Washington, D.C.
The Subcommittee on Standards and Security of the National Committee on Vital and Health Statistics held a public meeting on March 20-21, 2001, at the Hubert H. Humphrey Building in Washington, D.C.
The Subcommittee on Standards and Security held hearings on March 19-20, 2001 in an ongoing process focused on HIPAA administrative simplification. During the two days, the Subcommittee heard 15 perspectives and talked with three panels populated by representatives of: standards development organizations (SDOs) and leading message format standard developers; vendors and public health, hospital and other healthcare providers. Testifiers critiqued the framework for proceeding with patient medical record information (PMRI) standards, criteria for selecting standards, a questionnaire for SDOs and areas of focus.
Mr. Dickinson addressed key provisions of the HIPAA privacy and security regulations. He identified where gaps remain in HIPAA inter-application interchange requirements and the need to devise and implement standards-based interchange solutions uniformly across all applications in the healthcare provider enterprise, enabling a common HIPAA privacy and security infrastructure and, ideally, a single point of administration. He pointed out where interchange standards exist, draft standards or implementation guides are in progress, and/or substantive gaps remain.
Dr. Huff advised keeping costs appropriate to benefits by relying on modest advances in existing standards. Standardization needs to be evolutionary, allotting time to achieve returns on investments. The first priority in selecting PMRI message format standards is improving efficiency and effectiveness of healthcare: data comparability, data security and integrity, and interoperability are essential. He recommended adopting standards based on readiness and potential value to society. Dr. Huff advised asking SDOs the extent the standard met the needs of veterinary medicine and about public domain and commercial tools and materials for implementation and training. He suggested additions to the proposed transaction list, cited the need for implementation guides, and advised rewarding those supplying data according to approved standards.
Ms. Gilbertson said NCPDP’s aims to work with an existing standard, with modification if necessary and possible. NCPDP believes the HL7 standard is the best fit. She said the biggest difficulty with interoperability is agreeing on the data elements, values and meanings. She noted the need for a common language and discussed a prescriber/pharmacist interface ANS standard, SCRIPT. She recommended a narrower scope for the first phase of selecting PMRI standards, focused on specific business functions.
Dr. Behlen expressed concern about the selection process’s scope and specificity. He noted the difficulty of measuring and weighing the criteria and a lack of questions about data permanence and migration. He advised engaging SDOs to converge on a set of interoperable standards and cautioned that pressure to achieve rapid consensus yields too much optionality--the enemy of interoperability. Dr. Behlen recommended NCVHS narrow the scope of PRMI standards selection, focusing on transactions sending results to the permanent patient record. He encouraged the Federal Government to provide fellowships to participate in standards work. And he encouraged public licensing and support of free vocabularies.
Ms. Kratz emphasized quality of service requirements for PMRI standards and end-to-end performance issues. She noted message format standards do not adequately support emerging business models the criteria address. And she encouraged broader scope in the first phase of standard selection: otherwise new business models and contemporary engineering advances might be excluded. She suggested: (1) issues of end-to-end performance are indicators of interoperability, (2) characteristics of data quality, accountability and integrity should address end-to-end performance criteria (measurement criteria should be included), and (3) best practice guidelines could identify viable implementations. Useful indicators of market acceptance vary between communities. She recommended identifiable cost include costs (and savings) for implementation and she remarked on the significance of return on investment (ROI) figures. Noting the proposed list of transactions did not consider business processes, she recommended first identifying areas PMRI standards must address. And she suggested identification of workflow processes might lead the way to standard selection.
Mr. Steindel noted CDC’s requirements for reporting specifically describe data quality, accountability and integrity. He underscored the significance of interoperability and data comparability in public health’s ability to communicate across geographic borders and make rapid decisions. He also expressed CDC’s concern that the standards have reasonable implementation costs. CDC’s highest priorities for PMRI transactions are: admission/ discharge/transfer, laboratory order and diagnostic study information, laboratory and diagnostic study results, and treatment/medication orders. He recommended SDOs should be charged with producing a standard that meets most users' needs in a timely fashion.
Review of Testimony - Subcommittee
Dr. McDonald noted none of the targets for security inhibit the immediate goals discussed for communication of data among those with the right to it. Dr. Huff advocated incremental improvements, starting with HL7 Version 2.4 and implementation guides allowing the specification of terminology and stylistic variables that make the data comparable and complete. Dr. McDonald observed that “completeness” and “comparability” were hard targets to meet in real life and suggested more definable endpoints. Dr. Huff said implementation guides for some common data areas could be developed in a year and an implementation program could evolve in five-to-seven years. Ms. Gilbertson noted SCRIPT had an implementation guide, an ANSI-accredited version, and was actively used in the industry. Dr. Huff considered it basically a text field needing additional vocabulary and description, as well as standardized coding sets, to make it truly "standard" with common semantics. Dr. Fitzmaurice said the discussion on the need for a coding system was right on target; VA, FDA, NLM and others were working on developing a system and vocabulary. Ms. Gilbertson said NCPDP would welcome the code sets Dr. Huff described; they would fit with patient safety and clinical as well as business processing. Dr. Behlen suggested a demonstration project could attract commercial resources and draw vendor participation. Testifiers gave highest priority to interoperability, data quality, and data comparability as criteria. Mr. Blair noted that HL7 Version 3 was established on a formal reference information model tightly tied to vocabularies: ultimately, there would be strong semantics. Ms. Kratz recommended the group look at work underway in Europe, Australia, and Japan on electronic health records (EHR) using advanced technologies and cited telemedicine as a business model. Testifiers agree on the importance of international awareness and standards.
PANEL 2: Testimony of Vendors and Providers
Dr. Chute emphasized: a focus on understanding barriers to accessing existing and proposed standards, a plea for computable algorithmic formats, and the importance of interstandard comparability in the context of shared models. He expressed concern about consistency and whether or not standards would adopt principles conveyed in the criteria. He called for public and/or government underwriting of standards development and affordable, or publicly available, standards. If this wasn’t economically feasible, he urged regulatory actions so standard specifications do not create monopoly environments and exploitive pricing. Dr. Chute said patient records standards should be fully specified and implementable at a computer data level, with implicit, executable logic systems that contain internal rule sets and computable criteria so human judgment is not needed at every stage of data element entry, interpretation, mapping, and recognition. In a shallow model environment, a core clinical concept could be described and associated in a consistent templated way with notions of severity, anatomy, and pernicity that could be intelligently inherited by domain-specific consideration.
Dr. Cimino said he was pleased to see that the patient is a central feature of the document. He noted the importance of maintaining the original meaning of the data when sharing information across systems. He suggested another criterion was employing high-quality terminologies. Dr. Cimino recommended expanding queries on data quality to address duplication of data capture, a data model that considers all the needs of all users, and criteria for selecting terminologies. He advised asking how many people participated in the standard’s development and about how its effectiveness in meeting their needs is evaluated. He also suggested adding alerts and reminders to the proposed list of transactions.
Dr. Rozen expressed concern that this was an old, failed paradigm that is not patient-centric and ignores advances in communication technology. He noted concepts of a patient/consumer health record, patient access, or storage of information are not incorporated within the PMRI. Issues of patient access as well as ownership, control, and 7 x 24 availability of healthcare information needed to be addressed. Compliance should be weighted to facilitate comparison between competing systems; demonstration projects could identify implementations and standards that truly add value for the patient. Noting that most health information resides with patients and their families, Dr. Rozen said current and evolving clinical processes need to be recognized so systems implemented can benefit from these changes. He suggested adding questions about standards yet to be evolved, standards to facilitate posting and remote accessing on a hospitals website, what happens to functionality when the keeper of the PMRI changes, and authentication in the absence of a single unique personal identifier. To facilitate criteria for selection, he advised: a review of the definition of the PMRI, timely communication, and weighting the criteria to facilitate selection based on compliance with standards.
Mr. Rubin stated that open interface specification and other initiatives provide loose coupling needed for disparate organizations to interoperate while doing business autonomously. He urged the group to recognize issues extending beyond national boundaries. For maximum PMRI utility, organizations must be capable of extending the standard to address "external" interests without sacrificing the fundamental interoperability offered by PMRI interchange. Noting much attention is paid to "customization," he cautioned this capability must be differentiated from "extensibility." Mr. Rubin urged the group to broaden its view of integration technologies to include alternatives such as middleware solutions and to distinguish between concern for care of the Patient Medical Record and its transport. He advised weighting criteria toward interoperability, data comparability, flexibility, and relationship to other standards. He also proposed altering consistency with other standards and shifting emphasis from alignment to synergy, so touch-points between standards are explored and understood. Market acceptance should be viewed in the context of a changing landscape. And he advised expanding the scope of questions addressing timely standards development and flexibility to include: separation of concerns, vehicle for comparison of standards, and open registration authority.
Mr. Seliger recommended including dependence, relevance, and relationship of the standard among the questionnaire’s lead inquiries. He suggested that the extent the data is accessible to the end user and what is required for the end user to make sense of them are important aspects of data quality, accountability, and integrity. He proposed giving context management standards high priority, noting they could minimize, if not obviate, certain interfaces. And he observed technology neutral standards transcend programming language and system implementation. Who uses the standard is an important criterion that transcends numerical percentage of market acceptance. He recommended clarifying that the SDO publishes conformance statements independent of commercially available tooling. He noted the good practices to be found in CCAL and information model development.
Mr. alSafadi concurred with the questionnaire’s emphasis on conformance, noted links to an SDOs’ ability to create software and standards usable internationally, and recommended that the National Institute on Standards and Technologies (NIST) increase its activities in software diagnostics and conformance testing. He advised expanding the evaluation of data quality to include providing and adopting definitions for character models, digital representation of characters, transcoding, normalization, and adoption of a reference-processing model for that element and data. He recommended that the emphasis on consistency with other standards focus on technology, the Internet, and international standards. Predicting future standards will be jointly developed, he suggested asking about processes to expedite development when multiple SDOs are involved. He suggested adding to the proposed list of transactions: (1) medical images and waveforms to radiological images and (2) the standard being developed for structured reporting that Phillips supports as DICOM and HL7 collaboration (laboratory and diagnostic systems to EMR and HIS systems).
Review of Testimony - Subcommittee
Speakers had said the market shouldn't decide and suggested considering emerging technology. Dr. McDonald asked who decided. Mr. Seliger advised looking at market forces across domains: trying promising things from other vertical segments. The group noted SDOs had difficulty tracking market acceptance; the questionnaire offered different ways to represent their degree of market acceptance. Dr. McDonald proposed helping things already “ripping along;” Government should not decide. Mr. alSafadi said technology companies would listen when healthcare presented itself as a group; when interaction and liaisons between SDOs were embodied in organizations and technology, benefits could accrue. He expressed concern at the impedance mismatch between healthcare SDOs and other Internet SDOs. Dr. Chute said a selected standard could be kept from becoming a monopoly by public buy-out (making it “public domain”) or regulating so revenues are only applied to the standard’s development, maintenance, and improvement.
Dr. Rozen noted the greatest difficulty is semantic interpretation (how to interpret a patient’s expressions into medical terminology). He described Wellmed’s four-year effort with standards organizations on a consumer health catharsis that, in its infancy, has over 23,000 terms. Mr. alSafadi advised learning how each SDO sees the scope of its message format standard and the way it relates to others. What process was used to expedite the standards development cycle when more than one SDO was involved? He said what was missing, beyond a statement of liaison, was a formal process that engages collocation of meetings, gathering and communications. Many mechanisms could be employed; Mr. alSafadi advised determining what each SDO put in place. Mr. Seliger reported that VHA tried a "crosswalk" of international models: activities were too disparate; but asking organizations to classify themselves took them further. He encouraged objective metrics that let SDOs classify themselves. Dr. McDonald recommended SDOs tap overlapping expertise by dividing up work and cross publishing. He observed that the marketplace “sorted it out.” And he said the questionnaire was meant to elicit first cuts at standard suggestions. He noted their inclination not to mandate with the force of the administrative standards, but encourage standards far enough along to be relevant. Dr. Rozen said Wellmed worked in a unique, cutting area that changed rapidly and would outpace any standardization effort. But he expressed concern that this felt like the old model, and urged them to take it to a higher plane.
Dr. Chute noted the desirability of overriding meta-standards to facilitate communication and the lack of an adequate forum where standards issues could be on a bilateral or a multi-lateral basis, with consensus. He questioned the preferability of the emergent model of a mega-standard de facto taking over interstandard interoperability issues. Dr. McDonald noted the need for a controlling organization. He said the choices were collaborate or let the market play it out. Dr. Chute asked about a forum where this could be accelerated. The group noted they were still adopting a HIPAA standard that everybody had agreed was ready in 1996. The group said they will look at standards they are told are closest to prime time, apply criteria, and justify a recommendation. People could build off that. The market will either coalesce and begin adopting faster or come up with something better. Mr. Blair noted suggestions and different perspectives they could incorporate, based on testimonies. He outlined the first phase of the process: using this input, the group will modify the questionnaire and distribute it to SDOs for feedback in June. A matrix or tables will be compiled and analyzed and, in September/October, SDOs will be asked for clarification and additional information. A report will be distributed to stakeholders in January and recommendations submitted by February. The group still had to determine whether the next phase was medical terminology, content, or middleware. There would be more solicitations for guidance. Nine months from now, things would crystallize. Dr. Cimino said the issues were enumerated well and organizations were ready to respond. Dr. Chute concurred, urging the group to consider desiderata. Dr. Rozen said the path chosen was clear; they should move forward, telling the Committee PMRI should be re-examined. Without another look at the vision, patient input could be lost.
Mr. Rubin asked about the process for soliciting from vendors and provider organizations. Mr. Blair said they hoped a focus on market acceptance would identify standards vendors and users valued. Mr. Rubin pointed out that a vendor’s opinion was not a direct or certain indicator of acceptance. Dr. Cohn noted they would hear from provider groups and vendors the next day. He noted they were trying to identify standards that already had “road testing and needed a push.” People couldn't be asked to “blue sky it.” Mr. Rubin suggested the Committee contact organizations participating in SDOs, inquiring at the executive level about the standards. Dr. McDonald suggested asking which subset SDOs thought “most right,” inviting considerations beyond the list. Noting the ISO reference model for open distributed processing, Mr. Seliger emphasized the need for a terminology that could build an objective metrics piecing together disparate input. Dr. Rozen expressed confidence that seeking public opinion and reaching consensus would enable them to advice the Secretary.
Review of Testimony about Elements of the Proposed Process
Final rules were published on privacy and the transaction standards and code sets. Others slated for this year include: national provider identifier and system, employer identifier number, and security standards. Proposed rules for claims attachments and a unique identifier are underway. The Committee's PMRI recommendations were presented to the Data Council and agencies initiated projects. The PMRI standard and an appropriate vehicle for implementation remain to be determined; steps other than a mandate could be taken. Tentative recommendations will be determined within eight months. In terms of triage, an update on recommendations underway to “go and do good work” will be given at the full Committee meeting. Recommendations on criteria employed in regulatory actions will be evaluated in light of the proposed regulation. The Data Standards Committee will serve as focal point for evaluations and referrals not linked to a specific agency. Concern was expressed about proceeding with PMRI until there is a privacy framework. A report on progress made on PMRI recommendations will be made at the Subcommittee or NCVHS meeting. The full Committee approved the annual report to Congress on HIPAA.
Mr. Scanlon reported some agencies had plans for the increased research recommended. The recommendations circulated when budgets were unknown and there had been a reluctance to respond. Next round, they would get a clearer picture of what agencies could support. The group discussed mechanical difficulties of surveying users and the need for input from credible industry sources beyond SDOs. Dr. McDonald suggested random sampling of institutions. Mr. Blair proposed hearing from the broader provider community in September. Testifiers had different priorities, but the group noted everyone approved the criteria. Dominant priorities were interoperability and comparability. The subcommittee heard that government-chosen standards could lead to a monopoly and that other ways might operate just as well. They noted a precedent (X12) for sponsoring or subsidizing a standard. The group also heard that the cost is not the standard, but implementation. They discussed looking at the testimony to see how they might adjust their scope, while not giving up on the “low hanging fruit.” Something could be done in a year--and more after that.
The group considered adding questions about patient or consumer information to the survey and noted to ask CDC and HL7 about a mature standard for the transfer of information about immunization transactions. They discussed expanding the questionnaire 10-15 percent, so it did not overwhelm some 24 SDOs targeted. The group talked about hiring a consultant to pull together the analytic work. They noted Government rules might require clearances for a survey and considered a pre-questionnaire and hearing as another way to gain input. Dr. McDonald asked about a way to sample request for proposals (RFPs) from institutions. Mr. Blair suggested getting guidance directly from consultants that prepare them. Dr. Cohn reported SDOs had asked to give feedback on changes to the standards at the May 31st-June 1st meeting. Hearings will probably be scheduled in the summer for consultants to talk about clinical standards. The group discussed a hearing about enforcement and compliance of PMRI standards, hopefully after a notice of rules on enforcement. Time was set aside in September; three other topics are also candidates for discussion. The group noted their intention to address patient identifiers when the privacy and security areas are handled and appropriated funding is no longer blocked. They will query about August and October hearing dates. September and November are full Committee meetings. The survey will be distributed to SDOs in May, analyzed in the summer, and reviewed in September. Clarification and additional answers could be sought from SDOs in the fall. Recommendations for individual transactions and the report would be distributed in January for critiques and submitted for full Committee review in February. Dr. McDonald proposed that SDOs could come up with a way to take care of the few standards where SDOs “bumped up at the edges.” Participants discussed gradations of recommendations. Implementing a mandate that all data must come in one format from all sources might take six years; useful things with lower impact might be done in six months. The group discussed encouraging the Secretary to take action to analyze and fill gaps.
PANEL 3: Testimony of Standards Developing Organizations About Elements of the Proposed Process
Ms. Fuller urged the Committee to create a feedback loop between SDOs, vendors, and other implementers. She recommended the group: encourage formal ROI studies, promote case studies and best practices, adopt official or approved standards for certain classes of messages, require SDOs to monitor implementation and update standards based on feedback, identify overlapping standards and work to reduce competition between SDOs. She suggested SDOs evaluate their compliance with the criteria and be encouraged to address perceived barriers to adoption of their protocol. She noted the data and document sets integral to the claims attachment standard need to be better defined and called for additional standards in the areas of medical terminologies, data quality, accountability and integrity.
Ms. Stevens suggested asking if SDOs support alternative vocabularies adopted for different jurisdictions and functional domains--and how they are reconciled. She pointed out the market acceptance was linked to both level of participation in its development (Does the SDO's developmental structure support and encourage input from stakeholders?) and an application vendors’ commitment to the standard. She recommended vendor and provider questionnaires. She advised adding to the proposed list of transactions EMR exchange between systems and identification reconciliation from one ADT, HIS or EMR system to another. In determining the flexibility to respond to new requirements, Ms. Stevens noted the standard should be based on a thorough, stable information model, and the message specifications should be staple, but able to change with business needs. Implementation technology must be able to adjust rapidly to technological improvements without negatively impacting message specifications or the underlying model. Ms. Stevens said it was critical to assess the impact on the industry when the standard is selected. Noting the complexity and size of information standards, she advised publication in a database, XML or tool format to assist implementation.
Mr. Arges expressed concern about any additional regulatory efforts. Massive, complex and costly information system changes already challenged hospitals’ viability. Finding necessary funds to adopt changes would be crucial. Even more troublesome, many proposed changes offer little immediate administrative relief, yet compete with day-to-day expenditures for basic patient care services. He recommended a study examining financing options that looks at ongoing costs and challenges. He also encouraged exploring more fully the data gathering and development process. The functional goal was to free up the caregivers' administrative workload and ultimately yield complete documented and standardized medical records. He suggested case studies could show how some hospitals approach adoption, the context, method of financing, challenges faced and benefits gained. And he emphasized the importance of demonstration programs that test the validity and quantify costs associated with adoption and regulatory relief and financial support for hospitals--especially with unfunded mandates.
Review of Testimony -- Next Steps
The group had heard that the provider community had limited resources and was weary with unfunded mandates: they should look at incentives and keep in line with standards that had market acceptance and could act as a catalyst to mitigate implementation costs. Ms. Fyffe found it troubling in this environment to advocate increased use of technology and urged the group to brainstorm suggestions for this cost problem. Noting that Recommendation 5 indicated HHS should provide support for cost benefit studies for PMRI standards and Recommendation 9 indicated HHS should explore incentives to provide for equitable distribution of costs to help providers implement, the group discussed looking more at cost issues and burden on providers and making recommendations, primarily targeted at Congress. Nobody “out there” was opposed to PMRI standards. There was some resistance to changes, but everyone saw potential benefits. Ultimately, financial aspects were central; fiscal reality had to be part of any recommendations. They had to identify solutions. Dr. Fitzmaurice noted they heard the need to start from models and for tools to implement standards and research into costs and benefits. With open process and hearings, he said they could get their first set of recommendations within a year.
Dr. Zubeldia wondered about loan arrangements to get providers installed and ready to go. Dr. Shortliffe speculated about having some hospitals declared disaster areas so they could get loans from FEMA. Ms. Greenberg suggested a recommendation from the Committee to enter some data sets in the Meta Data Registry. Dr. Cohn proposed the USHIK presentation and a discussion for a full Committee meeting. Mr. Blair suggested they also have presentations on the Good Electronic Health Record (GEHR) and Reference Information Model (RIM). The group also agreed to have a presentation on HL7 RIM along with Denise Koo’s discussion on the public health data conceptual model and NED at the June meeting. Ms. Stevens urged the Committee to bring stakeholders together to support inter-SDO activities. The group decided at this point they would not ask SDOs about information models. Trying to reconcile information models was an important, huge, task they would hold to a later date. They discussed asking about standards that should be recommended for the mediation approach, the master patient index, and one that enables a transaction to reconcile multiple identifiers, facilitating patient identification.
Dr. Cohn convened this hearing of the Subcommittee on Standards and Security of NCVHS, noting that the National Committee is the main public advisory committee to the U.S. Department of Health and Human Services (DHHS) on issues health and national health information policy. Dr. Cohn stated that the focus of these hearings is on PMRI standards and next steps following up on the report to the Secretary on Uniform Standards for PMRI.
Mr. Blair provided background on the hearings. Last summer’s report to the Secretary, as called for within HIPAA’s Administrative Simplification Provisions, included recommendations and a set of guiding principles that conveyed an overall framework for proceeding with PMRI standards. The Subcommittee indicated in the report that they would give recommendations for the first phase about a year from now. The group had refined the guiding principles, defined criteria for selecting standards, a list of areas suggested as candidates for message format standards that will be the “low-hanging fruit” in this first phase of the PMRI standards process, and a draft questionnaire for SDOs. The group invited the testifiers to critique that framework, criteria, questionnaire and areas of focus.
PANEL 1: Testimony of Standards Developing Organizations About Elements of the Proposed Process
Unlike most the panelists, who represented SDOs, Mr. Dickinson addressed key provisions of the HIPAA privacy regulation and the proposed HIPAA security regulation. Noting there was relatively little focus on inter-application interchange requirements related to privacy and security, he advised that these implementations are best devised by adopting industry standard specifications. Focusing on the health care provider enterprise, he noted significant gaps remain to be addressed with data interchange standards. He described the need to devise and implement standards-based interchange solutions uniformly across all applications, enabling a common HIPAA privacy and security infrastructure and, ideally, a single point of administration.
He characterized a typical healthcare provider enterprise as comprised of multiple sources and points of access, data stores and interchange points for individually identifiable health information. Mr. Dickinson identified interchange requirements to enable a common enterprise-wide infrastructure that fully engages and uniformly implements HIPAA privacy and security across all applications. He pointed out where interchange standards exist, draft standards or implementation guides are in progress, and/or substantive gaps still remain in HIPAA inter-application interchange requirements to uniformly enable privacy and security across the healthcare provider enterprise.
The first inter-application interchange requirement identified is for a master registry of security policy domains within the health provider enterprise, each with a unique security policy. These may coincide with physical locations or business units (e.g., facilities, departments, services, specialties) within the organization. Interchange requirements include the application function associated with each domain and security classifications, the health record in that subset pertinent to the domain, classifications for access and use, and health parties, agents and roles accountable.
The second requirement Mr. Dickinson singled out is the master registry of application functions involved in the origination, amendment, access, use, display, translation, disclosure, transmittal, receipt, de-identification archive or destruction of individually identifiable health information across the enterprise, including equivalent or comparable functions existing in multiple application domains. Interchange requirements include security classifications.
The third requirement he noted is the master registry of the health record, subsets (identifying discreet data store within the enterprise and including legally sequestered subsets--e.g., psychotherapy notes, information from confidential sources or pertaining to legal actions) and corresponding security classifications. Interchange requirements again include security classifications.
Mr. Dickinson said the fourth requirement with gaps is a master registry (including security clearances) of healthcare parties, agents, and roles accountable for: (1) the origination, amendment, use, and disclosure of individually identifiable health information; (2) the provision, performance, and completion of healthcare services; and (3) agents accountable for the origination and translation of individually identifiable health information. Interchange requirements include organizations, business units, and agents accountable, and their related information. HL7 v2.4 enables exchange of identifiers, demographics, licenses, and credentials for individuals. Remaining gaps include: authorized roles, authentication details and digital certificates for that individual, clearances for application function and information access and use, as well as the interchange of accountable role information (e.g., role ID), and clearances for application function and information access and use related to that role.
Mr. Dickinson said a multi-SDO taskforce coordinated by ANSI HISB is looking at gaps in data interchange or message authentication, which includes user-authentication, data source/ origination authentication, data validation authentication, and data interchange authentication.
Data interchange standards are not adequately in place in chain of trust that: ensures auditability and traceability in terms of the flow of individually identifiable health information between points of origination, use, disclosure, report or claims submission.
Other interchange requirements with gaps include: (1) audit trails for data states allow us to ensure data integrity in terms of its persistence, permanence, and non-alterability. An informative ballot is pending in HL7; (2) audit trails for security events (e.g., successful and unsuccessful sign-on, user signoff, and various system faults reported in a security event audit trail) that presumably will be part of the uniform ballot coming from HL7; (3) requirement for sequestered record sets (including psychotherapy notes, information from confidential sources or pertaining to legal actions) and how they might be transmitted; (4) notice of provider privacy practices call for a notation of the patient's receipt of the privacy notice; (5) need to interchange patient consents for routine use of health information as well as a need for standards to document authorizations for disclosure of designated record sets for particular purposes--patient authorizations include: the explicit description of information, purpose for disclosure, names or identification of those authorized to make and receive the disclosure, and pertinent expiration dates or events; (6) four possible items of record keeping associated with denial of amendment requests (the patient's initial request for amendment, the covered entity’s denial, the individual’s statement of disagreement with that denial and the covered entity rebuttal) must be attached to further disclosures of the designated record set.
Dr. Huff noted that Intermountain Healthcare, a 22-hospital healthcare provider in Utah and Southern Idaho, has interfaced many of the ancillary and patient care systems Mr. Dickinson mentioned to provide electronic medical records. Each day, some 60 different kinds of interfaces convey over 500 point-to-point interface instances that result in about 900,000 individual pieces of patient data being added to the electronic medical records.
Dr. Huff said fine-grain structured standards enable having the data needed to understand what is happening with patients and appropriately allocate. He noted the cost will be substantial and the benefits to society do not, in and of themselves, justify investments by individual organizations to restructure systems and procedures. Dr. Huff said, at some point, standards relying on modest advances to existing standards need to be mandated. Otherwise, cost will be inappropriate to the benefit. Standardization needs to be evolutionary, with time allotted for organizations to achieve returns on investments as they pass through the adoption stages.
Dr. Huff suggested that improving the efficiency and effectiveness of healthcare should be the first priority in selecting PMRI message format standards. Data comparability, data security and integrity, and interoperability were essential. Other key criteria include: (1) implemented standards already working in a broad range of production clinical environments; (2) cost-effective standards measured in terms of market acceptance, availability of public domain and commercial tools to support and implement them, with education and training in place; (3) sustainable standards with stable organizational support, consensus participation, and change management (flexible responses to new requirements, timely corrections and enhancements, and ability to preserve business knowledge despite rapidly changing technology). He noted the importance of process, a formal information model, and an ability to transfer knowledge version-to-version.
Dr. Huff said the proposed questions for the SDOs were appropriate and suggested asking about the availability of public domain and commercial tools for implementation; training materials, tutorials and other educational tools; and the extent the standard met the needs of veterinary medicine. Technical training should focus on the standard and how it works, as well as the fields and their meaning. He defined practical training as how to implement: programming tools, the environment, and software sources. He noted difficulty understanding the level of granularity to use in responding, and he advised using separate questionnaires for each transaction domain. He also suggested the proposed set of PMRI standards could serve as a guide for transaction domains.
Dr. Huff suggested three additions to the proposed transaction list: patient’s chief complaint, problem list, or list of diagnoses; visual integration; and procedure scheduling (e.g., radiology, surgery, endoscopy). He prioritized the list: ADT; standard orders (clinical lab, anatomic pathology, radiology, ancillary services); standard results; in-patient medication orders to pharmacy systems; clinical documents; chief complaint, problem, diagnoses; images; visual integration; data from bedside instruments and monitoring systems; orders for outpatient medications to pharmacy systems; procedure scheduling; and charge capture information to billing systems.
Noting existing standards do not integrate clinical terminology in a tightly coupled manner, Dr. Huff cited the need for implementation guides. CDC had already produced an immunization implementation guide. Others need to be done for laboratory and radiology orders and results, medication orders, and microbiology culture results.
Dr. Huff suggested the best initial implementation strategy might be to reward those supplying data according to approved standards rather than mandating usage. He suggested the government could pay a per- case or per-transaction premium. Once broadly implemented standards proved effective, a transition plan could be mandated. Meanwhile, necessary infrastructure (e.g., implementation guides, technical and practical training) could be developed.
Dr. Huff recommended priorities for adopting standards based on readiness and their potential value to society. He suggested beginning with reporting clinical data to third parties (e.g., HIPAA claims attachments), and reporting to governmental departments, offices, and agencies (infectious disease reporting and immunization information to state and Federal agencies, tumor registries, HCFA chart review, as well as adverse drug events and clinical trials information to the Food and Drug Administration). Noting the pertinence of early reporting of certain veterinary data to human health and well-being, he recommended adopting next standards for reporting veterinary data to governmental departments and agencies. Other priority areas he ranked high included: (1) reporting clinical trials data to private companies, (2) reporting to national professional databases (e.g., cardiology including myocardial infarction and open heart surgery; mother's and newborns including birth defects and prenatal and neonatal care; and endoscopy); (3) data exchange between healthcare enterprises; and (4) data exchange within a single healthcare enterprise. Dr. Huff advised the greatest “bang” for the least work and cost would come from adopting these standards.
Dr. Huff emphasized the need for a plan for evolution including new standards, new versions of standards and a well-defined process for prototyping and implementing the next generation of standards in production systems before general adoption. He said it should be done systematically, inline with regulations, and mandating their use once there was proven success.
Ms. Gilbertson, Director of Standards Development for the ANSI-accredited SDO, said NCPDP’s goal is not to develop a new standard but to work with an existing standard and, if necessary and possible, request modifications to meet their needs.
Areas of interest to pharmacy include: sharing of prescription information with other health care providers (including other pharmacies), allergies, lab results and diseases. NCPDP believes the HL7 standard is the best fit. The NCPDP Work Group is studying a Patient Profile standard developed by HealthNet in British Columbia and intends to work with HL7 to develop a standard implementation that meets the needs of all Health Care providers.
The most difficult aspect in terms of interoperability, Ms. Gilbertson observed, is not exchange of information between computer systems, but agreeing on the data elements, values and meanings. She noted the need for a common language derived either by defining one language or providing a mapping/translation correlation. She said code sets, identifiers and values are far more important than structure or the transportation mechanism.
NCPDP does not have applicable standards in use specifically for PMRI. NCPDP, an ANSI SDO, is responsive to the industry's needs using the Standard Operating Procedures to ensure industry consensus and approval processes.
Ms. Gilbertson discussed a prescriber/pharmacist interface ANS standard, SCRIPT, which NCPDP created that supports new prescriptions sent from the prescriber to a pharmacy, compliance transactions, refill requests and responses from pharmacies notifying the prescriber to monitor patients. She said NCPDP felt that this business function should be included in PMRI, but that SCRIPT should be recognized as the standard. She reported NCPDP was not sure that intra-institutional charge information, which appeared to be a billing event covered under HIPAA transactions, fit into PMRI. She noted that there might be a need for a simple request that gives patient demographic information, pharmacy and medical allergies. Ms. Gilbertson emphasized the importance of looking at the business functions of other transactions; she said the standards should encompass separate business needs, and separate functions should be under individual implementation timeframes/schedules. PMRI should not be grouped into one timeframe unless analysis indicated all the business entities were tied together and needed to converse.
NCPDP recommended a narrower scope for the first phase of selecting PMRI standards. Noting that healthcare industries could not share all their PMRI information soon, she suggested choosing specific business functions and developing sharing of information electronically, adding other business functions after learning from these initial experiences.
She noted a questionnaire is a fine way to get the word out and feedback. But she cautioned that each response came from a different perspective and might entail totally different goals. She also expressed concern about how updated information would be managed and maintained among different groups (e.g., doctor, pharmacy, hospital). Each standard must address what systems did with updated information--did they believe it or wait until they saw the patient? What if differing sources gave differing information? Ms. Gilbertson noted that today’s flaws and inconsistencies would not be fixed just by moving electronic; the foundation must be stable regardless of the medium.
Speaking personally rather than officially, Dr. Behlen considered the criteria questionnaire and transaction set reasonable and well thought out, but he expressed concern about the selection process’s scope and specificity. He observed that the criteria for selection were valid, but noted it would be difficult to measure and weigh these criteria in choosing a standard to recommend under HIPPA. In prioritizing, Dr. Behlen said he would place data quality near the top.
Dr. Behlen noted a lack of questions about data permanence and migration. How can systems implementing the standard migrate data to future systems? How can loss of information content in multiple migrations over many years be avoided? How does the standard define completeness in transfer of a patient's record? Does it specify a storage format for archival data? If it includes such a specification, is it based on a general-use file system or standard-specific media-- can stored data be transferred from one media form to another with generic file-copying operations? He said it was a good questionnaire and he thought the survey and analysis might help HHS plan an approach to promoting standards for PMRI. But he doubted it would lead to selection of a winning standard.
He remarked that the proposed transaction set is reasonably complete. Noting that images are also produced in endoscopy, pathology, dermatology, cardiology and other disciplines, he proposed expanding "radiological images" to "medical images and waveforms." He pointed out that images and waveforms produced in diagnostic procedures could be subject to longer retention requirements as they become part of the medical record.
Dr. Behlen commented on a need to clarify the specificity of the selection of PMRI standards. Was the goal to specify a fixed set of transactions and select a single "winner" for each, or was it to engage a number of SDOs to converge on a set of interoperable standards? He recommended the latter, noting the process is not as simple as defining transactions and selecting the best for each. The various standards mostly occupy different application domains. He said the hard work ahead was bridging between standards and interfacing between systems serving these application domains.
Dr. Behlen suggested the Federal Government might provide fellowships to participate in standards work, perhaps awarded through the NLM by peer review. Travel fellowships could help offset reduced provider information technology budgets and standards-related grants might encourage participation by academic informatics. He noted that public licensing and support of free vocabularies would save considerable effort and remove an impediment.
While acknowledging that SDOs could benefit from help that enabled acceleration of the process, Dr. Behlen cautioned that pressure to achieve rapid consensus yields too much optionality--the enemy of interoperability. He emphasized that high interoperability requires little optionality, which often takes many rounds of negotiation and use-case review.
Dr. Behlen recommended that, at this point, NCVHS narrow the scope of PRMI standards selection, focusing on transactions sending results to the permanent patient record, and administrative functions managing results within the permanent record. Once standards were selected, profiles or templates needed to be defined to improve interoperability within each standard implementation. He said this was an ambitious project, and one with the greatest long-term impact.
Suggesting that the effects on implementation practice of implementing HIPAA claims requirements will ripple through the healthcare system in unpredictable ways, Dr. Behlen advised the group to wait for results before attempting to standardize transactions handling transient data.
Dr. Behlen noted the significant progress made by CEN in Europe and GEHR in Australia. And he remarked that standardization of storage formats, as well as message services, was a powerful asset in the maintenance of long-term records.
Ms. Kratz stated that feasible EHR systems require a broader focus than terminology and current message based standards. She emphasized the importance of considering quality of service requirements for PMRI standards and end-to-end performance issues. She noted PMRI standard criteria need to address emerging business models for the information economy, and message format standards do not adequately support emerging business models. She advised that PMRI standards criteria should not constrain the development of new market avenues of potential benefit to the medical domain.
Noting that the indicators of interoperability section of the draft questionnaire seemed to focus on customization of applications and implementation guidelines, Ms. Kratz remarked on the need for a definition of implementation guides as this varies within healthcare SDOs’ communities. She said issues of end-to-end performance are indicators of interoperability and play into the customization issues identified in the PRMI report. Where formal conformance criteria are not available, best practice guidelines prove to be an effective mechanism to identify viable implementations.
Ms. Kratz noted that the indicators of data comparability questions implicitly assumed coded vocabularies would offer functional semantics to enable data archive and retrieval. She advised that questions were too prescriptive in the use of tables referenced by message structures. Criteria should be included to address object-oriented frameworks or they should be more general.
Ms. Kratz said characteristics of data quality, accountability and integrity should address end-to-end performance criteria and measurement criteria should be included. Measurement criteria should address: (1) the definition of a standardized set of comparable measurement criteria implemented on a range of platforms; (2) existing measurement, monitoring and diagnostic tools; (3) rigorous documentation and analysis of core measurement infrastructure; (4) identification of application architecture and system infrastructure based on implementation experience (iterative performance testing identifying patterns of success); (5) advertising current best practices (basis for conformance criteria) to relevant communities.
Useful indicators of market acceptance, Ms. Kratz advised, varied with the community queried. She questioned if targeting only one community produced a meaningful number.
Predicting questions about consistency with other standards would bring forth an interesting array of Memorandums of Understanding and liaison relationships between SDOs, Ms. Kratz said the questionnaire might also address the ability to fast track a specification between SDOs to identify where they work together.
Ms. Kratz suggested the section on identifiable cost might consider costs (and savings) for implementation as well as development of the standard, and she noted the significance of ROI figures.
Ms. Kratz pointed out that standard organizations that deal in "Internet time" require standards to have a commercially viable working implementation, while ANSI does not. In the timely standards development procedures segment, she suggested PMRI candidates should provide reference implementations of the standards under evaluation, with associated tool sets and conformance criteria or best practice guidelines. Many SDOs, she observed, use a fast track or request for consideration process to expedite commercial implementations that are defacto industry standards through the ballot process.
She recommended broadening the scope of the flexibility to respond to new requirements section to include more "big picture" criteria for HCI, intelligent interfaces, middleware, operating systems, computational resources, network and mass storage, as these all have dependencies for end-to-end application requirements.
Ms. Kratz noted the proposed list of PMRI transactions to be considered for HIPAA standardization does not consider business processes. She recommended it begin by identifying, not existing standards, but areas PMRI standards need to address. She suggested identification of workflow processes might provide a path to standard selection and said user authentication and patient identification are good starting points.
Ms. Kratz mentioned a collaboratory on EHRs formed by Internet2 Health Sciences to address requirements of complex academic health systems that seeks international collaboration, based on open source sharing of intellectual properties called Open Electronic Health Records. Participants evaluated GEHR, launched by the European Community with an information architecture that has enabled independent implementations of EHR systems in different countries. The communication model of GEHR is flexible and scalable.
Contrary to other panelists, Ms. Kratz said a broader scope should be considered in the first phase of PMRI standard selection. By excluding appropriate standards, new business models and contemporary engineering advances may inadvertently be excluded from PMRI selections. This could prove detrimental to the industry.
She reported that EHR has growing international consensus as the terminology of choice. ISO TC215 has a specification in process on requirements for an EHR reference architecture, which recommends EHR as the terminology of choice.
Mr. Steindel, a Supervisory Health Scientist with the Division of Laboratory Systems, Public Health Practice Program Office, CDC commented as CDC liaison to several SDOs.
CDC’s requirements for reporting specifically describe data quality, accountability and integrity, and its programs frequently include information based on analysis of data primarily collected, through the Morbidity and Mortality Weekly Reports. He noted that CDC’s use of data quality does not generally refer to the accuracy of data for administrative purposes (though that data is used in vital statistics programs), but to the clinical accuracy of the data, extending that description to include other attributes of the event described in various case-definition/description documents.
Mr. Steindel noted the importance of data operability and interoperability in the ability of public health to communicate across geographic borders and make rapid decisions. This requirement might be days with a swimming pool outbreak of Ecoli 0157--or instantaneous with bioterrorism.
He said CDC successfully used XML to relate clinical concepts in the HL7 RIM to its developing Public Health Conceptual Data Model. Interface engines put systems in place.
Mr. Steindel emphasized several points about interoperability and data comparability: (1) CDC (which is primarily looking at SNOMED and LOINC vocabularies) needs semantic precision in the data and an ability to relate vocabularies is as important as the vocabulary used; (2) dealing with consistently changing scales and norms of measurements, CDC needs a system that allows the ability to identify scales and when and how to interrelate them; (3) concepts expressed in CDC’s case definitions are complex and can change rapidly: a standard must have flexibility to express new concepts and capacity to maintain and extend them; (4) many of CDC’s concepts refer to a group, while clinical vocabularies focus on the individual. CDC is working with HL7 to include the concept of the group in the developing RIM. Mr. Steindel said CDC envisioned this being done electronically and was focusing on XML to provide syntax.
Mr. Steindel conveyed CDC’s concern about the selection of standards with reasonable implementation costs, so the reporting of public health information does not require burdensome additional expenditures by their partners and data sources. He noted they would like external interfaces of PMRI systems so they could develop once to common standards and utilize this repeatedly for efficiency. Their expectation is that widespread use of standards and electronic reporting should materially reduce cost and time required for public health reporting.
Mr. Steindel said CDC has real needs for all of the PMRI transactions mentioned on the proposed list. The first four candidates are CDC’s highest priority: admission/discharge/transfer, laboratory order and diagnostic study information, laboratory and diagnostic study results, and treatment/medication orders.
Mr. Steindel recommended that SDOs should be charged with producing a standard that meets most users' needs in a timely fashion. He said PMRI is vital to NEDSS’s developing closer integration between public health and the health care systems, which lead to improved provision of healthcare as well as public health.
Review of Testimony - Subcommittee
Dr. McDonald noted that, while there are no perfect security standards, depending upon one’s goal, it is not hard to match HIPAA's security requirements with current non-medically-specified technology and use certificates or secure ID at a remote site plus SSL, with the local host responsible for usual security constraints. National, universal authentication is more difficult. But none of the targets for security inhibit the immediate kinds of goals discussed for the communication of data among those with the right to see it.
Dr. Cohn said he heard general agreement with the criteria. Dr. Cohn observed that the main themes of the PMRI report and criteria: interoperability, comparability, data security, data accountability and integrity lead towards the things HL7 is trying to achieve in Version 3. Version 2 has market acceptance, but not as rich a set of characteristics for interoperability, comparability and quality. He asked how this could be reconciled or balanced.
Dr. Huff advocated incremental improvements in existing standards. He encouraged starting with Version 2.4 and implementation guides that allow the specification of terminology and stylistic variables that make the data comparable and complete. CDC’s implementation guide for immunization has strong semantics. A subgroup is working on the representation of microbiology data. An implementation guide for laboratory, for example, could provide a more specific description of the set of LOINC codes allowed and other stylistic matters. He noted the proposed standard for ex-emalification would enable adding XML encoding. The richness of Version 3 could be accessed in a few years. It would be more mature, in place in production systems, and there would be better tools and education. What was evolutionary, Dr. Huff remarked, was that the information content remained the same and attached to the common vocabularies, adding continuity between V2.4 and 3.0. Demonstrations had shown that, even now, interface engines are easily capable of taking in a 2.4 message and sending out a Version 3 message.
Dr. McDonald observed that “completeness” and “comparability” were hard targets to meet in real life and he encouraged focusing on more definable endpoints. People “cheated”: they didn’t do everything by the standard. “Good structured stuff” ended up in a free text comment field, instead of in the standard.
Dr. Huff said implementation guides for some common data areas (e.g., orders, lab orders, radiology orders, lab results, radiology results, medication orders) could be developed in a year. If transmitting infectious data to the CDC was the next priority after the HIPAA claims attachment, some leading institutions could implement it over a year or two, getting the bugs out. And then an implementation program could be initiated, leading to a mandate two or three years later. Elements could be moved along as separate parallel projects. Dr. Huff predicted it would evolve in five to seven years.
Ms. Gilbertson clarified that SCRIPT, while not yet broad based, is actively used in the industry. In 1996, NCPDP reviewers commented it was for business functions, noting patient information was involved, but not, at that point, actual transmission of PMRI. There is an implementation guide; one version is ANSI-accredited. Dr. Huff remarked that it had many coded fields, but was not strongly coded in terms of the exact product distributed, dose or route forms. Basically, it was a text field. Ms. Gilbertson said it had the ability to identify NDC codes, but noted NDCs did not mean anything today in most prescriber systems. She remarked on the capability for a drug identification field with the text name coming from the physician’s database. Dr. Huff said it needed additional vocabulary and description, as well as standardized coding sets, to make it truly "standard" with common semantics. Dr. Fitzmaurice noted that Ms. Gilbertson had said she felt the SCRIPT standard should be recognized for its business functions, not included in PMRI. Yet, she also mentioned compliance transactions notifying the prescriber to monitor patients using prescriptions. He said many in the group felt the time to remind about other drugs a patient uses is when the physician prescribes medication. He said their discussion on the need for a coding system was right on target. VA, FDA, NLM and others were working on developing a system and vocabulary (or at least incorporating another drug vocabulary) to gain more medical information when decisions were made. Ms. Gilbertson agreed that NCPDP would welcome the code sets Dr. Huff had described; they would fit not only with the business processing, but also with patient safety and the clinical processing of information.
Dr. Fitzmaurice asked if anyone was proposing that PMRI standards include the concept of storage standards and how data content was retrievable over time. Dr. Behlen noted one of the orthodox creeds of standards work was not to tell people how to do things internally. These definitions were solely for interchange. When he talked about storage standards, he was talking about offline, defined ways of packaging message content into something that did not require a specialized “living” computer system in order to read it. He suggested that a demonstration project (a working connectathon showing proven configurations) could attract commercial resources and draw vendor participation.
Noting Mr. Steindel indicated that data quality and interoperability had high priority, Mr. Blair asked how the others weighted the criteria. Ms. Kratz gave interoperability highest priority. Dr. Behlen chose data quality. Dr. Huff said the standard had to meet the need it was targeted for, already be working in a wide variety of production systems, and be cost-effective. Public domain and commercial tools, education, and training had to be available. ANSI standards, open-consensus process, and a stable organization were needed to support and evolve it over time. And there needed to be a way to manage change.
Mr. Blair noted that the justification and rationalization for Version 3 was that it was established on a formal RIM, tied tightly to vocabularies so ultimately there would be strong semantics. He pointed out that information and semantics would last longer than any particular technology. XML was big today; in five years it would be something else. An understanding of the business needs, the semantics of the data fields, and the terminology coupled to them had to be maintained and supported. Ms. Gilbertson concurred: data comparability was the most important criterion on her list, but the business cases, business functions and common set of vocabularies were essential. She also noted the importance of addressing the circuitry and network of the Internet in the near future.
Ms. Kratz recommended the Committee look at the Healthcare Information for Australia report, which has a commission study by Dr. Sam Hurd on the GEHR and another on several implementations the University College, London has where different business processes are collating under a common GEHR architecture. Ms. Kratz commented on work underway in Europe, Australia, and Japan on EHRs using advanced technologies. She mentioned a conversation with a stateside physician in clinical practice who asked when the government was going to do something that could actually help him in his practice. Dr. Behlen hoped the mandate would be for an ANSI standard; he said questions on international standards are important for component markets like imaging where over half the members have vendor operations outside the United States. Noting situations along the borders with Canada and Mexico, Dr. Huff agreed the international aspect is important.
Ms. Kratz cited telemedicine as a business model; through its applications, academic sites gather unique datasets for disease states that, otherwise, physicians in training in this country could not see. Dr. McDonald cautioned not to identify international with ISO. Internet is the international standard--not ISO. A number of international organizations were doing standardization. He also observed that healthcare is mostly a local event that differs greatly because of the economics: e.g., doctors in Japan see their patients every day and notes are only 5 or 10 words. He noted the law gives preference to the ANSI SDOs. Dr. Behlen remarked that in standards work international awareness could avoid mistakes (e.g., assuming at a base level that any alphabetic character is eight bits). Dr. Zubeldia noted that current SDO standards authorized by the Secretary do not allow international characters.
PANEL 2: Testimony of Vendors and Providers
Dr. Chute emphasized three points: a re-emphasis on understanding barriers to accessing existing and proposed standards, a plea for computable algorithmic formats, and the importance of interstandard comparability in the context of shared models.
He said the criteria probably had appropriate depth, breadth, and detail as an information-gathering instrument, and noted the desirability of templateable information. He expressed two concerns: consistency (e.g., was any arbitrary, constant element characterized consistently across providers, systems, and environments?) and whether the criteria were to be used for inquiry or guidance--should standards adopt the principles conveyed in the criteria? Dr. Chute recommended utilizing the US Health Information Knowledge Base (USHIKB) to manage and manipulate information gleaned from the SDOs.
Noting that standards are slow, expensive, difficult, and painful, Dr. Chute discussed what could be done to minimize barriers to adopting standards. He encouraged public and/or government underwriting of standards development. If that proved economically unfeasible, he suggested regulatory actions so that standard specifications do not create monopoly environments that generate exploitive pricing consequences in the US health system. Affordable or publicly available standards should be preferred. Once identified as a PMRI standard, steps needed to be taken on pricing restrictions within specified ranges so revenue would not be reprogrammed away from the development and maintenance of the underlying standard.
Dr. Chute emphasized that patient records standards should be fully specified and implementable at a computer data level, contain implicit, executable logic systems that contain internal rule sets and computable criteria so human judgment is not needed at every stage of data element entry, interpretation, mapping, and recognition. Dr. Chute urged that, to the extent practical, public and patient medical record standards embody internal consistency and logic rules.
Addressing cross standard interoperability and consistency, he noted that Dr. Huff recently proposed the concept of a shallow information model. The hope was that in a shallow model environment, one could have a corps clinical concept described and then associated in a consistent templated way with notions of severity, anatomy, and pernicity--and that these notions could be intelligently inherited by domain-specific consideration.
He said a first step is to compile the existing metadata about standards and he suggested USHIKB could enable SDOs to liaison shared information models and/or XML mark-up standards toward consensus and compliance. He advised that the selection criteria for implementation be the SDOs’ willingness to play in these interstandard consensus environments.
Dr. Chute said the group’s description and first steps towards these criteria were excellent. He recommended discussion about the issue of guidance and prescription, actively discouraging barriers to standards access, encouraging computable formats for data content, and engendering interstandard agreements for interoperability and consistency.
Dr. Cimino, speaking as a member of the team developing the clinical information system at New York Presbyterian Hospital and as co-chair of the HL7 Vocabulary Technical Committee, observed that little attention had been paid to the fact that they are talking about taking care of sick people. He said he was pleased to see this is a central feature of the document.
Dr. Cimino suggested one criterion for message standards is whether the data put into the message is “dumbed down” to facilitate the transfer. He emphasized that maintaining the original meaning of the data is crucial for sharing information across systems.
Another criterion he proposed was employing high-quality terminologies. If the data in the message are not captured in high-quality terminology, usefulness suffers.
He also recommended considering how the government can support creation of the standard.
Dr. Cimino suggested adding three questions to queries on data quality. When is duplication of data capture alleviated--does the data need to be recaptured every time a message is filled in? Is there a data model that considers all users and all of their needs? What criteria are used to select terminologies? He also advised asking SDOs the number of people among all interested parties that participate and about the effectiveness of the standard in meeting the needs of interested parties is evaluated.
Dr. Cimino suggested two additions to the proposed list of transactions: (1) alerts and reminders (e.g., messages telling patients they need a vaccination or a physician about a problem with the drug ordered), and (2) logic and guidelines to generate alerts and reminders.
Dr. Rozen expressed concern that the proposed PMRI Standard focuses on an old paradigm that is not patient-centric, ignores advances in communication technology and thin browser interfaces, and is mired in a decades-old failed vision. Dr. Rozen noted that the concept of a patient/consumer health record, patient access, or storage of information are not incorporated within the PMRI. Issues of patient access, ownership and control (important in the HIPAA Privacy and Security Regulations) are largely omitted. And the failure of distributed healthcare information to be available on a 7 x 24 basis is not adequately addressed. Dr. Rozen advised that compliance with the proposed standards should be weighted to facilitate comparison between competing systems. He suggested encouraging demonstration projects to identify implementations and standards that truly add value to the patient.
Dr. Rozen recommended adding two questions to the draft Criteria for Selection of PMRI Message Format Standards. Does this adequately address the need for information to be posted by healthcare stakeholders for access by others who have proper authentication and authorization? Does it provide a mechanism for secure patient communication with healthcare system stakeholders, including physicians in a distributed patient environment?
Under Indicators of Interoperability in the draft questionnaire, Dr. Rozen recommended asking about the standard’s capability with thin browser interfaces. He suggested SDOs should identify usability testing for implementing their standard in Indicators of Data Comparability. And he recommended including two other areas. The SDOs should document education and on-going training required for maintaining efficient implementation of this standard. They should also address: (1) having the data available 7 x 24, (2) authentication standards, (3) standards for documenting access authorization, and (4) mechanisms for providing access in distributed healthcare environments.
Responding to the proposed list of PMRI transactions, Dr. Rozen observed that most health information resides with patients and their families. While initial focus is on transaction standards, he said current and evolving clinical processes need to be recognized so systems implemented can benefit from these changes. Dr. Rozen recommended: (1) patient information from Personal Health Records including health risk analysis, trends of the patient, and interests and behavioral characteristics; (2) mechanisms for integrating information from remote monitoring of parameters for managing chronic illnesses; and (3) focus within the record on communication between patient and physician.
Dr. Rozen also recommended adding four questions. Are there additional functions (e.g., communication) where criteria for use exist but standards have yet to be evolved? What standards will be adopted to facilitate posting and remote accessing of information on a hospital's Web site? Do these standards permit variation in who is the keeper of the PMRI--does functionality change when the keeper changes? How does authentication occur in the absence of a single unique personal identifier?
Dr. Rozen offered three suggestions to facilitate criteria for selection. He encouraged a review of the definition of the PMRI prior to finalization of the standards selection criteria. He advised that: (1) clinical and transactional data integration must occur with real-time message processes, (2) the patient should be at the center of healthcare management and information, (3) patient health records can be a mechanism for maintaining one’s lifelong health record, (4) integration of self-reported patient information (e.g., Health Risk Assessments) enrich the system’s medical record, and (5) data interchange of professionally-sourced medical information to the personal health record can improve health by encouraging compliance and monitoring.
Secondly, he emphasized that timely communication is core to PMRI functionality; the PMRI is no longer a flat file. Dr. Rozen noted: (1) communication capability needs to be at the core of functionality; (2) data interchange requires semantic standards through use of standard terminology as well as syntax standards; and (3) mechanisms for authentication and authorization must be part of proposed standards.
Finally, Dr. Rozen recommended weighting the criteria to facilitate selection based on compliance with standards. He also advised focusing more attention on emerging technologies that facilitate real-time posting of data and interaction.
Mr. Rubin, the Veterans Health Administration Enterprise Architecture Office representative to the Object Management Group and HL7's standards organizations and part of the Government Computer-based Patient Record Project’s reference modeling activities, encouraged the Subcommittee to remain steadfast while recognizing there will never be consensus on technology, platform, product, or business practices. Initiatives such as open interface specification provide the loose coupling needed to allow disparate organizations to continue to interoperate while doing business autonomously. He noted this work also addresses internal interoperability issues.
In considering standards, he encouraged the Committee to recognize issues extending beyond national boundaries. For maximum PMRI utility, organizations must be capable of extending the standard to address these "external" interests without sacrificing the fundamental interoperability offered by PMRI interchange. Noting much attention is paid to "customization," Mr. Rubin cautioned this capability must be differentiated from "extensibility." He viewed customization as sacrificing the core standard to bend to local interests (e.g., optionality) at the expense of interoperability. Extensibility, he described as the addition of information to support interests beyond the scope of the core standard. An extensibility mechanism benefited PMRI’s long-term utility.
Mr. Rubin urged the Subcommittee to broaden the questionnaire and its view of integration technologies to extend beyond message-based integration and include alternatives such as middleware solutions. And he emphasized the importance of acknowledging a separation of concerns between the care of the patient medical record and its transport. Without this, the PMRI would not be able to take advantage of evolving and maturing technology solutions.
Mr. Rubin said the criteria are well suited for selecting the PMRI standards. But he pointed out that weighting criteria to emphasize areas that are particularly important would add significant value. He recommended placing heightened importance on interoperability, data comparability as defined in the report, flexibility, and relationship to other standards. He also proposed altering consistency with other standards and shifting emphasis from alignment to synergy, so touch-points between standards can be explored and understood, allowing each standard to be used to its advantage.
He noted market acceptance is an important criterion, but he emphasized it must be viewed in the context of a changing landscape. Emerging technologies should not be discounted solely because of minimal healthcare penetration; standards with significant market share should not be blindly embraced at the risk of prolonging a potentially aging standard.
Considering data quality and accountability, Mr. Rubin noted a need to transmit relevant metadata in conjunction with the data. Metadata will provide insight into the steward and source system of the transmitted information and facilitate shared meaning and understanding.
Mr. Rubin noted the questionnaire does not expressly differentiate among categories of interoperability introduced in the Uniform Data Standards report: basic, functional, and semantic.
He urged the Committee to identify some objective metric to assist in differentiating standards and relevant touch-points. He suggested existing public work such as the ISO Reference Model for Open Distributed Processing might provide seed material.
Mr. Rubin cautioned that the questionnaire would not sufficiently identify reasonable expected costs to use the standard. He noted the cost of implementing standards extends well beyond licensing and that the cost of initial integration, off-the-shelf products, development, and recurring maintenance (e.g., keeping current with the standard’s evolution) are likely to be costs of primary interest. He recommended soliciting some notional cost model for organizations of a specified size with respect to organizational experience with the standard.
Mr. Rubin recommended expanding the scope of questions addressing timely standards development and flexibility concerns. He suggested asking about the average time for general availability of commercial and conformance tools to support an adopted standard and about relationships in place to expedite promulgation of standards through other standards organizations. He also advised inquiring about: (1) how new terms and code sets are supported within the standard--any changes required to the standard or deployed systems, (2) extensibility mechanisms offered to allow for the inclusion of new terminologies, (3) relationships or dependencies that exist between the standard and lower-level transfer protocols, (4) how extensions beyond the scope of the standards are supported.
Commenting on the proposed list of PMRI transactions, Mr. Rubin conveyed his colleagues’ concern that identified transaction sets did not appear to address the JCAHO ambulatory care requirement for "a summary list of all significant diagnoses, procedures, drug allergies and medications.”
He also recommended three areas as worthy of consideration in the PMRI selection process: separation of concerns, vehicle for comparison of standards, and open registration authority
Mr. Rubin noted that, in considering the interoperability of PMRI, two concerns are addressed: (1) the content itself--identifying information of interest and appropriate authoritative sources for describing it, and (2) addressing how content is transferred. He expressed concern that, given the framework for SDO responses and the questionnaire’s slant, alternatives that merit consideration (e.g., middleware, service-based architectures) may be deemed out-of-scope. And he urged the Committee to broaden the questionnaire to ensure that these alternatives are considered.
Mr. Rubin reiterated that NCVHS is in a pivotal position to identify an objective framework, useable industry wide, to compare and understand standards. He advised viewing such a tool--not as a measure of quality, completeness, or judgment of the standards themselves--but as a roadmap for the industry to gauge how standards fit together. Touch points between technologies, message formats, and middleware could be identified, providing guidance to understanding the landscape of standards and to leveraging each to its strengths.
Noting the importance of flexibility so organizations can carry out their business practices within the general framework of the PMRI standards, Mr. Rubin said a public forum is necessary to allow for open registration of extensions. This activity, most likely a natural extension of the ANSI HISB Metadata Registry work, would allow localities and organizations to relate their extensions to the core standard and to define relationships to other extensions.
Mr. Seliger identified two parts to the problem. The bulk of these hearings were about Data transmission and data messaging, but there was also the aspect of usability and accessibility between the user and the computer information system. As Dr. McDonald had pointed out, without accessibility and usability one had low-quality information that no coding schemes or automatic processing algorithm could improve. And if people did not use the information systems, the datasets would contain holes. Mr. Seliger proposed that interaction and the set of standards that govern, facilitate, and improve their quality be given priority. He noted that context management standards could minimize, if not obviate, certain interfaces.
He noted that CCAL, a group charted by HL7, publishes under the auspices of HL7 a series of standards for context management. (Context management enables applications at the clinical desktop or at any other point of use devices to share data with each other and the end user. He proposed that context management standards be given high priority.
Mr. Seliger recommended that the issue of dependence, relevance, and relationship of the standard should be among the questionnaire’s lead topics. He said he was pleased to see questions about a conformance standard and test tools and recommended clarifying that the SDO publishes conformance statements independent of commercially available tooling.
Mr. Seliger advised that determining the extent the data is accessible to the end user and what is required for the end user to make sense of the data are important aspects of data quality, accountability, and integrity.
He suggested who uses the standard is an important criterion that transcends numerical percentage of market acceptance: big vendors and prominent institutions are good leading indicators.
Mr. Seliger suggested another dimension to a standard is how easy it can be implemented. Many standards lack specificity, leaving details unresolved that could lead to disparate implementations that do not interoperate, even at fundamental levels. He recommended “teasing out” technical characteristics about how the standard is specified and what it requires. He noted that technology agnostic standards can be powerful in terms of flexibility, yet run into practical interoperability problems, but technology neutral standards transcend programming language and system implementation. He encouraged the Committee to look at the good practices to be found in CCAL and information model development (e.g., HL7’s work with Version 3) and consider them in its framework.
Mr. alSafadi, HL7 Image Management SIG co-chair, chair of the Biomedical Imaging Working Group CORBAmed, and member of the World Wide Web Consortium Query Working Group and Protocol Working Group, reflected he did not know how to weigh and compare the criteria or choose a standard to recommend under HIPAA. He agreed with the sequence and with the importance of interoperability, data comparability, data quality, accountability and integrity.
Mr. alSafadi concurred with the questionnaire’s emphasis on conformance and recommended that NIST increase its activities in software diagnostics and conformance testing. He suggested that NIST, in collaboration with SDOs, disseminate knowledge on testing and quality assurance activities, jointly develop test suites (tests and diagnostic tools) to complement the SDO's specifications, and provide facilities for workshops and forums.
He advised expanding the evaluation of data quality to include providing and adopting definitions for character models, digital representation of characters, transcoding, normalization, and adoption of a reference-processing model for that element and data.
Noting the section on consistency with other standards, Mr. alSafadi said he hoped the emphasis was on technology, the Internet, and international standards. He encouraged SDOs to work to ensure that these technologies meet healthcare concerns. He noted that the W3C is working on recommendations for XML protocol, query, transformation, device capabilities and schema.
Mr. alSafadi emphasized the link between healthcare SDOs relationships with international organizations and their ability to create software and standards that could be used by the international community.
He observed that the questionnaire mostly addressed the timely standards development intra-standard procedures, and he suggested exploring what processes are used to expedite the standards development cycle when more than one SDO is involved. He predicted that in the future more standards would be jointly developed.
Mr. alSafadi suggested adding “medical images and waveforms” (e.g., still dynamic sequences and real-time streaming or video) to radiological images in the proposed list for PMRI transactions.
He also recommended the standard being developed for structured reporting that Phillips supports as DICOM and HL7 collaboration be considered in terms of the laboratory and diagnostic study results for laboratory and diagnostic systems to EMR and HIS systems.
Review of Testimony - Subcommittee
Dr. McDonald noted several speakers had said the market shouldn't decide; that they should consider emerging technology. He asked who decided. Mr. Seliger clarified that his point was to look at market forces across domains: when something in another vertical segment looked promising they needed to be able to try it. He noted there had been a lot of interest in OMG about how some of the specifications that came out of the healthcare domain taskforce were being used. Some standards could be downloaded from the Internet and there was no way of knowing who applied them. Mr. Blair noted other SDOs had difficulty keeping track of market acceptance. ASTM had mentioned the same thing; even HL7 did not know the actual number. The questionnaire gave SDO's different ways to represent the degree of market acceptance. Dr. McDonald remarked they couldn't foist things in development; it was early and the goal wasn't to make a success out of the standards. He proposed that they help things already “ripping along.” Government should not be deciding.
Responding to a comment that technology companies were not receptive to healthcare, Mr. alSafadi said technology companies would love to listen when healthcare presented itself as a group. Interaction and liaisons between SDOs had to be set and represented in the organizations and technology, and then benefits could accrue to the healthcare system as a whole. He expressed concern at the impedance mismatch between healthcare SDOs and other Internet SDOs. In healthcare, it takes two-three years before the first implementations. In the Internet domain, it also takes that long to have a standard--but meanwhile implementations are already trying to capture the market.
Asked about how to keep a selected standard from becoming a monopoly, Dr. Chute replied it boiled down to either a public buy-out, making it practically public domain so it might evolve from there, or some regulatory arena where revenues from sales of mandated standards could legitimately be used for development, maintenance, and improvement of the standard, but not for other purposes.
Dr. Rozen was asked if his comment about integrating patient-reported information into the system’s medical record was a pitch for standardization of patient-reported information--and, if so, who should do it. He replied he would never propose the government do it. But he asserted the need for recognition at a higher level that patient-reported information needs to exist within the patient's medical record, be it a PMRI or a composite PMRI/consumer health record. He noted the greatest difficulty is semantic interpretation: how one interprets a patient’s expressions into medical terminology. He said they worked with the UMLS and other standards organizations over four years on a consumer health catharsis that, still in its infancy, has over 23,000 terms. When such information is brought into the medical record, both need to use the same language.
Asked to comment about the issues and what ought to be known about how SDOs see themselves functioning in a cross-SDO environment, Mr. alSafadi replied he would like to learn what each SDO saw as the scope of its message format standard and how it relates to the others. Dr. Cohn said the question he heard was about what process was used to expedite the standards development cycle when more than one SDO was involved. Mr. alSafadi concurred. What was missing, other than a statement of liaison, was a formal process that engages either collocation of meetings, gathering and soliciting information, or notifying about events. Many mechanisms could be employed and he wanted to see what was being put in place.
Mr. Seliger said that VHA, looking to leverage as much that's been done as possible, tried to do a "crosswalk" of international models. Just looking at standards, activities were so disparate it was fruitless. Asking organizations to classify themselves took them further. He echoed Mr. alSafadi. Any objective metric that let SDOs classify themselves would be of value, both to the group and consumers.
Dr. McDonald commented on the attractiveness, when there was overlapping expertise, of literally dividing up the work and cross publishing as NCCLS and HL7 did. Smart people figured out what was really going where and joined in. The marketplace sorted it out and decided. Whatever borders one came up with today, they’ll be crashes at the edge tomorrow. The purpose of the questionnaire was to elicit first cuts at standard suggestions.
He noted the group’s assessment that this would not be mandated with the force of the administrative standards; they would encourage standards far enough along to be relevant.
Mr. Blair asked Dr. Rozen if, knowing they didn't want to standardize much of the evolution of care data communications, but to enable interoperability and comparability, he realized he had a free hand to be a leader and a pioneer. Dr. Rozen said Wellmed worked in a unique, cutting area that was changing rapidly and would outpace any standardization effort. The derived standards were useful. Their consumer health thesaurus maps through the UMLS. They use the drug databases. HTTP is their communication media. They use XML7 with HLA where available. But he expressed concern that this still felt like the old model, and he urged them to take it to a higher plane. Availability of information was a critical factor throughout the standards process. Whatever standard was selected needed information available 7 x 24. What was so frustrating in the healthcare industry was that information was truly not available; it couldn't even be posted. He said the most important criterion was interoperability.
Responding to Dr. McDonald’s observation about interstandard collaboration and cooperation, Dr. Chute noted that he had asserted the desirability of overriding meta-standards to facilitate communication. He agreed that smart people went with the action. He pointed out that resulted in a single SDO increasingly dominating the health information industry, for want of an adequate forum where standards issues could be either on a bilateral or a multi-lateral basis, with consensus about who was best qualified to undertake a given strategy. He queried whether the emergent model of a mega-standard de facto taking over interstandard interoperability issues was a preferable situation. Dr. McDonald responded that there was a need for some controlling organization and the choices were collaborate or let the markets play it out. He suggested it was working out fairly well; at least people could talk to each other about the conflicts. And they couldn't re-publish in a different way and create confusion. Newer models brought powerful, big organizations and voluntarily collaboration. Dr. Chute asked if there was a forum where this could be accelerated. Dr. Fitzmaurice remarked that it sounded easy to ask the industry to identify standards ready for prime time, begin and, in a couple years, mandate. They started that with HIPAA in 1996 and were still getting it in place and adopting a standard everybody agreed was ready in 1996. At some point, Dr. Fitzmaurice concluded, it helped to have a prodding hand. They would look at standards they were told were closest to prime time, apply criteria and justify a recommendation. People could build off that. The market would either coalesce and begin adopting faster, or say it had something better. Either way, they would have better information about standards and greater national awareness.
Mr. Blair noted suggestions and different perspectives the Subcommittee could begin to incorporate based on the testimonies. He outlined the first phase of the process for recommending standards for PMRI, asking for comment on these plans for the next 12 months. The group would modify the questionnaire, based on this input, and distribute it the SDOs, looking for feedback in June. They would compile this information in matrix or tables, analyze it and ask SDOs in September/October for clarification and additional information. A report would be distributed to SDOs and other medical informatics stakeholders in January so they could submit recommendations (some circumspect way of serving as catalyst to move the industry forward, being careful to not overstep what was applicable) by February. They still had to determine whether the next phase would be medical terminology, content, or middleware. There would be more solicitations for guidance. Nine months from now, maybe things would crystallize.
Dr. Cimino said the group had enumerated the issues well and organizations were ready to respond. Dr. Chute concurred. He urged them to consider desiderata: what they sought was implicit in the questions; everyone might benefit from making it explicit. Dr. Rozen said the path they’d chosen was clear and they should move forward. They should also send a message to the whole Committee that the PMRI should be re-examined. He cautioned that, without another look at the vision, patient input could be lost.
Suggesting they were missing half the equation, Mr. Rubin asked about the process for soliciting from vendors who provide the applications and provider organizations that use them. Mr. Blair said they hoped the focus on market acceptance would identify standards vendors and users valued. Mr. Rubin said that was a necessary, but insufficient step. A vendor’s opinion was not necessarily a direct or accurate indicator of whether a product was accepted in the market or industry. Dr. Cohn agreed, noting that the next day they would hear from provider groups and vendors. He pointed out they were trying to identify standards that already had “road testing and needed a push.” They couldn't ask people to “blue sky it.” It was a delicate balance.
Mr. Rubin suggested that the Committee contact organizations participating in SDOs, inquiring at the executive level about satisfaction with the standards. He noted that they would want to quantify the size and constituency of the membership. Dr. McDonald suggested asking SDOs which subset they thought was “most right,” inviting considerations not already on the list. Mr. Seliger emphasized the need for a terminology that could be used to build an objective metrics to take disparate input from standards groups and piece it together in a way that made sense. He noted the ISO reference model for open distributed processing identifies viewpoints from which standards can be compared.
Dr. Rozen observed that the Subcommittee was not putting out a standard without consulting the public. He expressed confidence that this process of seeking out public opinion and reaching consensus would enable them to advice the Secretary. Dr. McDonald reflected that it was not such a complicated market. There were a small number of players; a couple at the edges might “duke it out,” but you could “count it with about five fingers.”
Review of Testimony about Elements of the Proposed Process
Mr. Scanlon reported on the status of other HIPAA standards. A final rule for the transaction standards and code sets was published in October and two-year implementation and compliance dates are forthcoming. A final rule was published on Privacy; the effective date was extended to April 14th and comments will be received through March. The Secretary and others will then evaluate how to resolve any issues. Three other final rules are slated to be published later this year on: the national provider identifier and system, the employer identifier number, and security standards. Decisions on the privacy rule will influence the security rule; the intent is for them to reinforce and support each other. A clearer picture on privacy is expected near the end of the month. Two new proposed rules deal with the proposed standards under development within HHS. A draft of the rule dealing with claims attachments is circulating within HHS. Work is underway on the rule dealing with a unique identifier for health plans and payers.
The Committee's PMRI recommendations were presented to the Data Council and circulated. Almost every agency initiated worthwhile projects with work underway supporting the recommendations. A number of ideas also have been proposed about how HHS might move collectively. These were brought to the Council's Data Standards Committee, the focal point for expertise on data standards and the PMRI. The agencies were asked to look at what they'd reported in the light of current situations and indicate their progress and future work. Mr. Scanlon anticipated further progress and additional projects. On broader recommendations, he noted the transition period left uncertainty for everyone. They would have to see what the specific PMRI standard and appropriate vehicle for implementation might be. A number of steps could be taken to move such a possible standard along other than a mandate that literally meant, aside from the voluntary ANSI and other standards process, making it the equivalent of United States law and invoking a process of requirements and regulatory and administrative procedures.
In terms of triage, Mr. Scanlon noted specific recommendations that said “go and do good work” were underway. There would be an update on these activities that had consensus by the next full Committee meeting. Recommendations relating to the criteria to be employed in regulatory actions would have to be evaluated in light of the specific proposed regulation. The group had agreed that the Data Standards Committee would serve as the focal point for initial evaluation and referral for anything not linked to a specific agency.
Mr. Scanlon observed that all of the standards had begun to relate to each other and remarked that was good, though when a rule bogged down, it affected others. He noted concern about how far to proceed and project in the PMRI area until the privacy framework was more certain. Mr. Scanlon said there was considerable progress on the PMRI recommendations; he will try to have a report for the next Subcommittee or NCVHS meeting. Anything dealing with policy and potential regulation remained uncertain until they heard from the new leadership. He noted that the full Committee approved the remaining changes to the annual report to Congress on HIPAA at the February meeting. A sentence or two had to come from the Privacy Subcommittee, and then it would go to the Executive Committee.
Mr. Scanlon remarked that, depending on the nature of the recommendation, the Committee might require a regulation, though not necessarily a HIPAA-type regulation, for certain Federal programs (e.g., the inclusion of certain standards in Medicare, ARC’s research grants, or CDC’s development of the National Surveillance System). Approaches in other areas might be more R&D and evaluative efforts to gain familiarity and experience in some of the standards. Or the Committee might conclude, after hearing from everyone, that several PMRI-type standards would make a big difference in moving forward and the best framework was something like HIPAA or some way the department could use it. Dr. McDonald noted they could come up with at least tentative recommendations in the next eight months and formalize recommendations within the year.
Dr. Yasnoff asked about the status and path for the increased research in this area that they had recommended. Mr. Scanlon reported some agencies had said they were planning this. They had circulated the recommendations when budgets for the year were unknown, and people had been reluctant to respond. Next round, people would know their budgets; they hoped to get a clearer picture then of what agencies could support. Dr. Yasnoff asked if there was any opportunity for the Secretary to recommend increased funding in the next round? Mr. Scanlon said whatever would be done, had to be within the context of what they already had. Something longer term (two-three years away) could be factored into the FY '03 planning.
Dr. McDonald remarked that a number of testifiers had raised other issues: wanting to influence the standards groups or learn things they could use for other purposes. He noted that the Subcommittee’s goal at this stage was to gather information about and make PICS--not to change behavior. He asked if they should change their goal or go forward.
Dr. McDonald cautioned that surveying users might mechanically be difficult. Some groups have large memberships and their selection process might have undue influence. Mr. Blair emphasized the importance of seeking input from credible industry sources beyond the SDOs, to validate the areas and the standards and determine if they are ready for “prime time.” Dr. McDonald suggested using mailing lists from a couple associations to do a random sampling of institutions, but questioned whether they had the mechanism or staff. Mr. Blair said they could hear from representatives of the provider community in September and get their reactions, based on the most recent data. They could invite a couple large managed-care organizations, a few hospitals, several group practices, maybe a long-term care institution and others for diversity. Representatives of the clinician at the AMA level would have awareness.
Dr. Fitzmaurice noted that testifiers had different priorities, but all felt the criteria were good. The most dominant priorities were interoperability and comparability. Nobody said a question was bad and comments asked for more detail or offered additional questions in areas ranging from a single standard for communicating security and privacy policies to all systems within an enterprise, and other practices including modeling and architecture. They also heard that when the government chooses a standard, it could create a monopoly. Noting that, right now, their role was to speed up the process, he asked about the role of the government in deferring monopoly status on a given standard or buying up that standard to make it free and available to everyone. The group noted there was a precedent for sponsoring or subsidizing a standard; X12 received a relatively modest fee to make it publicly available without the usual fee.
Mr. Scanlon remarked that the criterion for HIPAA was for a low-cost distribution mechanism, not free and not necessarily a monopoly, but using existing standards without cost being a barrier. He noted that for some standards that worked well; for others it was difficult to resolve. Dr. Fitzmaurice said his point was choosing a standard could confer a monopoly and there were other ways of doing this that might operate just as well. Dr. Cohn recalled that a number of testifiers commented that the cost is not the standard, but it's implementation. They also heard some people thought the standards were more expansive or a bit different than described. Dr. Fitzmaurice agreed on the cost of standards. He said they also heard that when good experts are involved, it may not take many people to make a good, industry-accepted standard; but you can't always get the best experts to come into a SDO and take time to complete the standard. That might be part of the cost of developing a good standard. Dr. McDonald agreed with Dr. Cohn. They should look at the testimony to see how they might adjust their scope, but not give up on the low hanging fruit. They could get something done in a year, and do more after that.
Mr. Blair said he was pleased with many of the suggestions, but he cautioned that one could get diminishing returns adding questions. He noted that, when they did the earlier survey, people didn't want to answer questions about implementation cost because they knew they were heavy and felt it would reveal their standard in a disadvantages manner. He doubted that they would get useful information on that question. Dr. McDonald reflected that had heard a lot of suggestions that had orthoganly different motives. If they were going to hold to their original purpose, they had to be careful about expanding the questions. Mr. Blair said he felt comfortable expanding the questionnaire 10-15 percent, so it did not overwhelm the SDOs. The group discussed engaging about 24 SDOs in the survey. He approved of Dr. McDonald’s recommendation that SDOs should look at a particular transaction, filling out the questionnaire for the transaction, not their SDO.
The group considered adding a question about patient or consumer information. No one knew of a SDO that had come up with a standard for transactions between a personal health record and an institution. Mr. Scanlon noted models and standardized health risk appraisals for individuals with specific measurement aspects were being created in the public health world standardized more or less in terms of the content in the questionnaire.
The group noted to ask CDC and HL7 about a mature standard for the transfer of information about immunization transactions from a local site to an immunization registry or between registries that sounded like a candidate for low hanging fruit.
Mr. Blair said he could cull through the testimony, revise the questionnaire, and e-mail it for critiques, but that he hoped they might hire someone to assist in pulling together a matrix or spreadsheet. Dr. Fitzmaurice said that he could probably help to some degree. Mr. Scanlon said resources were available for consultants; Margaret Yen was suggested for the analytic work. Dr. Fitzmaurice commented on the government rules for sending out surveys to more than ten recipients; OMB and other clearances might be necessary. The group discussed that a pre-questionnaire and a hearing might be another way to get the information. Dr. Cohn noted the Subcommittee’s inclination to be leaner rather than ask everything possible, making this easy on everyone.
Mr. Blair remarked that the day’s discussion had shown what a peculiar market this was, and emphasized the need for an industry overview in subsequent hearings. He reflected that they still did not have a sense of how the market really worked or the nature of any of its failures. He suggested they needed someone who followed the industry in terms of IT and could talk about how healthcare responds to IT issues, makes decisions, and invests. Dr. Cohn remarked that much that they talked about were not failures in standards, but the need when, making major software investments, for a standard with credence that would be around and supported. He portrayed the Subcommittee as providing additional reassurance to people choosing systems and standards that they had the full faith and support of the US Government behind them.
Dr. McDonald asked about a way to sample request for proposals (RFPs) from institutions. Mr. Blair suggested getting guidance directly from consultants that prepare them. Dr. Cohn reported the SDOs had requested to meet May 31st-June 1st for feedback on changes to the standards. (The time allotted on May 1-2 was canceled.) He questioned whether there would be time for consultants to talk about clinical standards. Hearings would probably be scheduled in the summer. Dr. McDonald noted the hearings should be targeted to give a sense of distributional numbers. People could give percentages if they thought that was competitive information.
Dr. Cohn noted that they anticipated a hearing related to PMRI standards. Time lines were unclear, but the intent was to talk during the summer or fall about enforcement and compliance. Dr. Fitzmaurice suggested they might want to wait until there is a notice of rules on enforcement. The Subcommittee set aside time in September; four topics were high candidates for discussion.
Dr. McDonald noted that they had said they would come back to patient identifiers when the privacy and security areas were handled. It was hard to move ahead without it. Mr. Scanlon pointed out that HHS was still forbidden to spend appropriated funding, including Committee meetings, on that issue. Dr. Cohn suggested they might come out with a letter once they thought the issue was handled.
Dr. McDonald said they would query about August and October dates for hearings. September and November were full Committee meetings. Mr. Blair said they would distribute the survey to the SDOs in May and analyze it in July/August. The Subcommittee could review it in September and clarify and additional questions and concerns with the SDOs in October or November. In November and December they'd work out recommendations for individual transactions and write the report. A draft would be distributed in January for critiques and be submitted for the full Committee review in February.
Dr. McDonald noted the need for time to think about the low hanging fruit, set criteria, what they wanted to consider next, and the approach they would recommend the government take with these standards. The group agreed that they would not choose to mandate most the standards, but clarify that they met “all the pieces” and when people decided to do it, they would do it that way. Dr. McDonald proposed that SDOs could come up with a way to take care of the few standards where SDOs “bumped up at the edges” (e.g., NCPDP and HL7 overlapping over prescription writing, DICOM and reporting overlapping).
Dr. Cohn expressed concern about going head with a plan that wasn’t firmer and tighter than Dr. Huff’s estimate of a six-year implementation. Dr. McDonald pointed out that it depended on the level of absolute--free text was easy. As the codes found an accepted life, it would become more absolute. Dr. Huff clarified that the six-year implementation he mentioned was the time period in which they could mandate, for instance, that all microbiology data came to CDC in that format. It took into account people who were not even automated yet. Tomorrow, they could send the stream of microbiology data from Intermountain Healthcare to CDC and it would all be the same Hl7 transactions used to put it into their electronic medical record. What was lacking was the standard terminology in those transactions. But there were things that could be done quicker. There was an implementation guide for immunizations; they could do that. People were already working on the microbiology guide. If HL7 made some definitive statements about use of LOINC codes and lab data, 90 percent of the laboratory guide already existed. Mandating that all data must come in that format from all sources took a long time and had huge impact on institutions. Things that were useful and had lower impact could be done in six months or a year. Mr. Blair commented that it would take six years to put through an NPRM process. Dr. McDonald remarked that what Dr. Huff described in a six-year time frame was the ideal, including Dr. Chute's perfect way. He noted the SCRIPT standard with its free text or code was easier and could be used on a lot of standards “yesterday.” They were dealing with a spectrum of expectation and the further along that spectrum they went, the harder it would be. Dr. Huff encouraged them to pick easier things that can be done faster. Dr. Cohn agreed there were gradations in terms of recommendations they could make.
Dr. Fitzmaurice noted that they had heard again about gaps and suggested one follow-up activity to the report might be to encourage the Secretary and others to take action to analyze and fill them.
Members expressed hope that they could begin work in August. They needed counsel on legal ramifications and would have to wrestle with issues where their views differed. Many agreed that, probably, they would not come out with an iron clad PMRI, as they had for the administrative standards. Instead, they had to come up with incentives; all government agencies would have to do this with their healthcare data.
PANEL 3: Testimony of Public Health, Hospital and Others About Elements of the Proposed Process
Given the plethora of record keeping systems extant in the healthcare industry, Ms. Fuller said messaging standards are absolutely critical to the implementation of electronic patient records. She said the Subcommittee's approach to messaging should yield a description of the current state of standards development and deployment and the extent to which SDOs are succeeding in creating one messaging protocol that can be applied to a given set of data without required modification. She said SDOs should be encouraged to address perceived barriers to adoption of their protocol and she urged the Committee to create a feedback loop between SDOs, vendors, and other implementers.
Ms. Fuller advised that the Committee could best assist the promotion of standards by: encouraging formal ROI studies, promoting case studies and best practices, adopting official or approved standards for certain classes of messages, requiring SDOs to monitor implementation and update standards based on feedback, identifying overlapping standards and working to reduce competition between SDOs.
She said AHIMA found the eight criteria appropriate for consideration in developing messaging standards and suggested establishing measurement standards for each criterion, incorporating SDOs responses. The SDOs might even self-grade their responses to the criteria, giving detailed answers to document their rating.
Ms. Fuller noted the questionnaire was appropriate to the categories, though information might be difficult to assemble. She observed that having the SDOs self-evaluate and complete the assessment grids raised the question of whether the outcome justified the work required.
Ms. Fuller said the proposed list of PMRI transactions was reasonable, reflecting systems most affected by current standards. But she advised it might be beneficial to consider the data and documents integral to the claims attachment standard. In their current form, the messaging standards did not address the completeness of the data for either valuation of the claim or the privacy standard relative to minimum necessary information. Until the data or document sets were better defined, significant variability would exist in the data sent in support of claims.
Ms. Fuller recommended pursuing additional standards in the areas of medical terminologies, data quality, accountability and integrity. She also said AHIMA strongly supports collecting information once and making it available for multiple purposes, both for efficiency and its effect on data quality. To advance this premise, she recommended focusing more attention on creating clinically relevant data sets that improve documentation and enable automation. She noted that work on the definition of data sets has not been well coordinated and there is only limited data consistency across these sets at the data element, definition, and category levels.
Ms. Fuller advised that the Committee can advance the message standards work by: (1) formalizing the feedback loop between SDOs and getting implementers to ensure the development and maintenance of high quality standards; (2) adopting a series of activities that promote the use of standards beyond the establishment of regulation; (3) using information gleaned from the survey to identify overlapping standards development and a process to facilitate communication between SDOs, thus reducing unnecessary duplication; and (4) having SDOs evaluate their compliance with the criteria.
She encouraged the Committee to advance the standards and terminologies data sets and structured documents by: convening hearings about how a process could be established to engage data set, document, and terminology development activities in the SDO infrastructure, working with organizations whose work links with claims attachments (e.g., emergency records, operative reports, discharge summaries, consultations, and therapy notes), and establishing criteria for non-SDOs developing clinically specific data sets, document structures, and terminologies.
Ms. Stevens said McKessonHBOC felt the SDO selection criteria were appropriate. Noting the increased complexity and size of information standards, she emphasized the importance of publication in a database, XML or tool format to assist implementation. And she suggested including a question about publication methods. Ms. Stevens approved of the questionnaire and suggested some clarifications and additional questions. The indicators of interoperability questions relied on the term "user, and it wasn't clear whether the reference was to a vendor/developer or a health care enterprise implementing an application. She noted that the questions of customizations also needed to be more specific, as the design of some standards allow for adjustments to satisfy differing implementation requirements. When user customization was necessary, she suggested asked about the circumstances, how it was implemented (e.g., optional fields, interface engines), and how it could be reconciled to future releases.
Ms. Stevens noted that in discussing issues of data comparability, one encountered alternative vocabularies that have been adopted for different jurisdictions and functional domains. She proposed asking if the SDO supports these and, if so, how they are reconciled. In the questions regarding market acceptance, she pointed out the importance of determining supporting application vendors’ commitment to the standard. Were tools designed and available to help in the implementation: e.g., integration engines, compliance tools, specification management tools?
While timely standards development procedures are critical, Ms. Stevens noted that the level of participation in its development often indicates the validity and market acceptance of a standard. She suggested adding a question to establish the SDO's structure supporting and encouraging input from all stakeholders.
In determining the flexibility to respond to new requirements, Ms. Stevens recommended adding a question regarding the separation of the model on which the standard is developed, the actual messaging specifications, and the technology used to implement the standard (e.g., XML) message. The standard should be based on a thorough, stable information model and the message specifications should also be staple, but able to change with business needs. Implementation technology must be able to rapidly adjust to technological improvements without negatively impacting message specifications or the underlying model. She also recommended that NCVHS look at creating vendor and provider questionnaires.
Ms. Stevens recommended two additions to the proposed list of transactions: EMR exchange between systems (e.g., hospital discharge summaries, physician history and physicals), and identification reconciliation (a patient merge transaction) from one ADT, HIS or EMR system to another.
Ms. Stevens advised that in considering the scope for this first phase, it is critical to assess the impact on the industry when the standard is selected. Any changes to applications must include an evaluation of the cost, both financial and operational, against expected benefits. The impact on the vendor community would be significant if all of the proposed transactions were mandated, resulting in prohibitively high costs to vendors and users.
At this time, the NPRI (and primary) message format supported by McKessonHBOC is the HL7 2.X standard. The range of versions supported for these interfaces is 2.1 to 2.4 with most live interfaces currently at the 2.2 level. The primary barrier to upgrading has been the operational impact on organizations. With any software upgrade, extensive testing is necessary; when testing interfaces, at least two systems are impacted (often from different vendors). It is expensive, time consuming, and disruptive. Ms. Stevens suggested the scope be defined in incremental phases, starting with domains most compatible with HIPAA that will provide maximum benefits through standardization and that most complement existing HIPAA administrative, financial, patient, and demographic management standards. She said medications management including treatment/medication orders and prescription information should also be a primary concern.
Ms. Stevens recommended that NCVHS focus first on transactions related to patient demographics and identification including admission/discharge/transfer as well as identification reconciliation or merge and EMR exchange transactions. She noted this could focus the industry on improving messaging for PMRI foundation domains. Subsequent phases could introduce remaining selected transactions.
Ms. Stevens also recommended that the Committee consider the international implications of defining an American national standard. She recommended that the group ensure that the work is synchronized with efforts toward international standards in healthcare informatics and that new standards are presented to international forums for adoption worldwide.
Ms. Stevens emphasized that development of PMRI depends not only on the development of a messaging transaction, but also on the adoption of uniform data standards. Anticipating these activities will extend to the required definition of a data standard in the form of a common information model, McKessonHBOC recommended that the Committee select a standard build upon a single comprehensive information model so that additional sets can be added as required without changes to the underlying model or impact on existing transaction sets.
Although AHA generally supports efforts towards the development and eventual adoption of these standards, Mr. Arges expressed concern about any additional regulatory efforts at this time. A number of massive, complex and costly information system changes over the past few years challenged hospitals’ viability. Finding necessary funds to adopt changes would be crucial. Even more troublesome was the thought that many changes offer little immediate administrative relief. While these changes might ultimately yield improvements to the entire process, they compete with day-to-day expenditures for basic patient care services. Mr. Arges said it was vitally important that the recommendations include a study examining financing options that looks at ongoing costs and challenges.
Mr. Arges encouraged the group to explore more fully the data gathering and development process. How are caregivers and others involved in the review of gathering, developing and posting the information processes? Which data standards are they familiar with and how well trained are they in their use in applications? Which technologies do providers readily have available and are these tools capable of complementing the way they deliver care?
He noted that an important functional goal for the adoption of new approaches is that they must free up the caregivers' administrative work load and ultimately yield complete documented and standardized medical records. He emphasized that the report should place additional emphasis on revamping the flow of administrative tasks. Streamlining administrative tasks should be a top priority. For example, today, hospitals develop the appropriate diagnosis or procedure codes at the end of the patient's episode of care, abstracting numerous pieces of information collected throughout the patient's day. Designing this process to work with the caregiver would improve the way information flows. The process needs to be examined. Which pieces at the front end require standard templates that aid the caregiver in patient evaluation and management? What key components do caregivers need to record during the evaluation? Can typical treatment options and tests ordered be included in the template?
Mr. Arges suggested case studies could show how some hospitals approach adoption, the context, method of financing, challenges faced and benefits gained.
He emphasized it is also important to launch demonstration programs that test the validity and quantify costs associated with adoption and that hospitals need regulatory relief and financial support, especially for unfunded mandates, like HIPAA's administrative simplification provisions.
Review of Testimony - Subcommittee
Ms. Stevens reported that McKessonHBOC was very familiar with the information model HL7 developed and was building a web technology based on that model and looking at integrating it into their products. She noted that 2.4 is being introduced with an updated XML rendition. She said McKessonHBOC was trying to balance advantages of new versions of HL7 against minimizing disruption for customers. Most customers are on 2.2, with a “heavy peppering” of 2.3 and 2.1. Within the next year, she predicted most interfaces would be 2.3 and 2.31. She urged the group to only look at 2.4 for a recommendation; several products were already coded with 2.4 interfaces and the recommendation would be an impetus to others. She noted it would be a year or two before hospitals completed upgrades and implementations and these interfaces were operating.
Dr. Cohn observed that the requirements of hospitals probably had less to do with standards than various groups requiring data sets. He asked if the Subcommittee needed to look into data sets. Ms. Fuller said AHIMA had looked at the data definitions from all the data sets mentioned in her testimony, following how patient information moves across healthcare. Even in identifier information, there was little consistency. Patients now move through a more fluid health care delivery system--with a more fluid information system, data could be generated, replicated, and transmitted reasonably. Mr. Arges advised making sure that many of the data elements are common and that the initial building blocks of the data set have a specific application. He recommended that the Committee pick the data set and data elements that define a particular scope of activity. Mr. Arges said hospitals needed to know vendors had guidance, so when they were ready to upgrade, someone would be out there to support them and they had a way to evaluate what the market offered.
Noting Ms. Fuller mentioned limited data consistency, USHIKB, and asking SDOs to submit data dictionaries as input for the meta-data registry, Dr. Fitzmaurice considered that the government might collect and match up data dictionaries to stimulate movement toward central definitions. Ms. Fuller said that was exactly what she was saying.
Responding to two issues heard in Mr. Arges’ testimony--performance elements and unfunded mandates--Dr. Fitzmaurice observed that the government could propose, as part of a research program, studying PMRI systems, the contracting process, and specifications related to satisfaction with outcomes. Noting the extent of uncertainty, he said some analysis might be done about the use of systems hospitals might obtain that looked not only at the realm of unfunded mandates but also the benefits of the system. Mr. Arges observed that better patient care was the ultimate end game. Standardization of data elements would be helpful; interoperability was essential. So was conveying to a broad audience a sense of why, how, and under what circumstances elements of electronic patient medical records unique to the marketplace worked well. Hospital providers could gain a better sense of what they might achieve as part of their outcome. With outcomes specifically identified, facilities looking to upgrade would know what to target and anticipate and could initiate staff training and other resources.
Dr. Shortliffe recalled that Dr. McDonald cautioned the group when they embarked on this venture not to put insurmountable pressure on institutions and organizations. Dr. Shortliffe observed that, as he considered day-to-day complexities, it was precisely the absence of such standards that accounted for much of the turmoil in the hospital environment. Rationalizing interactions between standards organizations for consistency and creating guidelines to move industry in the right direction would benefit hospitals in the areas where they feel the greatest stress. Noting the hospital industry’s (and healthcare’s) relatively small investment in information technology solutions, he asked about other mechanisms to encourage hospitals to invest in computing and information technology to support the kinds of medical record systems they were discussing.
Acknowledging the gap between what other economic sectors spend on information systems, Mr. Arges pointed out that healthcare is a model with many not-for-profit organizations doing community service and operating on a different level with weak margins. He said Congress needs to recognize that healthcare (one of the largest sectors of the economy), needs an infusion of resources to jump-start into the new information age. Mr. Arges compared today to the 1930s-1940s when Hill-Burton was created and funded bricks and mortars for facilities. Dr. Shortliffe asked if having a mechanism so hospitals could apply, individually or as consortia, for funding to tackle these problems would make the standards recommendations more palatable. Definitely, Mr. Arges replied; they heard people needed guidance.
Mr. Blair noted the group never intended to standardize electronic patient record systems, but looked to facilitate communications, interoperability, comparable data and data quality so vendors could continue to produce the best systems they could without limitation, sharing data that required standardization. There was a diversity of standards in the report; health care was multifaceted: they meant to proceed in a “straight step manner.” He said he felt Mr. Arges agreed with that direction, but because of the financial conditions of most hospitals, was concerned that a poorly thought out direction might bring additional financial burdens or disruption. He asked if conformance testing and some certification for vendors would “move the ball forward.” Mr. Arges said demonstration testing to ascertain the validity of these assumptions and a certifying body that could evaluate different vendor systems and substantiate that they conform would be helpful. Dr. Zubeldia remarked on the importance of a “yardstick” to measure compliance.
He also commented that he assumed that PMRI standards, like HL7, would actually be a savings for hospitals. There were be costs to implementing, but there would also be savings long term. Mr. Arges noted the benefit and importance of identifying a certain set of messaging standards and activities and giving providers the flexibility to move that way when it was right for them. He suggested that, if grant monies were available, providers would act sooner. He observed that these entities provide a service and many small facilities are somewhat disenfranchised; this is not the competitive business model found in other industries. Dr. Zubeldia asked if there was a benefit to knowing the target and that when, hospitals chose to leap, all landing on the same standard. Mr. Arges said this put market forces in play; vendors could build systems that brought solutions--and perhaps be more cost effectively because of the target. Dr. Zubeldia asked about ways to determine who was on target.
Ms. Stevens said McKessonHBOC had mixed success with compliance measuring. While it accurately determined whether a required field was supported in a message, she pointed out the question really was, was it necessary to support that field, given the specific application? Some very specific applications have narrow need. A product might fail a compliance test, yet be completely adequate, providing information and satisfying customers. Product compliance must be designed to measure, not compliance to a standard, but both successful implementation of the standard and successful integration of information. If one does not get the information needed, there is no point in being 100% HL7 2.4 compliant. Dr. Zubeldia agreed. A general practitioner might be perfectly compliant and satisfied with a product that stores office visits and little more. An ambulance company or supplier has different needs. All might be completely compliant, and completely incompatible with each other. The provider’s compliance profile must meet the payer’s requirements. Compliance profiles are flexible and can change. The group noted that this issue had to do with the current HIPAA standards and the concept of the maximum data set; when they met with the DSMOs they would consider clinical as well as administrative standard issues.
Noting that vendors wanted to be able to point to a good ROI story, Ms. Fuller suggested focusing on applications and vendors for ROI studies. She also encouraged looking beyond financial models at patient care. Studies about efficacy of information systems and reducing medical errors were another way to view ROI. Mr. Arges recommended looking at best practices, documenting change in terms of outcome. He agreed that financial, quality, and patient satisfaction were all part of the process. Dr. Huff described studies published in the literature showing how InterMountain Health saw ROI and higher quality care because of information systems they implemented. Studies showed that using systems to ensure prophylactic antibiotics were given before surgery reduced the risk of post-op infection by 2-4 percent; despite increasing costs, the cost of antibiotic use was reduced by using system to indicate drugs most effective in given areas for presumptive treatment of infections.
Dr. Cohn suggested the problem was not the cost/benefit of clinical information systems, but the disconnect between long-term ROI and short-term costs. Every quarter matters in the American economy, which is not structured to value information systems that take three-to-five years to implement and show an R&I. Mr. Arges concurred: hospital battles were month-to-month. An ROI five years down the road was a very hard sell against a new MRI with more immediate impact.
Mr. Arges suggested that additional reimbursement for increased quality of care (e.g., MedPac’s recent effort with Medicare) was another way that might increase hospital investments in information technology. But he noted it was hard to: quantify additional costs, tweak revenue with reimbursement formulas ratcheted down so far, or convince health plans and other third party payers to step up investment.
Review of Testimony -- Next Steps
Recalling that one of their original recommendations for clinical data standards dealt with financial support to encourage implementation, Dr. Cohn said they needed to look more at cost issues and burden on providers. He agreed with Dr. Shortliffe; they needed to take note of the financial situation and, if they could, make recommendations, primarily targeted at Congress, for coping with it. Ms. Fyffe reiterated that a third of the hospitals in this country are in the red; another third break even. She found it troubling in this environment to advocate increased use of technology and urged the group to brainstorm suggestions for this cost problem. She expressed confidence that a positive cost benefit analysis could be done and urged further discussion at hearings or meetings. Mr. Blair noted that Recommendation 5 indicated HHS should provide support for cost benefit studies for PMRI standards and Recommendation 9 indicated HHS should explore incentives to provide for equitable distribution of costs to help the provider portion of the delivery system implement PMRI standards.
Mr. Blair said he was pleased that the testifiers pointed out additional things to pursue in selecting PMRI standards and cautioned that the provider community had limited resources and was weary with unfunded mandates. He heard that they should look at incentives and keep in line with standards that had market acceptance and could act as a catalyst to mitigate implementation costs. Dr. Shortliffe observed that nobody out there was opposed to PMRI standards. The vendors, hospitals, other providers (and patients, who want their medical records mobile) saw potential benefits. Resistance was related to logistics and feelings of ownership threatened by alternative thinking. But, ultimately, financial aspects were central and fiscal reality had to be part of any recommendations; they had to identify solutions.
Dr. Fitzmaurice observed that some people (e.g., DICOM and manufacturers and vendors of equipment with built-in standards) were resistant to changes in standards. This was another strong reason to get to the table and participate in standards that work for a broader range of equipment. He heard that they should start from models and acknowledged he had a great likeness for phrasing things in terms of where they fit within a simple model. Data dictionaries, compliance and certification standards required this. Noting he heard that the cost issue and burden on the provider of standards had to be considered; he said they might want to target Congress in their financial recommendations for action.
Dr. Fitzmaurice emphasized that they were well on their way on the PMRI standards: no one said it wasn’t a good idea. Leadership would be critical. He heard they needed tools to implement standards (e.g., Meta Data Registry, conformance testing with qualifications, help with hospitals’ reporting and analytical burdens--patient safety, HEDUS measures) and research into costs and benefits. He said, with open process and hearings, they could get their first set of recommendations within a year.
Dr. Yasnoff observed that people seemed supportive of standards and the process outlined. He acknowledged Ms. Stevens for discussing transfer of electronic medical records as a transaction, which he noted was important given the tremendous movement of patients between health plans. Dr. Cohn recalled she also brought up identification reconciliation, an issue they could discuss but not spend money on. Noting that providers get the primary reward in this environment, Dr. Zubeldia wondered about loan arrangements to get providers installed and ready to go. Dr. Shortliffe speculated about having some hospitals declared disaster areas so they could get loans from FEMA.
Ms. Greenberg remarked it was unlikely that they would be able to spend much time trying to rationalize data sets, and suggested a recommendation from the Committee to enter some in the Meta Data Registry. She noted that would point out where the differences were and provide incentives to move towards rationalization.
Dr. Cohn noted that the full Committee had not talked about data sets and proposed that the USHIK presentation might go to them with a discussion. Mr. Blair suggested they also have presentations on GEHR and RIM. All the stakeholders had to come to some understanding and recommendations before there could be convergence on a model. Noting this had implications for the Subcommittee on Populations’ work, Ms. Greenberg approved of having it before the full Committee.
Dr. Cohn questioned if value was added moving from the data sets into the data model issues. Mr. Blair said it would be good for the Subcommittee to understand the objectives of each information model and data set and the difficulties and issues around trying to get common definitions, because of different use cases and purposes.
Ms. Stevens advised that one thing the Committee could do to make a quantum leap would be to bring stakeholders together to support inter-SDO activities (e.g., building an information model). Noting that Denise Koo’s discussion on the public health data conceptual model and NED was rescheduled for the June meeting, Ms. Greenberg suggested it would be a good for the presentation on HL7 RIM. CDC's goal was to rationalize or synchronize with the RIM. Dr. Cohn remarked that the CDC effort was one of the easier to understand, because they went to great effort to define every piece. The group agreed that conceptually it made sense to do this.
Dr. Fitzmaurice asked if it would be useful to inquire if the SDOs had an information model. Mr. Blair pointed out that certain message format SDOs are building RIMs, which hopefully will eventually enable them to produce more clinically specific standards with greater continuity in a faster, more stable way. He suggested they might want to separate that out, especially if they were going to talk about Version 2.3 or 2.4, which was not based on a RIM. He noted its importance for future directions, but cautioned it might be seen as giving imputed unfair advantage. Dr. Huff agreed that trying to reconcile information models would be an important, but huge, task. He favored it, but could understand holding it to a later date. Dr. Cohn said he heard that they would not ask about information models at this point.
Noting that various mediation approaches were other ways to handle information reconciliation besides a unique patient identifier, Dr. Cohn asked if there was a mature standard that should be recommended for the mediation approach. Ms. Greenberg asked if there were standards that needed to be advanced for the master patient index. Mr. Blair suggested including it in the questionnaire.
Ms. Stevens remarked that several billion Americans probably already have a patient identifier; creating a standard identifier made the problem more acute by introducing another number into the mix. She said her testimony emphasized a transaction to reconcile multiple identifiers. Several products used fuzzy logic to identify duplicate persons, but she knew of no organization that had developed a standard. Dr. Huff mentioned transactions within HL7 for merging patient identifiers and merging transactions to support these actions. He said he heard the need for a separate line item from ADT (discharge transfer transactions) for managing master member indexes and merging duplicate records. InterMountain Health Care had trouble supporting transactions necessary for keeping records consistent across systems. Their contracts now stipulated support for that set of transactions. The group discussed asking about a standard that enables a transaction that facilitates patient identification. Three components were involved: internal identification determining the accounts needed to be corrected, reconciliation, and merging or splitting of the account(s). Dr Fitzmaurice and Mr. Blair would produce and circulate a draft.
Mr. Blair suggested they might want a questionnaire for the CORBA MED patient identification standard. Dr. Huff clarified it is just a different technology for implementing the same transaction set; its fields and transactions are based on and its semantics dependent upon the HL7 standard.
Thanking everyone, Dr. Cohn adjourned the meeting at 12:15 p.m.
I hereby certify that, to the best of my knowledge, the foregoing summary of minutes is accurate and complete.
/s/ Simon P. Cohn, M.D. 6/3/2003
_______________________________________________
Chair Date