United States General Accounting Office GAO Report to Ranking Minority Member, Committee on Commerce, House of Representatives May 1997 MEDICAID MANAGED CARE Challenge of Holding Plans Accountable Requires Greater State Effort GAO/HEHS-97-86 United States GAO General Accounting Office Washington, D.C. 20548 Health, Education, and Human Services Division B-270335 May 16, 1997 The Honorable John D. Dingell Ranking Minority Member Committee on Commerce House of Representatives Dear Mr. Dingell: Over the past decade, Medicaid expenditures have soared. By fiscal year 1996, they reached $160 billion—nearly quadrupling fiscal year 1986 expenditures. Although the annual growth rate abated significantly in 1996, Medicaid expenditures continue to exert strong pressure on federal and state budgets. To help bring these costs under control, states increasingly are mandating significant numbers of their Medicaid population to enroll in managed care programs. By emphasizing primary and preventive care and treatment, it is hoped that managed care will improve beneficiary health care while curbing health care costs. As of June 1996, about 11 million Medicaid beneficiaries were enrolled in “capitated” managed care programs.1 Under a capitated managed care model, states contract with managed care plans, such as health maintenance organizations (HMO), and pay them a monthly, or capitated, fee per Medicaid enrollee to provide most medical services—which are coordinated through primary care physicians. This model, with its fixed prospective payment for a package of services, creates an incentive for plans to provide preventive and primary care and to ensure that only necessary medical services are provided. However, managed care also can create an incentive to underserve or even deny beneficiaries access to needed care since plans and, in some cases, providers can profit from not delivering services. Moreover, Medicaid beneficiaries required to enroll in managed care may find it difficult to seek alternative care if they find that plan providers fail to meet their needs. Because of your concern about these issues, we reviewed state efforts to hold managed care plans accountable for meeting Medicaid program goals and for providing beneficiaries enrolled in capitated managed care plans the care they need. As agreed with your office, we focused our study on the difficulties that purchasers, including states, have in monitoring 1 Based on the most current data available from the Health Care Financing Administration (HCFA) on managed care enrollment. Another 4 million individuals were enrolled in noncapitated managed care programs. Of the total 15 million managed care enrollees, about 2 million were enrolled in more than one plan, according to HCFA. Page 1 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 managed care programs and on state efforts to (1) ensure Medicaid beneficiaries have access to appropriate providers, (2) assess the adequacy of medical care provided through contracted plans, and (3) determine beneficiary satisfaction with plan performance. To understand the types of issues states face in ensuring accountability and quality in their capitated Medicaid managed care programs and the steps taken to address these issues, we visited four states—Arizona, Pennsylvania, Tennessee, and Wisconsin. At the time of our review, these four states collectively had almost 1.9 million Medicaid beneficiaries enrolled in their managed care programs. To analyze and illustrate state actions, we focused the scope of our work on 10 core accountability measures or processes deemed essential by HCFA and experts we contacted.2 We reviewed these states’ contracts with managed care plans and other plan requirements, as well as their efforts to monitor plan performance. Appendix I provides more detailed information on our scope and methodology. Ensuring that managed care plans provide enrollees the care that they Results in Brief need is a formidable task for private and public purchasers alike. In establishing their managed care programs, purchasers can require contracted plans to meet certain conditions—such as maintaining adequate provider networks and complying with data collection requirements—that help to hold them accountable for providing enrollees with appropriate care. However, establishing criteria for these conditions and monitoring plan compliance are often difficult because of a lack of population-based standards or benchmarks for what constitutes appropriate care or expected outcomes. In addition, individual-level data on patient care, such as those that are generated in a claims-based fee-for-service system, are not readily available. For states, establishing standards of care and tapping into alternative information sources on service utilization to assess the care that Medicaid beneficiaries receive can be a challenge. The four states that we visited—Arizona, Pennsylvania, Tennessee, and Wisconsin—have built access and data collection requirements into their contracts with managed care plans. We found, however, that plan compliance with the requirements we reviewed does not necessarily ensure that beneficiaries are receiving the care that they need. 2 Our work did not include a complete assessment of each state’s entire quality assurance process. Page 2 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 A number of these states’ requirements aim to ensure managed care plans develop and maintain provider networks that are sufficient to meet the needs of Medicaid beneficiaries. Some are criterion-based, such as patient-to-primary-care-physician ratios. For example, two states required that plans not exceed a maximum patient-to-primary-care-physician ratio of 2,500 to 1. Compliance with such a requirement, however, does not necessarily demonstrate that a network is sufficient to meet the needs of Medicaid beneficiaries. Patient-to-primary-care-physician ratios generally do not consider the number of networks a primary care physician participates in or a physician’s capacity or willingness to see Medicaid patients. Of the states that we visited, only Arizona required physicians to report their work load in full-time-equivalent terms and identified primary care physicians who participate in more than one plan and could be counted more than once. The four states also require plans to provide a full range of specialty services, even if this means beneficiaries must be referred to providers outside the plan’s network. However, because there are no established standards for specialists, these states have not specified the types and numbers of specialists to include in plan networks, making it difficult for these states to measure the adequacy of plan specialist networks before awarding a contract. Once plans have a contract, states can monitor the numbers and types of specialists participating in the network, but this does not necessarily indicate whether beneficiaries actually gain access to specialty care when they need it. Given the difficulties associated with gauging the adequacy of a provider network, the four states that we visited have taken additional steps to assess the adequacy of the medical care that beneficiaries enrolled in managed care receive. For example, each state has looked at aggregated statistics on the use of specific services. Some have found that, compared with fee-for-service, Medicaid managed care recipients were more likely to receive certain preventive and diagnostic services, such as childhood immunizations and cancer screenings. Arizona, Tennessee, and Pennsylvania also have invested in developing encounter data—the individual-level data on all services provided to all patients. Encounter data can enable states to conduct their own analyses on a wider array of services than is possible using aggregated statistics. These analyses allow states to examine patterns of care across plans, such as differences in service delivery by selected types of services, beneficiary groups, and providers. To date, Arizona has made the most use of its encounter data, including using them as the state begins to develop quality indicators. Tennessee’s early efforts primarily focused on developing and validating its encounter data; more recently, the state has begun to use these data to Page 3 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 assess service utilization patterns. Pennsylvania’s use of encounter data was even more limited. All four states also use data from plan-conducted clinical studies and state-conducted medical record audits to help assess patient care. Improved plan and state methodologies, however, could increase the usefulness of the data collected from these reviews. The four states that we visited also have sought to assess the adequacy of patient care by tapping into information provided directly by Medicaid beneficiaries enrolled in managed care, such as patient satisfaction surveys and data gathered from grievance processes. While it is important to gauge patients’ satisfaction with the care they receive, satisfaction data generally are not reliable measures of quality; most people lack the knowledge needed to adequately evaluate the appropriateness of the care they receive—or do not receive. In addition, newcomers to managed care may not fully understand how the system operates to effectively access services, advocate on their own behalf, or register dissatisfaction with their plan or provider. This is especially true for individuals with diverse language and cultural needs. Regardless, we found that if the states we visited improved certain methodologies for designing satisfaction surveys and stratified their survey and grievance data, they would have a better understanding of the needs and concerns of their Medicaid beneficiaries enrolled in managed care—especially those with special needs or chronic illnesses, who may experience problems in accessing services but whose numbers are too small to show up in analyses of broad-based data. Medicaid, a joint federal-state health financing program for the poor, Background provides health care for about 37 million low-income people.3 In fiscal year 1996, Medicaid expenditures accounted for more than 20 percent of state budgets. To help control expenditures and expand access to health care, 36 states have mandated enrollment for some portion of their Medicaid population in managed care programs. As of June 1995, nearly 14 percent of Medicaid eligibles were enrolled in capitated programs. Under a capitated managed care system, states pay contracted plans a monthly per-enrollee amount before services are delivered—a distinct departure from the traditional claims-based fee-for-service system in which providers are paid for each service as bills are submitted. In turn, the plans employ or subcontract with primary care physicians, who 3 Medicaid was established in 1965 as title XIX of the Social Security Act (42 U.S.C. 1396 et seq.). Medicaid is administered at the state level, with federal oversight by HCFA within the Department of Health and Human Services (HHS). Page 4 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 coordinate the delivery of health services.4 Some plans pay their subcontracted providers on a fee-for-service basis for care provided, while others pass certain financial risks on to providers by linking the providers’ revenues or profits to the total number of services provided to plan enrollees. While capitated managed care has strong cost-containment incentives, it also provides incentives for plans and providers to limit services—not only must plans and providers absorb all costs that exceed the capitation rate, they profit if the capitation rate exceeds their costs. Nationwide, most states initially implemented Medicaid capitated managed care programs by allowing beneficiaries to enroll on a voluntary basis in limited geographic areas. These programs were largely targeted to low-income families who received financial assistance under Aid to Families With Dependent Children (AFDC) and pregnant women and children who qualified for Medicaid. Increasingly, states are mandating beneficiary enrollment and expanding their programs to more geographic areas. In addition, they are beginning to include more populations with specialized needs, such as blind or disabled individuals who qualify for Medicaid under the Supplemental Security Income (SSI) program. As we reported in July 1996, 17 states had extended their Medicaid managed care programs to these more vulnerable populations.5 States must comply with certain federal statutory requirements for the development and oversight of their managed care programs. HCFA can waive some of these requirements—such as a beneficiary’s freedom to choose any provider—to enable states to restrict beneficiaries to the providers participating in a managed care network. Waivers also allow states to expand the scope of their programs to populations not otherwise eligible for Medicaid. These waivers are of two types: program or demonstration.6 Program waivers allow states to require beneficiaries to join a managed care plan, but beneficiaries are generally allowed to switch plans every 30 days. Demonstration waivers provide states with greater flexibility, and while they are more difficult to obtain than program waivers, they have been granted more frequently in recent years. States request demonstration waivers to establish mandatory programs that lock beneficiaries into one plan for periods of up to 12 months or to expand 4 Primary care physicians may be general internal medicine practitioners, family and general practitioners, pediatricians, or obstetricians and gynecologists. 5 See Medicaid Managed Care: Serving the Disabled Challenges State Programs (GAO/HEHS-96-136, July 31, 1996). 6 Program waivers are authorized under section 1915 of the Social Security Act. Demonstration waivers are also known as section 1115 waivers, after the section of the Social Security Act that authorizes them. Page 5 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 eligibility to uninsured populations. Table 1 compares various characteristics of the two waiver types. Table 1: Characteristics of Managed Care Programs Under Program and Program waivers Demonstration waivers Demonstration Waivers General characteristics Allow for waiver of certain Medicaid Allow for waiver of a broader range of requirements. Medicaid requirements. Waivers renewed for 2- to 5-year periods. Generally not renewed.a Generally used to establish primary care More recently used to establish broad case management programs and home-and changes in Medicaid programs. community-based service programs. Characteristics pertaining to capitated managed care Plans must comply with 25% requirement for Plans may enroll Medicaid patients private enrollment. exclusively. Full range of mandatory services must be Benefit package may be modified.b offered. Beneficiaries may be mandated to enroll in Beneficiaries may be locked in for up to 12 plan and can be locked in for no longer than months. 1 month.c a The Congress has authorized renewal of some demonstration waivers. b To date, only Oregon has been permitted to modify the benefit package for traditional Medicaid beneficiaries. Other states have been permitted to offer a modified package only to those newly eligible for Medicaid coverage under the demonstration. c Lock-in is up to 6 months for capitated plans meeting certain federal requirements. At the time of our review, Arizona and Tennessee had demonstration waivers for their mandatory statewide programs, which served both AFDC and SSI populations. Wisconsin had a program waiver for its mandatory program, which served only the AFDC populations in 5 of its 72 counties. In contrast, Pennsylvania had voluntary and mandatory managed care programs. The voluntary program—the larger program at the time of our review—served both AFDC and SSI populations in 13 of its 90 counties.7 This program required no federal waiver. To ensure that states comply with statutory and HCFA requirements, HCFA reviews state contracts with managed care plans. It also monitors state programs through independent evaluations and periodic reviews of state-submitted information on expenditures, medical services, and 7 At the time of our review, 475,000 beneficiaries were enrolled in Pennsylvania’s voluntary managed care program. With the February 1997 expansion of its mandatory program, 177,000 beneficiaries were participating in the voluntary program as of April 1, 1997; 66,000 are projected to be in the voluntary program by 1998. Page 6 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 enrollment data, which HCFA requires all states to report. The nature of HCFA’s requirements and oversight role depends on the waiver type. For example, under a demonstration waiver, HCFA develops terms and conditions that vary by state, depending on the provisions being waived. (For a more detailed discussion of federal regulations and HCFA requirements for waiver programs, see app. II.) Purchasers of managed care face a number of difficulties in ensuring Difficulties in enrollees receive the care that they need. In contrast to fee-for-service Monitoring Managed care—where the incentive is to oversupply services to increase Care Stem From revenues—capitated managed care, with its fixed payment system, contains incentives to provide fewer services to maximize short-term Limited Standards of profits. Assessing how well the care delivered matches beneficiary needs Care and Data on is difficult because few aggregate or population-based utilization standards or benchmarks on delivery of care patterns have been Service Utilization established for managed care. Benchmarks derived from providers and patients in the fee-for-service sector may not be appropriate since service utilization patterns are expected to change under managed care. Even where standards of care do exist for selected conditions or procedures, such as for prenatal care or childhood immunizations, monitoring the actual delivery of such services in capitated managed care is difficult because data on service utilization do not flow as readily as in a fee-for-service environment. In fee-for-service care, data on service delivery are captured in individual claims, which are submitted for payment as services are provided. In managed care, however, purchasers prospectively pay plans a monthly fee for services not yet provided. Unless plans capture service data in another form, it is difficult to identify the services actually provided. Without specific data collection requirements, providers may lack the incentive to accurately report individual-level data on all services provided since payment is not linked to documentation of the care provided. To learn more about what transpires in the delivery of managed care services—and to help ensure that the health care services they contract for are appropriately provided—private and public purchasers also rely on other measurement tools and data sources. Among these are reviews of patient medical records and surveys of patients on the care they receive from plans. Reviews of medical records can help purchasers assess the care provided to individual patients. These assessments, however, are time-consuming and costly and are generally undertaken for small Page 7 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 numbers of patients. Information from patients, while more accessible than medical records, can be problematic as well. Patients typically are not in a position to know what specific care or services they need for a given condition and often cannot assess the appropriateness of the care they receive—or do not receive. In addition, patients new to managed care may confuse differences in the way the system is meant to operate with deficiencies in the care provided. Problems associated with obtaining meaningful patient survey information may be even more pronounced for those in the surveyed population with unique language or cultural needs or who are unaccustomed to receiving routine health care in a structured system. Educating and informing prospective and newly enrolled beneficiaries about managed care and helping them learn how to use the system—as some states have done in their Medicaid managed care programs—can mitigate these problems.8 Most states are grappling with these and other issues associated with adapting and developing systems and processes for managed care—a relatively new health care environment. According to some experts, many states are struggling to maintain the staff needed to establish and oversee their programs, since frequent turnover of staff with managed care expertise is common. It is not surprising, then, that states are at various stages in their program development and monitoring efforts. The four states that we visited have taken a number of steps to overcome these various challenges and improve the odds that their money is well spent in their managed care programs. These states have established contract requirements that aim to ensure that participating managed care plans have the capacity to provide adequate care to enrolled Medicaid beneficiaries. Prominent among these requirements are standards for plans’ provider networks. In monitoring participating plans, these states obtain information to assess actual services delivered from various sources, including plan-collected and -submitted data, state reviews of patient medical records, and beneficiaries’ reports on their experiences. Before a contract is awarded, managed care plans must demonstrate that Measures to Assess their provider networks are sufficient to meet the anticipated needs of Beneficiaries’ Access enrolled Medicaid beneficiaries. In an attempt to measure provider to Care Are Still network sufficiency, the states we visited—Arizona, Pennsylvania, Tennessee, and Wisconsin—have focused on quantitative or other Evolving measures related to primary care physicians and specialist care. But just 8 See Medicaid: States’ Efforts to Educate and Enroll Beneficiaries in Managed Care (GAO/HEHS-96-184, Sept. 17, 1996). Page 8 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 as there are few standards for health service utilization, there are few standards for what constitutes a sufficient provider network. Three of the four states have established a specific number of primary care physicians that a plan must have, and all require plans to provide a full range of specialty services. The states also have relied on criteria that measure beneficiaries’ ability to reach their primary care physician within a reasonable time, in terms of maximum travel distances and waiting times. After contract award, the states use various monitoring techniques to determine the extent to which provider practices are in fact open to Medicaid beneficiaries. These measures, however, do not necessarily ensure that beneficiaries have access to the care that they need. Whether these measures provide meaningful information on beneficiary access is largely dependent on whether state monitoring efforts are independent and systematic and go beyond plan-reported, paper-based indications of compliance. Compliance With Primary One criterion that states have established in an effort to ensure a sufficient Care Physician provider network relates to the availability of primary care physicians, Requirements Provides expressed as a ratio of enrolled beneficiaries per primary care physician.9 At the time of our review, Arizona and Tennessee used a maximum Incomplete Information patient-to-primary-care-physician ratio of 2,500 to 1, as required by the About Network Adequacy conditions of their demonstration waivers, and Pennsylvania required plans to meet a ratio of 1,600 to 1.10 Wisconsin did not have specific contractual requirements for plans but looked for a ratio of approximately 1,200 to 1. To monitor plan compliance with these ratios, the states require plans to submit updated provider listings either annually, to coincide with contract renewal, or as frequently as monthly. The states also require plans to report all changes to the network as they occur and to note in their provider directories given to beneficiaries those providers who currently do not accept new patients.11 The states that we visited have found that plans in their managed care programs have complied with their patient-to-primary-care-physician ratios. But compliance with these ratios may not indicate actual physician capacity or Medicaid beneficiaries’ access to care. We believe that the 9 In 1994, there was one primary care physician for every 1,173 United States citizens (based on GAO analysis of the HHS Area Resource File). 10 Beginning in October 1997, Arizona will require plans to have maximum patient-to-primary-care-physician ratios of 1,800 to 1 for adults and 1,200 to 1 for children under age 13. 11 Typically, plan directories are updated annually; consequently, beneficiaries must directly contact the managed care plan for more current information on physician availability. Page 9 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 number of primary care physicians and their availability to treat patients may be overstated for two reasons. First, if the state reviews a plan’s network capacity by looking at only that plan rather than looking at all plans collectively, providers who participate in more than one plan may be counted more than once. Second, if the state does not use full-time- equivalency data to determine network capacity, network physicians’ other lines of business, such as treating Medicare or privately insured patients, are not taken into account. Of the states that we visited, only Arizona reviews provider participation across plans and assesses provider capacity in full-time-equivalency terms. For patient-to-primary-care-physician ratios to be an effective measure of patient access to care, states must also ensure that plan physicians are actually available to treat Medicaid beneficiaries. Ratios simply indicate the number of physicians that have contracted with a plan; they do not indicate the number of patients physicians are willing to treat and the extent to which physicians actually provide services. Arizona and Pennsylvania independently assess the extent to which physician practices are open to Medicaid beneficiaries by periodically or randomly calling physician offices to determine whether they are accepting new Medicaid patients. When done on a systematic basis, these checks can better ensure that provider practices are open to Medicaid beneficiaries. To improve oversight of its provider networks, Arizona increased plans’ quarterly provider reporting requirements in September 1996. Previously, plans were required to report provider names and the number of beneficiaries seen by providers. Plans must now submit additional data on provider access, such as the maximum number of Medicaid beneficiaries that a provider will accept, the total number of beneficiaries currently assigned to a plan, the providers who are accepting new members, specialty services that are available, and foreign languages spoken by providers. Specialist Mix Is Not Patients often require more specialized care than their primary care Specified in Plan Contracts physician can provide—such as oncological, urological, or pediatric subspecialty care. Yet, assessing beneficiary access to such care is even more difficult than assessing access to primary care physicians. And as states move more beneficiaries with special needs or chronic conditions into managed care, ensuring beneficiary access to appropriate specialty services will become even more critical. Page 10 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 Assessing the availability of specialty providers within a network is problematic for at least two reasons. First, there are no criteria or standards—in fee-for-service or managed care—for the number and mix of specialists needed to serve a population or for when and how often beneficiaries should be referred to specialists. Second, some specialists—especially those that are used infrequently—often are not included in the network and can only be accessed outside the network. Since these specialists are not in the network, states cannot readily assess their availability to beneficiaries when needed. The four states that we visited require that contracted plans provide a full range of specialty services, even if this requires referring beneficiaries to providers outside the plan’s network. Without recognized standards, requirements for specialists are often vague and expressed in terms of an objective—such as, “provide access to necessary specialty care”—that cannot be measured before the contract award. While the four states do not specify in the contracts the types and numbers of specialists that plans must include in their networks, they do count the number and type of specialists available in any one plan. This is done by reviewing the listing of providers in a plan network during the contracting and contract renewal processes. Each of these states also reviews plan listings periodically and requires plans to report all changes in their specialty networks. For example, Tennessee officials told us that the state uses a zip-code-based computer program to check the location of specialists each quarter—or more frequently if inquiries or questions come up on a particular provider type. States then rely on the judgment of their experienced contract staff—which may include health care professionals—to determine whether the plans’ specialist networks are adequate. This type of review, however, does not inform states of the specialist services that Medicaid beneficiaries may use or request, especially if the services involve out-of-network referrals. In the absence of accepted standards for specialty care, states have relied on other information sources—such as data on service utilization and beneficiary satisfaction—to help them monitor the appropriateness of care provided. Maximum Travel and As additional criteria for assessing the adequacy of provider networks, Waiting Requirements Are states commonly stipulate the maximum time and distance beneficiaries Additional Criteria for must travel to their primary care physician’s office, the number of days they must wait for an appointment to see their physician, and the time they Assessing Beneficiary wait in the physician’s office to be examined. These measures were Access to Care Page 11 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 developed on the premise that certain factors—such as lengthy travel times and distances, a physician’s failure to schedule appointments in a timely manner, and excessive wait times in the physician’s office—can discourage patients from seeking care. In general, the states we visited established maximum travel requirements of 20 miles or 30 minutes for urban areas and 30 miles or 30 minutes for rural areas. For their specialist networks, the four states do not have or grant some exceptions to travel requirements, since significant portions of their Medicaid population can reside in rural areas where certain types of specialty care may not be available. These states also require providers to meet state-established standards for scheduling appointments and attending to beneficiaries in a timely manner.12 To determine whether plans are complying with the time and distance travel requirements, each state we visited reviews documentation on the location of provider sites prior to contracting with a plan and on a periodic schedule after contracting. Arizona and Tennessee, for example, use a computer-based zip-code program. This allows them to determine whether the locations of the primary care physicians and other providers in the network, such as dentists and hospitals, meet contractual time and distance requirements. In addition, each state requires plans to provide transportation for beneficiaries who require medical attention and who cannot get to their provider’s location on their own, such as the elderly. These four states use various approaches—involving plan-provided data or independent checks—to monitor provider compliance with wait-time requirements. For example, each state requires plans to include appointment standards in its subcontracts with providers and to review logs to ensure that providers comply. During their periodic compliance reviews, the states review the plans’ procedures to ensure that providers have complied with the standards; the states also review selected sign-in and appointment books. For example, Tennessee found during one quarterly review of plan networks that some plans were not complying with the contractual travel requirements for their dental networks. According to state officials, the plans were notified of the deficiencies, and most took actions to resolve them. In at least one case, a plan did not take action and the state withheld 10 percent of the plan’s capitation payment 12 Generally, the four states’ requirements to schedule appointments with primary care physicians are same-day appointments for emergency care, within 24 to 48 hours for urgent or sick care, and 2 to 3 weeks for routine preventive care. Requirements for in-office waiting times are generally 30 to 45 minutes. Some states also specify appointment standards for specialty referrals and dental services, as well as maternity, mental health, and substance abuse services. Page 12 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 until it was in compliance. States also have independently assessed provider compliance by periodically or randomly calling provider offices to schedule appointments. However, such periodic calls may not be systematic enough to provide the information required to identify problems. As one means of gaining additional insight into beneficiary experience with providers, Arizona and Tennessee have included questions in their beneficiary satisfaction surveys that are specifically related to wait times. Recognizing the challenges associated with measuring the capacity of plan Systems for Assessing provider networks and their ability to ensure adequate care, states can use Adequacy of Medical other accountability measures and processes to assess the actual care Care Provided Can Be Medicaid beneficiaries receive. To identify average levels or recent changes in the use of services by beneficiaries enrolled in managed care, Further Improved states can use plan reports of utilization statistics, which summarize selected services provided to specific populations. For more extensive analyses, states can use encounter data, which are individual-level data for each service provided to each enrollee. Encounter data allow states to identify the care received by any individual and the provision of any procedure. In addition, states can conduct, or require plans to conduct, other analyses—such as clinical studies and medical record audits—that review the full medical records to assess the appropriateness of the care received by a sample of beneficiaries.13 The four states that we visited have taken different approaches in implementing their data collection methods. For example, Pennsylvania and Wisconsin have relied primarily on collecting and using aggregated utilization statistics to measure the adequacy of care. Arizona and Tennessee also have required plans to collect and submit encounter data. However, both states have found that developing reliable and useful encounter data has required lengthy and continuing investment. As a result, these two states have primarily used their encounter data to identify services that may be over- or underutilized and health care areas on which to focus their studies and audits. Arizona has begun to use its encounter data to develop standards for measuring the quality of the care provided.14 For all four states, we found that certain improvements in their 13 For specific federal regulations and guidance related to these accountability measures and processes, see table II.1. 14 Arizona further uses its encounter data to estimate the cost of serving beneficiaries in each county and to set capitation rates by county. See Arizona Medicaid: Competition Among Managed Care Plans Lowers Program Costs (GAO/HEHS-96-2, Oct. 4, 1995). Page 13 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 methodologies for conducting clinical studies and medical record audits could increase the usefulness of these reviews. States Challenged to Although utilization statistics and encounter data both capture patient use Develop Effective of services, utilization statistics are summary data that are generally relied Utilization Statistics and on to show the frequency with which a service is accessed by a specific population. As such, there are several difficulties associated with using Encounter Data for these summary statistics in assessing services provided under managed Monitoring Managed Care care. For example, utilization statistics are often compiled for only specific types of services, such as mammograms or childhood immunizations. Consequently, utilization statistics cannot be used to determine the full range of services that beneficiaries may receive. For example, in a prior study, we found that analyses of utilization statistics on early and periodic screening, diagnosis, and treatment (EPSDT) could not determine the actual number of EPSDT-eligible children who received required screenings or whether children with post-screen referrals actually received follow-up diagnosis and treatment.15 Utilization statistics also have generally not been used to assess the performance of individual providers. To obtain utilization statistics for other sets of services or services supplied by individual providers, states would need to modify their reporting requirements—which could be problematic for plans if they have to alter their data systems or provider reporting requirements to meet new state requirements. Unlike utilization statistics, which aggregate service use by a population, encounter data document all services that individual patients receive. While encounter data have certain limitations, these data provide states more flexibility to detect problems in beneficiary care by identifying patterns of service use by individual beneficiaries and services provided by individual providers. For example, these data can be used to assess the participation of any provider or group of providers and analyze patterns of care for specific diagnoses or procedures. With encounter data, states also can explore service delivery beyond what is captured by utilization statistics. These statistics allow plans and providers to “teach to the test,” that is, focus on service delivery areas that they know will be measured, perhaps to the exclusion of other services. Since encounter data encompass all services for all beneficiary populations, all plan-provided services are subject to state review, and the services being reviewed can 15 See Medicare Managed Care: More Competition and Oversight Would Improve California’s Expansion Plan (GAO/HEHS-95-87, Apr. 28, 1995). Page 14 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 be changed periodically without adding to the administrative burden of plans. Certain analyses of encounter data, however, are currently constrained by the limited number of benchmarks or standards against which states can measure the care provided for certain conditions. For example, while recognized standards exist for prenatal care and childhood immunizations, many diagnoses have multiple alternative treatments or therapies. Vocal chord stress, for instance, might be treated with medication, voice therapy, or surgery. Individual-level encounter data cannot show the appropriateness of many treatments provided to beneficiaries, nor can they provide information about beneficiaries who do not seek treatment. Regardless of these constraints, encounter data are valuable in that they support a wider array of analyses than do utilization statistics. Moreover, they provide the potential for supporting even more analyses as additional benchmarks are developed. To ensure that the utilization statistics and encounter data are usable, data collection standards must be established and plans must be monitored for compliance with these standards. Validating the accuracy and completeness of encounter data requires additional measures—especially in a fully capitated system where provider payment is not directly linked to the documentation of each service provided. Although the data collection efforts in the four states that we visited varied considerably—due, in part, to the federal guidelines and requirements associated with their waiver type—overall, their use of encounter data to identify problems in beneficiary care has been minimal. Wisconsin requires contracted plans to collect and submit on a quarterly basis utilization statistics on 59 types of health care services, including maternal and child health, mental health, and emergency room visits.16 Using fee-for-service experience as the benchmark, the state analyzes—and publicly discloses in periodic reports—plan-submitted statistics on certain services, such as childhood immunizations; lead testing; mammograms; and dental, vision, and hearing examinations. For example, the state found that certain preventive services, such as Pap smears and childhood immunizations, increased in managed care, whereas emergency room visits decreased—a redistribution of service settings that is consistent with the goals of managed care. In addition, the state found 16 For each indicator for which summary statistics are reported, plans also report a complete individual-level patient history file, which documents all services that the patient has received. This system allows Wisconsin to determine the actual number of beneficiaries who receive a service and to conduct more extensive analyses on areas of concern. Page 15 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 that beneficiary use of dental services was less in managed care than in fee-for-service care and took action to improve beneficiary access to these services. In relying on utilization statistics rather than on encounter data, however, Wisconsin cannot easily or independently assess the care that Medicaid beneficiaries receive beyond the 59 services. Such analyses would require encounter data for 100 percent of the services provided, which the state has opted not to collect in an effort to limit the administrative burden that collecting encounter data can place on the state and the plans. Pennsylvania requires each plan participating in its voluntary managed care program to collect both utilization statistics and encounter data. But the state does not prescribe a method for collecting and validating the encounter data and does not require plans to routinely submit these data to the state for review and analysis.17 In April 1996, the state required all plans to submit EPSDT data for a specific time period to respond to concerns regarding Medicaid beneficiaries’ access to this service. As part of their demonstration waiver requirements,18 Tennessee and Arizona require plans to collect and submit encounter data on 100 percent of services, potentially giving these states the greatest flexibility in their analyses. Tennessee’s efforts to date have focused on providing technical assistance to help plans configure their computer systems and data elements to ensure accuracy. HCFA is also providing the state with technical assistance in developing methodologies to analyze its encounter data. Despite these start-up issues, the state has begun to analyze its encounter data and to review provider practice patterns to identify potential over- or underutilization of care. Of the states we visited, Arizona has the most comprehensive system for validating and analyzing plan-generated encounter data and has made the most use of these analyses to guide state actions. To ensure that its encounter data are accurate, complete, and timely, the state conducts two validation studies of plan-submitted encounter data each contract year. The first study takes a random sample of medical records and manually compares these records with the encounter data submitted by the plans. The second study compares the number of inpatient maternity 17 Pennsylvania requires plans to submit encounter data for Medicaid beneficiaries enrolled in the state’s mandatory managed care program. At the time of our review, 75,000 beneficiaries were enrolled in mandatory managed care. Nearly 500,000 are now enrolled as a result of the state’s February 1997 expansion. 18 HCFA requires all states with a demonstration waiver to collect encounter data for 100 percent of services. Page 16 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 hospitalizations with newborn reports from hospitals. The state also uses its encounter data to evaluate individual plan performance. Furthermore, beginning October 1997, Arizona will require plans to compile encounter data on specific measures, including mammography screening, cervical cancer screening, children’s dental services, and well-child care.19 In essence, Arizona will require plans to use their encounter data to compute certain utilization statistics. This should make the information available to the state sooner. Arizona also is creating a quality management system that uses outcome-based standards and, over the past several years, has been creating a baseline for these standards. The system will seek to produce data that could indicate whether preventive care—such as dental visits for children, mammograms, and Pap smears—prevent more serious health problems. In the three states with mandatory managed care programs, plans may be sanctioned if they do not comply with utilization statistics or encounter data requirements. Tennessee assesses a 10-percent withhold on capitation payments for each month that a plan does not comply with data submission requirements. If the problem is not corrected within 6 months, the state keeps the withheld funds. Arizona imposes a financial penalty based on the number of data errors identified in its plans’ encounter data. Wisconsin may impose financial penalties if plans do not meet accuracy and timeliness requirements. Over the last 2 years, the amount collected in penalties by Tennessee and Arizona has been minimal, and Wisconsin has not assessed any penalties. More Targeted Clinical Recognizing the limitations of utilization statistics and encounter data, Studies and Medical each of the four states that we visited uses reviews of samples of Record Audits Are Needed individual patient’s medical records to determine whether appropriate and adequate care has been provided. Each state requires contracted plans to to Assess Impact of conduct at least one clinical study each year. A clinical study focuses on Medicaid Managed Care certain aspects of health care services, such as maternal health, to answer questions about the quality and appropriateness of care that has been provided. Each state also conducts its own medical record audits—as required by federal regulation—either internally or through a contracted external review organization, such as the State Peer Review Organization. Medical record audits also have the potential to assess the appropriateness 19 The required measures are based on Medicaid Health Plan Employer Data Information Set (HEDIS). Medicaid HEDIS provides guidance that states may use to measure, improve, and report on health plan performance. Medicaid HEDIS was the collaborative effort of representatives from state Medicaid agencies, managed care plans, the National Committee on Quality Assurance (NCQA), professional health and welfare organizations, beneficiary advocacy groups, HCFA, the U.S. Public Health Service, and others. Page 17 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 of the care provided as well as determine whether patients’ medical records properly document the health care and services that they received. The states’ current approaches to conducting these studies and audits could be improved to yield more useful findings. For example, we found that clinical studies often focus on the Medicaid population as a whole and less frequently target populations that may not fare as well under managed care, such as people with disabilities or chronic illnesses.20 In addition, the states’ sample sizes for its medical record audits appear to be insufficient to enable states to draw conclusions about the adequacy of the documentation of beneficiary care, particularly for certain populations or conditions too small to show up in pure random samples. Managed Care Plans’ Clinical Pennsylvania, Tennessee, and Wisconsin allow plans to select a topic to Studies study from one or more health care areas that the state identifies.21 In selecting study areas, the three states rely on the professional judgment and experience of staff in the state Medicaid agency and on independent expert opinion the states may seek. While studies of this nature can provide states with valuable information on plan performance and a baseline for evaluating subsequent plan performance, we found that the plan-conducted studies had several limitations. For example, only Wisconsin specified study areas that included conditions for which beneficiaries might require more care than the general Medicaid population, such as children with special care needs. Problems in the care for such beneficiaries may not be detected in studying samples of the overall Medicaid population because, relative to the larger population, their numbers are too small. In addition, allowing plans to select their clinical study topics gives them the latitude to select a topic where improvement may be needed, but it also allows them to select a topic that would yield positive results about plan performance. Finally, allowing plans to select topics to study does not enable states to compare results across plans for certain conditions or topics. We also found that the states we visited conducted limited reviews of plan methodologies for clinical studies. Pennsylvania, for example, reviews plan methodologies during periodic quality assurance audits—after the plans’ clinical studies have been completed. Wisconsin recently revised its process for reviewing plan methodologies. Previously, the state required 20 See “Differences in 4-Year Health Outcomes for Elderly and Poor, Chronically Ill Patients Treated in HMO and Fee-for-Service Systems: Results From the Medical Outcomes Study,” The Journal of the American Medical Association, Vol. 276, No. 13 (Oct. 2, 1996). 21 Arizona requires plans to submit a topic for state approval. Page 18 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 plans to submit a one-paragraph description of their methodologies, on which the state would base its approval. For the contract year beginning October 1996, Wisconsin began to require plans to submit detailed descriptions of the study topic and the methodology for conducting the study. To validate the results of the plans’ clinical studies, Pennsylvania and Tennessee review a sample of patient records during compliance reviews for their annual contracts. However, these states generally pull only a handful of records to verify the clinical study results. States’ Medical Record Audits Medical record audits document problems with patient medical records, such as incomplete patient histories, lack of indication of follow-up care, and illegibility and unavailability of records. These audits also can help identify underlying causes of service delivery or access problems. If a plan’s medical records are inadequate or indicate a service delivery problem, a state can require the plan to take corrective action. All four states that we visited use utilization statistics, encounter data, or both to focus their audits of Medicaid beneficiary medical records. For example, Wisconsin found through its analyses of utilization statistics that use of dental services was infrequent and, therefore, conducted a medical record audit to determine why Medicaid beneficiaries were not getting dental care. While these states’ periodic audits of beneficiary records have revealed weaknesses in the documentation of beneficiary care, we found that their audit methodologies often yielded results that were not statistically valid or may not have been sufficient to identify problems experienced by different groups of Medicaid beneficiaries. Specifically, the samples of records that the states used to conduct their audits generally were not stratified by specific conditions or populations—which could result in an underrepresentation of Medicaid beneficiaries with special needs—and the sample sizes may not have been adequate to identify areas that warrant further investigation. For example, Wisconsin uses a random sample of about 2 percent of cases for medical record audits. Arizona bases its sample sizes on the number of beneficiaries enrolled in a plan, with the sample sizes ranging from 30 to 100 patient records. Although the state agrees these sample sizes are not statistically valid, it believes that these audits, when combined with other periodic on-site reviews, are sufficient to identify best practices as well as problem areas to target for corrective action. Page 19 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 In assessing the performance of managed care plans, indicators of States Could Learn beneficiary satisfaction can complement other analyses of provider More From Improved network capacity and the services provided. To gauge the extent to which Design and Analysis beneficiaries are satisfied with Medicaid managed care and their managed care plan, the states that we visited review the results of beneficiary of Data on Beneficiary satisfaction surveys, grievance data, and the rates at which beneficiaries Experience With choose to switch plans or, if enrolled in a voluntary program, leave managed care altogether. While analyses of such data have helped these Managed Care states identify problems that Medicaid beneficiaries have with managed care, such as difficulty in scheduling appointments or accessing specialists, certain improvements in the design of their surveys and data collection methods could enhance the usefulness of their analyses. Improved Beneficiary Although the results of patient satisfaction surveys may not be the best Survey Designs Could Yield indicator of quality care, periodic satisfaction surveys—administered More Useful Data statewide or at the plan level—can help measure the degree to which Medicaid beneficiaries are happy with the providers and services offered in their managed care plan. To ensure that survey findings are reliable and useful in identifying areas that need systemwide improvement, the survey design and process must be methodologically sound, however.22 In addition, the survey must be designed to address several difficulties inherent in surveying the Medicaid population, such as effectively administering a survey instrument in multiple languages and overcoming historically low response rates. Plan-Conducted Beneficiary Pennsylvania, Tennessee, and Wisconsin require plans to conduct periodic Surveys surveys to assess beneficiary satisfaction with network providers and services. Pennsylvania and Tennessee review the plans’ survey methodologies during their annual contract compliance reviews, but they do not routinely examine the survey methods before the surveys are conducted. Wisconsin requires plans to describe in detail the methodology they intend to use before the contract is awarded. Once the contract is awarded, the state reviews and approves each plan’s survey methodology and instrument before it is administered. However, none of these states prescribes a methodology for conducting satisfaction surveys. Consequently, these states cannot compare survey results across plans. 22 Medicaid HEDIS provides states with technical guidance on designing a satisfaction survey for Medicaid beneficiaries. This guidance includes how to identify topics for surveys and the trade-offs between phone and mail surveys, such as data collection costs, time required for data collection, response rates, sample sizes, and overall quality of data. In addition, HHS is developing beneficiary satisfaction surveys for both general and special needs populations, which should be of further help to the states once completed. Page 20 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 Statewide Surveys Statewide surveys allow states to compare results across plans for various access and quality measures, such as use of specialty services, average waiting time for physician office visits, and beneficiary perception of the quality of care provided. As a condition of its demonstration waiver, Tennessee is required to conduct statewide satisfaction surveys annually. Arizona—on its own initiative—will soon complete a statewide beneficiary satisfaction survey. Pennsylvania and Wisconsin plan to conduct surveys as part of their planned program expansions. Since implementing its managed care program in 1994, Tennessee has conducted two annual statewide surveys. HCFA required that the state’s beneficiary satisfaction survey include questions on referrals to specialists, average waiting time for physician office visits, and reasons for disenrollment. Tennessee has used these surveys to identify trends in service use. For example, the state found that between 1993—the last year of fee-for-service care—and 1995, hospital use decreased 6 percent, and visits to doctors’ offices increased 8 percent—a redistribution of service settings that is hoped for in managed care. In addition, its 1995 survey showed that 75 percent of enrollees were satisfied with the care they received through their managed care plan compared with a 61-percent satisfaction rate in 1994. Although this showed improvement in overall beneficiary satisfaction, the state also could use these survey results to further explore remaining causes of dissatisfaction for one fourth of the beneficiaries. Arizona is issuing the results of its third statewide beneficiary survey in spring 1997.23 The comprehensive survey included questions on the use of health services, time elapsed in getting an appointment with a physician and in waiting in the physician’s office, problems with access to specialty care, and an overall rating of the plan and quality of care. The state conducted telephone interviews, which allowed interviewers to verify that they were speaking with the appropriate beneficiary and to ask appropriate follow-up questions. Of the current 450,000 beneficiaries enrolled in the program, over 14,000 were interviewed. The state intends to use the survey results to provide feedback to plans. 23 Arizona’s first survey, administered in 1989 by an advocacy group, was a comprehensive look at all health services used. The second, administered in 1995 by the Arizona State University Survey Research Laboratory, was a focused survey on prenatal and maternity care. Page 21 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 Targeted Analyses of To satisfy a federal requirement for operating a Medicaid managed care Grievance Data Help program, states must ensure that participating plans have an internal Identify Areas That Need grievance process through which beneficiaries can report their dissatisfaction with plan providers, services, and benefits. Through these Improvement grievance processes, the states that we visited have been able to identify and address a number of beneficiary concerns. Some states also look at individual beneficiary grievances to identify specific and localized problems. Other opportunities for analyzing grievance data, however, exist. For example, monitoring the volume of grievances filed—particularly across plans—could reveal previously unidentified problems. Even a low number of grievances could indicate that beneficiaries do not understand the grievance process. Arizona requires beneficiaries to submit grievances directly to the plan. Pennsylvania and Wisconsin have no such requirement but encourage this practice; they also allow beneficiaries to submit grievances directly to the state. Tennessee requires beneficiaries to submit grievances directly to the state. After receiving a grievance, the plans must provide beneficiaries with resolution and action in a reasonable time frame, ranging from 30 to 90 days. If a beneficiary is not satisfied with a plan’s decision, the beneficiary can appeal to the state. Most grievances are resolved at the plan level, however, according to officials in the states we visited. At a minimum, the plans that directly receive grievances are required to periodically report to the state the number and type of grievances they received—such as denial of requests for out-of-plan services or difficulty in locating a provider or in scheduling an appointment—and the status of these cases. To probe beyond such aggregated information, which may mask specific or localized problems, Arizona and Wisconsin informed us that they review each grievance that plans receive. In addition to the grievance process, each state has developed other means for beneficiaries to voice their concerns. For example, Tennessee has a toll-free information hotline to respond to beneficiary questions and concerns. Tennessee also sponsors hotlines run by advocacy groups to answer questions posed by beneficiaries with special needs, such as persons with acquired immune deficiency syndrome (AIDS) or human immunodeficiency virus (HIV), hemophiliacs, and persons with disabilities, as well as the general Medicaid population. In addition to state-run hotlines, Wisconsin requires each plan to have a beneficiary advocate who serves as a liaison between the state, the plan, and the beneficiary. The plan advocate identifies major areas of concern, such as lack of access to Page 22 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 mental health care, and works with the plan and the state to correct the problem. This can obviate the need for beneficiaries to register grievances. States’ Analyses of Beneficiaries who disenroll from a managed care plan may do so because Disenrollment Data Could of dissatisfaction with the care they receive through the plan. Therefore, Help Identify Problem collecting and analyzing data on disenrollments can provide important insights into plan performance. In a voluntary program, such as the one we Areas visited in Pennsylvania, beneficiaries can switch plans or return to fee-for-service care. In mandatory programs—such as those in Arizona, Tennessee, and Wisconsin—beneficiaries can switch plans during open seasons, which occur every 6 or 12 months.24 The states we visited, however, generally do not conduct routine disenrollment studies. According to officials in these states, they would conduct a disenrollment study if a significant number of disenrollments were detected.25 They believe that disenrollments—especially in low numbers—could signify a number of occurrences other than beneficiary dissatisfaction or problems with the plan. For example, in 1992, Arizona conducted a disenrollment study and found that most of the beneficiaries who changed plans during open enrollment—which was less than 5 percent of all beneficiaries in managed care—did so for reasons other than plan dissatisfaction. Specifically, the state found that some beneficiaries disenrolled because they wanted to continue to see a provider who was no longer in their plan’s network. Others switched to have all family members in one plan, and still others wanted to enroll in a plan where provider location was more convenient. Unless it sees a substantial change in enrollment rates during an annual open season, Arizona has no plans to conduct another study. More analyses of these disenrollment data—even if the rate at which beneficiaries leave or switch plans is low—could reveal significant problems. Disenrollments concentrated in an area or among people having similar needs, such as people with AIDS, may indicate a potential problem in a plan. Also, any plan having higher disenrollment rates than other plans may merit scrutiny to determine the reason. 24 Under a demonstration waiver, states can obtain federal authorization to require beneficiaries to remain enrolled in a specific health plan for 12 months. Under certain circumstances, such as relocation, beneficiaries can change plans at other times. 25 In 1995, disenrollment rates in Arizona and Tennessee—the two demonstration waiver states where enrollment only changes annually—were 4 and 6 percent, respectively. Page 23 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 In view of the billions of dollars that are being paid prospectively to Observations managed care plans and the questions about the degree to which managed care is meeting the health care needs of Medicaid beneficiaries, there is a new demand for public accountability. The continuing trend toward expansion of mandatory, capitated Medicaid managed care programs requires that states have the ability to adequately oversee their contracts with health plans and ensure that states get what they are paying for. However, developing systems to hold plans accountable for ensuring that Medicaid beneficiaries receive the care that they need has been a challenge for states—especially since there are few benchmarks and standards against which states can measure beneficiary access to network providers and the appropriateness of the care provided. The four states that we visited have made progress toward developing accountability measures to ensure that beneficiaries have access to quality care. As they expand or refine their Medicaid managed care programs, these states continue to scale the steep learning curve to becoming an effective purchaser of managed care. Yet, to instill greater public confidence that managed care can effectively and efficiently meet the health care needs of Medicaid beneficiaries, more effort is needed. For example, to varying extents, these states could improve their methodologies for collecting and analyzing data—especially encounter data—on beneficiary care. They could better target their clinical studies, medical record audits, beneficiary satisfaction surveys, and reviews of grievance data on specific services and beneficiary groups—particularly those with special needs or conditions whose numbers may be too small to show up in broad-scale surveys or studies. The need for these improvements takes on even greater importance for those states planning to expand their managed care programs to other geographic areas or populations, such as people with disabilities or other special needs. We provided a draft of this report to the Administrator, HCFA. We also Agency Comments provided a draft to Medicaid officials in each of the four states we visited and Our Evaluation and to independent experts and researchers from the Center for Health Care Strategies, Medical College of Virginia, and National Academy for State Health Policy. Each provided technical or clarifying comments, which we incorporated as appropriate. In addition to technical comments, each state informed us of recent or planned initiatives for ensuring plan and provider accountability in their programs. For example, in February 1997, Pennsylvania implemented Page 24 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 HealthChoices—a mandatory managed care program projected to serve more than a half million Medicaid beneficiaries in the state by 1997. With this new program, Pennsylvania plans to improve those accountability measures used under its voluntary program or adopt new accountability measures, such as disenrollment studies. HCFA, Arizona, and Wisconsin also noted that the draft did not discuss all accountability or quality assurance measures that were in use at the time of our review and were concerned that the account of state efforts was incomplete. In the final report, we state that the focus of our work was on 10 key measures or processes that states commonly use to assess plan accountability—not on each state’s entire quality assurance process. Arizona and Wisconsin also commented that the draft report did not reflect in all respects their experience with managed care. For example, they disagreed with our premise that since managed care plans receive prospective capitated payments, there is a financial incentive to limit or not provide needed services. HCFA echoed this comment. The two states pointed out, for example, that the use of certain preventive services increased when they moved to managed care. In response to their comments, we revised the report to include examples of their experience in the changing patterns of care under managed care. Nevertheless, in a prepaid capitated system, the incentive remains to provide fewer services in order to maximize short-term profits, as the HCFA Administrator recently testified before the Senate Committee on Finance. Concern about plans and providers having a short-term focus is exacerbated by the fact that significant numbers of beneficiaries frequently gain and lose Medicaid eligibility within a short period of time. Arizona also made a number of comparisons between managed care and fee-for-service and suggested that our report include such comparisons. The purpose of our report was not to weigh the merits of one system against those of another. Rather, we set out to identify potential problems Medicaid beneficiaries may have in accessing services through managed care and state efforts to address these access issues. Several reviewers, including HCFA, agreed with our conclusion that certain measures of physician capacity do not adequately ensure beneficiary access to care. It was suggested that we report on other important criteria states use to assess the adequacy of provider networks—specifically, beneficiary travel and waiting times. The final report reflects additional information on this issue. Page 25 GAO/HEHS-97-86 Medicaid Managed Care Accountability B-270335 Finally, the experts we consulted generally agreed with the accuracy and comprehensiveness of our presentation of the issues. They also emphasized that the transformation of existing state systems and processes to an effective managed care program—especially one with meaningful oversight mechanisms—requires great change accompanied by continuous refinements and adaptations. Each state—with varying levels of experience with managed care, resources, and in-house expertise—understandably approaches this evolutionary process with varying strategies and time frames. Even as states confront their many challenges in implementing managed care, strong and consistent accountability systems remain integral to their success in meeting the needs of Medicaid beneficiaries. This perspective is more fully reflected in the final report. As arranged with your office, unless you announce its contents earlier, we plan no further distribution of this report until 30 days after the date of this letter. At that time, we will send copies of this report to the Secretary of Health and Human Services, the Administrator of HCFA, state officials in the four states we visited, appropriate congressional committees, and other interested parties. We will also make copies available to others upon request. Please contact me on (202) 512-7114 or Kathryn G. Allen on (202) 512-7059 if you or your staff have any questions. Major contributors to this report are listed in appendix III. Sincerely yours, William J. Scanlon Director, Health Financing and Systems Issues Page 26 GAO/HEHS-97-86 Medicaid Managed Care Accountability Page 27 GAO/HEHS-97-86 Medicaid Managed Care Accountability Contents Letter 1 Appendix I 30 Scope and Methodology Appendix II 33 Federal and State Oversight Roles of Medicaid Managed Care Appendix III 36 Major Contributors to This Report Tables Table 1: Characteristics of Managed Care Programs Under 6 Program and Demonstration Waivers Table I.1: Selected Accountability Measures and Processes 30 Table I.2: Description of the Four States’ Medicaid Managed Care 31 Programs That GAO Reviewed Table II.1: Federal Regulations, Requirements, and Guidance for 35 Selected Accountability Measures and Processes Within States’ Medicaid Managed Care Programs Abbreviations AFDC Aid to Families With Dependent Children AIDS acquired immune deficiency syndrome EPSDT early and periodic screening, diagnosis, and treatment HCFA Health Care Financing Administration HEDIS Health Plan Employer Data Information Set HHS Department of Health and Human Services HIV human immunodeficiency virus HMO health maintenance organization NCQA National Committee on Quality Assurance SSI Supplemental Security Income Page 28 GAO/HEHS-97-86 Medicaid Managed Care Accountability Page 29 GAO/HEHS-97-86 Medicaid Managed Care Accountability Appendix I Scope and Methodology To ensure quality in their Medicaid managed care programs, states use a number of quality assurance and oversight mechanisms. For example, some states require accreditation of plans by independent entities, such as the National Committee on Quality Assurance. Most states build into their contracts with managed care plans a variety of accountability measures and processes. Once states have contracted with plans, they monitor the plans to ensure that they comply with these requirements. Based on our review of federal requirements and guidelines and state contracts and supporting documents, we identified a number of accountability measures and processes that states commonly include in their contracts with managed care plans. States can rely on various controls to assess compliance with some of these measures—including plan licensing, physician credentialing, and fiscal solvency requirements. Some accountability measures are more difficult to develop controls for and, therefore, are more difficult to monitor. To assess state efforts to hold managed care plans accountable for meeting Medicaid program goals and providing beneficiaries enrolled in capitated managed care plans the care they need, we narrowed the focus of our study to three areas: ensuring an adequate provider network, tracking the medical care provided to beneficiaries, and assessing beneficiary satisfaction. From among numerous quality assurance measures and processes, we identified 10 that states commonly use to monitor plan compliance in these accountability areas. Health Care Financing Administration (HCFA) officials and experts we contacted agreed that these measures and processes are essential to ensuring that plans meet the terms of their contracts. (See table I.1.) Table I.1: Selected Accountability Measures and Processes Accountability area Measure or process Adequacy of provider network —Patient-to-primary-care-physician ratio —Access to specialists —Travel distances and waiting times Adequacy of medical care —Utilization statistics —Encounter data —Clinical studies —Medical record audits Beneficiary satisfaction —Satisfaction surveys —Grievance procedures —Disenrollment data Page 30 GAO/HEHS-97-86 Medicaid Managed Care Accountability Appendix I Scope and Methodology To examine how these accountability measures and processes were implemented, we visited four states—Arizona, Pennsylvania, Tennessee, and Wisconsin—and reviewed their systems for ensuring access to quality care in their managed care programs. To select these states, we first identified a universe of 14 states that, as of June 1994, had more than 100,000 beneficiaries enrolled in capitated Medicaid managed care programs. We chose that date because we believed that, even for states just entering the managed care market, they would have had at least 2 years at the time we began our review to develop and implement their accountability systems. We then judgmentally selected four states that would provide a mix of experiences for a variety of factors. These factors included type of program (Medicaid demonstration waiver, program waiver, or voluntary nonwaiver), years of managed care experience, size of program, and geographic diversity. (See table I.2 for a brief description of each state’s Medicaid managed care program.) Table I.2: Description of the Four States’ Medicaid Managed Care State Program name and description Programs That GAO Reviewed Arizona Arizona Health Care Cost Containment System is a mandatory statewide demonstration program, operational since 1982, with Aid to Families With Dependent Children (AFDC) and Supplemental Security Income (SSI) enrollment of over 450,000 in 14 participating health plans. Arizona did not have a Medicaid program before the waiver. Pennsylvania The state’s voluntary nonwaiver program has been operational since 1972, with enrollment of about 435,000 AFDC and SSI beneficiaries in 18 counties in 11 health maintenance organizations (HMO) as of June 1996. (The scope of our work focused on the state’s voluntary program.) The state also had a mandatory program, Health Pass, in certain areas of Philadelphia from 1986 to January 31, 1997. On February 1, 1997, a new program waiver, HealthChoices, began in 5 Philadelphia-area counties; the voluntary program will continue in 13 counties. Within the next 2 years, the state plans to apply for a statewide demonstration waiver. Tennessee TennCare is a mandatory, statewide demonstration waiver program, operational since 1994, with enrollment of 849,000 AFDC and SSI Medicaid beneficiaries and over 334,000 uninsured persons in 12 participating HMOs at the time of our review.a Wisconsin The state’s mandatory program waiver for its AFDC population has been operational since 1984 in 5 counties, with enrollment of 138,000 in 11 participating HMOs. A modified waiver was approved October 1, 1996, to expand mandatory enrollment to 68 of the state’s 72 counties. a As of January 1997, Tennessee began to contract solely with HMOs. Before this, Tennessee also contracted with other health plans, such as preferred provider organizations. Page 31 GAO/HEHS-97-86 Medicaid Managed Care Accountability Appendix I Scope and Methodology For each state, we reviewed the contractual and other requirements the states have established with plans for these selected accountability indicators. We also interviewed officials from the state Medicaid, health, and insurance agencies to discuss specific contract requirements with managed care plans, state oversight activities and state actions available or taken as a result of monitoring, and state plans for changes in and expansions to their managed care programs. To identify federal requirements and guidance available to states to oversee their Medicaid managed care programs, we interviewed officials from HCFA’s central office and Chicago and Philadelphia regional offices, the Department of Health and Human Services’ (HHS) Office of Inspector General, and state-level advocacy groups. We also interviewed experts with The George Washington University Center for Health Policy Research, the National Association of Managed Care Regulators, the National Committee on Quality Assurance, and the National Association of Insurance Commissioners. Finally, we consulted with the following experts and researchers in the course of our work: Jane Horvath, National Academy for State Health Policy, Washington, D.C.; Robert Hurley, Medical College of Virginia, Richmond, Virginia; and Stephen Somers and Karen Brodsky, Center for Health Strategies, Princeton, New Jersey. We performed our work between October 1995 and March 1997 in accordance with generally accepted government auditing standards. Page 32 GAO/HEHS-97-86 Medicaid Managed Care Accountability Appendix II Federal and State Oversight Roles of Medicaid Managed Care By virtue of the mandated federal-state Medicaid partnership, states must meet certain federal requirements when developing their managed care programs. States may implement managed care programs under one of three options. The first option is a nonwaiver program that allows states to contract with managed care plans to deliver health care services to Medicaid beneficiaries who voluntarily participate. Certain conditions must be met, such as allowing beneficiaries the freedom to stay in a traditional fee-for-service system or enroll with a managed care plan from which they can disenroll at any time. Plans also must adhere to a “75-25 rule,” which prohibits participating managed care plans from enrolling 75 percent or more Medicaid and Medicare beneficiaries. The managed care program in Pennsylvania that we reviewed is a program of this type. The other two options for managed care—program and demonstration waivers—allow HCFA to waive certain provisions of the Medicaid statute, including beneficiaries’ freedom to choose from among participating providers. Under a program waiver, enrollment can be mandatory, but states are still required to ensure that plan enrollment of Medicaid and Medicare beneficiaries does not reach 75 percent and, in most cases, plans cannot lock in enrollment for more than 1 month. Wisconsin operates its mandatory managed care program under a program waiver. Under a demonstration waiver, states may be given permission to contract with plans that do not comply with the 75-25 rule and to exclusively enroll Medicaid beneficiaries. They also have been permitted to lock in beneficiary enrollment for up to 12 months. The managed care programs in Arizona and Tennessee are statewide mandatory programs operated under demonstration waivers. In addition, some states, such as Tennessee, have used demonstration waivers to expand eligibility to include non-Medicaid-eligible people who were formerly uninsured. Certain federal regulations, requirements, and guidance influence the development of state managed care programs and state monitoring of managed care plan performance. The extent of these requirements often depends on waiver type and can vary by state. In general, HCFA monitors the planning for and implementation of demonstration waivers more than for program waivers. The initial terms and conditions of approval for demonstration waivers are more detailed than for program waivers and are more specific in the content and timing of reporting requirements. For example, HCFA’s terms and conditions for a demonstration waiver have required that states specify in their contracts with plans a specific patient-to-primary-care-physician ratio that plans must meet. HCFA also requires that most states establish travel-related requirements for plan Page 33 GAO/HEHS-97-86 Medicaid Managed Care Accountability Appendix II Federal and State Oversight Roles of Medicaid Managed Care networks, such as maximum times and distances beneficiaries must travel to reach their primary care physician. In contrast, under a program waiver, HCFA suggests that states establish a patient-to-primary-care-physician ratio or that providers be located near beneficiaries. Under demonstration waivers, HCFA also requires states to provide an overall quality assurance monitoring plan and, as part of that plan, requires states to specify a minimum data set of encounter data. This minimum data set must receive prior HCFA approval. Program waivers, in contrast, have guidelines on quality assurance programs but not as many specific requirements. For example, HCFA does not require states to develop encounter data under program waivers. Table II.1 summarizes federal requirements and guidance by selected accountability measures and processes. Page 34 GAO/HEHS-97-86 Medicaid Managed Care Accountability Appendix II Federal and State Oversight Roles of Medicaid Managed Care Table II.1: Federal Regulations, Requirements, and Guidance for Additional HCFA Selected Accountability Measures and Accountability measure or requirements and Processes Within States’ Medicaid process Federal regulation guidancea Managed Care Programs Patient-to-primary-care- Plans must ensure that For demonstration waivers, physician ratio beneficiaries in managed states must meet maximum care have the same access 2,500 to 1 ratio. For to providers and services program waivers, HCFA as beneficiaries in suggests ratios be fee-for-service plans. evaluated. (42 C.F.R. 434.20(c)(2)) Availability of specialists Services are to be the same HCFA suggests that states as those provided under have a system for fee-for-service plans. (42 authorizing and C.F.R. 434.20(c)(2)) coordinating specialty services. Utilization statistics and Requires plans to maintain HCFA requires all states encounter data appropriate record systems with demonstration waivers for services provided to to collect 100% encounter enrollees. (42 C.F.R. data and requires all states 436.6(a)(7)) to quarterly report aggregated statistics on selected services. Clinical studies Plans must have an internal HCFA suggests states quality assurance system. conduct quality-of-care (42 C.F.R. 434.34) studies. Medical record audits States must annually HCFA requires states to conduct an audit of medical comply with federal records. (42 C.F.R. 434.53) regulation for medical record audits. Beneficiary satisfaction Plans must have an internal For demonstration waivers, surveys quality assurance system. a state may be required to (42 C.F.R. 434.34) conduct a survey as HCFA prescribes. For program waivers, HCFA suggests plans conduct periodic surveys. Grievance procedures Plans must have an internal HCFA requires states to grievance procedure report grievance data approved by the state that quarterly. provides for prompt resolution. (42 C.F.R. 434.32) Disenrollment studies States must monitor HCFA suggests states enrollment and termination analyze enrollment statistics. practices. (42 C.F.R. 434.63) a For demonstration waivers, additional requirements and guidance are in the terms and conditions that HCFA develops when it approves a state’s waiver request. These terms and conditions can vary by state. For risk-based managed care plans, such as program waivers, HCFA’s guidance is included in Monitoring Risk-Based Managed Care Plans: A Guide for State Medicaid Agencies, report prepared under contract for the Medicaid Bureau/HCFA by the Medicaid Management Institute of the American Public Welfare Association (Washington, D.C.: HHS, July 1993). Page 35 GAO/HEHS-97-86 Medicaid Managed Care Accountability Appendix III Major Contributors to This Report Kathryn G. Allen, Acting Associate Director, (202) 512-7059 Karyn L. Papineau, Evaluator-in-Charge, (202) 512-7155 Karin A. Lennon, Senior Evaluator Karen M. Sloan, Communications Analyst Betty J. Kirksey, Evaluator Daniel S. Meyer, Senior Evaluator (101385) Page 36 GAO/HEHS-97-86 Medicaid Managed Care Accountability Ordering Information The first copy of each GAO report and testimony is free. Additional copies are $2 each. Orders should be sent to the following address, accompanied by a check or money order made out to the Superintendent of Documents, when necessary. VISA and MasterCard credit cards are accepted, also. Orders for 100 or more copies to be mailed to a single address are discounted 25 percent. Orders by mail: U.S. General Accounting Office P.O. Box 6015 Gaithersburg, MD 20884-6015 or visit: Room 1100 700 4th St. NW (corner of 4th and G Sts. NW) U.S. General Accounting Office Washington, DC Orders may also be placed by calling (202) 512-6000 or by using fax number (301) 258-4066, or TDD (301) 413-0006. Each day, GAO issues a list of newly available reports and testimony. To receive facsimile copies of the daily list or any list from the past 30 days, please call (202) 512-6000 using a touchtone phone. A recorded menu will provide information on how to obtain these lists. For information on how to access GAO reports on the INTERNET, send an e-mail message with "info" in the body to: email@example.com or visit GAO’s World Wide Web Home Page at: http://www.gao.gov PRINTED ON RECYCLED PAPER United States Bulk Rate General Accounting Office Postage & Fees Paid Washington, D.C. 20548-0001 GAO Permit No. G100 Official Business Penalty for Private Use $300 Address Correction Requested
Medicaid Managed Care: Challenge of Holding Plans Accountable Requires Greater State Effort
Published by the Government Accountability Office on 1997-05-16.
Below is a raw (and likely hideous) rendition of the original report. (PDF)