Washington, D.C.
The Subcommittee on Population-Specific Issues of the National Committee on Vital and Health Statistics (NCVHS) was convened on Monday and Tuesday, January 12-13 in the Hubert H. Humphrey Building in Washington, D.C. The meeting was open to the public. Present:
Lisa Iezzoni, M.D., Chair
Hortensia Amaro, Ph.D.
Richard
Harding, M.D.
Vincent Mor, Ph.D.
George H. Van Amburg
M.
Elizabeth Ward
Carolyn M. Rimes
Olivia Carter-Pokras, Ph.D.
Dale Hitchcock
Ronald Manderscheid, Ph.D.
Kathryn Coltin, M.P.H., NCVHS Member
Robert Gellman, J.D., NCVHS
Member
Lynnette Araki, National Center for Health Statistics (NCHS)
Gail Janes, Ph.D., National Center for Chronic Disease Prevention and
Health Promotion
Marjorie Greenberg, NCVHS
James Scanlon, NCVHS
Barbara Hetzler, NCHS
Lorin Ranbom, Ohio Department of Human
Services
Angela Gonzalez Willis, Center for Mental Health Services
Michelle Lombardo, Health Care Financing Administration (HCFA)
Stanley
Nachimson, HCFA
Rhoda Abrams, Health Resources and Services
Administration (HRSA)
Cheryl Fish-Parcham, M.S.W., Families U.S.A.
Representative Lee Greenfield, Minnesota Legislature
Gretchen
Adeson, National Association of Children's Hospitals
Robert Brewer,
M.D., Nebraska Public Health Assessment Division
Robert Griss, Center
on Disability and Health
Jennifer Fiedelholtz, HRSA
Grace
Gorenflo, R.N., M.P.H., National Association of County and City Health
Officials
Janet Spinks, AQUAS, Inc.
Nancy Clarke, Oregon
Department of Human Services
Eileen Peterson, M.S.P.H., United
HealthCare
Joan Jacobs, Office of Minority Health
Joanne Lee,
HRSA
A. Michael Collins, University of Maryland at Baltimore
Donna
Melton, National Association of Children’s Hospitals and Related
Institutions
Dave Baldridge, National Indian Council on Aging
Michael
Millman, HRSA
Jason Goldwater, NCHS
Rachael Block, HCFA
Eric
Goplerud, Ph.D., Substance Abuse and Mental Health Services Administration
(SAMHSA)
Neva Kaye, National Academy for State Health Policy
Mary
Jo O’Brien, The Lewin Group
Harry Rosenberg, Ph.D., NCHS
Donald
Umlah, Ph.D., Health Partners of Arizona
James Weed, Ph.D., NCHS
Margaret Schmid, National Committee for Quality Assurance (NCQA)
The Subcommittee met for two days to hear from representatives of several stakeholder groups about what information they need on Medicaid managed care. The first panel represented federal agencies: Dr. Eric Goplerud from Substance Abuse and Mental Health Services Administration (SAMHSA), Rhoda Abrams from Health Resources and Services Administration (HRSA), and Dr. Gail Janes from the Centers for Disease Control and Prevention (CDC). Dr. Goplerud noted the difficulties caused by having different reporting requirements for different payors. The domains for which reporting is needed in the mental health/substance abuse area are access and availability of services, appropriateness and quality of care, and patient’s outcomes or results. Ms. Abrams stressed HRSA's concern for the uninsured, a group that over time includes past and future Medicaid enrollees. Large holes are opening in the safety net, and the traditional safety net delivery system is under stress.
Dr. Janes focused on CDC's interest in whether Medicaid recipients are receiving access to comprehensive care, including the full spectrum of preventive services, and whether data on the processes and outcomes of care are being collected, analyzed and benchmarked against the population at large. CDC programmatic dollars and technical expertise support data linkage initiatives and improvements in state information capacities. Like other presenters, she noted that there is no substitute for accurate, standardized individual level encounter and eligibility data collected from the provider. Concerns about the availability and quality of encounter data arose early in the discussion and continued as a major theme for every panel.
In the discussion period, Dr. Goplerud expressed strong concern about the devolution of responsibility to localities and its implications for data quality. There was discussion about possible uses of population-based data to supplement and link with encounter and enrollment data. It was noted that a barrier to data linkage is various organizations' difficulty in working together.
The second panel, on national perspectives, featured Neva Kaye of the National Academy for State Health Policy and Grace Gorenflo of the National Association of City and County Health Officials. Ms. Kaye cited a range of data sources that can supplement encounter data, such as complaints and grievances. She called attention to the problem of enrollment "churning," a recurring theme throughout the hearings. Ms. Gorenflo noted that the movement of Medicaid enrollees into managed care is resulting in a reduced role for local health departments as service providers. They continue to want information on how Medicaid managed care works and is working, however, including how well beneficiaries are being educated about managed care and whether appropriate outreach activities are continuing. She noted a potential role for local health departments in holding MCOs accountable. NACCHO's data collection efforts look primarily at health department structure and function.
One focus of the discussion period was alternatives to encounter data, which are needed because of the slow turnaround time for encounter data. The panelists were asked how the Subcommittee could move things forward. Ms. Kaye underscored the need for standardization, particularly of provider numbers, and for help in understanding enrollment churn. Ms. Gorenflo identified health outcomes as a major issue.
The next panel focused on legislative, consumer and advocacy issues. Speakers were Robert Griss, Director of the Center on Disability and Health in Washington, D.C.; Representative Lee Greenfield of the Minnesota legislature; and Cheryl Fish-Parcham of Families USA. Mr. Griss stressed the possibilities of using the Medicaid program as a vehicle for introducing public accountability into the larger health care system, and of applying the Medicaid experience to issues of quality assurance and coverage of uninsured populations. He also raised the possibility of creating a civil rights context for access to quality care drawing on the Americans with Disabilities Act. Mr. Greenfield is one of the seven Minnesota legislators who developed Minnesota Care. Based on that state's experience, he predicted that managed care will not save states money when used in rural areas. He noted the array of data that states need to assess Medicaid managed care, and observed that state agencies probably have to hire higher- salaried people to track and analyze data.
Ms. Fish-Parcham described a pilot project in Washington, D.C., with a group of Head Start parents and community organizations, aimed at monitoring Medicaid managed care in the District and helping policy makers educate consumers and involve them in the policy debates. The survey found problems similar to those in other states, notably concerning discontinuities of care. She offered several recommendations to the Subcommittee.
The state panel included three state health department representatives: Nancy Clarke of Oregon, Lorin Ranbom of Ohio, and Dr. Robert Brewer from Nebraska. Ms. Clarke said that the overarching questions Oregon seeks to answer through its data are whether Medicaid beneficiaries and other low-income people are getting the care and preventive services they need, and if this care is comparable to commercially insured managed care. Other questions for Oregon are how managed care can be used as a lever to improve the provision of preventive care to all Oregonians, and how managed care organizations can do population-based practice and to manage for quality intervention, rather than simply for cost. She expressed distress about recent BRFS cuts. She joined several other panelists in calling attention to staffing issues: Medicaid agency staff are generally not trained in epidemiology, and public health analysts do not know the dynamics of health plans. She recommended retraining existing analysts and helping them acquire these skills rather than competing in the market for talents that are in high demand.
Mr. Ranbom described a range of data uses in Ohio, including geocoding. Ohio has had good results with encounter data, which it has started to use for case mix analysis. He (and other panelists) described the penalties and incentives their states use to get providers to submit timely encounter data. Addressing the staffing issue, he said that nine people outside the agency were hired to manage the data collection and analysis process, with new job descriptions and higher pay levels to attract people with clinical, service provision and managed care experience. They were then trained to do data analysis using extremely large data sets. This has been a highly successful program.
Dr. Robert Brewer, on assignment to Nebraska from CDC, reported that the state's health and human services system was formed in January 1997 by combining five agencies into three. This has facilitated collaboration between the former health department and the Medicaid program. Major issues in the state are access to providers, particularly in rural areas, and the capacity of local staff to collect and analyze data. He said it would be very valuable to have data standardization, focusing specifically on encounter data and taking into account the needs of multiple users. In discussing ways of inducing plans to comply with reporting requirements, he suggested providing technical assistance, as well as threatening penalties. Nebraska has used the former, positive approach in the early stages of Medicaid managed care.
In the discussion period, the group talked about the limitations of HEDIS data and the fact that they are still superior to other data sources. They also touched on race and ethnicity issues for states, expansions in children's health coverage, and general data quality issues.
The Subcommittee heard a presentation from Dr. Harry Rosenberg of the National Center for Health Statistics (NCHS) on a new population standard for age adjustment statistics. He recounted the history of this issue, the problems caused by the use of two standards for age adjustment, and recent steps to remedy this situation. He then reviewed the recommendations of a recent workshop: that the population standard be changed from 1940 to the projected population for the year 2000, and that the single standard be used by all agencies presenting official data. The second recommendation is that public health agencies and the Department implement the population standard beginning with the published data for the year 1999.
Dr. Rosenberg enumerated several ways the proposal will be moved through the Public Health Service, including a forum in Atlanta at the Centers for Disease Control and Prevention. The implementation committee specified the types of instructional materials and tutorials that will be developed to publicize the new standard. In response to a question, he cited aspects of the likely impact of the adjustment on Healthy People 2000 data. He welcomed a suggestion to include the Social Security Administration in the review and comment procedures.
The Subcommittee agreed to write the Data Council in support of these changes, to reinforce the effort to get buy-in for the recommendations.
Next, Dr. James Weed of NCHS reported on the possible effects of the new OMB race/ethnicity standard on vital statistics. He stressed the difficulties in collecting multi-race information in birth and death records, due to the place and manner in which they are collected; he predicted that vital statistics may lose their continuity with the past as a result. Dr. Weed predicted a good deal of public interest once racial and ethnic minority groups realize the implications of the new standard for population data. NCHS is working on bridging studies for at least the next five to eight years.
Dr. Iezzoni noted the need for "a real back and forth" about how the federal and state systems are going to work out the problems. She invited both speakers to come back in a few months, possibly for a breakout session at the June NCVHS meeting.
Returning to the major topic of this meeting, the Subcommittee heard from another panel on national perspectives. It featured Rachel Block of Health Care Financing Administration (HCFA), Mary Jo O'Brien of the Lewin Group, and Margaret Schmid of National Committee for Quality Assurance (NCQA). Ms. Block began by stressing the uniqueness of each state program and the fact that only one or two offer comprehensive managed care. She joined others in calling for more standardization, and said that all state Medicaid systems will be in the same database beginning in 1999. Work is continuing on a report on key performance indicators, which will include selected measures of program performance covering enrollment, utilization quality, financing characteristics, and satisfaction. She noted particular concerns about the quality of race and ethnicity data and of diagnostic data. She called attention to a Philadelphia project exploring ways for health plans to assess birth outcomes at the community level, and testing an enhanced encounter form for that purpose. She suggested that the Subcommittee consider what tools other than HEDIS would be useful for evaluating managed care. Reiterating an earlier point that HEDIS does not provide the kind of timely, early warning data needed, she suggested that the consensus process that was used to come up with new specifications be replicated to solve that problem.
Ms. O'Brien reported on a recently-completed project by The Lewin Group for DHHS, looking at local, state and federal public health agency data capacities, at what data are and should be collected, and at how they are and should be collected. The project combined national studies, local case studies, and development of a research agenda. She also discussed some of her experience in the Minnesota public health sector.
Ms. Schmid then reported on NCQA's Medicaid HEDIS project. She anticipated that the Consumer Assessment of Health Plans Study (CAHPS) will be incorporated into future versions of HEDIS for the Medicaid population. NCQA has developed a more comprehensive database of HEDIS statistics on the Medicaid population which should be in place for the 1997 reporting year, with data available to participating states by the end of 1998. In addition, NCQA has been working to develop a method for applying HEDIS measures to Medicaid beneficiaries receiving care in fee-for-service and primary care case management programs. NCQA and the American Public Welfare Association (APWA) are seeking to increase the number of states voluntarily using HEDIS to measure performance, to overcome the problem of different reporting requirements in every state.
Discussion with Ms. Block touched on variations in state reporting requirements, the anticipated impact of the Balanced Budget Amendment, and the Quality Improvement System for Managed Care. With Ms. O'Brien, the Subcommittee discussed the possibility of sharing responsibility between managed care plans and the public health sector. In addition, Subcommittee members expressed interest in the validity of complaints and grievances as a quality indicator, and in ways of addressing the problem of churning.
The group discussed the problem of tracking differential access among racial and ethnic groups. HCFA's policy is that this is the states' responsibility. HEDIS data are not differentiated by racial/ethnic group, but Ms. Coltin noted that this analysis is possible using HCFA enrollment data.
The two final panels were composed of provider representatives, beginning with Dr. Donald Umlah of Health Partners of Arizona and Kathryn Coltin of Harvard Pilgrim Health Care. Dr. Umlah expressed strong concern that managed care organizations, to further state policy, will restrict utilization in order to control costs, rather than emphasizing care and prevention. He urged the Subcommittee to do what it can to institute measures at the outset to monitor quality and access. He then described the kind of data Arizona collects and how they are used. He noted that membership data do not include race/ethnicity, which would be useful. The group returned to this topic in the later discussion. His final point was that the state only began to use encounter data for quality analysis in recent years, although they have been reported since 1982.
Ms. Coltin identified a wide array of questions to be answered with managed care data, permitting comparisons of such factors as utilization, non-medical services, and compliance rates between managed care and fee-for-service coverage and between Medicaid and non-Medicaid enrollees. One set of questions concerns access to specialty services and outcomes in this area; another relates to patient satisfaction. In all cases, the questions arise about whether the appropriate data are being collected. After discussing the data collected for internal management, she described data that Harvard Pilgrim Health Care reports.
She then discussed data they would like to have but do not currently collect. This includes socioeconomic status, race/ethnicity and functional status. She said the greatest problem is in the area of diagnosis and claims data. Ms. Coltin described her institution's efforts to compensate and correct for information deficiencies, including data validation studies and development of computer-based patient records.
In the discussion period, she joined Dr. Umlah in saying that while race/ethnicity information has been avoided for fear of its discriminatory implications, her institution would welcome a government mandate to collect it. Dr. Umlah described a project to coordinate primary and mental health care in a single location and the challenges of getting these sectors of the system to work together.
The final panel was composed of Eileen Peterson of United HealthCare, Dr. Michael Collins of the University of Maryland at Baltimore, and Dave Baldridge of the National Indian Council on Aging. Ms. Peterson focused on the difficulty of deriving appropriate performance measures for the Medicaid population. She noted that HCFA's move to develop guidelines to standardize EPSDT collection and map it to administrative data deserves applause, and may lead to helpful efficiencies. Finding ways to standardize risk assessments for the Medicaid population is another problematic area. She concluded with five recommendations, including using existing data whenever possible, providing technical assistance and developing analytic expertise, and demonstrating how the required data and analysis can be used to improve people's health.
Dr. Collins' Center for Health Program Development and Management has helped the State of Maryland implement its Medicaid managed care program. He said Maryland is "betting heavily" on encounter data, which it regards as essential to program operation. He acknowledged that the data are not yet very good, but asserted that they will only be improved through use. Maryland uses encounter data for risk adjustment, which it calls health-based payment. In reviewing the obstacles to quality data, Dr. Collins called attention to limitations in provider capacity and the multiple demands on providers. He also noted the important role of vendors, and encouraged the federal government to alert them well in advance of anticipated system needs.
Finally, Mr. Baldridge described the efforts of the National Indian Council on Aging (NICA) to strengthen the information infrastructure for Indian tribal health care, particularly for elders. He noted that Indian health systems are the smallest and least advantaged end users of state data, and they experience the most severe problems with getting quality information. They are also contending with a "backlog of neglect," while learning how to work with capitation. These are the challenges NICA is trying to help them address, partly through a data bureau created with funding from several sources. The Council hopes within a couple of years to supply every tribe with federal and state data sufficient to meet their needs and at no cost to them. Mr. Baldridge said that State governments seem not to share the federal government's commitment to Indian health, but he predicted that nevertheless some good models may emerge. In response to a question, he affirmed that capacity building is a major priority for the Indian health care system.
In the final discussion period, the group began by airing concerns about proprietary data. The panelists predicted that this will be less and less of a problem as the trend continues toward standardization. The group then discussed possible ways of collecting information on non- medical or enabling services. Finally, they looked at the small-numbers problem regarding Indians and other small populations.
Dr. Iezzoni thanked Carolyn Rimes and Jason Goldwater for arranging the program, which brought together people from around the country and was organized around a series of topical panels grouped logically to address the list of questions mailed to panelists prior to the meeting. Self-introductions were made.
Foremost among the questions addressed to panelists was, What do you want to know about Medicaid managed care? Different perspectives were provided from representatives of government agencies, private payors, and patient advocacy groups.
Eric Goplerud, Ph.D., Substance Abuse and Mental Health Services Administration
Dr. Goplerud highlighted the performance measures to be collected and the kinds of standards needed to hold Medicaid managed behavioral health care programs accountable. He stressed the importance of knowing the uses of data before information is collected. Data should flow from core questions that states or other entities consider most important as they enter the managed care business. Perhaps these are questions of access or equity, expense, and accountability. Whatever the impetus for implementing Medicaid managed care, goals must be set in order to know whether the need has been met. This is true from the perspective of mental health and substance abuse services, and it probably applies to Medicaid in general. For example, a goal for health care reform could be to constrain rising costs.
To the question of accountability in privatizing public services, managed care is a reasonable answer, but managed care is not likely to redress the historical problems in behavioral health, i.e., under-funding and the concerns about care and protection of special and vulnerable populations. While managed care appears to work well when resources can be integrated or pooled, it confronts the problem of protecting vulnerable populations; and certain classes or types of services have not done particularly well in a competitive marketplace. Managed care does not appear to be capable of dealing with these related issues.
Performance measurement should match the identified problem and the attempted solution. Sara Rosenbaum and her colleagues at George Washington University have been studying Medicaid managed care contracts in general and Medicaid managed behavioral health care contracts in particular. They are issuing a complete volume on the great number of existing behavioral health performance measures, many of which conflict with each other.
The Mental Health Statistics Improvement Program, under the leadership of Ron Manderscheid, has developed a consumer-oriented report card that is currently undergoing extensive testing. The directors of state and national associations of mental health, substance abuse and Medicaid have identified for SAMHSA and HCFA one core area that they all agree is a major deficit: how to hold the Medicaid managed behavioral health care programs accountable. These state directors asked HCFA and SAMHSA for funding (and received it) to develop consensus around a set of core performance measures that could be used to get consistent reporting across mental health and substance abuse services in Medicaid managed behavioral health care contracts. A copy of the report and the set of indicators generated by the study was provided to all participants on this panel, along with a copy of the draft final report of the American College of Mental Health Administrators, which was also involved in developing mental health report cards relating to managed care.
All these reports agree that it is very expensive to gear up for a different set of reporting requirements for every purchaser, or to have competing or overlapping requirements. These reports ask if it is possible to ascertain the essentials in any kind of managed care system. NCQA, responding to pressures from purchasers and consumers, has developed a behavioral measures advisory committee to report to the Committee on Performance Measures. All four of these reporting entities agree on three domains in which reporting is needed: (1) access or availability, (2) appropriateness or quality of care, and (3) outcomes or results.
Dr. Goplerud was asked about cultural accessibility, cultural appropriateness and cultural definitions of outcomes in behavioral health care. CMHS, in a contract with the New York State Research Institute, convened work groups to look at a set of cultural competency guidelines and to develop performance indicators from the guidelines. Although cultural issues apply to any chronic disease, they become starker in the behavioral health area because of historic underfunding and the stigma attached to mental and behavioral illness.
Referring to the IOM recommendations about process, outcome and capacity measures, the question was raised whether any similar suggestions came out of the CMHS work groups. Dr. Goplerud noted that medical necessity and utilization review have been ways people have been excluded from care. The question of whether people are actually getting services is the place to look, and that is more of an access and utilization question than a capacity question.
One of the issues of deep concern in behavioral health is the number of well publicized acquisitions and mergers over the last few years; some firms have been aggressive and assumed massive debt that will require them to minimize on costs in order to keep stock prices up and to service debts. Most of the savings will come from decreasing access to the comprehensive array of services.
Rhoda Abrams, Health Resources and Services Administration
Medicaid managed care affects the entire low-income population, and HRSA is equally concerned about patients both on and off Medicaid. Her presentation emphasized what is happening to the uninsured, because that issue affects the outcomes of Medicaid services. Because availability of services has a direct impact on access to care and outcomes, HRSA looks intensively at what is happening to the service delivery system. Unlike HCFA, HRSA has no encounter data for the Medicaid population per se, but has various data reporting systems for the grant programs. The mission of HRSA is to assure equitable access to quality care for underserved and vulnerable populations, which by definition include the Medicaid population. HRSA's data are usually on the populations that are served, Medicaid or not. The agency works with HCFA to see where it can make a match between their respective data systems.
She reviewed the agency's programs, which include the maternal and child health program, the Ryan White AIDS program and rural health. The delivery systems under Medicaid managed care are of particular concern to HRSA because they affect the outcomes of populations served under Medicaid. These are the safety net providers who provide substantial services to uninsured and otherwise disadvantaged populations. Public and other hospitals are included, as well as community health centers, rural clinics, health departments, maternal and child health and AIDS health providers.
One problem with Medicaid managed care is that the safety net is under stress. Availability of care for the uninsured and for some of the Medicaid populations is now changing. Public hospitals, health centers, and health departments that were accustomed to serving a significant Medicaid population, now must contend with reduced Medicaid revenues for managed care. Reimbursement rates are down, and reductions have been made in the number of individuals eligible for Medicaid. The pattern is fairly consistent around the country. Data show that more people are uninsured, and these uninsured seem to be concentrated with fewer and fewer providers. Public hospitals are bearing a greater burden. Another safety net provider, health centers funded by HRSA, serve populations that mirror the Medicaid population--a heavily minority population, essentially below 200 percent of poverty. Most of the safety net providers are necessarily participating in managed care. Participation in managed care while increases services for the uninsured population, is burdening the delivery system.
As a result of changes in the delivery system, some of the safety net providers are beginning to limit the range of services in order to accommodate reduced revenues. In efforts to save medical services, they are making changes in the enabling services such as transportation, translation, health examinations and support services that are not necessarily funded by a Medicaid managed care program.
The extent to which managed care companies may have expanded the range of services is not known. Some state programs, as part of their contracting process, have required the provision of some of these services. However, this is not an across-the-board requirement and there are no data to show to what extent these services are being provided or withdrawn.
In tracking changes in access to primary care services in a limited set of programs, there appears to be a discrepancy between access to primary care and access to specialty, ancillary services, hospital and enabling services. She stressed the importance of considering not only those eligible for Medicaid, but the uninsured in assessing access.
The HEDIS program is beginning to look at outcomes, but the program is not likely to be widely implemented. HEDIS is an expensive and difficult process, although its measures target primary care. HRSA has a major HEDIS training program for all the providers funded by HRSA.
Data sources are a major difficulty with measuring performance. The states do not uniformly collect encounter data, thus limiting the monitoring and analysis of outcomes. Claims data under managed care are also limited. HCFA has grappled with enrollment data as a major source of information. While HCFA requires encounter data under 1115 waivers, these waivers exist only in a few states.
Asked if the absence of encounter data is a bookkeeping issue or a proprietary information issue, Dr. Abrams replied that it is about how the managed care companies manage internally . The companies may be reluctant to submit information because it is proprietary; they do not collect it; or it is too expensive to collect.
Dr. Gail Janes, Ph.D., National Center for Chronic Disease Prevention and Health Promotion
Dr. Janes focused on two overarching issues that concern CDC and their state partners: whether Medicaid recipients are receiving access to quality, comprehensive care, including the full spectrum of preventive services; and whether the processes and outcomes of Medicaid-funded care are being collected, analyzed and benchmarked against the population at large to insure parity in both health care access and health status.
She summarized several concerns of CDC: the indirect effects of Medicaid managed care on health care for the indigent and uninsured; the need for population-based data systems that would allow tracking of health needs and status of this population; and the role of public health as an active partner in the development and analysis of evaluative data systems designed to inform the Medicaid purchasing process. She emphasized the importance of an equal partnership between public health and Medicaid in efforts to insure the health of the Medicaid population.
Regarding the second overarching question, she said that CDC and its partners in state public health organizations do not support data collection initiatives with the primary goal of explicitly collecting data about Medicaid managed care. However, a number of initiatives can provide insights about the processes and outcomes of care and the status of Medicaid managed care recipients.
Dr. Janes briefly described the survey-based data systems of CDC. BRFSS, which is administered in all 50 states, has been piloted in several states as a measure of utilization of preventive services in a Medicaid population. Use of Medicaid enrollees as a sampling frame appears useful for tracking the health status, health risk behaviors, and health care utilization of this population.
Another CDC data system, SLAITS, could also provide state level measures of health status and health care utilization and access for Medicaid populations. Additionally, the traditional categoric data systems of public health can provide crucial inputs to the Medicaid purchasing process and are equally critical assessors of the Medicaid product, since many describe the outcomes of care. However, at this time the national immunization survey can not be used for tracking immunization status of the Medicaid population due to problems capturing relevant insurance information.
Recently, CDC programmatic dollars and technical expertise support some state-based data linkage initiatives, such as the move to increased standardization of data formats, data sets and coding and classification schemes is an essential part of this process. CDC also sponsors a wide variety of one-time funding initiatives designed to support new and innovative data initiatives within the states. An ongoing Assessment Initiative is currently supporting data integration pilots in six states, several of which are exploring the usefulness of Medicaid and vital statistics data for support to Medicaid purchasing and performance measurement. The Medicaid checklist project, out of CDC's Office of Managed Care, is helping to develop purchasing specifications and contract language for use by states in their Medicaid purchasing contracts.
In order to better assess the health status and health care utilization of the Medicaid population, CDC's Center for Chronic Disease Prevention and Health Promotion is examining the efficacy and validity of a variety of telephone survey mechanisms for determining the health insurance coverage status of respondents. Results indicate that use of telephone surveys to distinguish between the managed care and fee-for-service population must be done with caution. An issue that deserves great emphasis is the structure of a question that will accurately determine whether a person is covered by a managed care or fee-for-service program, since it is extremely difficult but essential to get this information.
The CDC data systems, comprehensive as they are, need to be augmented with accurate, standardized individual level encounter and eligibility data collected from the provider. CDC is also striving to support state-level initiatives that will broaden the usefulness of existing data systems through standardization and linkage of registry and administrative data, to encourage collaboration between Medicaid and public health.
Numerous state-based organizations are participating in this process. Dr. Janes reiterated CDC's special concern about access to care for Medicaid populations, especially the need for access to good preventive services and information, and the need for public health to be an active partner in determining what services are going to be available. To this end, CDC has used its resources to support some of the innovative activities in the states, and emphasizes the usefulness of its many data collection initiatives.
Discussion
Responding to a question about whether Medicaid has collaborated with registries to capture health outcomes, Dr. Janes cited initiatives in states such as Massachusetts and Washington that indicate that linkages can be done. Significant barriers are the traditional and historical suspicions and discomfiture about working with non-traditional partners. Dr. Janes also cited the role of many foundations, such as Robert Wood Johnson, in supporting states’ efforts to find new uses for existing health data systems.
Dr. Goplerud mentioned the potential for cost shifting, not only between public mental health or substance abuse and Medicaid, but between Medicaid health and managed care carveouts and the criminal justice system. There is a three state pilot funded by the Center for Mental Health Services and Center for Substance Abuse Treatment to integrate databases between Medicaid, mental health and substance abuse. This would enable the agencies to track cost shifting. He added that since the HCFA 1500 and the UB 92 have only a single diagnostic field, it would be helpful for them to include a substance abuse question in a “yes or no” format.
In response to the question of responsibility for overseeing the managed care carveouts or the Medicaid managed care, Dr. Abrams said the state itself is legally responsible. The extent of oversight varies from state to state. She added that HCFA has a monitoring effort.
Dr. Goplerud indicated that SAMHSA was concerned with the devolution of responsibility and authority to the counties and local jurisdictions. Dr. Goplerud said, "We are concerned that some 3,000 counties are doing independent contracting for managed behavioral health care or managed care, and with extremely small staffs." Despite the states' responsibility there will be a lot of contract negotiation and monitoring responsibility at the county level. In the absence of uniform and consistent requirements for reporting, contracting and benefit requirements and without considerable sophistication in monitoring these systems, it is uncertain what will happen to vulnerable populations.
A question was asked about using epidemiological data to project rates of consumption and establish benchmarks, as well as describing other ways to monitor whether higher levels of intensity services are being provided because of medical necessity. Dr. Goplerud said that for Medicaid populations, going back and looking at historical patterns is really problematic. Information is beginning to come from researchers. There are also benchmarks within each state of what happened prior to management of Medicaid, and in system dynamics modeling in alcohol behavioral health.
Dr. Abrams added that managed care companies have their own ways of tracking their costs and services, providing another possible source. The difficulty is in the changing availability of basic input data; the states vary in their requirements and analyses.
Dr. Janes stressed the need for states to have sufficient support to utilize their own data systems for tracking and benchmarking.
Commenting on the point about Medicaid beneficiaries going on and off eligibility, Dr. Iezzoni asked for perspectives about the implications of this for SAMHSA and CDC. Dr. Goplerud noted frustrations where rapid cycling makes it extremely difficult to collect outcome measures. Dr. Janes also acknowledged the problem as critical for CDC. Looking at administrative data for Medicaid recipients leaves out the Medicaid population of tomorrow, i.e., the uninsured. It would be particularly useful if some of the state-based surveys were to give information on the needs of certain sub-populations to help CDC plan for what will be needed tomorrow. It would be useful to have the money to increase the sample size of those surveys in order to gain more information at the county level.
Another Medicaid issue was raised by Dr. Goplerud, i.e., the population with chronic needs. They are in Medicaid because they are SSI eligible; there is little turn over, and the costs are high and rising. It is extremely important to focus on functioning and outcomes for this population especially in the area of behavioral health.
Neva Kaye, National Academy for State Health Policy
Ms. Kaye's organization (NASHP) helps states improve their health care policies, primarily by sharing information among one another. On the question of why states collect data, she noted that the reasons cited most often are: determine payment, help consumers select a plan, plan and manage programs, and to do research and evaluation.
The benchmark for determining whether payment is adequate is usually based on fee-for- service, which requires certain data. It is very difficult to figure out what kinds of information people need to help them select a plan, but states are, nonetheless, still struggling with this problem, and some are making progress.
The most frequently cited example is the need for data that can be used in program and plan management. NASHP does a survey every two years, and she discussed some of the 1996 data. She then briefly discussed different data sources: complaints and grievances, enrollee and provider surveys, utilization reports, care coordination systems, and non-Medicaid records.
Most states have found these data on complaints and grievances extremely useful in determining a plan's performance. It isn't feasible to depend only on utilization reporting to determine potential problems because of the time lapse before problems are identified. States depend on complaints and grievances, and on focused and random medical chart reviews.
States use enrollee and provider surveys to gauge satisfaction and access. Arizona has used the enrollee surveys to find out why people dis-enroll from plans. Utilization reporting is probably the most interesting source. The number of states collecting utilization information has increased, and many collect both encounter and utilization aggregate data.
Ms. Kaye touched on problems states have collecting encounter data, mainly the lack of uniform coding and obtaining information from subcontractors. Care coordination systems are a potentially good source of information for the wide range of services covered that states could incorporate into their care coordination databases.
Many states have started working with vital statistics. These provide complete information not just about Medicaid, but about the uninsured, making possible comparison with the commercially insured populations as well.
Two issues come up frequently in the states trying to use Medicaid data: the "enrollment churn," and small numbers (i.e., enrolled in the program). Data validation is another issue. Several states have started doing medical record reviews to compare encounter data to the medical record. One project specific to dual eligibility tries to examine resource for programs with dual eligibility to find out how data collection is working using Medicare and Medicaid data.
The states have been working to get information and to learn how to use it, but it would be very helpful if provider numbers and other factors were standardized. Another area where help is needed is in the use of the data collected, for example, adjustment enrollment churn. Finally, benchmarks are needed to define good performance.
Grace Gorenflo, R.N., M.P.H., National Association of City and County Health Officials (NACCHO)
Ms. Gorenflo spoke within the context of local health departments, focusing first on data collected at NACCHO. There are 2,888 local health departments across the nation and "if you've seen one, you've seen one." She made the distinction between local health departments and community health centers. Health departments were established primarily to provide population- based services to protect and promote the health of everyone in a given jurisdiction. Approximately 40 million people each year receive some personal health service from a local health department.
Nationwide, seven percent of health department budgets come from Medicaid, a number that increases in proportion to the size of the health department. What appears to be happening with Medicaid managed care for the past two years is that health departments, like some health centers, are getting out of the business of providing care to Medicaid patients. Medicaid patients are frequently enrolled in managed care arrangements. But the health departments have a vested interest in whether or not care is being received in other entities.
Most of the questions NACCHO would like to have answered revolve around cost, quality and access. First, are managed care organizations doing a good job of educating patients about the system? The degree to which patients are enrolled and educated adequately is a serious issue for those automatically enrolled patients. On the question of quality and health outcomes, managed care organizations vary widely. Health departments could have a role in monitoring and ensuring accountability.
She reviewed a number of other questions, such as whether sufficient providers are available and whether there are enough of them willing to accept lower reimbursement rates; whether Medicaid patients are actually getting their care; and whether health departments and other entities are providing outreach services. Traditionally, outreach has not been a function of managed care organizations. Another question pertains to non-medical health care: preventive care; handling of medical records with patient turnover; disease reporting; and so on.
The proliferation of managed care offers an opportunity not only to look at selected providers and to assess whether or not they are meeting public health requirements, and develop incentives for them to do more and better reporting. Finally, the uninsured are a great concern in view of the crumbling safety net.
Ms. Gorenflo said NACCHO looks primarily at health department structure and function. Its bi-annual national profile of local health departments has an excellent response rate. For the first time this year questions about Medicaid managed care were asked, and analysis of that data will be available within a few months.
NACCHO would like to take a closer and more objective look at current trends for local health departments, specifically the effects of managed care not only on the Medicaid population and the uninsured, but on the general population, since they provide many invisible services that protect the entire community, such as assuring a safe water and food supply. Anticipating that health departments will serve fewer and fewer Medicaid patients, NACCHO would like to look more closely at the ability of local health departments to provide these population-based services, which may be comprised.
Discussion
A concern was raised about alternatives to encounter data, given its slow turnaround time--e.g., will the agency be solvent long enough to be able to provide the services?
Ms. Kaye said that states rely best on complaints and grievances, which could be the first indication of a problem. States also get financial information, but again, there is usually a lag time. The approach a number of Medicaid agencies have taken to the possibility of provider default is to provide for continuity of care for enrollees. Medicaid agencies have not found a good way to obtain this information, therefore, there is fairly standard contract language for requirements when a subcontract is canceled.
Ms. Gorenflo mentioned that local health departments don't have "a lot of teeth with their managed care organization." An additional issue is that some health departments have in effect a conflict of interest because they are trying to work with managed care organizations as service providers themselves. Most health departments do not provide mental health and substance abuse services.
Ms. Greenberg noted that if 38 states have risk contracts, all but two of them are getting encounter data. Yet Ms. Abrams said that once patients get into managed care, often there is no encounter data, and that while 1115 waivers offer encounter data, it is not actually being collected. There seems to be something missing. Ms. Kaye commented that the disconnect is the completeness and validity of the data. Most of the states are still trying to make sure their data are accurate and complete.
Dr. Iezzoni asked for perspectives on the use of surveys like the BRFSS for looking at Medicaid managed care. Ms. Kaye said that many of the surveys such as BRFSS are CDC's proof of principle projects, a fairly new initiative of CDC, which is now trying to get the work out to the states.
Dr. Brewer explained that the BRFSS recently published an article on morbidity-mortality weekly reports that compared the Medicaid population with commercially covered populations in Michigan, in collaboration with NCQA. There is a lot of interest now in how to incorporate more behavioral risk factor information in member satisfaction surveys, and comparison of this information with state information.
Mike Millman of HRSA commented that the community health center survey mentioned by Ms. Abrams started with a probability sample of delivery systems. Perhaps the best way to collect data about the relationship between outcomes and the inputs, structures, processes and characteristics of the delivery system is on a population sample. There may be some utility to going back and forth between the two modes so that they complement each other, so that, ultimately, the goal would be to link that information to administrative data systems.
Asked if the panelists have any specific advice for the National Committee to give to Secretary Chillily that would help improve the information available about Medicaid managed care, Ms. Kaye called attention to standardization, particularly of provider numbers, and asked for help dealing with enrollment churn. Ms. Gorenflo identified the issue of health outcomes as a point of focus.
Robert Griss, Director, Center on Disability and Health
Mr. Griss shared some of the dilemmas health policy researchers and advocates face. Medicaid is an important program for people with disabilities, but most people with disabilities are not on Medicaid. Still, the way in which managed care deals with people with disabilities is critical for the total disabled population, and an important litmus test for the quality of health care for the entire population. He emphasized the need to examine the experience of the Medicaid population and more importantly, to use the Medicaid program as a vehicle for introducing public accountability into the larger health care system.
He compared a Medicaid managed care pilot in a Midwest state and commercial plans in that same state, a project done anonymously. In that project he looked at an HMO, a staff group model HMO and an IPA, each of which answered 46 pages of questions and enabled a look at the decision making process.
Mr. Griss said there was surprisingly little focus on disability on the radar screen of the commercial managed care plans, whose capitation payment is far lower than that from Medicare. Still, it is important that the Medicaid program be seen as a model in many ways--e.g., as the way of providing an effective EPSDT benefit to children. It is imperative that Medicaid experience be applied to issues of quality assurance and coverage of uninsured populations. A surprising omission was a needs assessment process in plans at the point of enrollment, including for commercial enrollees. Also, although non-clinical interventions are most likely to improve health status, the plans presently lack incentives to provide them.
Mr. Griss asserted that it should be possible to get benchmarks for people with disabilities for specific conditions that would indicate the extent to which the observed outcomes differ from the expected ones. It is possible that changes in health status for all people with disabilities can be aggregated and that linkage between those outcome measures with payment policies could be made so that plans are rewarded for improvements in health status.
He raised the possibility of creating a civil rights context for access to quality care, noting that the Americans with Disabilities Act provides an opportunity to do so. Currently, plans are able not only to choose which services will be sold and the limits to the benefit package, but also to choose the definition of medical necessity, as well as the utilization review criteria, and the economic profiling criteria for retention of providers.
These are some of the ways the Medicaid experience with managed care can really benefit people with disabilities.
Representative Lee Greenfield, Minnesota Legislature
One of the seven Minnesota legislators who developed Minnesota Care, Mr. Greenfield is in his 20th year in the Minnesota House. Since 1987 he has chaired the Finance Division, which funds the Department of Health and Human Services and Medicaid. He briefly reviewed the state's use of prepaid managed care plans for its Medicaid population. Early on, the state concentrated on families with children, and now has only families with children in the prepaid plans. The HMOs are by law nonprofits, and all are state-based companies licensed and regulated by the legislature.
The state's experience in metropolitan areas realized a cost saving using managed care, but that saving did not apply as the program expanded to rural areas where it is not easy to get providers to accept Medicaid rates. As prepaid managed care expanded into rural areas, the cost of service increased significantly with the increased access to needed services. This raises a serious question about cost saving in the future in rural areas.
The foremost question of the members of the state House of Representatives about Medicaid and managed care is financial: how do these health care results compare to those in fee- for-service clients and those of populations covered by health insurance? The state created a public-private partnership to gather such data and to disseminate information and they are developing some surveys on satisfaction.
A significant question is what happens to populations with disabilities. Counties are now developing pilot projects to serve these populations. There is not yet convincing evidence that there is any savings with managed care, but there may be better coordination of services.
In order to defend the move to viable prepaid health plans, it is necessary to gain outcome data as well as data from encounter interviews and satisfaction surveys. There is a mechanism for collecting complaints throughout the system and for collecting data for HCFA. Some diagnostic data is collected, and an attempt to develop risk adjustment measures for the Medicaid population requires more diagnostic and other data. Satisfaction surveys conducted two years ago show that managed care plans for the general population did better than indemnity health plans in terms of satisfaction rates, particularly in Minnesota Care (a program for working poor people). This year satisfaction surveys will be done only for Medicaid to see how these findings compare with those collected in the past.
State agencies probably have to hire higher-salaried people to track and analyze data. It is not likely to save money on administration if the job is done right. One way to make data collection less costly is to require it as part of doing business in health care in general. If done only for Medicaid, it would appear to be very expensive to collect the kind of data needed.
This year the program performed studies focusing on quality of care studies, using three HEDIS effectiveness of care measures: childhood immunizations, prenatal care and breast cancer screening. In Minnesota, despite some foot dragging, the administration, insurers, the health department, and Medicaid now use uniform forms and definitions. This makes it easier to manage data.
One of the sticking points raised in the questions sent to panelists is confidentiality. Had it not been for a Minnesota statute requiring that data collected by any state agency, or any agent for the state, identify individuals with an encrypted number, the right to collect data would have been lost a couple of years ago. This may be true for other states, as well. Because of people's concerns about health record confidentiality, the state Department of Health has been asked to develop regulations regarding access and use of data.
Mr. Greenfield concluded that managed care, in some version, will dominate American health care in the future, and the experience of Minnesota and other states suggests that the only way to make it work is to monitor it for quality and access. This can't be done without data. Asked about data costs, he estimated that the data institute spends more than a million dollars a year.
Cheryl Fish-Parcham, M.S.W., Families U.S.A.
Families USA provides technical assistance on Medicaid managed care and other health issues to consumer organizations in a number of states. A pilot project in Washington, D.C., with a group of Head Start parents and community organizations monitors Medicaid managed care in the District. She described it as a research project to help policy makers in the District make decisions as to how managed care could best work and how consumers could be educated and involved in policy debates.
Ms. Fish-Parcham described the methodology, which recruited 12 parents who then surveyed their peers as to the problems they faced in the last year under Medicaid managed care. Although the focus was on the District, the problems are similar to those in other areas. The survey found that the discontinuity of care is a problem in managed care: 37 percent of the respondents had to change doctors upon enrolling in managed care and the District has one of the worst default assignment rates in the country. Other problems reported concerned the length of the wait to see a doctor, appointments for urgent sick care and return calls in an emergency. There were misunderstandings about what services were covered and confusion about how to handle problems with managed care plans. Since those findings were reported, the District has made efforts to correct the problems, although some issues still exist.
There are also limitations in the data that states are collecting Medicaid Managed Care, e.g., Maryland, Virginia, and the District. Limitations include: first, consumer satisfaction surveys fail because they are left to the plans to administer. Second, voluntary dis-enrollment rates are not consistently collected by states, and the states do not consistently separate voluntary disenrollment from other sort of disenrollment, such as loss of Medicaid eligibility. Third, states use different ways of defining automatic assignments.
States are contracting for external quality reviews. The scope of these reviews vary and the lack of consistency prevents them from being useful measures. There are also difficulties in interpreting complaint data across states because people don't know how or where to complain. Ms. Fish-Parcham said that the lack of complaints does not indicate a lack of problems. States have not come up with good ways to aggregate available data and find patterns. Access measures are also inconsistent. As Neva Kaye mentioned earlier, there are no benchmarks to help states determine whether they have a good or a bad result.
Ms. Fish-Parcham concluded with recommendations for further information about important questions of concern to Families USA:
§ Continuity of care, both during initial enrollment of Medicaid populations and ongoing
§ Typical voluntary dis-enrollment rates in states that compile rates of monthly, semi-annual and annual intervals
§ Standardized instruments for assessing consumer satisfaction with additional questions specific to state programs
§ Treatment outcomes for health problems typical among TANIF related beneficiaries
§ Best practice monitoring instruments for chronically ill people
§ Data about behavioral health that would assist states in making decisions
§ Aggregations of complaint data to reveal patterns
§ When does cost sharing reduce participation and access?
Discussion
In response to Robert Griss's comments on civil rights, Dr. Mor asked if anyone has used the civil rights and ADA argument for equal access under the law in relation to health care. Mr. Griss replied that in Oregon there has been a lot of sensitivity to these issues. Also, HCFA has used ADA in challenging Medicaid waivers for New York, and in connection with the capacity of plans to provide AIDS treatments. Dr. Iezzoni took issue with Mr. Griss about aggregating the experience of people with a full range of disabilities, because of the small number problem. He indicated that it is useful for getting a benchmark.
Dr. Mor inquired if, under the Minnesota law on encryption, record linkage is still possible. Mr. Greenfield noted an exception where people object to have their social security number used. If those objections are accommodated, it is not clear what the cost would be and what matching problems would be created.
Dr. Iezzoni asked for suggestions in dealing with interviewing linguistic minorities on functional performance and consumer satisfaction. Mr. Greenfield explained that the plans are required to have people who can handle language differences in their provision of service and for surveys. In response to a question from Dr. Iezzoni about performance evaluation based on patient reporting, Mr. Griss said that the focus should be on outcomes that are most meaningful to the consumer, not necessarily just the medical criteria often used in acute care.
Nancy Clarke, Oregon Department of Human Services
Ms. Clarke outlined the state's policy and administrative structures. Oregon passed the Oregon health plan in 1993, resulting in a conversion to managed care for 85 percent of the state's Medicaid population. It was phased in through a series of reforms that rolled in more and more people with access to the minimum benefit package. It involved several agencies in addition to the public health department.
Her department's concerns relate to whether people in Oregon have access to medical care--especially disadvantaged populations, including Medicaid, but also rural and ethnically diverse populations. Specifically, are they getting the care and preventive services they need, and is the care comparable to commercially insured managed care? Other questions being addressed are how managed care can be used as a lever to improve the provision of preventive care to all Oregonians, and how managed care organizations can be helped to learn to do population-based practice and how intervention can be assessed for quality, rather than simply for cost. Yet another concern is preventing the shift to managed care from creating a giant hole in the safety net for people in the state.
The data currently collected now is through population telephone surveys and addresses the issue of increasing the availability of insurance. Recent cuts from CDC regarding BRFSS are distressing because of its extensive use in tracking this kind of data. Now attempts are being made to understand the underinsured, and to collect data on satisfaction surveys, a new area of inquiry. Presently the department is moving toward a one-time administrative survey of all the state Medicaid plans and most of the commercial plan in order to have comparable state data. The CAPS survey will be used this year. The state has had very high response rates, particularly from people who had no insurance before and who are satisfied now to have something.
Another category of data is eligibility and enrollment, now being used to look at which populations roll or churn a lot and which do not. This will enable us to better understand their health problems and to use the data for improving continuity of care.
There is also an extensive collection of qualitative data from site reviews by Medicaid, which looks intensively at plans for quality, especially the coordination process for people with exceptional or special needs. Additionally, there are data initiated by plans; the department's challenge is to be a savvy user of that data.
Their biggest failure in data collection relates to encounter data. "We had the naive notion that we could take a 12-year-old MMIS system and just tweak a little bit and we would be able to answer managed care questions, but that idea was seriously wrong." Their second area of failure in data analysis was in failing to anticipate the impact on safety net clinics. Without baseline data, there can be no follow-up, and data analysts are left to rely primarily on anecdotal data.
As for assessing and integrating outcomes in the absence of encounter data, the department collected specifically targeted information, e.g., a phone survey of mammograms, which was mildly successful. Such follow-up surveys work where there is a good agreement in advance about what the measure is going to be. Increasingly, the department is deferring to standards, such as HEDIS or any other standard that has been tested by experts in developing and testing measures.
Dr. Clarke mentioned "Project Prevention!" as an example of the way the state's 15 plans work together on a single issue, immunizations, to build a statewide immunization registry which is nearly completed. The next cooperative project is on tobacco cessation. The plans are working in other areas and will report best practices back to the department, but so far they have not done a very good job of reporting data on those areas.
Another exciting project is the assessment initiative, under a new CDC grant to help fund joint positions in the health department and Medicaid agency to work on encounter data. A BRFSS will be taken of the Medicaid population, matching birth and death rates.
There are good plans for using encounter data for cost analysis, but few plans are really using it for population management. Medicaid agency staff are generally not trained in epidemiology and public health analysts don't know the dynamics of plans. There is a need to retrain existing analysts and help them acquire these skills rather than trying to compete in the market for these talents that are now in such high demand. Anything that can be done to help in these training efforts would be gratifying.
Lorin Ranbom, Ohio Department of Human Services
Mr. Ranbom noted the convergence of ideas between his prepared comments and those of previous speakers. The Ohio Medicaid program has almost 400,000 people in managed care plans, mostly a maternal and infant and child population, with contracts with 14 managed care plans in the state.
Typically, data collection includes a lot of information about enrollees, which is geocoded so that those data can be mapped and analyzed geographically. A primary care physician registry enables review of the capacity of managed care plans. Many of the plans have providers that are not Medicaid providers, which makes analysis difficult. Locations are mapped of all of the enrollees in one plan and coded by their distance from their primary care provider, making it possible to compare plans.
Mr. Ranbom summarized the kinds of information collected for Medicaid, and identified the kinds of data used to monitor the health plan. Information on voluntary dis-enrollment comes from consumer surveys. Results with encounter data are good. Performance measures are based on some of the encounter data. Clinical quality of care studies, using both encounter data and medical records, are done in tandem. From year to year changes are made depending upon the focal areas of the plans. Recently, they have started to combine case mix analysis with encounter data.
Many performance measures used have been benchmarked against the fee-for-service delivery system. Mr. Ranbom stated that they have found encounter data to be less costly to collect and have not had problems with the capacity of managed care plan administrative system. A lot of the managed care plans have had to make major system changes in order to comply with the department's requirements. For some measures encounter data are reliable, and for other measures they are not. Medical record extraction is good for measuring the clinical processes of care.
The Ohio health department has set standards for timely submission of encounter data and medical records information for minimum claim volume.
The state's decision process for data collection takes a lot of leadership and vision for the work to be done correctly. The leadership in the Office of Medicaid was extremely forward looking. Nine people outside the agency were hired to manage the data collection and analysis process, for which new job descriptions were created with higher pay levels to attract people with clinical, service provision and managed care experience. They were then trained to do data analysis using extremely large data sets. This has been a highly successful program that has also involved staff in understanding some of the technical innovations and medical informatics. Consumer advocates and providers in managed care plans are also included in decision-making processes.
Concerning data collection, the agency's approach is to have equivalent measurement processes for both managed care plans and the fee-for-service delivery system. Logistical impediments include the variations in the capacity of managed care plan in administrative and clinical data systems and limitations in the UB 92 and HCFA 1500 electronic formats for transmitting data elements.
Mr. Ranbom concluded with three recommendations: (1) consider national review of the ICD-9 and HCPCS CPT-4 coding system with the intention of making changes that will enhance states' ability to measure delivery system performance; (2) review the state of the art of the administrative and clinical data systems of managed care plans and make recommendations for their improvement, e.g., standards and system certification; and (3) consider changes to the electronic format for health care transactions that will accommodate data transmitted from managed care plans to purchasers.
Robert Brewer, M.D., Nebraska Public Health Assessment Division
Dr. Brewer is on assignment from CDC as Nebraska's chronic disease epidemiologist. He has spent far more time on Medicaid managed care issues than anticipated. The Nebraska Health and Human Services system was formed in January 1997 by consolidating five agencies into three. This has facilitated collaboration between the former health department and the Medicaid program, which now has over 103,000 managed care enrollees in the state. However, there are only five full-time staff assigned to the Medicaid managed care program. It is important to keep in mind the small capacity in Nebraska and other states.
Dr. Brewer provided details on staffing and organizational structure. The main groups under the Medicaid managed care mandate are similar to those in other states. A major issue in Nebraska is access to participating providers, and the different issues in urban and rural areas. A related issue is whether technical expertise exists at the local level to do data collection. The data needed to answer the questions include encounter data, client satisfaction surveys, provider surveys, and enrollment surveys. An encounter data system is a cornerstone of the state's quality assurance activities. However, the state is only starting to capture that information, and compliance problems have already arisen. With one exception, the arrangement between the providers and the managed care plans is fee-for-service, which has serious implications for data collection.
In addition to member satisfaction surveys, enrollment surveys are administered, including a core assessment and health status questionnaire. Another survey that collects information on provider characteristics and provider experience working with health plans would be useful in comparing managed care with fee-for-service. Some 90 percent of the data from one of the plans and 80 percent of the data from the other plan passed edits. There is a plan to evaluate the completeness of reporting into the encounter data system and to expand member satisfaction surveys to assess the delivery of some of these services.
The state has specific performance goals related to access to covered services, and these goals are detailed in the contracts with managed care plans. However, the measurement of these performance goals is largely based on self-reports. Audits of the plans are planned. Dr. Brewer said it would be very valuable to have data standardization, focusing specifically on encounter data and taking into account the needs of multiple users.
As far as logistical impediments go, he said that public health and Medicaid programs in general have not worked together. Public health data systems have not been designed with a goal of linkage with Medicaid data, so it is necessary to "fish around" and to try and find variables that can be used in that linkage. Regarding confidentiality issues, there has not been much resistance encountered in Nebraska to reporting data based on these concerns.
He outlined contract provisions for enforcing data reporting requirements with the plans, but said they have not yet been used, in view of the fact that it is a startup program. In addition to the "stick approach" it is useful to look for ways to collaborate more effectively with the plans. One such attempt in Nebraska is to have the data manager, Medstat, go out and meet with the plans and to review their data collection procedures and to try to provide assistance where improvements are needed. Another step that can be taken is to provide quality indicators to the plans.
Dr. Brewer reiterated his call for consensus standards for encounter data that address the needs of multiple users with flexibility to allow states to add specific data elements that they find useful. Further, all people concerned need to study and report on methods to promote linkage of public health and Medicaid managed care data. He suggested that the Committee provide assistance, particularly guidance on methods to assess data quality.
Discussion
Ms. Ward asked Mr. Ranbom if plan comparison charts are given back to the plans and made public, and was told that it is up to the plans to review and decide whether there are any problems with their encounter data. At some point in the future, the information becomes public. Advocacy and community groups are interested in holding managed care plans accountable in their local areas and they rely on the department to provide information to them. Concerning publication, some of the data are suspect so they may put caveats on the information. The external quality review report is released publicly. Ms. Clarke reported that her department will be publishing HEDIS data for their commercial and Medicaid populations next year.
Noting the frequent references throughout the day to using a HEDIS standard, Dr. Mor asked Mr. Ranbom to speak to the issue of basic validity of the encounter data on which all of the HEDIS measures are constructed, and whether these data should be published if they haven't actually been analyzed for validity. Mr. Ranbom said that for some measures a validation process for encounter data is used for all of the plans. They do not publish measures that have not been validated. Dr. Brewer commented on the calculation of HEDIS measures in terms of a denominator problem. Ms. Clarke said that the HEDIS measures "as bad as they are, are better than the ones we made up ourselves." The approach taken in Oregon is that "our job is not to issue report cards, but to help everybody raise the bar constantly with quality improvement."
Switching to another subject, she observed that the downside of collaboration is that everybody "will take their marbles and go home if you try to enforce anything." So the question is how to keep improvement moving without having everybody leave.
Dr. Carter-Pokras, referring to civil rights laws, returned to questions about race and ethnicity, and each panelist commented on their respective states' work in this area. The subject does not get much attention in Oregon because of the state's very small non-white population. Ohio has a relatively large African-American population but extremely small Hispanic, American Indian, and Asian Pacific Islander populations, but a lot of analysis is done that can be cut by each of those groups. A problem is that many people do not feel that the ethnicity data collected at enrollment is very reliable. Nebraska makes an effort to collect information on race/ethnicity as part of the client file in the enrollment process and to match clients with physicians who are able to communicate when there is a linguistic need.
Dr. Carter-Pokras asked for more information about the children's health insurance initiative and what kind of changes in data systems are anticipated in states that have decided to expand the Medicaid program as a way of dealing with that initiative, and the panelists commented on this. Mr. Ranbom observed that one of the big problems in Ohio is going to be determining an adequate capitation rate for that population. As a result, these children will probably be in fee-for-service at least for a year before they are moved into managed care. In Nebraska, Dr. Brewer said, the population is going to be skewed more toward adolescents, where there is a major gap in coverage. There, a major concern is the availability of providers, particularly in rural areas of the state. Ms. Clarke reported that in Oregon they will do what they always do whenever there is a new pot of money: try to make both the clients and the providers a smooth and seamless part of the existing program.
Dr. Brewer noted that much of the Medicaid population is not going to be listed in data such as hospital discharge records as Medicaid recipients. This classification issue is one the Subcommittee may want to consider further.
Ms. Coltin observed that her health plan is trying to develop performance measures but is limited sometimes by the coding systems themselves as to what they can collect. Health plans have not typically been a very influential constituency in going to the AMA to advocate for changes in coding. To the extent that states are payors, perhaps they are likely to become more influential. She asked if other states have also encountered similar problems with inadequacies of the coding system. Ms. Clarke said that Oregon has a similar problem, but that ARC in conjunction with AMA has been a good advocate for creating dummy codes for certain medical conditions. Mr. Ranbom noted that many people do not know who is responsible for making coding changes.
Ms. Greenberg noted that the whole issue of procedure coding and involvement in standards will be addressed in a meeting NCHS and CDC are holding with people within the department and from state health departments and other state organizations who will talk about the implications of HIPAA for public health and health services research. The goal is to have a workshop later in the year for the different constituencies. The initial focus is on expanding the National Uniform Billing Committee and the National Uniform Claims Committee and conducting an educational campaign. Dan Friedman will participate on behalf of the National Committee.
The meeting recessed, to reconvene the following day.
Harry Rosenberg, Ph.D., National Center for Health Statistics
The first hour of the session was chaired by George Van Amburg. He noted that at least two federal agencies use two different standards for age adjustment on a regular basis, which creates great havoc. The National Center for Health Statistics (NCHS) has framed recommendations developed in workshops. The context and recommendations were presented by Harry Rosenberg, chief of the Mortality Branch at NCHS.
Dr. Rosenberg began his presentation with a brief background on the population standard for age adjusted statistics, which is about to undergo the first official change since 1943. Because the population of the United States is getting older and the risk of death is higher at older ages, the crude death rate is being pushed up artificially. The technique for countering that bias is called age adjustment, which helps statisticians understand the true risk of death for different racial groups in our nation.
Periodically since 1940, there has been agitation for change in the population standard. A 1991 recommendation by a NCHS-convened workshop that all federal agencies in the Public Health Service use the same standard was not followed. The National Cancer Institute used the 1970 standard, parts of CDC used the 1980 standard, and NCHS used the 1940 standard. Several recent criticisms in the press suggested that perhaps the boasts about improvements in mortality for heart disease and cancer might not be real. So, the NCHS convened a broadly-based second panel in 1997 that included George Van Amburg representing the National Committee. That second workshop looked at policy and implementation issues that addressed these questions: Should the United States or the Public Health Service use a single or multiple standards and which standard should be used? How often should the standards be updated? and How should the recommendations of the workshop be moved forward?
Nine recommendations were set forth by that workshop, and he described each of them. The first was that the population standard be changed from 1940 to the projected population of the year 2000 and that the single standard be used by all agencies presenting official data. The second was that public health agencies and the Department implement the population standard beginning with the published data for the year 1999. The third was that the agencies continue to use and publish their current standards until the official implementation year when the new common standard will be adopted, and that agencies implementing the new standard prior to 1999 simultaneously publish rates adjusted to both the old and the new standard.
Others included having NCHS convene a work group to evaluate the age adjustment standard at least every 10 years and an implementation committee to move these recommendations from a proposal to reality, and that the Center publicize the new standard widely.
Dr. Rosenberg reported that the implementation committee had its first meeting in early December. He enumerated several ways the proposal would be moved through the Public Health Service, including a forum to be convened in Atlanta at the Centers for Disease Control and Prevention. The implementation committee was specific about the types of instructional materials and tutorials that will be developed to publicize the new standard. It was suggested that the educational material on age adjustment be included on ICD-10, the 10th revision of the international classification of diseases.
Asked if there is a sense of what the implications are for achievement of the 2000 objectives, and the impact of these recommendations on data collected by race and ethnicity, Dr. Rosenberg replied that the Health People 2000 goals would remain as targets, and that the path toward the targets will not change much. With regard to differentials between the races, the 60 percent differential between the races will be reduced somewhat, and that general pattern will also hold for other ethnic groups with a distribution toward the younger population. The differences will be less marked for Asian population groups. The Social Security Administration will be kept informed, but the work of that agency will not be materially affected.
To the question of whether both Social Security and HCFA will use this approach to readjusting their summaries of the entire 55 plus population or the average Medicare beneficiary or Social Security recipient, Dr. Rosenberg said it was an excellent suggestion to include them in the review and comment procedures.
As for ways the committee could help in this process, Dr. Rosenberg said that if the committee felt comfortable with the process he has described it would help if it supports the recommendations at the next Data Council meeting. Their support for the proposals would help efforts to get a buy-in for improving health statistics. Subcommittee members seemed to agree that this is a desirable action item.
James Weed, Ph.D., National Center for Health Statistics
Noting that the preceding speaker had brought a solution, Dr. Weed said he would bring a problem. He reviewed the new OMB standards on race and ethnicity and their complex effects on vital statistics, especially those relating to self-identification, where individuals may report more than one racial identity.
He outlined some of the difficulties associated with collecting data on births and deaths from varied sources. There is nothing to prevent anybody from writing more than one race on a birth or death certificate, but currently only the first mentioned race is coded, so that only one race is reported to NCHS by the states.
One of the questions pertaining to birth and death certificates is whether vital statistics will continue using an open-ended format and still be in compliance with the standard, even if that is advisable. A second question is whether the states can change the list of races from the standard. A practical consideration is the amount of space on the certificate for open-ended questions. Another practical question arises from the fact that although the OMB directive directs that the term "multiracial" will not be used, three states currently require administrative forms to include "multiracial" as a reportable race. Since OMB does not override a state law, it remains to be seen whether those states will change their laws in order to conform to OMB standards, and whether the term "multiracial" will die away.
Collecting multiple race data in vital records is a gigantic problem for hospitals and funeral directors. Dr. Weed said it is fortunate that a new panel will soon hold its first meeting to evaluate and revise forms for birth and death certificates. Race is bound to cause a lot of discussion. Ultimately, the problem of using these standards to collect data on vital records is a state responsibility, and the federal system will have to work with states on these issues soon because the Census will use the new standards in the year 2000. Dr. Weed called attention to a copy of what a census dress rehearsal questionnaire will look like, and "that is going to create a basic incompatibility of vital statistics data with census data" since the census is already collecting multiple race data relatively efficiently. He estimated that four to eight percent of the population might mark more than one race, but by 2000, it is very unlikely the vital records are going to do the same. It will continue to be a problem in getting numerator and denominator to use the same base for race. Since the census vital statistics make inter-census estimates, the problem can be expected to continue into the next century.
Discussion
Dr. Iezzoni observed that there will have to be "a real back and forth" about how the federal and state systems are going to work out the problems involved in race and ethnicity issues for the 2000 standard.
If the OMB standards for tabulating race and ethnicity are followed closely, Dr. Weed was asked, how does one allocate individuals who identify multiple race to a single race, and what effect will that have on figures on mortality in data taken from the states? Dr. Weed said that minority races are likely to become smaller because individuals who are Native American and white, or black and white, or Asian and white are going to report both races and will not be in a single race group. Proportionally, minority racial groups will diminish more than the white group. This would probably mean that life expectancy rates will go down for all the minority groups.
There will also be a large impact on infant mortality rates. The implication is that there will be more multiple races reported on birth records than on the infant death records, which will change the denominator. However, one can be born into one race or ethnic group and die in a different one, so the problem is going to increase. One result may be that the race questions on death records will not be filled out because people just don't know, and then important information would be missing. The NCHS is working on bridging studies for at least the next five to eight years when the system will have to move on its own, and there are two or three OMB work groups looking at various tabulation issues.
To keep the subcommittee informed of progress in dealing with the many problems raised, Dr. Iezzoni invited both speakers to come back in a few months, possibly for a breakout session in the June full committee meeting.
Rachel Block, Health Care Financing Administration (HCFA)
Ms. Block observed that Medicaid is and will always be a highly dynamic program, and also that, "trite but true, every state is unique." By mid-1997, 15 million or approximately 50 percent of Medicaid beneficiaries were enrolled in managed care; and about 35 states have HMO risk contracts for some or all of their programs, which means that at least 15 are utilizing other kinds of managed care models. The trend toward mandatory enrollment continues.
One thing the Subcommittee needs to consider carefully is that only one or two states have comprehensive managed care models. Therefore, any data collection, monitoring or other statistical strategies for evaluating managed care need to define managed care. Given the high degree of variation, there is need for a model that encourages standardization where it is appropriate, while taking into account a wide degree of variation.
The emerging information technologies create opportunities in a highly competitive marketplace, but they also reinforce existing concerns about confidentiality and security. A person-level database will substitute for the HCFA 2082 reporting format. This will enhance the basic Medicaid data currently being collected, and also get all the states in the same system. HCFA is currently working on parallel specifications for encounter data submissions through the MSIS system. Now, data are collected and reported on two basic areas of Medicaid managed care: encounter data and the national summary. These data are collected only on an annual basis, but it is hoped that in the next year or two as states come into the MSIS system, there will be a more regular updating of basic enrollment data, at a minimum.
Work is continuing on a report on key performance indicators. These will include selected indicators of program performance covering four areas: enrollment, utilization quality, financing characteristics, and satisfaction. This will be tested through regional offices--information that can be given to the Subcommittee will be available in a couple of months. There is also an opportunity in the new children's health insurance program, to devise additional reporting requirements for those programs.
Ms. Block identified key areas of concern relating to the quality of race and ethnicity data, and welcomed any recommendations of simple and practical strategies to validate basic data collection at the front end of the eligibility process or through the health care delivery system. Also, questions were raised about the validation of the claims based diagnostic information.
She described a Philadelphia project exploring how health plans can assess birth outcomes at the community level, being conducted by a collaboration between the city health department and a health plan. They will test an enhanced encounter data form on which all the health plans in the city health department and clinics have basically agreed to collect standardized information on certain key risk factors. The next step is to think about how these enhanced encounter specifications can be applied to global or population-based conditions.
HCFA is interested in trying to use all the regulatory tools available, especially technical assistance and other kinds of leadership, to help states improve their capabilities in assessing managed care. Corresponding efforts to improve data quality will also be made.
An interesting question for the Subcommittee to consider is what tool other than HEDIS would be useful to evaluate managed care. She reiterated an earlier point that a gap in HEDIS is that it does not provide the kind of timely, early warning system. Also, if additional data are considered necessary for health plans to collect or report, they could be factored into the assessment of the capabilities of their information systems.
Mary Jo O'Brien, The Lewin Group
Prior to joining the Lewin Group, Ms. O'Brien was Commissioner of Health in Minnesota. Building on Ms. Block's comments, she reported on a project the Lewin Group has just completed with DHHS. The goal is to look at the capacity of local, state and federal public health agencies in terms of what data are and should be collected, and how.
State and local interviews identified 10 states to observe in terms of the relations between state and local health departments. The ultimate goal was to explore the impact of the public health infrastructure on health outcomes and costs. Ms. O'Brien emphasized that any strategy that seeks to characterize information must be employed over a period of time. Their data strategy involved three complementary steps:
§ a look at macro level national surveys
§ a micro level look at case studies of some communities
§ a research agenda
The intent of the local surveys was to mirror state surveys. The case study rationale is to know what best practices are and what is not working on the ground. She noted the importance of talking not just to the public health agencies but also the community at large, to broaden the definition of the public health community and its impact. The research agenda establishes a crucial link between the infrastructure and outcomes.
Ms. O'Brien recommended that a lead agency be assigned the responsibility for implementation of a coordinated data strategy and that implementation be funded.
Margaret Schmid, National Committee for Quality Assurance (NCQA)
Ms. Schmid manages the State-Federal Projects Unit of NCQA. She noted that NCQA collects Medicaid HEDIS data on behalf of HCFA, which requires all Medicaid managed care plan to report HEDIS. There are 51 HEDIS measures in the current reporting set applicable to the Medicaid population, organized into six domains. Ms. Schmid anticipated that the Consumer Assessment of Health Plans Study (CAHPS) will be incorporated into future versions of HEDIS for the Medicaid population.
In 1997, NCQA received approximately 30 Medicaid submissions. It is currently working with the APWA to obtain a grant from the Commonwealth Fund to develop a more comprehensive database of HEDIS statistics on the Medicaid population. This database should be in place for the 1997 reporting year, with data available to participating states by the end of 1998. In addition, NCQA has been working to develop a method for applying HEDIS measures to Medicaid beneficiaries receiving care in fee-for-service and primary care case management programs. The methodology for this effort will be released shortly, allowing states and policy makers to attempt comparisons of different systems of care available to Medicaid beneficiaries.
From NCQA's perspective, the primary barrier to collecting national data sets stems from the unique reporting requirements and processes in the Medicaid environment for every state. NCQA and APWA are seeking to increase the number of states voluntarily using HEDIS to measure performance. She reviewed some of the practical barriers to obtaining HEDIS data even at the state level: health plan relationships with providers, geographical differences in covered services in health plans, and carveout services.
The consolidation of the managed care industry has added enormous complexity to health plan information systems, and little guidance has been available to Medicaid agencies about how to collect and use these data. NCQA has been working over the last two years to assist states in implementing HEDIS for the Medicaid population. Ms. Schmid observed that nearly all of the issues in collecting genuinely comparable information about plan performance in the Medicaid environment are also facing the commercial and Medicare worlds.
Ms. Schmid said that while the challenges to collecting data are significant, progress made over the past three years has also been significant. She pointed to the work of the Subcommittee as invaluable in sorting out different approaches that can move the endeavor forward.
Discussion
Asked for the HCFA perspective on the variations in reporting requirements among state contracts, specifically regarding managed care data collection for state Medicaid programs, Ms. Block referred to HCFA's practice of requiring 1115 waiver states to collect encounter data. Project officers for those waivers have come to rely on routine reports from states, without necessarily getting all the raw data into HCFA. With freedom of choice waivers, she said there has not been a requirement for encounter data; however, there is an administrative requirement to support focused clinical studies in relevant public health categories.
With the Balanced Budget Amendment (BBA), HCFA faces some very challenging requirements that the states will have to begin to incorporate into all of their managed care programs. She explained that they are now in the early stages of development of regulations to support the implementation of the BBA. There is time to work on a standardized format, which is likely to build upon HEDIS. The other explicit requirement is that states submit encounter data through the MSIS system. As the agency is faced with managing data on a much larger scale, she will share information with the Subcommittee about the specifications they will be working on with the states.
Finally, Ms. Block mentioned the quality improvement system for managed care (QISMC). QISMC includes an explicit requirement that health plans provide specified performance measures to HCFA or to states. In addition, there will be specific areas of focused study for both Medicare and Medicaid. A draft of the standards will be soon be ready for public comment. It is not certain that QISMC, HEDIS or any other standardized data set will effectively address the question of how to get an early warning signal when either a plan or provider within a plan is providing less than adequate care.
Remarks addressed to Ms. O'Brien expressed concern about how responsibilities could be shared between the public health apparatus and managed care companies to stimulate the collection and analysis of public health data for various state populations. A related question was whether the analytic capabilities of both might be made more of a public resource. Ms. O'Brien noted that in Minnesota, they tried to involve the health plans in public health strategies. That is not within the scope of the current Lewin effort, but it could be looked at subsequently at the research levels. For something like that to happen for Medicaid, a mechanism would have to be developed at the state level. Returning to a general theme of the hearings, she noted the challenges of helping federal, state and local government agencies to work with and support one another.
In response to another question, Ms. Block said that all the mental health/substance abuse waivers she has seen in the last couple of years have included some mechanism and reporting requirement between the state and the delivery system. The basic threshold requirements for states to submit their data through the MSIS system are irrespective of program financing and delivery strategies. So any data relative to a mental health program, managed care or otherwise, would flow through that system. She noted that SAMHSA has recently been working with the Medstat group to analyze some of the MSIS data on an historical basis and to tie that in with diagnostic information. HCFA will be working in the next year to devise a specification for mental health encounters, and also for long-term care encounters.
A question was raised about the extent to which information on complaints and appeals reflects satisfaction or quality of care, and studies of the validity of this measure as a performance indicator were proposed. Ms. Block replied that complaints and grievances are a key issue to look at in health plans generally. HEDIS requires that states document how health plans inform enrollees about grievance procedures. A key issue is that the health plans generally don't do a very good job of collecting and reporting complaints and grievances; usually the information becomes reliable only in the event of a more formalized appeals process. From the health plan perspective, there may be need for a specific definition of a complaint or grievance. A lot of work is still needed just to define the terms, and patient surveys may be a better way of getting at the issue.
Dr. Iezzoni returned to the issue of churning and asked how patients go on and off Medicaid, how the safety net picks up their needs when they are off the Medicaid rolls, and whether there is a data gap that the Subcommittee should highlight. Ms. O'Brien said they are working with several states around different strategies of safety net providers, and specifically with Medicaid agencies as purchasers, to figure out how to stabilize, maintain, and secure the safety net in various communities. One of the issues is the willingness and capacities of different agencies to work together. The relationship between state and local levels is a critical factor, as some states are pushing more responsibilities down to the local level without providing sufficient resources.
Ms. Schmid said that some states are experimenting with the idea of using the state birth certificate registry as a source of information for certain data that health plans are responsible for having about their enrolled populations. Her group is part of a program financed by the Bureau of Primary Health Care to train community health centers--important traditional providers to the Medicaid population--to learn how to use HEDIS measures.
Ms. Block again cited the Philadelphia project. Partly because of the complexity of immunizations records, and shortcomings in the vital statistics data, they selected prenatal care as their first project.
Dr. Carter-Pokras asked whether HCFA is doing any focused study on differential access among racial and ethnic Medicaid subpopulations. She noted that states seem not to be doing analysis to determine these kind of differences in access to quality care. Ms. Block replied that HCFA expects the health plan to assess any significant population variables, of which race and ethnicity might be especially predictive of health outcomes or other significant quality issues. Perhaps, she suggested, the issue should be re-debated in terms of whether, in fact, the state Medicaid agencies or the health plans should be required to collect the data. She added that HCFA has no specific performance requirement in contracts for states to report information on access or quality indicators across ethnic, racial or gender groups. States may be incorporating that requirement into their contracts. HCFA expects states to address their overall strategy for assessing and assuring access and quality care across all populations enrolled in managed care, but it is left up to the states how they are going to deliver on that expectation. Their experience shows that when a state tries to break out race and ethnicity as it relates to either health care utilization or outcomes, it is generally within the context of focused studies.
The point was raised that HCFA might take testimony from the Office of Civil Rights on general data reporting requirements under all federally assisted health programs. Many states have relatively detailed anti-discrimination requirements that go well beyond anything the Office of Civil Rights has articulated as a discriminatory activity in managed care. No state, even those with strong anti-discrimination provisions, articulates in their contracts the data they expect the contractors to provide that will let them measure this; nor do they generally match the eligibility of a beneficiary against enrollment data. This is a situation where there are essentially no federal anti-discrimination standards or data requirements. Yet, many states are extremely sensitive to the issue of racial and ethnic discrimination.
Ms. Schmid said the HEDIS measures are designed to look at the level of services that are delivered across the enrolled population. Ms. Coltin noted that HCFA requires that health plans report not only the measures, but also Medicaid ID numbers of all members. HCFA's plan is to link those with their enrollment data, and then to look at these measures by race and ethnicity. By aggregating data across plans in regions or states, they will be able to get a picture as to whether any patterns emerge.
Donald Umlah, Ph.D., Health Partners of Arizona
Dr. Umlah described himself as an "outcomes nut" who is on the QISMC committee and is also working with the NCQA state policy development organization on the comparison study of multistate fee-for-service and managed care indicators. He is also with the Arizona Physicians IPA, a contract HMO with the largest Medicaid health plan in the state. Arizona is a highly penetrated, competitive managed care state that provides services to all of the entitled populations. He is in charge of quality measurement of prevention and wellness programs and disease state management.
He spoke first of his concern about new and expanding managed care Medicaid states, and of how his perspective has changed since the 1950s when managed care was sold as the way to keep people healthy. By and large, that has not been proven true. He theorized that when managed care moved in on fee-for-service medicine, the huge amounts of waste in the system shifted the focus to utilization. As a result, there was a tremendous push to control and deal with utilization issues rather than with preventive health. In immature markets, there is a great temptation to push very strongly on the utilization side rather than to focus on the quality side of data development and monitoring indicators. This may also be the case for Medicaid managed care. To the extent that this committee can influence that process, when a state is just starting with Medicaid managed care, it should focus on quality indicators and qualitative data as well as on utilization.
He then described the kind of data Arizona collects and how they are used. He described prior authorization as "the most real time, online data" they have to look at service utilization. Encounter and claims data typically come from the HCFA 1500 and UB-92. Those data are used to determine utilization patterns and are an enormous source for inputs to their quality outcomes and performance indicator system. Built around preventive health indicators, disease state management and other management indicators, it is a proactive utilization system that feeds the information system, and is also a major source for most of their quality reviews.
Another type of data is the membership information supplied entirely by the state. These data do not supply racial information, although the his group has requested it and would find it very helpful. However, the state does collect racial information in their eligibility system. Other sources of data are member satisfaction surveys as well as member focus groups. All member complaints are collected, categorized and trends noted for use in designing programmatic responses. Provider files are also used.
Dr. Umlah's group has been reporting encounters to the state since 1982. It has only been within the last three to five years that they have seen that data used at the state level as quality indicators. Regarding future issues and problems, he highlighted the interactions among the electronic medical record, the re-use of results data, and patient confidentiality.
Kathryn Coltin, Harvard Pilgrim Health Care
Ms. Coltin commented that among the many utilization-related questions addressed in data analysis is whether socioeconomic status affects utilization rates for patients in managed care versus those with fee-for-service coverage. She emphasized that this question is broader than Medicaid.
She raised numerous related questions: Are managed care organizations reducing utilization rates, and are they doing so for recommended services or only for discretionary services? This is where, she said, some of the fear about managed care comes in. And are reductions in recommended services occurring in both managed care fee-for-service? Are reductions in discretionary services greater for lower socio-economic status patients? Are there differences in compliance rates for recommended services between commercial and Medicaid enrollees in the same managed care organization, and does this vary across managed care organizations?
Ms. Coltin suggested determining whether differences between Medicaid and commercial rates have more to do with differences in patients or in provider behavior. Another consideration is what kinds of non-medical services are necessary--e.g., outreach, transportation, and social services, none of which are covered under the capitation payment that the Medicaid agency makes to the plan. She noted that managed care organizations are taking on some activities that were previously funded through mechanisms other than Medicaid and for which they are not being reimbursed.
Another set of questions deals with access to different types of services, in particular, specialty care services. Do risk adjusted outcomes for Medicaid enrollees in managed care organizations differ from those for commercial enrollees or from Medicaid fee-for-service patients? And in what ways do the outcomes vary?
Still another set of questions relate to comparison of patient satisfaction and quality of care or services provided to Medicaid managed care enrollees by the managed care organization' contracted suppliers of medical and social services. In looking at variability in care for different population subsets, patients with lower socio-economic status might be more vulnerable in those settings as well. For all these questions, there is a larger question of whether the appropriate data are being collected to answer them.
Ms. Coltin shifted her focus to the kinds of data they report out. To report on all of the HEDIS measures, they use several internal quality indicators that look at several conditions, e.g., asthma, depression. They also report satisfaction surveys; in fact, they have conducted so many satisfaction surveys that they are beginning to get declining response rates from patients.
Her group also reports to the state Medicaid agency's division of medical assistance on targeted quality indicators they have identified as part of their semi-annual review process. Among the many indicators they have looked at are rates of member participation in new member orientations, and new member screening and outreach.
She displayed an overhead copy of a HEDIS report to show how they compare commercial and Medicaid rates, breaking out the data by different delivery system components within their organization. They have not observed great systematic differences among physician compensation and its effect on quality of care, but they are trying to see if any of the incentives in the payment structures are having an impact on quality.
Ms. Coltin spoke of the kinds of data they would like or need but now don't have: sociodemographic data on race and ethnicity, the primary language spoken, health risks or needs, and better data on disease burden, risk behaviors, and functional status. Their biggest gap in the area of health services utilization is in pharmacy services in their Medicaid contract. Another data gap is in laboratory results. They would also like to see some sort of indicator for new diagnoses that appear on a claim that would flag a new problem.
One of the most important limitations of data is in the claims area, with diagnoses probably being the biggest problem. This lack leads to problems in identifying the appropriate target population for measures and intervention, and in evaluating disease severity. And as mentioned earlier, there is the problem of response rates and response bias in information collected from patients.
What are they doing about it? She briefly summarized some of the measures they are taking. They do a number of data validation studies, using algorithms to try to cross check diagnoses to make sure they are accurate. They are working with the state to link their data both to some of the DMA files in the case of pharmacy data. They are encouraging the adoption of computerized patient records. They are putting information and requirements around data quality into some of their contracts. Finally, they are engaging in strategic partnerships with suppliers, vendors, partners and coalitions.
Discussion
Ms. Coltin responded to a question about collecting race and ethnicity from commercial enrollees by saying that they do not have that kind of data on anyone since it was viewed as potentially discriminatory information. If it were federally legislated and everybody was doing it, she said they would "jump on that bandwagon," because of potential value of the information.
Asked whether in Arizona any managed care providers have undertaken studies to document changes in cost as a result of increasing coverage for mental health and substance abuse services, Dr. Umlah reported that they finished a study last year that found that people in the mental health system got less primary care overall than people in the control group. A secondary follow-up will co-locate primary care in a demonstration project within the same facility as a large at-risk provider for care of seriously mentally ill patients. They are working on the combination as a way of coordinating care across the two systems, overcoming the obstacles and getting these institutional cultures to work together.
Asked whether any managed care organizations have done or are doing studies to estimate costs of enabling services for different population, Ms. Coltin replied that it varies by state. They have looked at the costs of outreach because they have to do outreach to the commercially enrolled population as well, but they have not looked at the marginal costs of outreach to the Medicaid population.
Eileen Peterson, M.S.P.H., United HealthCare
Ms. Peterson leads United HealthCare's Center for Health Care Policy and Evaluation. She has worked for 10 years developing administrative data systems for population health management. She cited the public health principle of the greatest good for the greatest number and said that it applies to data as well as services. She emphasized the importance of good data and the need to be very selective about the kinds of data that one invests in and how one uses those data.
United HealthCare serves Medicaid populations in 10 states in a variety of health plan models for hundreds of thousands of people. The dominant model is the IPA health plan, for which they have good administrative data. They also have good out-patient pharmacy data. Some of their plans collect encounter data, the standards for which are the same as the fee-for-service data that matches the HCFA 1500 model. Her organization is committed to the collection of HEDIS data. In Medicaid, the biggest frustration for everyone involved has been to come up good performance measures for the population. For many of the measures, they have to go to medical records to get the needed data, and that is not only a burden, but also the data are often incomplete. The organization is involved with academic researchers in efforts to explore and develop new measures for data.
EPSDT data is another area of concern for health plans. The move by HCFA to develop guidelines to standardize EPSDT collection and to map it to administrative data deserves applause, and may lead to helpful synergies and efficiencies.
United HealthCare's health plans have become involved in collecting data for health risk assessment in both Medicaid and Medicare populations. A big issue is standardization of those tools and creation of information about how they can best be utilized in a real world setting.
Ms. Peterson enumerated five recommendations for the committee:
Michael Collins, University of Maryland at Baltimore
Mr. Collins is deputy executive director of the Center for Health Program Development and Management at the University of Maryland, Baltimore County. Their primary client is the Department of Health and Mental Hygiene of Maryland, a unified state health department that includes public health and Medicaid functions. His group played an important role in helping the state design and implement its current Medicaid managed care program called Health Choice, a program that began in July 1997 and now has enrolled about 333,000 persons in capitated managed care programs. It is a relatively comprehensive program with only a mental health carveout and a few tiny exceptions.
Maryland is "betting heavily" on encounter data, which it regards as essential to program operation. The plans have known from the beginning that they wouldn't be able to meet the state's quality standards without decent encounter data. Managed care is about management, and management is about information. But, too often people collect data they don't know how to use. The state needs encounter data above all for quality measurement. His group intends to replicate some of the HEDIS measures, and to look at some variations, particularly around eligibility issues. A group of stakeholders are working on special population-specific measures that can be analyzed using administrative data.
Maryland also uses encounter data and other claims from the mental health system for health-based payments (AKA risk adjustment). Mr. Collins reported that the state's encounter data are still quite poor, but they are working with HCFA to make their encounter data requirements as much like claims data as possible. Maryland also offered and paid for technical assistance to all the plans to help them meet their encounter data requirements. Things have not gone very well in the first two rounds of encounter data submission, in predictable ways: timeliness, completeness and accuracy, the last of which is the hardest to remedy.
Mr. Collins summarized other current concerns. Plans are having persistent problems in communicating with their providers and getting data from capitated providers. In general, information technology operations in managed care organizations are undercapitalized. Most providers are in several managed care organizations and they have difficulty keeping up with the differing data demands of the managed care organizations. Many of the contracts with providers lack regulations that require them to submit data.
He also called attention to the role of vendors of health information systems. Managed care organizations make large investments in systems that are hard to change once put in place. He suggested that the federal government give vendors notice of what will be required two years in advance, for their product development.
On another subject, he noted that many things can be done on the state level to audit accuracy before going to the medical record review level--for example, more and better automated measures. He affirmed the issue that surfaced in this meeting concerning the problems of linking numerator and denominator data. He suggested auditing Maryland's data collection on race and ethnicity, to assess the quality of the data.
Dave Baldridge, National Indian Council on Aging, Inc.
Mr. Baldridge is the executive director of the National Indian Council on Aging (NICA), based in New Mexico. The mission of the Council is to bring about improved comprehensive services for the nation's 200,000 older Indians and Alaska Natives, who have very little in the way of health services. The nonprofit organization was formed by a group of tribal chairmen. Based on results of a questionnaire sent to Indian service providers in 1992, they developed an agenda for the future. Three of the top five needs expressed by elders concerned health and long-term care.
He explained that Public Law 93638 covers more than 300 tribes that are contracting to use some portion of Indian Health Service monies to provide some or all of their own health care. This is a great exercise in tribal sovereignty, but it is also frightening, especially for the smaller tribes which lack adequate facilities and medical and clerical personnel, and face a backlog of three generations of neglect.
The NICA set out to empower the Indian health care system around three significant trends affecting Indian health care: decentralization of the Indian health care system; the move toward standardization of data; and state managed care Medicaid. In the current fragmented state systems, Indians are the smallest end users. The 38 states where Indians reside are benefitting from HCFA data, but the tribes dwelling in those states are not.
Traditionally, the Indian health care delivery system is responsible not only for the health and direct care that an HMO might provide, but also for public health and maintenance of facilities. The tribes are now at a new negotiating table with new players and dealing with a new term: capitation. The interface between the Indian traditional delivery system and the new managed care system is extremely complex for both clients and the Indian providers. The tribes are moving at various speeds toward managed care, depending upon their contracts with the host state. For some, it is working well, but in other states (Arizona, New Mexico and Oklahoma), HCFA Medicaid data are not available. Getting data seems to be a major issue, although these 300 tribes have as legitimate a need for good health data as the states and managed care organizations.
Based on this analysis, NICA created a data bureau to facilitate access to HCFA and other data for the Indian delivery system, the tribes, and the urban clinics for Indians. So far, they have received federal grants totaling more than a million dollars, starting with a grant from the Administration on Aging. They have developed a cooperative agreement with HCFA, which involved strenuous consideration of the Privacy Act. NICA will be involved both in ascertaining what data are needed by the tribes and describing the need to HCFA, and in helping the tribes interpret and analyze the data provided by HCFA. It is hoped that within a couple of years, the Council will be able to supply every tribe with federal and state data sufficient to meet their needs at no cost to the tribes.
The Council recently procured a grant from the Administration for Native Americans to supply data from HCFA or IHS to the 40 urban Indian clinics. The data will include GIS maps of the cities in which the clinics operate, making it possible to look at such factors as health and socio-economic status, census data, and proximity to health services. This ground-breaking project has attracted a lot of attention.
The council also won a competitive five-year grant to revamp the entire IHS data system, converting it to an interactive GIS system that can be easily accessed by computer. During the third year of the grant, they plan to create an interactive atlas of American Indian elder health.
Mr. Baldridge expressed concern that state data are not going to help because the states do not seem to feel a responsibility for Indian health care that is comparable with that shown by the federal government. Nevertheless, he said it is possible that some good models will evolve.
Discussion
Mr. Baldridge affirmed that the data work the council is doing is for all of IHS, although funding is under an aging entity. He noted that aging has implications for all generations. Furthermore, once data for ages 55 plus have been obtained, it is not so difficult to get the same data for younger populations. Asked about the impact of increased reporting requirements and sophisticated data systems on resource-poor providers, Mr. Baldridge said they would be "thrilled" to be able to build capacity. They are concerned that one of the first things to suffer will be data acquisition by the tribes that simply won't be able to afford to acquire data.
In response to another question, he said the IHS staff has not paid much attention to race and ethnicity data because it is so overwhelmed with other issues. He noted that the hundreds of recognized tribes, which the IHS allow to determine their own membership qualifications. More than half the Indian population in this country is urban. Some Indians feel they would be discriminated against by reporting ethnicity so they hide it from government officials. As a result, many urban Indians are assigned to the nearest HMO rather than referral to an Indian provider. Mr. Baldridge added that managed care puts Indian providers in as much or more jeopardy than it does the clients.
To another question about coding for alternative or complementary native healing practices, Ms. Ward offered that alternative medicine is a minuscule fraction of what is actually provided. The RPMS system is ahead of many of the other systems, and it has many of the same codes built in. She noted that this is a subject the Subcommittee might consider further.
The implications of proprietary data were raised -- in particular, questions about the willingness of proprietary organizations to share both data and methods they are developing, including for evaluating quality. Ms. Peterson predicted that as standards progress and become more accepted, there will be a falling away from that proprietary perspective, opening many possibilities. Right now, a major concern for managed care organizations is their responsibility to safeguard privacy and confidentiality. Mr. Collins commented that there is a lot of movement in the right direction, and the use of "black box" systems of quality measurement are almost inconceivable now. What will stay strongly proprietary, he said, is the implementation of those measures into software.
Ms. Peterson suggested thinking about what the private sector has historically been good at--namely, finding opportunities where innovation is rewarded. Perhaps the private sector could be motivated to implement within standardized software systems that would make it easier for people to access and disseminate data. Mr. Collins referred to David Blumenthal's recent article in the Journal of the American Medical Association on the subject as a good argument for that approach.
The panel was asked for thoughts on how to collect data on enabling services. For example, when HRSA has doing case studies of Medicaid managed care plans, they were unable to report on the utilization of these services, yet they are hallmarks of safety net providers. The only systematic effort seems to be the Indian Health Service's use of community health representatives who collected detailed data in this area.
Ms. Coltin said that her managed care organization uses mostly ad hoc systems to collect data on enabling services. For instance, with child care logs, most services are tracked, but not in ways that are readily tied back to an individual user. Mr. Collins noted that since these items are, for the most part, not paid for on a bit-by-bit basis, encounter data are not a very good source of information. It was noted that it is difficult to make the case for these services because nobody is tracking who needs them and who uses them.
Ms. Block asked about identifying alternate means of assessing quality improvement efforts for populations or health needs too small to be picked up by HEDIS. Ms. Coltin described Harvard Pilgrim's approach, breaking data down into different kinds of aggregations of groups, e.g., care sites or physicians. Ms. Peterson said that her organization can go across multiple health plans; they have recently submitted a paper to Pediatrics that looked at the special health care needs of a population of children. She noted that Medicaid HEDIS must address the issue of comparability as well as the question of whether providers have the infrastructure to report the measures. Dr. Amaro noted that some of the national data systems have the same small-numbers problem.
Dr. Iezzoni adjourned the meeting at 2:55 p.m.
I hereby certify that, to the best of my knowledge, the
foregoing summary of minutes is accurate and complete.
/s/ Lisa I. Iezzoni May 5, 1998
Lisa I. Iezzoni, Chair Date