The Subcommittee on Population-Specific Issues of the National Committee on Vital and Health Statistics was convened on Monday, July 21 in the Hubert H. Humphrey Building in Washington, D.C. The meeting, which was chaired by Dr. Lisa Iezzoni, was open to the public. Present:
Subcommittee members
Lisa Iezzoni, M.D., M.S., Chair
Hortensia Amaro, Ph.D.
Richard Harding, M.D.
Vincent Mor, Ph.D.
M. Elizabeth Ward
George Van Amburg
Staff
Carolyn Rimes
Olivia Carter-Pokras, Ph.D.
Others
Jennifer Madans, National Center for Health Statistics (NCHS)
Robert Gellman, J.D.
Rachel Block, Health Care Financing Administration (HCFA)
Stanley Nachimson, HCFA
James Hadley, HCFA
William Clark, HCFA
Lisa Herz, The Medstat Group
Shruti Rajan, Urban Institute
Edward Sondik, NCHS
Kathleen McGugan, Ernst and Young
Lynnette Araki, NCHS
Lauren Craig, NCQA
Rachel Block, HCFA
Beth Wing
Sanjaz Pandezz, University of Maryland at Baltimore College (UMBC)
Lisa Herz, The Medstat Group
Nicholas Meyers, American Psychiatric Association
Michael Tate, American Dental Association
Noemi Villafranca, HCFA
James Lubitz, HCFA
Michelle Hertsch, HCFA
Kathe Fox, The Medstat Group
Mike Barry, Public Health Foundation
Mary Jo Braid, UMBC
Anna Fallieras, The Lewin Group
Ruth Perot, Summit Health Coalition
Mike Nolin, UMBC
Brad Shiverick, AHCA
Elie Smith, AHCA
Violet Wu, Health Policy and Strategy Assistance
Jane Harman, NCHS
Leslie Cooper, National Institute on Drug Abuse
Audrey Burwell, NCHS
In preparation for its year-long project on Medicaid managed care, the Subcommittee heard presentations from HCFA staffers and Medstat and Urban Institute researchers about the connections among Medicaid, managed care, and dual eligibility. Members agreed on a Subcommittee work plan, including the initial stages of the Medicaid managed care project. They discussed the OMB directive on race and ethnicity and asked Drs. Amaro and Carter- Pokras to develop comments reflecting their discussion. After it is approved by the Subcommittee, the document will be forwarded to the full Committee for approval and then sent to OMB and the Data Council. Finally, the Subcommittee discussed, with Dr. Edward Sondik and Dr. Jennifer Madans of NCHS, ways of improving data on the health status of racial/ethnic groups.
· January 12-13
· February 9-10
· April 21-22 (Boston site visit)
· July 14-15 (wrap-up meeting for Medicaid managed care project)
6. Dr. Amaro and Dr. Carter-Pokras were authorized to develop comments on the OMB directive on race and ethnicity, based on today's discussion. The draft will be circulated among Subcommittee members, and once approved, will be sent to all NCVHS members for action. The final document will be sent both to OMB and to the co-chairs of the Data Council.
7. The Subcommittee accepted Dr. Sondik's invitation to advise the Center on race/ethnicity health data on an ongoing basis. He promised to keep the Subcommittee apprised of developments in this area, including the Center's efforts with the Indian Health Service. The Subcommittee plans to develop its work plan in this area once it has new members who can help carry it out.
Following introductions, Dr. Iezzoni thanked Ms. Rimes for organizing this meeting, which begins the Subcommittee's year-long study of the data issues involved in monitoring the impact of Medicaid managed care. She welcomed the invited speakers.
Ms. Block, who directs the Data and Systems Group, began with a broad picture of trends and issues. She noted the significant increase in the number of Medicaid beneficiaries enrolled in managed care (up from 1-2 million before 1992 to 13.3 million in 1996) and the corresponding increase in the number of state waiver applications. More than half of Medicaid beneficiaries in 1996 were enrolled in an HMO or other form of comprehensive risk arrangement. States take different approaches to administering and locating authority for these programs. A few states represent the bulk of enrolled Medicaid managed care populations. Details are available on HCFA's Web site (WWW.HCFA.GOV).
Ms. Block highlighted three issues. First, Medicaid beneficiaries are involved in several types of managed care arrangements, and each state's Medicaid managed care program is unique. Second, the characteristics that make people eligible for these programs (e.g., income, family characteristics, disability status) tend to involve them in other programs and funding streams which also deserve consideration. States are in a unique position to organize their data to provide this broader picture -- for example, as Maryland does in its HIV Information System. Finally, states are starting to be interested in bringing managed care models to their long- term care systems. Two issues are paramount in the long-term care context, which Ms. Block called the cutting edge in respect to Medicaid managed care. One concerns dual eligibles, notably the constraints imposed on states by the different rules in the two programs, and the development, despite constraints, of new and interesting models. The other issue stems from the fact that long-term care involves different services and providers and different kinds of information, thus suggesting the need for new ways of evaluating the impact of managed care for that population. Much attention will now focus on these new models, but it will also be important to continue research and evaluation efforts focusing on Medicaid's core population: women and children.
Dr. Nachimson described the basic Medicaid data and state systems. Most Medicaid data come from claim and eligibility records. HCFA has not yet mandated the collection of encounter data by state Medicaid programs, although 1115 waiver states collect some. HCFA does require every program to do an annual summary data report, the 2082 -- a point-in-time enrollment estimate reported on hard copy. Summarized data are published annually and are available on HCFA's Web site. The 2082 has not been changed to accommodate the shift to managed care. There will be some changes for FY 1988, but still no service information for people in prepaid managed care.
States do not have to complete the 2082 if they participate in the Medicaid Statistical Information System (MSIS), which collects detailed person-level data on eligibility, utilization and payments. 30 now do so. The system can receive encounter data through dummy (shadow) claims. A spinoff of the MSIS system is the State Medicaid Research Files (SMRF), which have been made more useful for researchers. HCFA only gets encounter data from a few states. They are trying to figure out how best to collect and compile the information, with the aim of making plan comparisons. HCFA is also working on revising the requirements for managed care to include data requirements, including encounter data.
Asked if she thought national evaluation were possible given the variations among states, Ms. Block said she sees a combination of national and state evaluations as the ideal, using widely accepted national health status goals and standardized instruments for the former. HCFA should work with states to set their own goals, which should be complementary to the national ones.
Asked for suggestions for the Subcommittee's forthcoming project, she recommended doing case studies of two states. The Subcommittee could help by identifying basic characteristics and issues, suggesting ways for states to manage their information systems, and specifically identifying approaches to encounter data.
Mr. Van Amburg asked about carve-outs, and Ms. Block said this issue should be added to the major ones she identified in her initial remarks. One aspect is whether carve-outs are done through additional capitated arrangements or remain fee-for-service. She cited family planning as a public health focus deserving of special attention.
Asked about the possibility of coding the characteristics of various state approaches, she said HCFA's program information at present is in narrative form. However, other sources such as the Urban Institute have additional information on this.
Dr. Harding asked how it can be assured that the information reported by behavioral health carve-out programs represents reality. Ms. Block noted that behavioral health is a new area for HCFA; there is a need either to hold states accountable to broadly accepted standards, or to use ones generated by the community. The latter (local standards) are more likely for mental health, at least in the short term. She affirmed the need as well for independent external review of program outcome reporting, and she expressed hope that the nation's accrediting bodies will address the vacuum in this area. She encouraged the Subcommittee to include this concern in its final report. A broad issue here is the need for the appropriate purchaser and regulatory oversight of the data system and services being purchased, and clarification of the purchaser's role. States vary greatly in this regard, and recommendations in this area would be timely. Dr. Iezzoni noted that this problem suggests the need for auditable data, which in turn gets into complex privacy and confidentiality issues.
Dr. Mor returned to Ms. Block's point about integrating health data with those from other social programs, and he asked for examples of states that are doing this. The speakers cited South Carolina, Illinois, and Maryland. RWJ Foundation is funding this. ASPE has also put together a registry of state-level efforts to integrate data, and the registry is on their WEB site.
MEDSTAT has worked with state and federal agencies on Medicaid and Medicaid managed care, and it also builds decision support systems for state Medicaid programs. Dr. Herz focused on the State Health Care Reform Monitoring Project, a three-year HSFA sponsored project that provides technical assistance to waiver states on using encounter data for program monitoring and evaluation. The TA covers both the uses of encounter data and technical issues in collection and measurement. With respect to performance measures for special populations, she called attention to the AHCPR's CONQUEST database as a comprehensive system that might be useful to the Committee.
States are struggling with how to use the encounter data they are getting and how to integrate them with fee-for-service experience. States sometimes do not know what encounter data to collect or how to process them. Complex system adjustments are needed, for which states have limited resources. Another problem is the delay in receiving encounter data from plans, which is partly the result of inadequate plan RFPs and contracts. Dr. Herz noted that George Washington University has just released a comprehensive report on contractual issues. Contracts may not include adequate data requirements, or the requirements may not be enforced.
From the perspective of providers and plans, she noted that information systems have limitations, and the lack of standardization is a huge burden on plans. Plans may have trouble reporting to the states because of multiple systems set up for administrative purposes. For their part, providers may have been promised that with managed care there would be less of a paperwork burden, and they are having a hard time adjusting to the fact that this is not the case. Moreover, incentives to submit encounters are not great in the absence of a link to payment. A major issue concerns carve-outs, with states sometimes asking that special populations be allowed to use traditional providers. These providers may create MCOs in order to provide services, but may have limited program management experience. Eligibility issues are another challenge for providers. Finally, the leverage over providers with respect to data requirements varies by plan type, with little leverage possible over providers who are not in exclusive contracts.
Dr. Herz said that while state and plan issues would not disappear with national data standards, such standards would help reduce the burden and improve the quality and completeness of data. She noted that many states have made considerable progress using encounter data -- for example, Minnesota (where plan CEOs were responsive to an appeal from the Medicaid Director) and Tennessee (which is conducting detailed analyses and has a good Web site).
Dr. Herz encouraged the Subcommittee to articulate the importance of encounter data to monitor care for special populations. Data standards will be important in this context. She recommended adding a family identifier to the core data elements, to match children to their parents and facilitate program evaluation. Conversations with AFDC and TANF administrators could be useful on this subject. AFDC had case identifiers that linked children and mothers. Detailed race/ethnicity specifications also are important for monitoring health status and care outcomes. Finally, states need help on standard methods for assessing and ensuring encounter data quality and the creation of a national repository for performance benchmarks.
Dr. Iezzoni noted that the Subcommittee will want copies of all the reports mentioned by the presenters (e.g., the GW study). Ms. Greenberg said GW is working with CDC on contracting specifications for achieving public health and prevention goals, with data requirements being one facet.
Asked about the idea of a national repository for performance benchmarks, Dr. Herz said no one is working on this at present. States and plans could benefit from knowing the range of experience and how to focus their efforts to achieve better performance.
This office, formerly the Office of Research and Demonstration, is responsible for evaluation, including of the 1115 waiver programs. The task is now contracted to large research firms (Mathematica, Urban Institute and Systemetrics). Two evaluations have looked at five states each, and the Oregon demonstration was evaluated separately. Each five-state evaluation costs about $6 million each, over five years. HCFA is the primary sponsor, with significant funding from ASPE and SAMHSA. In a mental health context, longitudinal focus groups are being used to look at the use of both covered and uncovered services. All evaluations are focusing on the impacts of the expansion of eligibility and the increased use of managed care. Case study interviews are being done with each state. A report has been completed on the first set of interviews for three states.
Encounter data, which waiver states are required to submit, are the "guts" of these evaluations. States generally collect more extensive data than they are required to report. However, Dr. Hadley said he did not "feel great about encounter data" because HCFA is getting "a trickle" three years into the five-year evaluation. Some adjustments are necessary. States vary in terms of data flow. HCFA is studying the possibility of contracts to work with states on encounter databases for future evaluations. The major focus is getting a good core set of encounter data. The means and focus of evaluation are also being reconsidered. One problem is that state data are not comparable.
In response to a question, Dr. Hadley said Mathematica would be willing to talk with the Subcommittee about the data systems they are evaluating. He suggested contacting Sue Dodds, but cautioned that it is hard to generalize about the states' data systems. He indicated that States are often running the data collection and verification activities on a shoestring. After further questioning about terminology, it was established that these are not "demonstrations" in the usual sense. Dr. Iezzoni said the Subcommittee would not use that term in reference to them. Dr. Hadley pointed out that non-demonstration (non-waiver) states have no data requirements. To a question, he said there is no special funding for the demonstration states to do evaluations, although they had to develop new data systems and practices.
To other questions, he said access to care is being measured primarily through beneficiary surveys. Oversampling of at-risk populations varies from state to state. The encounter data being provided are more extensive than the McData minimum data set.
Dr. Hadley said the Subcommittee should receive reports from this project as they come out. Dr. Iezzoni said they also want to hear from the investigators.
Dr. Clark is with the Division of Health Systems Research in the Office of Strategic Planning. He is also project officer for several states requesting waivers to test integrated approaches to managed care for people who are dually entitled.
There are two comprehensive models of care for such people -- ONLOC and the Minnesota Senior Health Options Program, together serving at most 3,000 people. (There are some 6 million dual enrollees in the U.S.) ONLOC combines Medicare and Medicaid funding and gives plans freedom to decide what services to provide. They focus on adult day care as the
core service site, using a multi disciplinary team that meets weekly. Housing is part of the overall concept. The model is sometimes called "life care for the poor."
Dr. Clark characterized a heavily institutionalized scenario in which an "entrenched bureaucratic tangle" cannot be undone until statutes are changed. Waivers cannot solve the problem, in his view. To this is often added beneficiary confusion over coverage and provider breakdowns. He contrasted this with a picture of emerging new approaches to serving dually entitled beneficiaries in which Medicare and Medicaid are complementary and symbiotic. Questions arise around who "owns" the population, the state or the federal government.
The major policy issues under discussion in this context are continuity of care, beneficiary choice, and administrative oversight. Financing is another issue, along with the question of how HIPAA data requirements will impact on the design of systems in this area. Dr. Clark asserted that waiver planning should take into account any anticipated standardized encounter data reporting.
Regarding HCFA's data activities with the states, he said HCFA has merged Medicare and Medicaid claims for 12 states and is working toward assessing each data set and the extent to which they can be merged. He said data sets such as the OASIS and the MDS, among others, also should be considered for integration, along with mortality data, other public health data, and studies such as the Framingham Study.
Finally, Dr. Clark noted the "frightening" experience of the 1115 waivers and the problems with inconsistent encounter data. With the dual eligible population, the challenge is to design the programs correctly at the outset.
Ms. Rajan said "the duals" are disproportionately non-white, female, under age 65, and with disabilities. The lack of coordination between the federal programs serving these individuals is a problem. She presented the results of her Institute's HCFA-funded research, which focuses on home health care for duals and the interactions between Medicare and Medicaid programs in this context. Home care is one of the few service areas where the two programs overlap.
In phase 1, which will be presented today, the researchers looked at spending patterns across states and did case studies in five states. Phase 2 involved a multi variate analysis using the Medicare CBS. Ms. Rajan noted the importance of this topic due to the vulnerability and diversity of the population, their disproportionate use of the Medicare home health benefit, the dearth of research on the subject, and the growing interest in innovative programs in this area. The Medicare home health benefit provides post-acute care, while the Medicaid home care program is oriented to chronic care and involves a home health benefit, personal care, and waiver programs. The research found that spending for home health care by both Medicare and Medicaid increased significantly between 1988 and 1993, indicating no overall trade-off between
the programs. There is wide variety among the states in the amount of spending. For most individual states, the pattern does suggest a tradeoff between the Medicare and Medicaid programs.
Regarding the data gaps in this area, Ms. Rajan stressed the need for uniform individual- level Medicaid expenditure and utilization information. Another gap is detailed information from every state on the program dimensions that affect the interactions between the programs.
Ms. Rajan observed that states have very little experience with Medicaid managed care for the population in question, and plans are reluctant to capitate long-term care.
Returning to the point about integrating HIPAA requirements, Dr. Clark said HCFA has not progressed to the point of addressing this issue in its reviews.
Dr. Iezzoni thanked all the speakers for their contributions.
Members reviewed revisions made to the overall Subcommittee work plan and passed a motion approving it. It will be presented at the September meeting of the full Committee.
The Subcommittee developed a conceptual framework and list of priority activities on the basis of which staff will develop a scope of work for a contractor. They also agreed to meet with HCFA staff and potential contractors to discuss what questions states are asking, what data systems offer answers, and where the gaps are.
Dr. Iezzoni began the discussion by explaining the camera metaphor, which shows the relationships among different levels and categories of questions about the health and health care of the American people. She then used this schema to ask Subcommittee members a series of questions on the Medicaid managed care project. Their answers are summarized below.
1. What do we want to know about in respect to Medicaid managed care?
· whether people in managed care are getting equivalent, better, or worse care than those in fee-for service
· status of carve-out services
· provision of dependent services
· impacts on access, quality, cost, health status, and prevention
· impact on providers -- health plans and direct patient services
· impact on minority providers
· impact of disappointed enrollee expectations created by marketing
· benefits and costs of managed care
· integration and continuity of care
· macro-level economic questions
· impact on public health and related social programs
2. What populations are we concerned about?
· infants, children and mothers
· adolescents and young adults
· racial and ethnic groups
· providers
· persons with substance abuse and mental health problems
· migrant populations
· homeless people
· rural residents, both patients and providers
· the long-term care population, including people with disabilities and the elderly
· people with chronic health conditions
· the medically fragile
· AIDS/HIV patients
· civil rights monitors
3. What data systems should be a focus of concern?
Here, there were references to consumer surveys and possible uses of encounter data, questions of how to gather data on "non-traditional" providers, and discussion of how to use case studies to get at the impact on public health. The Committee's core data elements were cited as a good frame of reference. Members agreed on the need to also use data sources from other social support services, at all levels of government.
This general topic prompted the observation that the federal government does not seem to know what questions to ask, and/or where to look for answers. Several members supported the assertion that the need to report massive encounter data has not yet been sufficiently demonstrated, and that the benefits must be shown and questions clarified to justify the risks and effort of generating the data. Mr. Van Amburg offered a "radical" challenge to explain why it is necessary to amass an extensive encounter data system. Ms. Greenberg noted that there is ambivalence within HCFA about this issue.
The group discussed this question and alternative approaches at some length. It was asserted that the medical record, not encounter data, is the source of analysis, and furthermore that information systems will obviate the need for centralized data storage. Members envisioned a scenario in which providers collect the data they need in order to care for patients and evaluate the results, with the data residing with providers and obtained as needed for special studies rather than routinely reported. Surveillance and reporting would be done as needed, for example on prevention-related services.
It was noted that the morning's presentations indicate that despite these realities, the federal government is thinking in terms of a massive encounter data system handled by it and the states.
Using the principles outlined in the foregoing discussion as the framework, the group then considered how to focus the scope of work for the contractor. The final product is to be a report on key questions and issues, data sources, and data gaps that concludes with recommendations about how to fill critical gaps. Members agreed on a general approach that does not assume the warehousing of data, but rather matches questions to benefits and evaluates where it makes the most sense to house data.
After reviewing what two states they might want to study and visit, with Massachusetts and Arizona as leading candidates, they agreed to meet on September 29-30 with representatives of HCFA and possible contractors to further hone their thinking. Members outlined the following questions as guides to the initial discussions:
· What questions do individual states have?
· What data elements are needed to answer them?
· What data systems, if any, exist as sources?
Decisions about next steps will be made following the September discussions. The Subcommittee deferred the territory issue that was originally planned for discussion at the September 29-30 meeting.
Members agreed to use their breakout session during the September 8-9 NCVHS meeting to look at how the Census Bureau, the Social Security Administration and other programs are defining "disability."
The following 1998 dates were reserved for Subcommittee activities, including meetings and site visits in Boston and Arizona:
· January 12-13
· February 9-10
· April 21-22 (Boston site visit)
· July 14-15 (wrap-up meeting for Medicaid managed care project)
Dr. Amaro led a discussion of the recently released OMB Directive on Race and Ethnicity, on the basis of which she and Dr. Carter-Pokras will draft Subcommittee comments. Generally, members were positive about the recommendations and the underlying analysis.
Dr. Iezzoni expressed concern about the Middle Eastern category and suggested proposing further research on that category.
Dr. Carter-Pokras said another area that needs further research is the status of indigenous people born in South and Central America. They are affected by the ruling of the OMB Director that only original peoples in North America may use the American Indian designation.
The group discussed the ramifications of the proposed changes, which they generally viewed as reasonable. It was noted that the Subcommittee should recommend that the effect of the "mark all that apply" change be evaluated in a health context.
They then agreed on a process for drafting and reviewing the comments, which will be submitted to the full Committee at its September 8 meeting -- also the final day of the comment period. Dr. Iezzoni asked that staff check Committee records to be sure the recommendations are consistent with earlier ones by Dr. David Williams and the Subcommittee on Minority and Other Special Populations.
Members discussed the importance of structuring the changes to enhance the ability to collect data on subpopulations and do so in a way that can be collapsed back into the basic OMB categories.
Dr. Iezzoni welcomed Dr. Ed Sondik and Jennifer Madans. Dr. Amaro described recent discussions about how to improve the type and amount of data on racial and ethnic groups, especially those for whom traditional sampling approaches do not work well. This concern, which has been articulated but not acted on for many years, has become a priority for Dr. Sondik.
Dr. Sondik said the issue arises on a regular basis. The Center must find a way, with its limited resources, to get better information on these populations. He described a few things being considered. One is to use a portion of the NHANES every year to address the problems of a specific population, starting for a year or two with New York Puerto Ricans. Another, called the HANES RV, involves taking a smaller vehicle to more remote areas to gather data on special populations. In general, Dr. Sondik asked the Subcommittee to consider advising the Department and the Center on a strategy for this issue.
Dr. Carter-Pokras reported that a policy is under consideration to require all data Department data systems to collect racial/ethnic data or provide a good rationale for not doing so. Currently, informal evidence suggests that roughly one third of Department data systems do not use the OMB categories.
Members responded positively to Dr. Sondik's general objectives and the specific ideas he mentioned. He said the Center is considering federal/state partnerships on some of these activities, such as an enhanced telephone survey and an expanded HIS. One idea is to give states the option of adding on questions and possibly changing the sampling strategy. He added that survey integration makes it possible to use questions from NHIS, SLAITS, NHANES and MEPS; and to employ statistical matching to SIP and CPS.
Dr. Amaro called these ideas a good compromise between the too-expensive and infeasible ideal of oversampling subpopulations on a regular basis, on the one hand, and doing nothing, on the other. Dr. Madans said the plan is to embed all these special efforts within the national data collection to ensure comparability with national data.
Dr. Sondik described the value he found in the Healthy People Hispanic review, which by its nature has a broad purview and which brought to bear the perspectives and expertise of a wide range of people. Moreover, the focus was on the health needs of the population, not on the mechanism for collecting information. This, he said, should be a model for other efforts. In addition, the country should get a sense of what is known and needs to be known about every group, rather than waiting for a crisis to mobilize activity. He noted that the subgroups that need study are not only racial and ethnic but also socioeconomic, among other variables.
Dr. Mor observed that the nature of geographically concentrated population subgroups calls into question the two-stage HIS sampling strategy. Similarly, managed care is evolving at different rates in different areas of the country. He asked whether the Census could provide health and disability information that could be cross walked to more detailed information. He also cited the American Community Survey as a possible source. Dr. Madans said that if the data sharing bill is passed, the American Community Survey could be used as a sampling frame for the HIS. However, it is not scheduled until around 2005. Dr. Sondik noted that the Department also is considering how to take health care surveys out of a facilities-based frame of reference.
In response to a question, Dr. Madans discussed efforts to design confidentiality protections so that contextual data can be made available to researchers.
Returning to Dr. Sondik's request for guidance from the Committee, Dr. Amaro asked what specific things would be helpful, and he asked them to consider studying the extent of the gaps and recommending how to address them. Both Dr. Iezzoni and Dr. Amaro assured Dr. Sondik that these issues are central to the Subcommittee's charge. Once new members are appointed, the Subcommittee will develop its work plan in this area. Dr. Sondik promised to keep them apprised of developments in this area, including relevant meetings.
Dr. Iezzoni thanked Drs. Sondik and Madans, and adjourned the meeting.
I hereby certify that, to the best of my knowledge, the
foregoing summary of minutes is accurate and complete.
/s/ Lisa I. Iezzoni November 7, 1997
Lisa I. Iezzoni, M.D., M.S., Chair Date