure that they can talk to each other. Mr. Pratt commented that it would be very helpful if HL7 defined the minimum criteria needed to claim conformance.
Dr. Dolin remarked that encouraging people who are developing their DTDs to look at the RIM and the PRA and strive for consistency with these models is what will achieve greater semantic interoperability. Mr. Meyer agreed, stressing that the point is for everyone to work together.
Dr. Cohn observed that the latest session had gotten close to the key issue of the business case. Noting that the intent of HIPAA is to improve the efficiency and effectiveness of the health care system, he asked if the HL7 RIM will promote that end. Mr. Shafarman laid out the case for infrastructure in this regard, commenting on the ways that standards can improve efficiency and create new forms of clinical decision support. He noted that patient records gather information from many sources and thus need standards. Mr. Meyer added that vendors need to accommodate a host of systems in hospitals. Mr. Pratt pointed out the potential benefits through telemedicine from reaching remote populations.
Dr. Beeler noted the scope of change and progress just since HIPAA was enacted, particularly the growing relevance of HL7 and the coooperation between it and X12. A central question is, What can we do to to make standards an easier process within business? The business case is not always obvious to people, even though it exists. What incentives are appropriate to encourage standards participation among vendors and providers, and especially the latter?
Another participant (unidentified) observed that standards are a means, not an end; thus, what is important is to clarify goals, and then standards will help get there.
Asked for more commentary on his ideas for conformance testing, Mr. Pratt suggested starting with HL7 Version 3, meanwhile specifying a minimum standard for Version 2, ballotted by the membership. The reason for using an outside body such as NIST would be to remove any question of bias.
Mr. Harrington said that with an unambiguous specification and no optional elements, it is possible to do conformance testing. He once chaired a NIST group on conformance testing, and agrees that NIST is a good body to do this. He supports the idea of interoperability testing, for which well understood paradigms exist.
Mr. Shafarman pointed to AWG conformance profiles and CDC's ELR project implementation guidelines as resources available now for verifiable conformance tests even in "the Version 2 world," along with the HL7 claims attachment SIG methodology. What is lacking is tooling and a formal methodology, something that the conformance SIG might develop. Mr. Harrington said vendors need to help, and the SIG is helping create the marketplace by defining profiles. Citing the experience of HL7 Canada in trying to get the process started, Mr. Meyers endorsed the idea of an external agency, adding that the process still requires one vendor to be the first to attempt compliance.
Mr. Beatty noted that AHCT is setting up a facility for compliance testing of HIPAA- related transactions, evidence that the industry is coming forward with compliance verification. He added that such verification is separate from compliance accreditation.
Dr. Fitzmaurice asked what is available for sole or small-group practitioners in standards for communication. Mr. Pratt predicted the increasing availability of such tools, as telemedicine becomes viable. Mr. Meyer asserted that most physicians already have some kind of network affiliation and are using the Internet for research, to follow the literature, and so on. Mr. Shafarman agreed that the Internet is the "natural channel" in this regard. Dr. McDonald remarked that the work on standards is in fact being done with both "the big guys" and the small practitioner in mind, and vendors have been supplying practice management systems for years.
In conclusion, the co-Chairs thanked the panelists for helping the Work Group understand these issues more fully. The meeting was recessed until the following day.
Dr. Bormel worked on the UMLS at the National Library of Medicine prior to his present post at the Cerner Corporation. The latter organization provides clinical and management information and knowledge systems for health care organizations. He began by underscoring four points made the previous day:
The overlapping components of patient information captured in a clinical encounter are descriptive documentation, care management, clinical performance measures, and process. Instead of pretending it is not the case, the industry should recognize that automating this amount of information will take considerable work and time and then should figure out how to minimize these factors and properly incentivize the desired behavior. There are three types of text in patient care: uncodified, "only codified," and fully contextualized granular information. People generally are comfortable with only one type.
Reaching the desired outcomes requires leadership, incentives, and tools. Dr. Bormel showed some data comparisons to illustrate a point that patient medical record information is critical for generating more meaningful clinical and administrative data. What is needed is sets of appropriately contextualized granular data, plus measures of data that show variations. He regards medical management as bringing together information from administrative, clinical, and financial sources to improve quality, lower costs, reduce inappropriate variation, and meet reporting requirements.
He outlined a schematic for understanding the health care world, with acuity, cost, time, and population as variables. The issues for PMRI differ as a function of these variables. Comparability will make it possible to inform stakeholders with substantive measures. A major issue is making the process acceptable to physicians.
Clinical and cost profiling depend on the use of common definitions and consistent terminology, among other factors. Patient record information can reveal "the why" and the clinical context of health care in ways not possible with billing data.
Regarding definitions for PMRI, Dr. Bormel stressed that from the perspectives of prevention and population-based health, it is necessary to distinguish between patient medical record information and person health information. He noted that the latter is more concordant with the computer-based health record defined by NCVHS. Trying to integrate personal, patient and population types of the CHR would result in huge costs and a failure to operationalize basic business and clinical objectives.
Having providers record appropriately codified information would improve quality. This process will only work, however, when the benefit to them exceeds the implementation burden.
Dr. Sutherland cited the findings of case control studies that health care automation can triple compliance, cut costs in half, and improve clinical outcomes by 50-100 percent. He offered other statistics suggesting that adverse treatment events can be reduced through automation to control the process. (Reference: Demanding Medical Excellence, which compares health care to the airline industry in terms of safety procedures.) More than 60 percent of adverse events can be prevented with online information processing. Meanwhile, only 5 percent of U.S. physicians have any form of real automation in clinical practice.
The infrastructure exists to implement a universal, sharable medical record; what is needed now is agreement on content and structure, plus a commitment to do it. It is important to recognize the potentialities in information processing and not stay fixated on current constraints. Internet use is driving the change. The core issues are connectivity, integration, scalability, robustness, and the capability to rapidly evolve the systems. Comparability of data is the overriding issue. People are being driven to an Internet-based architecture. They want their vendors to provide Web-based and XML technologies.
In the move to comparable PMRI, there are still few interoperable products on the market although there is progress in that direction. Vendors need to support component-based standards as well as health care standards. The standards that work well are "small and pragmatic, and build from the bottom up." Industry needs to focus its research on XML in the context of a component model.
Dr. Sutherland outlined the idea of a comprehensive, distributed clinical medical system using Web-based XML technology, which he regards as the future for the medical record. Getting there begins with getting comparable data and accurately identifying patients. He observed that object models will evolve with use and must be implemented in order to develop. He also noted the natural tendency to develop specialized models, an inevitable process in health care which is a regionalized or "cottage industry" with local forces at work. The government wants a universal model across vendors to assure good sharing of quality data. Thus, the focus must be on interoperability, facilitating the emergence of XML vocabularies, and supporting stability.
Dr. Sutherland observed that epidemiological clinical trials have taught many lessons about data collection that could be applied to the health care environment. The government might encourage closer collaboration between NIH and the Library of Medicine for research on the patient medical record and technology transfer.
Care providers must be direct users of the computer record for quality data to result. What is needed is to embed clinical data acquisition into clinical work flow and to provide value for clinicians so they are willing to use computers. He suggested, for example, paying clinicians to use a CPR and/or charging a fee for paper submissions. Also needed is agreement on codification of medical vocabulary standards. The best role for government is in supporting development of standards and providing incentives for quality. Currently there are no standard criteria for quality; these should be developed in consultation with data users with domain expertise. Also, the existence of 50 state laws on patient medical record privacy will create devastating confusion and must be corrected.
Dr. Sutherland concluded by observing that the government facilitated the Internet, and looking at that model might yield lessons that could be applied to creating an online patient medical record system. Web standards enabling distributed message documents and components could have a dramatic impact on data quality.
In response to a question from Dr. Fitzmaurice, Dr. Sutherland commented that while there is growing use of computers by young physicians and med students, this does not solve the issue of common vocabularies, which need mechanisms for bringing expert groups to agreement. He agreed that there is a need for coordination and "a push" to get medical professionals and specialty groups working on vocabularies.
Ms. Fyffe asked what steps should be taken first to achieve transmission of basic patient information. Dr. Bormel cited the work of NCQA on HEDIS as a good example of quality-focused clear thinking. They discussed a scenario for the exchange of PMRI between sites of care in a hypothetical patient's two states of residence. With the patient's consent, basic medical record information could be transmitted in a summary frame in the appropriate DTD over the Web. To get there, the emphasis in the short run should be on quality frames for transmitting essential clinically relevant information. He added that while use of the Internet is very important, the need to create Web technology and standards should not distract from the equal priority of developing such things as computer reminder standards for improving the quality of care. Dr. Sutherland agreed that the emphasis on and off the Web should not be on the "whizzywigs," but on basic medical data.
The group discussed the high cost of adverse drug reactions, one of the health risks that could be reduced through automation of information management. Dr. Yasnoff cited a "very nice" meta-analysis that found that 5.1 percent of all hospitalizations are due to adverse outpatient drug events. The consensus in the field is that one-third of those events are preventable with adequate information, thus savi ng between $12 and $20 billion. The group juxtaposed this fact with the low total investment in information systems by the health care industry.
Returning to the scenario raised by Ms. Fyffe, Mr. Meyer remarked that the crucial issue is not transport but access to a secure Web-based medical record by all providers involved in a patient's care.
In conclusion, Dr. Cohn proposed that the Work Group organize a briefing on the recent literature on the costs and opportunities related to quality, i.e., the links among quality, standards, and PMRI.
REVIEW: WHAT DID WE LEARN FROM THE TESTIMONIES?
Work Group members were asked to identify the major points and themes that they heard from the four panels, and to note anything the Work Group should follow up on.
Ms. Frawley:
Dr. McDonald:
Dr. Cohn:
Dr. Yasnoff:
(Mr. Gellman asked for an explanation of the overall goals for this process, particularly to whom the Committee's recommendations will be made. Mr. Blair stated that the Work Group is staying close to the mission defined by HIPAA. After studying uniform data standards for PMRI, it will make legislative proposals to the HHS Secretary on that topic by August, 2000.)
Dr. Greberman:
Dr. Kolodner:
Col. Ray:
Dr. Cohn engaged Ms. Fyffe in a discussion of whether the Work Group is making progress in defining what patient medical record information is. They related this issue to basic core information that is accessible by all of a patient's providers in different sites. Ms. Fyffe suggested that the Work Group soon outline the skeleton of its report to Congress "so we don't waver from our goal."
Mr. Blair:
The co-chairs circulated a "first cut" of questions for panelists for the May hearings and asked for comments, on the basis of which the document will be revised. Dr. Yasnoff suggested adding questions about pricing and licensing, and named individuals who could educate the Work Group. He urged that the presenters be given clear directions about sticking to their allotted time.
Among other things, Ms. Humphreys suggested adding a first question about the intended purpose of the medical terminology, followed by a question about its current use.
Because of the number and complexity of the questions, Mr. Gellman suggested focusing them on the issues on which the Committee must make decisions, setting priorities among them, eliminating some of them, and/or inviting written answers.
It was noted that the last strategy would work for descriptions of the various terminologies. Mr. Blair observed that the Work Group already has lengthy descriptions of more than 30 of them. The group agreed to distribute those write-ups to the panelists and ask them to update the information, which will then be provided to Work Group members. Mr. Solbrig suggested leveraging a recent survey by HL7.
Dr. Greberman suggested reinforcing the representation from international activities. He will send specific recommendations on how to do that.
Ms. Humphreys suggested asking terminology users to identify their highest priority data elements or pieces of PMRI for standardization. Dr. Cohn noted that this gets at the Work Group's question about near-term priorities.
Work Group members and staff discussed ideas for possible testifiers for the May hearing. Dr. Fitzmaurice reviewed the reasons for beginning with an educational panel and topics they might cover. Ms. Humphreys noted that some of this material already has been presented in previous meetings, by the same experts. Members envisioned Dr. Chute presenting the recommendations of the Second National Terminology Conference, and Mr. Cimino [?] presenting the desiderata for medical terminology.
The issue of self-interest versus objectivity was raised in respect to presenters. Dr. Cohn speculated that some experts will have specific interests, while others will not. Ms. Humphreys suggested aiming for a representative group of people with experience using and analyzing many different terminologies.
Additional names of possible testifiers were suggested, and alternative formats were discussed. Two stated aims were to minimize confusion and to afford opportunities for discussion. Members felt it would be most useful if presenters could stay for the entire time to follow themes through the meeting and be in ongoing dialogue with the Work Group.
The names of vendors that should be included were suggseted. Members expressed interest in hearing from is small and medium-sized developers who are at a disadvantage relative to the large vendors and may have chosen not to "put terminology in." There was also interest in hearing from businesses helping individuals to record personal health information on Web-based systems (i.e., personal medical record information systems). Other categories mentioned: someone managing and collecting patient data for clinical trials; practice management vendors. The overarching point here is that there is a wide spectrum of software vendors whose work is relevant to this broad topic. Members seemed to favor using a broad definition.
Regarding practice management vendors, Ms. Fyffe observed that the Work Group should become familiar with the perspective of small practices that are happy with their vendors (of which there are about 1,000), to gain insight into how to get them on board with a standard computer- based record. She agreed that their perspective is important on many issues besides terminology, and asserted that their perspective is important whether or not they have clnical systems.
Mr. Blair observed that people working on electronic medical records tend to view practice management vendors, whose eclectic activities often do not include electronic health records, as "entirely separate vendors."
Dr. Cohn agreed on the desirability of talking with this group and challenged the notion that administrative and clinical data are entirely different and mutually exclusive. He suggested hearing from users of clinical terminologies and also from those who have not adopted them. Ms. Humphreys suggested people who have built their own, as well.
Members discussed possible ways of gaining access to the knowledge of Ms. Humphreys, who is unable to attend due to a scheduling conflict. Dr. Fitzmaurice suggested asking her to review the transcript and offer her observations on themes and findings.
The co-chairs asked members to send them additional ideas for the May session by e-mail. Further planning of panels and questions will take place by phone and e-mail.
Members agreed with his suggestion that the Work Group on Data Quality be approached about a joint project on data quality.
The group discussed holding one hearing outside Washington, D.C., and agreed to make this decision after the May hearing with the added information about the participation of West Coast terminology developers. Dr. Cohn said his reason for another location is to get the thoughts of small and medium-sized users of standards, who are less likely to be involved in SDOs or able to get to Washington. Dr. Greberman suggested talking with the Office for Advancement of Telehealth about getting their fundees in for these discussions. The Work Group will revisit this issue.
Dr. Cohn then suggested taking a look at the NCVHS/HHS guiding principles for selection of administrative standards in terms of their applicability to the Work Group's work on clinical standards. These principles are published at the beginning of the NPRMs. The group welcomed this idea and will pursue it at the June meeting.
Mr. Blair expressed his concerns about the Work Group's name and the ways it lends itself to misinterpretation, particularly the assumption that the Group will recommend the mandated use of computer-based patient record systems. Actually, the group's mission is uniform data standards for patient medical record information and its electronic exchange. Thus, the name should more closely fit the mission. He proposed "Workgroup on Patient Medical Record Information," the language used in the law. The word "Standards" could be added at the end. Dr. Fitzmaurice responded that this connotes just the right focus, and Ms. Fyffe agreed. Dr. Kolodner and Dr. Greberman also supported the proposed name change.
Ms. Humphreys commented that although "medical" is in the charge to the Work Group, the word "health" may be more appropriate becasue it encompasses public health and other types of health data. This led to a discussion of the group's mission and long-term objectives.
Mr. Blair asked for an informal show of hands (members and staff) regarding the proposed changes, with the following results:
Dr. Cohn explained that the proposal would be referred to the Executive Subcommittee and, if approved, presented to the full Committee in June.
In response to a question, Dr. Cohn said he doubted that the Work Group would have an interim report ready to submit to the full Committee at its June meeting. If anything comes out of the May hearing that needs urgent communication to the Secretary, the Work Group can vet a draft letter at its June meeting. He reminded the group that June 22, the day before the full NCVHS meeting, is also scheduled for hearings. There were no responses to his request for comments on anything that should be brought to the full Committee's attention in June.
The meeting was then adjourned.
I hereby certify that, to the best of my knowledge, the foregoing
summary of minutes is accurate and complete.
/s/ Jeffrey S. Blair February 2000
_______________________________________________________
Chair Date