The National Committee on Vital and Health Statistics was convened on Wednesday and Thursday, February 3 and 4, in the Hubert H. Humphrey Building in Washington, D.C. The meeting was open to the public. Present:
Dr. Detmer announced that he would be leaving the National Committee to take a post at the Judge Institute of Management Studies at Cambridge University, beginning in April.
Mr. Scanlon reported on the following topics:
· the OMB initiative on discrimination tracking and measurement, coordinated by the Data Council Work Group on Race and Ethnicity Data;
· plans for an OMB proposed rule implementing a recent law requiring the federal government to amend Circular A-110 governing federal research grants to ensure that all data under the award are available to the public through FOIA;
· impending HCFA guidelines on Internet security and the appropriate use of the Internet to transmit Privacy Act-protected HCFA data and other sensitive information;
· the imminent release of the new web-based HHS directory of data systems.
· Dr. Braithwaite reported that the draft of the proposed rule for claims attachment standards should be ready for clearance by the end of May and may be published by July. He gave status reports on other proposed and final rules under development.
Dr. Detmer emphasized the importance of the fruitful collaborative relationship that has evolved between NCVHS and the Data Council.
Ms. Araki is taking a position at HRSA. Dr. Detmer presented her with a certificate detailing her accomplishments and expressing the Committee's thanks.
Canada's "infostructure" serves a health system that is managed by the provinces and territories, with broad principles set at the national level. A recent national initiative elicited input from 500 national stakeholders about their information needs, highlighting needs and gaps similar to those faced in the U.S. The areas in which people are working with an eye toward national integration and modernization include standards development, physical infrastructure, core software, and privacy-confidentiality-security. Ms. Zelmer said the overarching goal is a flexible system with relevant information that is integratable around a person, a community, or an issue.
The foundation of Australia's information infrastructure is the National Health Information Agreement, initiated in 1993. Signatories include the commonwealth and the six states and two territories. This commitment to support and promote the process is supported by a development plan that provides a list of priorities and a work program that implements work in the priority areas. The broad aims of the Agreement are consistent with those outlined in the NCVHS vision statement on the U.S. NHII. The immediate priorities include improving health and service delivery for indigenous peoples and developing an integrated national health and welfare information model. Dr. Reece stressed the need for all countries to work together on a common template, in order to compare data and share knowledge.
Work on England's National Health Service information model began in 1981 and eventually was informed by a clinical process view that regarded data as necessary to improve health. The English government funded several projects to look at the detail of the health care process, including administration and planning, and ultimately it funded development of what became the Health Care Model, a process model. It will soon be posted on the Web (as are the Australian and Canadian models). Dr. Jones emphasized that the investment in information technology must be regarded as essential overhead, as it is in the U.K., and he joined his colleagues in calling for a combined international effort to build the information infrastructure.
In the discussion, Committee members raised questions about fragmentation and the approach to linkages, data quality and incentives, the business case for the electronic patient record, and proprietary barriers.
Dr. John Eisenberg, Co-Chair of the Data Council, conveyed the appreciation and good wishes of the Data Council to Dr. Detmer and presented him with a certificate and letter from Dr. Shalala. He announced that the Secretary has appointed Dr. Lumpkin as Acting Chair of the Committee.
The Committee's views were elicited on the 1996-1998 report and the report to Congress on HIPAA. Mr. Gellman stated his view that in regard to the unique individual identifier, the reports overstate the Committee's privacy commitment and understate its support for the identifier. After a brief discussion, the Committee accepted Ms. Frawley's suggestion that she, Dr. Lumpkin and Mr. Gellman work on this section of the reports and try to agree on language, and that failing this, Mr. Gellman can attach his statement. Other modifications were suggested regarding Year 2000 concerns and security. Dr. Detmer remarked on what the reports reveal about all the Committee has accomplished in three years.
The Committee returned to this agenda item on day two, approving an approach to the unique identifier issue and a process for finalizing and approving both reports. (See action items, above, and detailed meeting summary, below.)
Members were asked to review draft purchasing specifications and to provide comments to Ms. Rimes by February 22. The revised draft will be vetted with various organizations and individuals around the country. The final language will be part of the Subcommittee report submitted to the Committee in June.
Ms. Frawley circulated a letter to the President of the National Association of Insurance Commissioners, drafted by the Subcommittee on Privacy and Confidentiality. The Subcommittee reviewed the model act in the light of its report to the Secretary and the Secretary's recommendations to Congress and concluded that it cannot endorse the model act for reasons enumerated in the letter. NCVHS members offered suggestions for revisions.
The modifications were incorporated into a new draft that the Committee reviewed and approved on day two of this meeting.
The topic of risk adjustment has increased currency now that the Balanced Budget Act requires MCOs to provide encounter data to HCFA for risk-adjusted capitation rates. HCFA is implementing a diagnosis-based risk adjustment method to replace the AAPCC, to be implemented in 2000. Initially it will use only inpatient principal diagnoses. The hope is that when fully implemented, health-based payment will change the incentives for managed care and improve diagnosis coding practices and data availability. The reliability and validity of ambulatory diagnosis also should improve.
The panelists spoke on their perspectives on rate adjustment methodologies. After expressing some optimism about the potential benefits of the new risk adjustment mechanisms when they are fully implemented, Mr. Knudson stressed the critical data flow path between providers and plans and analyzed some of its impediments. He recommended a study to distinguish the more fundamental problems from the trivial ones, along with a look at training programs and monitoring to understand coding validity and reliability.
Mr. Anderson focused on things that make it difficult for plans to respond to Medicare data requirements, such as the simple fact that substantive change is never easy. He pointed out that organizations should be expected to behave in their financial self-interest on this subject, which has become very politicized to the detriment of the data agenda. He called for a new national agreement on vision and direction for risk adjusters. He supports HCFA's "directionally correct" first step and hopes Medicare will move forward swiftly, give plenty of notice, and engage the industry in dialogue. He stressed the high stakes for payment and concluded with several recommendations for the National Committee's role in this process.
Ms. Tudor noted the evidence of selection bias of 10 to 14 percent in the enrollment of Medicare beneficiaries, and she stated that Medicare wants to compensate plans appropriately if they enroll sicker people. She explained why HCFA had to begin with only inpatient data and she briefly described the new system, which adds health status to demographic factors. The plan is to move to a comprehensive risk adjustment mechanism by 2004. She described some of the testing and analysis done in the start-up period and expressed hope that many of the problems will be resolved in time to set up a more efficient system for outpatient data.
Dr. Boesz presented the business realities, noting the costs of data collection and the nervousness of health care business managers about what they regard as growing and unpredictable demands for data. This uncertainty is aggravated by HCFA's projection of a 7 percent reduction in payment, raising the question of whether companies can stay in the Medicare business. One determinant is what kind of business partner the government will be in the future. She described some of the dynamics of the company's relationships with providers, whose discomfort with the new requirements is exacerbated by the rapid rate of implementation. An extended deadline would provide some relief.
In the discussion, Committee members raised questions about external evaluation of the system, the likelihood that it will actually influence behavior, and the tension between these incentives and the push for quality improvement.
Dr. Friedman observed that the functions of public health surveillance will continue to be crucial in any emerging national health information infrastructure. He cited lack of integration as a notable weakness of current surveillance systems.
Dr. Koo said CDC wants the Committee's help in articulating the centrality of public health surveillance in public health practice. She gave an overview of public health surveillance as it undergirds and informs practice. The National Notifiable Disease Surveillance System moves information from counties to states to the CDC. Generally, CDC has separate information systems for surveillance (STDs, AIDS, TB, etc.) because of the categorical funding of public health. The result is a multiplicity of information systems with variable data input and access methods, making it difficult to evaluate problems across programs. To remedy these problems, with HIPAA as a motivator, CDC has started working more with standards development organizations (SDOs), especially those concerned with clinically relevant standards. Dr. Koo gave some examples of work with SDOs. Another major project is the surveillance systems integration project.
Dr. Birkhead pointed out that all surveillance and all interventions are local. After describing some of the problems with today's fragmented and duplicative public health surveillance system, he discussed a proposed conceptual umbrella being advanced by CSTE and praised CDC's new push for integration. He concluded with a caveat about using administrative data as a basis for inferring public health surveillance conclusions, stressing instead the value of direct provider reporting and the provision of clinical and lab information.
Dr. Davis said public health surveillance activities have expanded dramatically to include a range of health outcomes and determinants, including occupational health and the impact of work on health. In this context, the goals of surveillance are to identify the magnitude and distribution of health and safety problems and particularly high-risk industries, occupations and work sites where intervention is warranted. She discussed some of the challenges of collecting data in this area. Weaknesses in national employer-generated systems point to the need for a strong state role in occupational health surveillance. In recent years, many states have been building surveillance systems with funding by the National Institute for Occupational Safety and Health, CDC, using several data sources. She described her department's activities around work-related injuries among minors.
In the discussion, Committee members raised points about using public health surveillance to elucidate the impact of diverse factors on health; the problem of unstable denominators in epidemiology; and privacy issues in the use of record-level data.
Immunization registries are computerized information systems about immunizations in children. Universal coverage is approached by encouraging community and state registries to exchange information through a nationwide network. The goal is for all private and public providers of immunizations to participate in the Registry, compiling complete immunization histories for each uniquely identified patient.
To address technical challenges and make provider participation as easy as possible, CDC started working with HL7 in 1995 to develop standardized transactions for transmitting the immunization core data set. Version 2.3 was issued in April 1997, with an implementation guide. Several activities will be announced in March that were stimulated by support from President Clinton and Secretary Shalala. Ms. Abernathy described a broad-based collaboration to create a common protocol for immunization registration. The first draft of the new implementation guide is being reviewed by participants, and will be published on the Web and provided to vendors when finalized.
Mr. Tait called his presentation a case study in establishing a population-based registry across a broad spectrum of the state, using multiple technologies to collect information. The registry now contains about half the State's immunization events. The immunization component is being integrated with the State's Cornerstone public health information system. An aggressive strategy has been mounted to present a variety of ways to participate in the voluntary system, using multiple technologies for both automated and non-automated providers. The Department has implemented an EDI-interface engine to support ANSI and HL7 standards, and exchanged HL7-based EDI immunization data with the States of Georgia and Arizona.
Forty Shared Medical Systems customers perform immunization registry and tracking functions using its systems, many under the auspices of county health departments. SMS has worked with HL7 code sets to increase the specificity of information storage. Dr. Eisenberg said his company welcomes the movement to further standardization of terminology, viewing it as beneficial for both vendors and customers. He outlined several attributes of an effective immunization information transmission process. Like all of his fellow panelists, he stressed the importance of a national standard and implementation guide.
Dr. Tang began his testimony with a vision for the Year 2010, from which he derived a perspective on standards. He said his scenario is technically feasible today; the chief preconditions for realizing this potential are privacy and confidentiality legislation, unique ways of identifying each individual, and data standards. He outlined a phased-in approach for physician use of a national registry, noting the standardization and other requirements to achieve each phase.
Discussion with the Committee focused on parent opt-outs from the registries, patient/parent access to information, privacy protections in state law, and further roles for the federal government. Dr. Lumpkin praised the CDC's assistance to states in EDI development and suggested that the Subcommittee on Privacy and Confidentiality look at the multiplicity of state policies on informed consent and their effect on information flow when a patient moves from one state to another.
Each of the work groups and subcommittees reported on recent activities and plans for the near future. (See the brief summaries in the minutes for details.)
The Committee approved charges and work plans for the Work Group on Quality and the Work Group on the National Health Information Infrastructure.
Ms. Greenberg reported on the flurry of activity following last November's workshop. A list of priorities and action items has been developed, and meetings in conjunction with a recent NAHDO meeting led to creation of work groups around specific tasks in the work plan. The organizers want to make this activity as inclusive as possible, and they are committed to active participation in standards development by public health and health services research. Ms. Greenberg asked for suggestions of organizations interested in learning more about HIPAA and willing to include a briefing on a meeting agenda. A user-friendly data dictionary is expected by summer.
The group approved plans for a 50th anniversary event in June 2000. Mr. Scanlon announced that in late February, the Department will issue a Federal Register notice soliciting nominations for membership on the NCVHS to fill vacancies that are expected to occur as of June 1999.
Dr. Detmer reiterated his thanks and passed the gavel to Dr. Lumpkin. Dr. Lumpkin acknowledged the accomplishments of the Committee under Dr. Detmer's leadership. He then challenged the Committee to take the lead in stimulating action toward a rational health information system for the nation. Transformation of the system can be achieved by making this domain meaningful to those who care about the nation's health.
Following introductions, Dr. Detmer announced that he would be leaving the National Committee to take a post at the Judge Institute of Management Studies at Cambridge University, beginning in April. He expressed his gratitude to the Committee and staff for their hard work and support.
The President's Initiative on Race has generated an effort to understand and document the role of discrimination in U.S. society. The Office of Management and Budget is coordinating an effort to improve the current statistical capabilities to measure and track discrimination in five sectors of the economy: criminal justice, the labor market, education, health care, and housing. HHS is responsible for the healthcare area. It and the other departments will develop a work plan for a literature review and review of the state of the measurement art. A research conference at the National Academy of Sciences in October will help with the latter and set an overall theme.
The Department's initiative on improving discrimination measurement began in January. The Data Council has asked its Working Group on Race and Ethnicity Data to work on the issue. One of its first steps will be a review of published and unpublished works on the statistical capability for measuring discrimination in the health area. Race and ethnicity will be the focus.
The 1998 Omnibus Appropriations Act included a provision requiring the federal government to amend Circular A-110 governing federal research grants to ensure that all data under the award are available to the public through FOIA. Previously, research grantee data were not considered subject to FOIA. The provision allows a user fee in connection with the request. The requirement has caused considerable concern, and agreement on the need for a set of procedures on how research data should be shared. Concerns focus on confidentiality and cost issues and on the effects of using what is likely to be "a blunt instrument." OMB is developing an NPRM and will invite comments on implementation. It has expressed interest in the research community's perspectives. Mr. Scanlon noted the exemptions and protections allowed under FOIA. The Department will develop a Department-wide approach to this issue, once the OMB language is released.
Turning to Internet policy, he noted that HCFA issued policy guidelines last November on security and the appropriate use of the Internet to transmit Privacy Act protected information and other sensitive information. (See the HCFA Web site <www.HCFA.gov>.) Mr. Streimer explained that this applies only to HCFA data. The agency is still working on implementation issues. In the near term, the procedures will mostly be used for exchanging data with researchers, and not for claims related transactions.
Finally, Mr. Scanlon said the Data Council is in the final stages of updating the HHS Directory, which will be posted on the Data Council Web site and published in hard copy. It includes virtually all HHS data collection systems, with details on each, and links to specific agencies and data.
On the FOIA issue, Mr. Gellman said that although there has been "panic in the streets or maybe the labs" on this provision, he does not think it is a serious problem. He has compiled materials on the topic from many sources and will send them to people on request.
Dr. Iezzoni noted the possible implications for changing informed consent procedures. Mr. Scanlon said that generally, identifiable information is not released with a FOIA request. Mr. Gellman affirmed that protection for patient identifiable information is strong and clear in FOIA, and this change does not pose a threat. There is more ambiguity in regard to how FOIA principles apply to non-federal organizations.
Expanding on his comments in response to a question from Mr. Blair, Mr. Streimer described a pilot project sponsored by standards development organizations to look at claims- related transactions over the Internet. HCFA will not draft instructions for that activity until this pilot is completed. Ms. Frawley asked Mr. Streimer to convey to HCFA the thanks of many in the industry "that this has finally gotten out on the street." She called it "a good first step."
The Department has received more than 28,000 comments on the four NPRMs on HIPAA requirements that it published in 1998. It is reviewing and analyzing them and working out how to respond. The aim is that the answers to the comments will be conveyed through a draft of the final rule. The final rules are expected to be issued in late 1999.
Meanwhile, HHS is working on proposed rules on the plan identifier and the claims attachment standard. The latter is now in review; the former is in its second review after modifications made following the first review. Both are expected by mid-1999.
The first report of injury transaction awaits a consensus document from the industry; an NPRM in this area is expected by late 1999. All work on the unique identifier for individuals is on hold, pending a privacy law or regulations.
Dr. Starfield asked about the thinking about standards for race, ethnicity, and social position data under HIPAA, especially as to where they would be obtained. Dr. Braithwaite said it is assumed that HHS will adopt consensual industry standards. HCFA programs require particular data, and providers and the standards development process determine how they are provided. He noted the OMB requirements on race and ethnicity data.
Dr. Detmer commented on the benefits of the close working relationship between NCVHS and the Data Council in addressing health information policy issues. This has included the Committee's roles in holding hearings and in keeping the broad sense of context so the focus doesn't narrow too tightly onto HIPAA efforts. He expressed hope that this relationship would continue and flourish.
Dr. Detmer announced that after 12 years of service, Lynnette Araki is leaving NCVHS to take a position with the Associate Administrator for Planning, Evaluation and Legislation at the Health Resources and Services Administration. With the Committee's approval, he presented Ms. Araki with a certificate detailing her many key roles and expressing the Committee's admiration, gratitude and good wishes. Ms. Araki thanked the Committee and said she had greatly enjoyed working with them.
Dr. Detmer observed that the inadequacy of the health component of the national information infrastructure initiative prompted this Committee to develop a vision and strategy for closing that gap. Models can be found in the infrastructures in Australia, Canada and the U.K. He called attention to the charge of the new Work Group on the National Health Information Infrastructure and the domains identified in the Committee's concept paper on the NHII. One of the areas of activity is analyzing relevant models to see what can be learned from them and what opportunities there are for collaboration.
He then welcomed the following guests:
Ms. Zelmer said Canada now officially uses the neologism "Infostructure" for its health information system. Noting that information infrastructures are a function of the health systems they serve, she described Canada's system, which is publicly funded, privately delivered, and managed by the provinces and territories. Broad principles are set at the national level about accessibility and portability. There is interest in increased integration of the health systems, with health information seen as a "key enabler" for health reform. Deinstitutionalization in mental health and regionalization are trends that heighten the need for good information on the health of Canadians and on their health system.
Ms. Zelmer outlined these attributes of the current infostructure:
There are these problems with the data: they are incomplete, fragmented, and difficult to share; they are not always optimally used; and usable information does not consistently reach those who need it. Home care is an area about which little is known in Canada, and information on race and ethnicity variables is poor. Also, the data are collected in "stovepipes."
To address these limitations, the Ministry of Health 18 months ago established a National Advisory Council on Health Infostructure. This Council, the Canadian Institute for Health Information, and Statistics Canada talked to 500 people in various settings around Canada about their information needs. This highlighted the need for information on communities, the cost-effectiveness of interventions, and outcomes, among other things, and overall the need for consistent standards across the continuum of care. Ms. Zelmer noted that activities in these areas are proliferating worldwide.
The areas in which people are working, with an eye toward national integration and modernization, include standards development, physical infrastructure, core software, and privacy/confidentiality/security. The goal is a flexible system with relevant information that is integratable around a person, a community, or an issue.
Ms. Zelmer concluded by outlining a desired evolution: from standards, to data, to information, to knowledge, to wisdom -- all toward the goal of better population health.
Dr. Reece referred to the web site of the Australian Institute of Health and Welfare <www.AIHW.gov.au>, where all its publications are available. She described the architecture of AIHW's information infrastructure in terms of its plan, players, processes, and products, each of which is critical to its success. The foundation is the National Health Information Agreement, initiated in 1993. Signatories include the Commonwealth, the six states and two territories and the two major statistics organizations, Australian Bureau of Statistics and the Australian Institute of Health and Welfare. This is a commitment to support and promote the process. It is supported by a development plan that provides a list of priorities, and a work program that implements work in the priority areas. Participants range from the Health Minister to local agencies.
Dr. Reece noted that the broad aims of the Agreement are consistent with those outlined in the NCVHS vision statement on the U.S. NHII. They include the collection and interpretation of nationally relevant health information; agreement on definitions and standards and on guidelines for coordination of access to information; and improved access to uniform information by community groups, health professionals, and government and non-government organizations. The immediate priorities include improving health and service delivery for indigenous peoples and developing an integrated national health and welfare information model. Australia also is seeking the ability to link records and to improve health outcomes information, with special focus on severe mental illness and on surveillance of potentially modifiable major disease risk factors.
The process is overseen by the Australian Health Ministers Council and the Health Ministry. The Information Management Group directs implementation of the information agreement, oversees the work program, and recommends the priority areas. Much of the "grunt work" gets done by the National Health Data Committee, which develops and maintains a data dictionary and the national health information model (NHIM) and promotes metadata standards.
Turning to the products, Dr. Reece identified three components: the NHIM (the conceptual structure outlining what the country needs to know); the National Health Data Dictionary (the vehicle for all data definitions); and the National Health Information Knowledgebase (the electronic query tool and repository for metadata). She showed diagrams of the NHIM and the NHIK and discussed the relationships among all three components, stressing that the definitions are critical. The NHIK includes information on former definitions of data elements.
Dr. Reece concluded with "a brief plug for a common template," noting along with Dr. Detmer the degree of duplication in the various countries' activities when much of the information needed is the same everywhere. Progress in this area, which is being pursued by international standards organizations, will make a big difference and make it possible to compare data and share knowledge with other nations.
Dr. Jones' team of data modelers and analysts works for the Information Management Group of the National Health Service. He began his talk by reviewing the history of England's Health Care Model. Work on the NHS model began in 1981 and eventually was informed by a clinical process view that regarded data as necessary to improve health. The English government funded several projects to look at the detail of the health care process, including administration and planning. The initial focus was on building systems, and the first product was a clinical process model and "common basic specification" using object-oriented analytical techniques.
Experts in health, information, and government then advised the government to support the building of the model and to make it free to everyone. The result is called the Health Care Model. Dr. Jones described aspects of the model, which essentially represents the processes done in the delivery and administration of care. It is in a built-in OMT, and will eventually be moved to UML. It will soon be posted on the Web.
The model is independent of technology and can be used to assess current technologies and potential standards. From the clinician's point of view, the model serves as a guide in gathering information from the patient, from the knowledge base, and from other data sources about the patient. From the national infrastructure vantage point, it indicates where standards meet the care process as well as administrative and management processes. It also produces an audit trail. It can be used to manage projects, develop national standards, and target research. It also links to an accreditation mechanism for institutions that will be mandated in a year, and it helps system designers in developing integration in that the process model is a kind of road map. The model is independent of organization structure.
The model creates partnerships between/among bodies with shared interest or responsibility for data elements, facilitating the national information strategy. The partnerships bring together suppliers, technicians, and users.
One of the risks of not having such a model are potentially inconsistent standards. The risk of having the model is that "it will become a straightjacket" that inhibits innovation and fails to respond as needs change. The difficulty of getting political and user endorsements is a challenge -- and Dr. Jones noted the difference between the two types of endorsements. He decried the existence of "competing models" covering the same area, and urged that whoever maintains the model, knowledge and resources should be pooled.
The U.K. government has a long-term information strategy (posted on the government's Web site) with a modernization fund to improve the NHS that dedicates one-fifth (1 billion pounds) to information technology modernization over seven years. Dr. Jones stressed that IT should always be regarded as overhead and "top-sliced," because "you can't afford not to do it." Both a data model and a process model are essential. Finally, he echoed Dr. Reece and Ms. Zelmer in urging that countries pool resources and skills for mutual benefit.
Dr. Deering, lead staff for the NHII Work Group, responded at Dr. Detmer's invitation and was the first of many to thank the presenters. Noting that they would take part in the Work Group's afternoon session, she commented on the balance between breadth and detail in the combined presentations. Now the Work Group has the challenge of mapping all this input to its goals for the U.S.
Dr. Lumpkin asked for the thinking behind the term "health care" in the name of the U.K. model. Dr. Jones replied that in the U.K., "health care" spans the hospital, community, and social care.
Dr. Starfield noted that while the U.S. has played a seminal role in the evolution of health statistics, it is dogged by fragmentation. She asked what we can learn from other nations about how to think about linkages across the continuum of care, about the social issue of terminal care, and about the unique identifier. Ms. Zelmer cited Canada's success in using the National Population Health Survey as a mechanism for requesting consent for data linkage. The rate at which consent is granted was very high, and this has permitted a number of linkages. In addition, each of the jurisdictions has a unique identifier, and a national one is being worked on.
Dr. Reece said Australia is in the investigative stage with regard to record linkage. With the continuum of health care, the important thing is to incorporate community services in the thinking, and this requires data standards that permit compilation of information from many sources.
The U.K. is in the process of implementing an NHS number nationwide. Dr. Jones encouraged the Committee to talk to the people working on that project about their experiences. He described aspects of England's national information strategy, noting that in that model, the longitudinal health record affords the linkages. This will be tested over the next three years in several beacon sites.
Mr. Blair noted the less centralized way in which IT develops in the U.S. health system and the need for a business case for electronic patient records (EPR); he asked if England had generated such evidence. Dr. Jones said one purpose of the aforementioned testing is to "get a handle on its value." He pointed out that English practitioners have funding for acquiring a computer but that doesn't help them to use it, and the culture has not moved along with the technology.
Dr. Reece said that those in "beacon locations" in Australia that have been implementing the EPR are enthusiastic about it. A problem is that it becomes proprietary in each location and different systems proliferate. In changing the culture, Australian planners think it will help to pay practitioners to collect data and to do so using a standard mechanism.
Dr. Cohn raised the issue of having good data quality for the data model. Ms. Zelmer said Canada is working on this issue, coping particularly with the difficulties of getting data not linked to payment. They also are working on cross-validation of data sets.
Dr. Reece pointed to the critical role of a commitment by the major stakeholders in the National Health Information Agreement. Also, the government provides some motivation in that funding is contingent on data submission. Moreover, stakeholders ultimately get data back for use in policy decisions, and closing that loop is important. For information of state and local interest, the national government encourages people to post data standards for locally- based definitions on the Knowledgebase to reduce duplication of effort and illustrate emerging standards. In that vein, Ms. Zelmer said the publication this year in Canada of an extensive national health report has greatly raised visibility of the role and purpose of data.
Dr. Jones said that until recently, practitioners were not clear about the clinical value of collecting data, a process that was originally framed in the U.K. in terms of business management. A push for data quality has to be driven by the needs of the user community.
Dr. McDonald hailed the fact that the Australian model is nested and does not look like "a circuit diagram." Dr. Reece remarked that communication/education about the process is very important. They found that taking away the lines in the diagram did not hamper its utility and made it much more understandable. She urged people to look at the model electronically, where the relationships are more clear. Dr. Jones said the English model is in fact "a spider model"; but the way to understand it is through the process model, not through the data structures.
Noting the importance of standards as step one, Dr. McDonald asked Ms. Zelmer about cross-province data integration. She described the Partnership for Health Infomatics and Telematics, a national public/private sector partnership. Data definitions in that context relate to the National Health Data Model and are based on the HL-7 data model.
Dr. Mor noted the problem of proprietary structures and systems of data in the U.S. and asked who owns data in the presenters' countries and how that affects the discussion about "who can do what with what kind of data." All the presenters described quite complex conditions in this respect in their countries, focusing on the statutory limits on uses of data and voluntary agreements to share data rather than on legal ownership.
Dr. Detmer brought the session to an end, with thanks to the presenters.
Dr. John Eisenberg, Co-Chair of the Data Council, publicly expressed his thanks and those of his co-Chair Dr. Hamburg for Dr. Detmer's leadership of the National Committee since mid-1996. He then read a letter of "gratitude and appreciation" for Dr. Detmer from the Secretary. The letter, which is part of a certificate presented to Dr. Detmer, acknowledges the "major contributions to the advancement of health information and privacy policy" made by the Committee during Dr. Detmer's tenure as Chair and its valuable role as "a bridge to the industry, research, and public health communities."
Dr. Eisenberg announced that the Secretary has appointed Dr. Lumpkin as Acting Chair of the Committee.
Ms. Greenberg commented that the forthcoming three-year report covering all the Committee's activities takes a more integrative approach than past annual reports, which were organized in terms of subcommittee activity. Background documents are referenced on the Web site rather than attached to the report, which also contains a foreword by Dr. Detmer and the NHII paper. The report will be printed and also posted on the Web.
The Committee's comments were elicited on both the three-year report and on the Second Annual Report to Congress on HIPAA implementation.
Mr. Gellman stated his dissatisfaction with the discussions of the unique patient identifier in both documents, which in his view do not fairly reflect what went on. He asserted that while the reports suggest that the Committee was primarily concerned about privacy implications, in fact the Committee reaffirmed its support for earlier NCVHS recommendations and espoused the goal of a unique patient identifier without formal analysis of costs or benefits. He acknowledged, however, that the September 1997 NCVHS recommendation states that it would be premature to select and implement an identifier in the absence of legislation to assure the confidentiality of individually identifiable health information and to preserve an individual's right to privacy.
He continued that in 1998, the Committee voted to proceed with hearings on the issue prior to the availability of a white paper promised by the Department. He asserted that these actions "sparked a national wave of opposition" that prompted Congress to enact a moratorium on adoption of a patient identifier. Thus, Mr. Gellman asserted, a fair description would show that the Committee's interest in privacy was secondary to its interest in having a patient identifier. He stated his intention to submit this statement as dissenting views to both reports if the final versions do not adequately address his concerns.
Dr. Detmer mentioned some options for addressing this and asked the Committee, and particularly the Subcommittee on Privacy and Confidentiality, how they wished to proceed. Ms. Coltin advocated trying to arrive at a description of activities that everyone believes is accurate. She challenged the accuracy of some of Mr. Gellman's statements (e.g., that the Committee never took testimony on the issue), noting that over the years there has been a good deal of testimony on the issue. Ms. Frawley suggested that she, Dr. Lumpkin and Mr. Gellman work on this section of the reports and try to agree on language; failing this, Mr. Gellman can attach his statement. The Committee approved this approach.
Moving on, Dr. Lumpkin suggested that the Second Annual Report to Congress on HIPAA make clear that the timing of HIPAA implementation makes moot the concerns expressed about the Year 2000 transition.
Dr. Detmer agreed with Ms. Greenberg that the HIPAA report should cover the Committee's support for the workshop on the Implications of HIPAA for Public Health and Health Services Research.
In addition, the language on security practices in the HIPAA report will be corrected to correspond with the Committee's original letter to the Secretary on the subject.
Finally, Dr. Detmer commented on the impressive collective effort reflected in the 1996-1998 report and the second HIPAA report to Congress. The Committee will revisit this agenda item in day two of this meeting.
Dr. Iezzoni said that the Subcommittee on Populations, as part of its initiative on Medicaid managed care, asked the contractor (Sara Rosenbaum of George Washington University) to draft model contract language around data and data reporting specifications for states to draw from on a voluntary basis in developing contracts with managed care organizations. She circulated copies of the draft language and asked members to provide comments to Ms. Rimes by February 22. The draft will be revised based on Committee comments and vetted with various organizations and individuals around the country. The final language will be part of the Subcommittee report submitted to the Committee in June.
Ms. Coltin represented the Subcommittee in the discussions in which the specifications were developed. She explained that these information and data purchasing specifications will be part of a broader menu of possible contract provisions to assist states. The set is intended to help meet the data needs of state Medicaid agencies and state health departments. A set of underlying principles clarify (among other things) who is accountable for what information and try to limit the demand for data from MCOs in recognition of their primary role as care providers. Ms. Coltin asked for comments on whether the principles seem a fair basis for evaluating the provisions, and if so, whether the provisions adhere to the principles.
Mr. Scanlon noted that the analysis is somewhat based on existing Medicaid state contracts and hearings, and he reiterated that state use of the language is voluntary. Ms. Greenberg explained that the contractor was already working on aspects of the project with CDC and SAMHSA; the NCVHS contribution is to support an overarching effort to look at information requirements across different types of MCO services.
Ms. Frawley referred to a letter to the President of the National Association of Insurance Commissioners, developed by the Subcommittee on Privacy and Confidentiality. The group was briefed in November on the Health Information Privacy Model Act approved in September 1998 by NAIC (comprising the insurance commissioners of all states, Washington, D.C., and the territories). The model act covers protected health information and the activities of insurance carriers. The insurance commissioners can try and enact it through their state legislatures.
The Privacy/Confidentiality Subcommittee looked at the model act in the light of its report to the Secretary and the Secretary's recommendations to Congress. It concluded that it cannot endorse the model act for a number of reasons, the most serious of which concern the treatment of 1) redisclosure of protected health information and 2) health services research and the use of IRBs. The draft letter focuses on these key issues. The Subcommittee would like to send it to all 55 commissioners and to the staff that briefed the Subcommittee. Ms. Frawley briefly described the model act.
Members suggested a few ways to strengthen the letter, which will be revised and acted upon in day two of this meeting (see below).
Dr. Detmer observed that the Committee has had a longstanding interest in risk adjustment, a topic of increased currency since the Balanced Budget Act required MCOs to provide encounter data to HCFA for risk-adjusted capitation rates. The panelists were asked to speak on their perspectives on rate adjustment methodologies.
Mr. Knutson directs a health research organization affiliated with a large multi-specialty group practice and health system. Within his group, he works mostly on risk adjustment and is developing the Medicaid managed care risk adjustment method for the State of Minnesota. He also has analyzed risk in some Medicare-Choice sites. Formerly, he managed an HMO, so he has experience with trying to produce data from "inside the black box."
He noted that HCFA is implementing a diagnosis-based risk adjustment method to replace the AAPCC, to be implemented in 2000. Initially they will use only inpatient principal diagnoses because of the unavailability of ambulatory diagnoses, fully recognizing the inadequacy of using only inpatient data. The diagnostic cost group system and the principal inpatient diagnosis model classify beneficiaries on the basis of hospitalizations and predict expenses in a given period.
Turning to the inherent data issues, Mr. Knutson said the hope is that when fully implemented, health-based payment will change the incentives for managed care and improve diagnosis coding practices and data availability. The reliability and validity of ambulatory diagnosis should also improve. Currently, there are serious data flow and data quality problems any time payment is not imminently connected to data; "dollars really need to be at stake before the problems are seriously addressed." However, HCFA has deferred addressing this on the ambulatory side because of the data issues.
The data flow between providers and plans is the most critical path. Providers are now taking more risks, including delegated responsibility to process claims. In these complex relationships there is fertile ground for errors in data flow, influencing the MCO's ability to produce data for risk adjustment. Mr. Knutson explained that there are three arrangements for claims processing: by the plan, by the provider, and mixed. These types commonly co-exist within the same plan or market, adding to the complexity. In addition, the data are used tactically in contract negotiations.
Despite these complications, there are examples of the use of diagnosis data for risk adjusted payments to plans, e.g., in Colorado and Maryland Medicaid programs. There are also examples of diagnosis-based risk adjustment on the private purchaser side. One is the Health Care Action Group in the Twin Cities, a purchasing alliance that has introduced capitation-like incentives. Every physician, lab, DME supplier, and hospital bills the purchaser's single administrator directly rather than running data through HMO systems. Better cash flow is a strong incentive.
Mr. Knutson noted that there are many reasons why data stops at different points between provider and purchaser -- some "trivial or even banal."
Regarding data quality and what can be expected under the new risk adjustment system, he listed more ambulatory ICD codes per encounter can be expected, more modifiers, and probably more strategic coding to focus on particular codes. (He likened this to "a code arms race.") Meanwhile, system developers will keep trying to make systems that are less amenable to this. With coding rules and training on how to code legitimately, data quality should improve with this incentive. Mr. Knutson recommended that plans and providers be given training to hasten their learning, something that Colorado is already doing.
Finally, he suggested that organizations that care about data for the public good try to understand the process engineering aspects of data flow from providers to HCFA. Data issues are intermingled with other strategic and tactical issues that confound the flow. A study that would distinguish the more fundamental problems from the trivial ones would be useful, along with a look at training programs and monitoring to understand coding validity and reliability.
Mr. Anderson said he would focus on the things that make it difficult for his and other plans to respond to Medicare data requirements. Kaiser has long supported improving risk- adjustment methods for Medicare beyond the "crude" actuarially-adjusted per capita cost method, "even if it might in many cases result in reduced payments to Kaiser."
He discussed the following barriers to successful data submission in Medicare:
On the first, he noted the complexity of Kaiser and the difficulty of convincing managers of the need to do new things and commit new resources, such as the challenge of convincing hospitals of Kaiser's responsibility for submitting data to HCFA. New systems were needed on short notice, and HCFA's responses to requests for help are slow and some of its own systems are confused. Add to this the Y2K problem.
On the second factor, he explained Kaiser's concern that the new coding and reporting system not undermine existing behavioral incentives to provide appropriate care. He described the range of sizes, degrees of complexity, and different ways of doing business among its affiliates. In general, providers are becoming wary about the whole subject of risk-adjustment.
Third, Mr. Anderson said he supports HCFA's decision to start now with something that is "directionally correct," even though a full encounter-based model will take years to implement. It is working hard to deal with the resultant incentive issues, but in ways that are controversial. He hopes Medicare will move forward swiftly, give plenty of notice, and engage the industry in dialogue.
Fourth, he emphasized the high stakes in the payment issue. Because of the estimates of overpayment to Medicare health plans, plans fear increased uncertainty during the transition. He praised HCFA's adoption of a modest phase-in of the new payment method over five years, and noted that there is strong motivation to be sure everything is counted correctly because mistakes can be costly. Because of these uncertainties and anxieties, many organizations have become strident opponents of implementation, a political opposition that will complicate the discussion.
Mr. Anderson pointed out the broader context for risk adjustment and the applications beyond payment. The message, he said, is "pay attention," as these may be the most important areas for future evolution of the methods.
He encouraged the National Committee to help ensure a focus so that data requirements are limited to essential purposes, that unnecessary data are not collected, and that there is coordination among the users. He called attention to the implications for providers and their role in ensuring the accuracy and completeness of records, and urged the Committee to help them develop strategies to encourage proper and complete coding. He also recommended developing standardized definitions and methods of transmission and placing confidentiality and privacy in the proper perspective so that risk adjustment and progress toward evidence- based care are possible. Finally, he encouraged the Committee to help shape the national agenda for risk adjusters, exploring applications beyond payments and building consensus about how to support their use.
Studies by several organizations have indicated a selection bias of 10 to 14 percent in the enrollment of Medicare beneficiaries. Medicare wants to compensate plans appropriately if they enroll sicker people. The BBA required implementation of a risk-adjusted payment methodology by January 1, 2000, giving HCFA no choice but to start with a system based only on inpatient hospital data-- "the only place we could have started." It published the methodology it plans to use on January 15, 1999 and released software on the Web. Its report to Congress will be released on March 1, 1999. A letter to plans with the estimated impact on their payment rates is scheduled for May 1, a deadline HCFA is trying to get extended.
Ms. Tudor briefly described the new system, which adds health status to demographic factors. Payment is based on each enrollee's risk factor, initially on the basis of the principal inpatient diagnostic cost group (PIP-DCG). The plan is to move to a comprehensive risk adjustment mechanism by 2004. She asserted that starting with inpatient data is a good first step, as inpatient admission is a proxy for greater severity. Initially, only a small percentage of enrollees will be risk-adjusted. She then described the PIP-DCG and other aspects of the new model and showed some examples with the effects on underpayment and overpayment. There will be a reconciliation once a year, early in the year, comparing capitated payments to actual costs and determining who should have gotten more, or less.
Ms. Tudor described some of the testing and analysis done in the start-up period, looking at predictive ratios, impact analysis, the completeness of the data, ratio of hospital inpatient encounters to average plan enrollment, comparison with HEDIS inpatient hospital rates, and other factors. Planning started in 1997 to give the plans time to get used to sending in data and working with it, and to provide advance payment estimates. Plans are now submitting data on which payment in 2000 will be based.
Implementation of the comprehensive model will require more lead time than the current model and is estimated to be between 39 months and 4 years. Outpatient data will be needed in 2000, with a model that will be adopted in 1999. Data were collected retroactively for the current inpatient model. The plans had a variety of ways to send the data and a helpful abbreviated version to use. Everything was done to ease the burden of plans. Ms. Tudor expressed hope that many of the problems will be resolved in time to set up a more efficient system for outpatient data.
Acknowledging the concerns about privacy, she expressed confidence that HCFA can protect "the factor." Plans will only receive information on the plan-wide total, not on individuals, and it is impossible to infer the diagnosis from the factor.
Dr. Boesz reviewed her twenty-plus years of experience with encounter data. She characterized the 1970s as interested in encounter data for understanding how to manage business; the 1980s, how to regulate the business; and the 1990s, how to pay for the care. Through it all, the business people in health care have wanted to use the minimum amount of data needed to make these decisions. Today, Aetna feels strongly it doesn't need a full UB92, even to do health risk assessments nor for patient care management. So a warehouse of that amount of data is costly and labor intensive, and "business folks get ... nervous when they hear all this." They want to know "what is all this comprehensive data going to do for us?" This uncertainty is aggravated by HCFA's projection of a 7 percent reduction in payment, raising the question of whether companies can stay in this business with "the Medicare product." One determinant is what kind of business partner the government is going to be in the future. These plus other factors led Aetna and other Medicare contractors to decide to exit some markets last fall.
She described some of the dynamics of the company's relationships with providers, who "push back rather vigorously" in regard to data collection and its cost implications. Companies have had to make rapid decisions rather than working through the issues, particularly because of the BBA. The possible extension of the May 1 deadline for the ACR submission could provide a bit of relief; another is the promise of a phase-in period for risk adjustment. Thus her company is unable to assure that they will stay in given markets, given current uncertainties. The Y2K issues inherent in technology add to the challenges.
One of the issues raised is that the data collection required could make the managed care environment feel like fee-for-service again (e.g., data intensive). Another concern is that the data collected for one purpose will be used for other purposes for which they are not appropriate. Dr. Boesz agreed with the proposition that if risk adjustment is going to be done, better data are needed; the problem is the instability created by these changes when implemented over a short period of time. For example, it is difficult to offer providers long- term contracts without knowing what the company will be paid by HCFA.
She recalled that in the mid-1970s, there was a similar concern about overpayment, called skimming, by federally-qualified HMOs, and eventually well and sick people become more evenly distributed across plans. She proposed that today, a better solution might be to concentrate on incentives to get more volume into MCOs. The industry is also concerned that employers will want to follow the government into risk-based adjuster payment. This, she asserted, would be a step back towards the notion of experience-based payments.
On the Committee's role, she noted the importance of setting standards for minimum data sets for use by Government agencies. For example, Government acceptance of HEDIS- type data sets would be an improvement. Promoting acceptance of the data recommendations of accreditation bodies is also important, along with proactiveness in the confidentiality area.
Clarifying the reasoning behind "the factor," Ms. Tudor explained that plans are to be paid based on the multiplication of the factor times the county in which the individual resides. To a follow-up question, Dr. Boesz said Aetna will pass the risk on to risk-sharing subgroups, as appropriate.
Mr. Anderson addressed the concern about being on the slope toward fee-for-service reimbursement or claims-based reimbursement. He pointed out that the effort here is to design an incentive-neutral adjustment system. In the short run, the incentive is in the wrong direction, but perhaps at the next stage countervailing incentives can be introduced. He is hopeful that this will evolve.
Dr. Newacheck noted the Committee's interest in the impact of the new system on costs, quality, and plan participation and he asked about HCFA's plans for internal and external evaluation. Ms. Tudor outlined some of HCFA's plans to evaluate impact, noting that the Medicare Plus Choice Evaluation is external.
Returning to Mr. Knutson's comments about training for coding, Ms. Frawley said that her organization, the American Health Information Management Association, is available as a resource and ally. It is already working with hospitals around a host of issues (e.g., fraud/abuse, DRGs, APCs, NPRMs, OASIS).
Dr. Cohn asked about the prospects for use of the UB92 short version and the HCFA 1500, and Ms. Tudor said HCFA will take into account HIPAA requirements.
The group discussed the predictive power of the adjuster and the likelihood that it will change people's behavior. Mr. Anderson stressed that predicting payments for populations, not individuals, is what is important, and he noted the possible need for some form of reinsurance to cover high-cost episodes. Mr. Knutson said that the predictive accuracy of the new models is far superior to that of the old actuarial model.
Ms. Coltin noted the apparent tension between the current stage of risk adjustment and HCFA's QISMC program, which is trying to build in incentives to improve quality of care. Improving quality is costly, and initially keeping people healthy would result in lower payment under Medicare risk adjustment. Ms. Tudor said this is why HCFA isn't putting much emphasis on risk adjustment in the next few years, until the comprehensive model is ready.
Dr. Detmer thanked the presenters and recessed the meeting, to reconvene in Subcommittees and Work Groups.
Dr. Friedman introduced this session, which he helped to organize, by observing that the functions of public health surveillance will continue to be crucial in any emerging national health information infrastructure. However, the nature of data and data collection will change. This session will give an overview of public health surveillance activities and related health information issues, with special attention to standards and integration. While multifaceted and largely effective in its primary mission, public health surveillance is also minimally integrated at the person level and over time. It is not unitary at the national or even state level, but rather a collection of systems that "sometimes don't fit together particularly well." He introduced the three panelists and turned the floor over to Dr. Koo.
Dr. Koo stressed CDC's desire to help NCVHS include public health surveillance in its visioning and activities to build the national health information infrastructure. It also wants the Committee's help in articulating the centrality of public health surveillance for public health practice. She noted that epidemiology, a basic science of public health, tries to determine what kind of people have a given disease as well as risk factors and opportunities for intervention, in contrast with physicians, who try to determine what kind of disease a given person has.
Public health surveillance is done to assess public health status, define priorities, evaluate programs, and stimulate research -- all part of public health practice. It needs to be systematic, ongoing, and linked to action. It uses many kinds of data. It is used to estimate the magnitude of a public health problem, follow trends, determine geographic distribution of an illness, portray the natural history of a disease, detect epidemics or define a problem, generate hypotheses, stimulate research, evaluate control measures, detect changes in health practices, and facilitate planning. Dr. Koo gave some examples in each of these areas. She noted that at the local level, surveillance data are also used to identify individual cases and to track contacts for communicable diseases in order to treat them.
For these varied purposes, many types of data are needed from many sources beyond medical data (e.g., police records and traffic accident records). Dr. Koo described the National Notifiable Disease Surveillance System. Reporting is mandated by state law or regulation, not at the federal level. States electronically report to CDC the data they receive from counties, usually on paper. Generally, CDC has separate information systems for surveillance (STDs, AIDS, TB, etc.) because of the categorical funding of public health. The result is a multiplicity of information systems with variable data input and access methods, making it difficult to evaluate problems across programs.
To remedy these problems, with HIPAA as a motivator, CDC has started working more with standards development organizations (SDOs), especially those concerned with clinically relevant standards. Dr. Koo gave some examples of work with SDOs. The CDC is very interested in computerized medical records. Public health officials want to capture electronic lab data as they are generated, and to get demographic data at the same time. Various state projects for integrated reporting from managed care and Medicaid have CDC funding. The Injury Center has worked with SDOs on emergency department data standardization, and Cancer Registry people are working with HL-7.
Another major project is the surveillance systems integration project, which includes core data for electronic reporting, a unified user interface, and development of a secure data communications pipeline using the Internet.
Dr. Koo noted the connections of these efforts to the National Committee's projects on the NHII and on health statistics for the 21st century --examples of the interdependence of public health surveillance and health information systems. She stressed that surveillance is integral to daily public health practice.
Dr. Birkhead said he would offer a state perspective on public health surveillance, noting that "all surveillance is local." Contrasting state/local surveillance with national efforts, he said the statutory basis for surveillance rests with states, along with the link to control. Thus, immediate access to real-time data is important. He gave examples of state/local surveillance and its links to control activities with respect to assuring appropriate treatment of cases, assuring that contacts are treated, removing the source of outbreaks in the community, and monitoring effectiveness in real time (for example, using the BRFSS). He noted that it is difficult to distinguish surveillance from control activities.
The National Notifiable Disease Surveillance System (NNDSS) is built from surveillance data at the local and state level, on a list of conditions recommended by CDC and CSTE. States decide what to monitor and collect standard data elements which are reported to CDC. Individual-level data go to CDC without identifying information. One issue now is the lack of integration of the national system, making it necessary for states to maintain four separate categorical software packages plus NNDSS reporting. This simply does not work at the local level, where one public health nurse may be responsible for all. He welcomed the push toward integration now taking place in CDC.
Dr. Birkhead showed a matrix with 17 data systems on one axis and 6 health/disease categories on the other. One data system often contains data on several public health categories, highlighting the potential benefits of standardization.
Also needed are adequate non-categorical resources for training, surveillance, and local infrastructure development. A related issue is staff: only 1,600 FTEs nationwide are devoted to public health surveillance for all infectious diseases, most of which go to HIV, STDs and TB, with only 64 for food- and water-borne diseases.
In 1994, CSTE released a proposal for a conceptual (virtual) national health surveillance system. It is an umbrella that would provide visibility and get people out of the categorical mindset. It is proposed as a federal/state/local collaboration, one result of which would be a single point of access for surveillance data on the WWW.
Finally, Dr. Birkhead offered a cautionary tale by describing a study in New York that matched gastroenteritis codes from a health care fiscal intermediary with the State's surveillance file. What showed up statistically as a cholera epidemic turned out, after investigation, to be the result of inverted code numbers. The tale points to the dangers of false positives as well as of missed true positives in data collection and reporting using administrative data, and it shows the merits of direct reporting from providers. Adding laboratory and other clinical data will make the information even more useful.
Recently, public health surveillance activities have expanded dramatically to include a range of health outcomes and determinants, one of which is occupational health and the impact of work on health. The health endpoints in which the Department of Health is interested are injuries, diseases, and hazards. Dr. Davis' remarks focused on occupational illnesses and injuries. She stressed that surveillance of hazards would be optimal and lend itself best to prevention efforts in the work area; however, this is in early stages of development and most current efforts derive from injuries and illness.
In this context, the goals of surveillance are to identify the magnitude and distribution of health and safety problems and particularly high-risk industries, occupations and work sites where intervention is warranted. In addition, public health workers are looking at population patterns -- finding, for example, that Latino workers are at high risk of overexposure to lead in the workplace.
National statistics on occupational injuries come from data from employers, who are required to maintain logs under the Occupational Safety and Health Act. The logs from a sample of employers generate national estimates. In most states, the sample is not large enough to generate state-specific information. This Bureau of Labor Statistics system does not capture the majority of occupational diseases, which are difficult to document, especially chronic diseases. Another concern is underreporting of injuries on the OSHA logs, estimated at 25 percent. Tying reporting to regulatory intervention creates an incentive to under-report.
Thus, there is increasing recognition of the need for a strong state role in occupational health surveillance. State data can be used to assess the BLS data, and they can also help link surveillance to intervention. In recent years, many states have been building surveillance systems with funding by the National Institute for Occupational Safety and Health, using several data sources including case reporting and data systems. Dr. Davis stressed that this combination of case-based and population-based reporting is important. There has been an effort to develop standardized core variables and data collection instruments and packages, to facilitate aggregation across the states.
Turning to the Massachusetts experience, she described efforts to build a surveillance system for work-related injuries to minors, based on the knowledge that many are at high risk and that injuries are largely ignored. Hospitals are now required to report work-related injuries among persons less than 18 years old to the health department, which received NIOSH funding to develop a model surveillance system. A sample of ten hospitals are now reporting on a monthly basis. These data are combined with data from the worker's compensation system claims and other sources. (Importantly, there is only a 3 percent overlap between ER cases and cases in the worker's comp system, pointing to the need for multiple data sources.) Serious cases are interviewed. Technical assistance is provided to employers, and educational materials to parents and teens. Egregious safety violations are referred to a regulatory agency for worksite follow-up.
Summary data are used to target broader-based intervention efforts. Distribution of injuries by industry has revealed the five highest-risk industries, whose workers have been targeted for education; and geographic distributions also have been computed, showing the areas at greatest risk. The data led to funding to develop a model community-based intervention program in the City of Brockton. They also have informed state and national policy on child labor.
Dr. Davis reinforced the points made by other presenters about integration and coordination as instruments of planning. She observed that planning for particular health issues needs to take place in a large public health surveillance context, which in turn is embedded in the larger context of comprehensive and coordinated health information systems.
Dr. Starfield asked about the prospects for using surveillance to look beyond specific diseases to general issues of poor health, and Dr. Birkhead cited some examples of this kind of effort. CDC is trying to develop new statistical and epidemiological methodologies for detecting trends. Looking at health community-wide is still hampered by the categorical nature of systems and the absence of key information on risk factors. Dr. Koo cited efforts to relate asthma admissions to air quality data as an example of what CDC is trying to do. Dr. Starfield explained her interest in elucidating the impact of a range of risk factors on individual health, and Dr. Birkhead said another constraint is confidentiality concerns.
Dr. Lumpkin pointed out the instability of denominators, especially in intercensal years, and the resulting problems in determining rates. (He noted that epidemiology is sometimes defined as "the science of searching for a denominator.") He then raised the issue of record-level versus aggregate data in reporting to CDC. Mr. Gellman broadened the question to ask what is collected and what is reported by the different surveillance systems. Dr. Birkhead said this varies; for example, the NNDSS uses record-level data, but without identifiers. The goals of the surveillance system are compromised if providers withhold data because of concerns about confidentiality, and there are efforts to develop a model state statute around confidentiality of surveillance data. Mr. Gellman pointed to the proliferation of narrow confidentiality laws and its result of "more confidentiality laws, not necessarily more confidentiality."
In response to a question, Dr. Birkhead said that confidentiality concerns often cause providers to fail to report. Dr. Davis added that in the occupational injury area, the burden of reporting is the major factor. Mr. Gellman predicted that talk of "linking everybody's computer systems together" will only heighten the public's confidentiality concerns. The panelists commented on the statistical adjustments to account for estimated underreporting. Dr. Friedman estimated that 90 percent of surveillance does require person-specific information because it is aimed at intervention; however, this information stops at the local or state health department level. Dr. Detmer commented that it would be very useful to public health surveillance to have hard information on security lapses and the "leakiness of the system." Mr. Gellman observed that this is hard to find, because most abuses are probably by insiders with legitimate access.
Asked to comment further on the use of claims data, Dr. Birkhead agreed with Dr. Cohn that they may be usefully leveraged for some public health surveillance purposes.
In conclusion, Dr. Friedman thanked the presenters and observed that the goal for these public health endeavors is to bring them together so they are more, not less, than the sum of their parts.
Ms. Abernathy stated that immunization registries are confidential, computerized information systems about immunizations in children. Universal coverage is approached by encouraging community and state registries to exchange information through a nationwide network. 60 CDC-funded immunization projects have current or planned registries. The goal is for all private and public providers of immunizations to participate in the Registry, to compile complete immunization histories for each uniquely identified patient.
Among other things, a comprehensive system could generate phone or mail reminders when immunizations are due, target at-risk populations, determine aggregate immunization levels for providers, and integrate immunizations with other preventive health programs as well as measuring changes in immunization rates over time.
Children are typically entered into a registry at birth or at their first contact with the healthcare system. Each immunization is sent to the registry. Providers can query the registry for a history of each patient. To do this, registries must receive and process information from many kinds of systems.
Providers are supportive of these goals, but do not want to purchase special equipment or devote additional staff time to participate. To address technical challenges, CDC started working with HL7 in 1995 to develop standardized transactions for transmitting the immunization core data set. Version 2.3 was issued in April 1997, with an implementation guide. The more advanced registries then began developing tools to exchange records using this standard.
In July 1997, President Clinton stated support for efforts to raise immunization levels, promising more federal support. Secretary Shalala was directed to work with the states on an integrated system, and a good deal of work ensued on several fronts. The activities will be announced nationally in a few months.
In the technological sphere, a stakeholder/expert group convened by NIP in August 1998 advised NIP to develop standards and benchmarks rather than trying to develop software for the registries. Floyd Eisenberg (see below) encouraged the registries to develop a common exchange protocol, and participants agreed to remove optionality from the messages. Five states plus San Bernardino, California then met in September to work out how to use HL7 in their documents, enabling a field-by-field comparison of the formatting of transactions that contributed to development of one common implementation. The cooperative effort has since been joined by the Indian Health Service, the Departments of Veterans Affairs and Defense, and Kaiser Permanente's national Clinical Information System. The first draft of the new implementation guide is being reviewed by participants, and will be published on the Web and provided to vendors when finalized. NIP also plans to educate providers about the systems.
Ms. Abernathy concluded that NIP supports the National Committee's work on standardization and on computer-based patient records.
Mr. Tait called his presentation a case study of establishing a population-based registry across a broad spectrum of the state, using multiple technologies to collect information. The effort has the strong support of the Department of Public Health, and the Registry now contains about half of the State's immunization events. In 1993, the State developed Cornerstone, an integrated system for public health, which replaced a WIC system and incorporated other family components and is connected by a local area network. It is installed in more than 270 locations in Illinois and has established control over data definition and content. The immunization component is being integrated with this system.
Local health departments and community health centers with preexisting immunization databases, typically linked to local billing systems, undermine the goal of a population-based registry and necessitate electronic interfaces. An effort began in 1996 to incorporate private sector immunizations. Participation is not mandated, and an aggressive strategy has been mounted to present a variety of ways of participating, using multiple technologies for both automated and non-automated providers.
Mr. Tait briefly described the Illinois Immunization System, or TOTS (Tracking Our Toddlers' Shots), which allows access to data and reminders via the phone as well as fax and EDI options. Because of the challenges of EDI in the absence of standardization, the Department implemented an EDI-interface engine to support ANSI and HL7 standards. It has since exchanged HL7-based EDI immunization data with Georgia and Arizona. Further work is underway with several states and CDC to clarify the standard and create guidelines for using it. Within Illinois, many local health departments and community health centers are unable to support EDI transactions and standards. Many partners have difficulty modifying their systems for the new HL7 transaction. It is important to continue to clarify the standards and develop good guidelines for their use, to encourage software vendors to build the necessary functionality into their products.
SMS serves the information technology needs of the health industry in North America, Europe, and Asia-Pacific. Forty of its customers perform immunization registry and tracking functions using SMS systems, many under the auspices of county health departments. SMS has worked with HL7 code sets to increase the specificity of information storage. Immunization records are entered at the point of care and stored in the clinical data repository. Dr. Eisenberg said his company welcomes the movement to further standardization of terminology, viewing it as beneficial for both vendors and customers. One benefit will be connecting the immunizations provided in physicians' offices and clinics not affiliated with health departments.
Dr. Eisenberg outlined several attributes of an effective immunization information transmission process, including perceived added value for the physician or business entity, having a critical mass of information in the registry, and neutrality in terms of productivity. Standardization is a necessity, along with patient authorization. Dr. Eisenberg also recommended that connectivity to a registry be simultaneous with eligibility and/or claims transmission, or occur after office hours for no additional fee. Finally, he reiterated his points about the importance of a national standard and implementation guide
Dr. Tang thanked Dr. Detmer for his service as Chair and wished him well. He began his testimony with a vision for the Year 2010, from which he derived a perspective on standards. In the vision, most physicians use CPRs and there is secure, authenticated access to comprehensive, composite health records. The clinician can with one click order an immunization, document it, and update the patient record and local and national registries. CDC has instant access to the registry and a rapid means of getting new guidelines to clinicians. The system also provides reminders.
This scenario is technically feasible today. The major challenges for realizing this potential are privacy and confidentiality legislation, the unique ways of identifying each individual, and data standards. Dr. Tang outlined possible phases of implementation for physicians' use of these services: first, access to national immunization registries; then, integration of computer-based patient records with the registries; and finally, computer-based decision support and reminders for immunization guidelines.
For phase one, he stressed that access to national networks of local registries must be secure, perhaps using a Web-based interface. The movement of consumers and patients necessitates nightly uploads of data, which could at the same time send HEDIS compliance reports to providers. Parents could get immunization records off the Internet. These processes depend on standards for data, transactions, and messages as well as content standards. The standard should include a standard template with core data sets. Unique identifiers are also needed, and Dr. Tang stated his "plea" for their adoption as soon as comprehensive privacy legislation is passed, so that public goods like the National Immunization Registry are possible.
In phase two, the need for provider-initiated activity is eliminated because of electronic querying of the provider's CPR system. This requires collaboration between registry operators and CPR vendors as well as precise standards. The result would be a reduced burden on the provider and integrated access and update capabilities.
Phase three would provide a way for CDC and other agencies to disseminate new recommendations and guidelines seamlessly through computer-based decision support. Dr. Tang described a research project that measured and compared the immunization compliance of internists with and without use of CPRs. The group with CPRs increased their immunization rate by 78 percent, perhaps because of computer-based reminders. He noted that 20 percent of the vaccines were given outside the clinic, pointing to the importance of regional registries.
While stressing that he does not question their value, Mr. Gellman took issue with Ms. Abernathy's characterization of registries as confidential. He cited the various legal ways in which confidentiality could be breached, including audits and subpoenas. She agreed, adding that CDC believes parents should know about registries if their children are in them and be allowed to opt out. CDC is working with parent focus groups to get a sense of what percentage of parents do object, since the vocal opposition at hearings may not be representative. Generally, parents support the state registries, and fewer than 2 percent opt out of them. Dr. Lumpkin pointed out that state laws vary as to informed consent and the release of this information.
Asked to comment on patient or parent access to immunization records, the panelists said this too varies by state and registry. Dr. Tang and Dr. Eisenberg spoke of patient/parent access to immunization records as part of their vision for the registry. Asked about smart cards, the panelists said they do not obviate the need for a registry behind them as the computing source; moreover, standards are needed to read the card.
Ms. Frawley asked about the number of states with statutes or regulations protecting immunization registry data. Ms. Abernathy said that Ms. Horlick (staff to the Privacy/Confidentiality Subcommittee) is a key part of a CDC survey of state activity in this area, some of the results of which are on the CDC Web site.
Asked to comment on his research on immunization rates comparing two groups, Dr. Tang said the report will be published in the March Journal of the American Medical Informatics Association. He also commented on the incentives for physicians to keep good records, including information on the reason for not immunizing a patient, e.g., an allergy. (There is a standard code for this.)
Asked if the government should be doing more, Dr. Eisenberg said it would be helpful to standardize coding for immunizations. Dr. Tang called for enforceable, auditable standardized quality measures.
In conclusion, Dr. Lumpkin applauded CDC's actions in helping states start to implement EDI. Vendors will be more interested in helping develop the provider side of the system if there is a single implementation of the HL7. An unresolved issue that the Subcommittee on Privacy and Confidentiality may want to address is the different state policies about informed consent and their effect on information flow. He noted that research shows that physicians do immunizations at a much lower rate than they think they do.
Dr. Detmer reported that the new heads of the major parts of HHS have been invited to meet with NCVHS to talk about their thinking and directions for the Department. The invitation was not accepted for the February meeting, but there are still hopes for the June meeting.
Dr. Cohn announced two Work Group-sponsored hearings: on March 29-30, on messaging format standards, patient medical record information, and data quality; and on May 17-18, on medical terminology issues. Mr. Blair added that the Work Group will produce a schedule of hearings for the rest of the year to review with the rest of the Committee.
Dr. Lumpkin said the group has been working on the mechanism for collection of race and ethnicity and other demographic information, which it hopes to pursue in the full Committee context in order to collaborate with the Subcommittee on Populations.
The Subcommittee will be working on a process for adopting a consensus for electronic data interchange for first report of injury, a topic also of interest to the Subcommittee on Privacy and Confidentiality.
Regarding a letter received from NAHDO, Dr. Detmer said he would inform NAHDO that the Committee will address the issue in a hearing.
The first draft of the report on the insular areas is being reviewed and revised and then will be sent to all meeting participants. The Subcommittee hopes to finalize its recommendations at a May meeting and bring them to the full Committee in June.
The Subcommittee also plans to finalize the report on the Medicaid Managed Care project in May, producing another action item for the June meeting.
On January 22, the Subcommittee held an educational day on “post-acute care”. In June it will begin an initiative on the continuum of care, a joint project of the Subcommittee and its Work Group on Quality. The Subcommittee will focus on measuring functional status, one of the uncompleted elements of the Committee's core data elements recommendations. The Work Group will emphasize quality issues in care venues outside acute care hospitals and standard office-based visits. It is still refining its work plan.
Dr. Iezzoni circulated the draft charge of the Work Group on Quality. The Committee passed a motion approving the charge and preliminary work plan.
In an all-day meeting on February 2, the Subcommittee met with six representatives from the pharmaceutical industry with the goal of understanding its information flow, with emphasis on consumer rights issues and privacy/confidentiality concerns.
The Subcommittee will hold a two-day meeting in April to which it will invite pharmacy benefits management companies for a half-day discussion of their role in this area. Another half-day will be spent on workers compensation and occupational health. The third issue to be addressed is employers' use of health information and the privacy protections in place, leaving a half-day for Subcommittee synthesis.
A session with the Medical Information Bureau is planned for June. At that point, having completed its work plan, the Subcommittee will be free to provide technical assistance to the Department on privacy rule-making in the event that Congress fails to pass privacy legislation by August.
Ms. Frawley and Ms. Horlick then reviewed the revisions made in the letter to the National Association of Insurance Commissioners in response to Committee comments received the previous day. The Committee passed a motion approving the letter. Ms. Frawley said the Subcommittee plans to initiate a dialogue with NAIC as a follow-up to the letter.
The Work Group met briefly on February 3, with the centerpiece being a visit from the three national representatives who spoke to the full Committee. After staff presented descriptions of the approaches to national health information infrastructures in Australia, Canada, and the U.K., the national representatives offered feedback that will be incorporated into the reports. A clear sentiment emerged about the importance of staying in touch with people in these countries, as well as others, in order to work iteratively with them on common issues. Dr. Detmer noted that the health information infrastructure is "not a one-time fixed target."
The group also agreed on the potential benefit of having on the Web a compilation of terms and definitions from a variety of sources.
Following this report, the full Committee passed a motion approving the charge and work plan of this work group.
Five commissioned papers in this area are underway, with drafts due by early April. Invitational discussion groups will be held in March and May on future major issues in health and health care and their implications for health statistics. A two-day session to look at definitional and boundary issues is scheduled for late March. More communication is planned between NCHS and CNSTAT to clarify plans for the workshop CNSTAT will host, possibly in September.
A new, fourth component to this project is a series of activities to elicit views on the entire subject, with plans for open forums, professional meetings, state visits by Dr. Sondik, and possibly a public hearing next fall. Dr. Friedman said this activity was inspired partly by the Canadian provincial hearings on health information needs. Another inspiration is the public process engaged in by the National Institute of Occupational Safety and Health, the National Occupational Research Agenda Process.
Ms. Greenberg hailed the "wonderful demonstration of synergy" in the morning's panel on public health surveillance, with its references to the national health information infrastructure, and in the case study of the national immunization registry with its references to standards issues. These are examples of what the public health community would like to do and support.
The November workshop on HIPAA, public health and health services research generated plans for a consortium to organize the public health and health services research communities around data standards issues and goals. Two conference calls took place in December at which opinion leaders agreed on a list of goals. Much of the initiative is coming from the states, who see the potential benefits of standardization for their state health data. The phone conferees developed priority action items in December and subsequently held meetings in conjunction with the NAHDO meeting, attended by approximately 50 people. Several task groups were created to operationalize the work plan. Now the interest is in making the work as inclusive as possible. There is a listserv and also a web site being maintained by The Lewin Group <www.lewin.com/hipaa>.
Ms. Greenberg asked for suggestions of organizations interested in learning more about HIPAA and willing to include a briefing on a meeting agenda. She stressed that there is a great deal of interest among public health people in participating in the standards process. In response to a question, she said an inventory of current content requirements for state hospital discharge databases and other encounter databases is expected by June, developed by NAHDO under contract to AHCPR. There are plans for implementing all of the items on the action plan.
Dr. Friedman remarked that this is the first time the different public health data-related organizations are sitting down together to discuss these issues, and it is being done voluntarily.
Ms. Frawley said that she, Dr. Lumpkin and Mr. Gellman had agreed that the discussion of the unique identifier in both the 1996-1998 report and the Second Annual Report to Congress on HIPAA should address two things: 1) the Committee's actions and recommendations independent of and prior to HIPAA, and 2) HIPAA's requirements of the Committee in this area and the Committee's response. Dr. Lumpkin suggested a statement to the effect that NCVHS has long identified the need for a unique health identifier.
Following discussion, the Committee passed a unanimous motion approving revision of the 1996-1998 report to reflect the fact that NCVHS identified the need for a unique health identifier long before HIPAA was enacted. Mr. Gellman took no exception to the motion, but reserved the right to submit a minority position when the report is released.
No other concerns were voiced in the discussion of the 1996-1998 report. Dr. Detmer said a complete NCVHS staff list would be added.
The report to Congress on HIPAA was reviewed by the Standards Subcommittee on February 3, and Dr. Lumpkin enumerated four areas where it recommends changes. Ms. Frawley said she would redraft the section on security so it is fully consistent with the Committee's letter to the Secretary on that subject.
The Committee unanimously approved the following process for completion and final approval of the 1996-1998 report and the report to Congress on HIPAA: The drafts, revised as indicated above, will be sent to the entire Committee for comments. Any concerns raised in that review will be adjudicated by the Executive Subcommittee.
Dr. Detmer reported the Executive Subcommittee recommendation of a half-day event in observance of the Committee's 50th anniversary, to be held in conjunction with the June 2000 NCVHS meeting. An effort will be made to get the Great Hall of the National Academy of Sciences for a reception, and to get foundation help in paying for it. The idea is to use the event to look both back to where the Committee has come from and forwards to where it is headed. He offered his help with the event.
The Committee agreed to this general approach and asked the Executive Subcommittee to take charge of planning.
Ms. Greenberg reported that Ms. Kanaan would be asked to prepare a 50-year report on the Committee, updating the one prepared for the 45th anniversary and incorporating elements of the three-year report just completed.
Mr. Scanlon announced that the Department will issue a Federal Register notice soliciting nominations for membership on the NCVHS to fill vacancies that are expected to occur as of June 1999.
Dr. Detmer concluded the meeting by expressing his personal thanks for "a wonderful professional experience" and passing the gavel to Dr. Lumpkin.
Dr. Lumpkin acknowledged the Committee's many accomplishments under Dr. Detmer's leadership. Looking ahead, he observed that health and health care in the U.S. are at a nexus point. Positively transforming the system will require a far greater investment in information systems than the current 2 percent of gross revenues. A major obstacle today is the lack of a rational approach to health information systems; and precipitating action to fill this void is the responsibility of the National Committee. In this move toward a vision, the work of the Work Group on the National Health Information Infrastructure can provide the structure. He proposed using the 50th anniversary event to unveil the vision -- the model of the infrastructure -- and to begin to coalesce people around it.
To be successful, he continued, the Committee needs "to expand the ranks of the converted; ...to be evangelistic"; and "to bring it down to real terms" that mean something for the health of real people, including those without health services living in housing developments a mile from the HHS Building. Communicating the message to those concerned about the nation's health will make transformation possible. He concluded that the Committee is "certainly up to the challenge."
Dr. Lumpkin then adjourned the meeting.
I hereby certify that, to the best of my knowledge, the foregoing summary of minutes is accurate and complete.
/s/ John R. Lumpkin 6/22/99
_______________________________________________________
Chair Date