Evolution of Medicaid managed care systems and eligibility expansions.

Introduction

Under the authority of section 1115 research and demonstration projects, States may restructure their Medicaid programs and serve as policy laboratories for the rest of the Nation. In 1993, the Federal Government began approving a new generation of section 1115 projects as a novel approach to fostering State health care reform initiatives. Two earlier section 1115 projects were in Arizona, which began in 1982 (McCall, 1997), and Oregon, which began in 1993 (Mitchell et al., 1998). The Federal approach to the review and approval for the earlier projects was quite different from the projects discussed in this article. Under a contract from HCFA, we participated in a multiyear evaluation of five of the earliest section 1115 initiatives approved in the new era of State flexibility: Tennessee's TennCare, Hawaii's QUEST, Rhode Island's RIte Care, Oklahoma's SoonerCare, and Maryland's HealthChoice. In this article, we review the programs' implementation, summarize key findings from several reports, provide new information about the latter stages of program implementation, and present analyses of changes in insurance coverage. We refer to the section 1115 projects as Medicaid projects, although they are more commonly called by their new State-specific names (TennCare, etc.) in the States themselves.

Although each State program had unique details, all involved large-scale shifts to mandatory capitated managed care, and three States also made major eligibility expansions. The States had many motivations but typically hoped managed care would save money (or at least stabilize expenditure growth), improve access to primary care, and, in some cases, permit a modest expansion of the range of benefits available. The first three States (Tennessee, Hawaii, and Rhode Island) were committed to reducing the number of uninsured people through Medicaid expansions, financed largely by expected managed care savings.

The implementation of the section 1115 projects involved a number of bold, rapidly paced initiatives at many different levels. The States made sweeping changes to their Medicaid program policies and operations. At the Federal level, HCFA approved and oversaw an unprecedented number of new section 1115 projects. (This experience helped HCFA in reviewing and approving State Children's Health Insurance Program [SCHIP] plans a few years later.) Finally, managed care plans and health care providers had to organize new networks and develop new contractual arrangements to serve the large number of new Medicaid members.

In this article, we review the lessons learned since 1994, when the first three programs began. The projects evolved through three stages: design, initial implementation, and maturity/policy refinement. In earlier reports, we focused on the first two stages of these programs (Wooldridge et al., 1996; Ku et al., 1998; Ku and Wall, 1997; Ormond and Goldenson, 1999). We give more attention to the third stage of the programs, after the initial birth pains subsided and the programs tried to solidify their policies and objectives. (Because it began in 1997, Maryland's program may not be at the same developmental level as the other States'. On the other hand, Maryland had much more pre-section 1115 Medicaid managed care experience.) We discuss how the five States (1) gradually changed eligibility and enrollment policies, (2) modified their managed care systems, and (3) evolved management strategies. We do not address the impact of the programs on health care utilization or the quality of care received by recipients but focus on the programmatic and policy aspects of implementation. Other parts of this broader evaluation project, now in progress, will address impacts of the projects and program expenditures. Some now complete include Moreno and Hoag (2001) and Moreno (1999). Impact reports have also been authored by other researchers (e.g., Conover and Davies, 2000; Griffin et al., 1999).

The policy environment for Medicaid managed care has changed recently. The passage of the Balanced Budget Act of 1997 empowers States to implement mandatory Medicaid managed care programs without Federal waivers, although there are still a number of legislative and regulatory requirements that States and plans must meet. Section 1115 waivers are still required for many types of eligibility expansions or other technical changes in program structure. Medicaid managed care is spreading, and the experience from these pioneering States may help guide later efforts.

Methods

We conducted periodic case study site visits to each State beginning shortly after initial implementation, interviewing State Medicaid and health department officials, managed care plans, health care providers (State medical, hospital, and primary care associations, as well as individual providers), advocates, and other stakeholders. Three rounds of site visits were made to each State, except for Maryland, which started much later and only had two rounds. In the initial year of the evaluation, we also conducted focus groups of consumers and providers in urban and rural sites. In general, the information in this article is current through fall 1999. Where noted, we were able to update data through the end of August 2000.

We supplement the case study information with insurance-status data from the March 1992 through 1999 Current Population Surveys (Campbell, 1999), which are discussed in the section on insurance trends.

Background

Key aspects of the section 1115 projects in the five States are summarized in Table 1, including the dates of implementation, the groups that are required to enroll in managed care, and eligibility expansions. All the projects focus on medical managed care and none include long-term care under capitation. Most rely on capitated care, although Oklahoma also developed a capitated primary care case management model for rural areas, in which specialty and inpatient care remain fee-for-service. Tennessee has the largest and broadest program, placing almost all Medicaid beneficiaries in managed care, while the other States focused on subsets of their Medicaid populations.

Table 1

Key Features of the Section 1115 Projects

Program Name	Tennessee TennCare	Hawaii QUEST	Rhode Island RIte Care	Oklahoma SoonerCare	Maryland HealthChoice
Implementation Date	January 1994	August 1994	August 1994	April 1996(July 1995 for HMOs in urban areas under a section 1915b waiver)	July 1997
Prior Experience with Medicaid Managed Care	One voluntary HMO.	One voluntary HMO.	One voluntary HMO.	None before 1915b waiver.	Voluntary HMOs and mandatory primary care case management.
Mandatory Managed Care Coverage	All, except those getting only Qualified/Specified Low-Income Medicare Beneficiary wraparound benefits.	Welfare, poverty-related, and expansion groups. Has proposed to expand to the disabled but not yet approved by HCFA.	Welfare, poverty-related, and expansion groups.	Initially, welfare and poverty-related groups. Urban areas use HMOs. Rural areas use primary care case management. In 1999, began phased-in mandatory managed care for disabled.	All but those dually enrolled in Medicare, institutionalized, spend-down medically needy and a few other exceptions.
Expansions in Medicaid Eligibility (Not Including SCHIP)	Initially, all uninsured could join, with State subsidies up to 400 percent of the FPL. In 1995, stopped enrolling new uninsured people, unless Medicaid-eligible. Eligibility was reopened for some in 1997.	At first, non-elderly, non-disabled people with incomes up to 300 percent of the FPL could join, with sliding scale premiums. Later, income level was lowered to 100 percent of the FPL for new applicants.	Initially expanded coverage for pregnant women and children up to age 6 to 250 percent of the FPL. Later this was expanded to children to age 8.	None.	None.
Managed Care Carve-Outs	Behavioral health managed care system began in 1996.	Separate dental managed care plans. Separate behavioral health plans for those with severe mental health problems.	None.	None.	Separate managed care plan for those with “rare and expensive conditions.” Behavioral health carve-out run by State.
Other Key Features	Charge premiums, deductibles, and copayments to some expansion groups.	Charged premiums to some expansion groups until eligibility was cut back.	Some expansion groups choose between premiums or copayments.	Separate programs in urban and rural areas. Rural system directly State-administered.	Some capitation payments are risk-adjusted, based on prior conditions.

NOTES: HMO is health maintenance organizaiton. HCFA is Health Care Financing Administration. SCHIP is State Children's Health Insurance Program. FPL is Federal poverty level.

SOURCES: The Urban Institute and Mathematica Policy Research, Inc.

Eligibility and Enrollment Policies

Eligibility Policies

The number of people participating in Medicaid managed care was affected by States' eligibility policies, as well as the initial phase-in periods. Figure 1 illustrates monthly managed care enrollment levels for the section 1115 project for each quarter from inception through July 1999. (Note that Tennessee's data use the right-hand axis because its enrollment levels are so much larger than those of the other four States.) Because the data include both managed care enrollment and eligibility, the first year of data in each program show large increases as managed care was phased in, but later caseload changes are attributable to overall program enrollment and eligibility changes. In their first years, both Tennessee and Hawaii also had large eligibility expansions along with the managed care shift, and Rhode Island had a smaller expansion. To some extent, QUEST substituted for prior State-funded programs. It replaced the State's general assistance medical care program and the State Health Insurance Program, which also served people with incomes up to 300 percent of the Federal poverty level (FPL), but which had capped funding. TennCare also replaced a small State program for uninsurable people, but most of the expansion was to people not already in State programs.

Figure 1:

Trends in the Number of People Enrolled in Managed Care: 1994-1999

After the first year, Tennessee encountered difficulties raising the State share of program expenses and stopped enrolling new uninsured people. It still accepted those eligible under old Medicaid criteria and those who were “uninsurable” (i.e., those denied private insurance because of medical conditions) and grandfathered uninsured people already on TennCare. The caseload fell from 1.3 million per month at the end of 1994 to 1.2 million by late 1995. In 1997, the State partially reopened eligibility for uninsured children and displaced workers, partly because the State expected Congress would pass SCHIP legislation, which could help finance child coverage.

In April 1997, TennCare was opened for children who did not have access to health insurance through their parents. In January 1998, it was opened to children from families with incomes below 200 percent of FPL, even if their parents did have access. However, although Federal approval for Tennessee's CHIP program was not granted until September 1999, CHIP matching rates were retroactively applied for children 15 to 18 years of age from families with incomes below 100 percent of the FPL, already covered under TennCare. TennCare caseloads began rising again that year. In 1999, Tennessee again experienced problems with program financing, and serious discussions about TennCare's future began, including options to scale back eligibility (Snyder, 1999). The State has been particularly concerned about increases in the number of uninsurable enrollees, who have higher health care costs (Snyder and Klausnitzer, 2000).

In Hawaii, an unexpected statewide recession led QUEST caseloads to rise almost 50 percent beyond the initial enrollment estimates of 110,000 people per month. Further, a lawsuit was filed under the Americans with Disabilities Act protesting the fact that QUEST eligibility expansions were not available to disabled people. In 1996, Hawaii began ratcheting down eligibility, eventually lowering the income standard to 100 percent of the FPL with an asset test, although there were higher limits for pregnant women and children, based on the Medicaid poverty-related eligibility expansions that existed before QUEST. The State also capped enrollment at 125,000 people. A small program, QUEST-Net, was developed to serve those who lost QUEST coverage; it had fewer benefits than QUEST and lower capitation rates. In July 2000, Hawaii implemented its SCHIP program, which serves uninsured children not otherwise eligible from families with incomes between 100 and 200 percent of the FPL.

Rhode Island's initial expansion under RIte Care was modest in comparison, raising eligibility for pregnant women and children under age 6 to include those from families with incomes up to 250 percent of the FPL. Later, as the State found that it could afford them, further expansions were made. In 1995, the age limit for children was increased to 8. In October 1997, the State implemented an SCHIP expansion for children through age 18 from families with incomes up to 250 percent of the FPL. It later expanded eligibility for children from families with incomes up to 185 percent of the FPL, providing coverage to parents using section 1931 authority. Section 1931 is part of the 1996 Federal welfare reform law that lets States use more liberal methods to count income or assets for families with children (Guyer and Mann, 1998). Childless couples and adults are not eligible. In 1999, RIte Care's enrollment began increasing. In summer 2000, the State found that the additional caseload was growing faster than anticipated, and the State adopted plans to modify eligibility criteria, particularly to avoid covering those with available employer-sponsored coverage (Rowland, 2000).

Oklahoma and Maryland's section 1115 projects did not include eligibility expansions, but both States later had SCHIP expansions to serve pregnant women and children from families with incomes up to 185 (Oklahoma) and 200 (Maryland) percent of the FPL. Like Rhode Island, both Oklahoma and Maryland were considering further insurance coverage expansions.

Enrollment Policies

In the first year of implementation, the most visible problems in each section 1115 project revolved around the enrollment of Medicaid beneficiaries into managed care plans and their subsequent assignment to primary care providers. (States typically require ongoing and new Medicaid participants to select a managed care plan within 2-3 weeks. If they do not make a choice, the State assigns them to a plan, using criteria that vary by State. Depending upon the State, the person may choose a primary care provider when choosing a plan or must select the provider soon after the plan is assigned, or a primary care provider is assigned by the plan.) All States went through an initial period of chaos and confusion. Common startup difficulties included insufficient patient education about how to choose health plans and primary care providers, errors and omissions in plan and provider directories, high automatic assignment rates, apparent assignments to plans or providers other than those requested, and delays in processing cases. The net effect was that many patients were assigned to unfamiliar or inconveniently located primary care providers and, conversely, many doctors complained that they lost most of their old Medicaid patients, while gaining many new patients. Both patients and providers complained about disruptions in the continuity of care. Although the disruptions were unfortunate, we are not aware of evidence that the quality of medical care was compromised.

All the States but Rhode Island used mass managed care enrollment procedures at startup, in which ongoing Medicaid recipients were mailed enrollment materials and told to enroll, by mail or phone, by a given date. Rhode Island used a phased-in approach, enrolling participants at the welfare offices when they were being recertified for Medicaid. The mass mailings were efficient, but many people did not understand or properly complete the forms, and applicants found it hard to get through busy telephone lines at the State offices, enrollment brokers, and health plans. The resulting confusion led to higher rates of automatic assignment to health plans, because people did not make their selections. On the other hand, a phased-in process has the disadvantage of building membership and managed care revenue slowly, which may particularly hurt new plans.

All the States found the first several months bumpy, but this period seemed more difficult in Tennessee and Hawaii because of their rapid implementation schedules. The States that implemented later, Oklahoma and Maryland, learned from the early States' travails, although these two still had some problems.

Most parties agreed that the confusion subsided after the first year or two. With familiarity, beneficiaries and health care providers became more able to maneuver the systems and became more inured to problems. Further, the State agencies and managed care plans fine-tuned operations and policies (e.g., assigning special staff to expedite enrollment of pregnant women and newborns into managed care plans or clarifying how plans may market to prospective members). After the initial mass enrollments were over, it was easier to manage the flow of new applicants and plan members. Some problems persist, such as insufficient enrollee education about managed care and continuing difficulties in verifying the current plan status and primary care provider of Medicaid patients, but the sense of exasperation faded as programs matured. Complaints are still voiced (particularly by physicians), but there are far fewer.

Insurance Trends

Did the Medicaid eligibility expansions increase the overall health insurance coverage of low-income people in these States? Data from the Current Population Surveys (CPS) for the years 1991 through 1998 provide basic insights. To reduce the effect of small sample sizes, we computed 2-year rolling averages (i.e., data labeled 1991-1992 are the averages of estimates from the March 1992 and 1993 CPS). (Detailed information on the formula used to compute the standard error for a State's insurance rate in a single year is available from the authors.) Data are presented for non-elderly people with incomes below 200 percent of the FPL (Table 2). We also analyzed data for those with incomes from 200 to 400 percent of the FPL, which includes the upper bounds of expansions for some States; for the sake of brevity, we do not include this table and discuss relevant findings in the text.

Table 2

Health Insurance Coverage of Non-Elderly People with Incomes Below 200 Percent of the Federal Poverty Level: 1992-1998

State	1992-1993	1993-1994	1994-1995	1995-1996	1996-1997	1997-1998	Significant Difference2
United States
Medicaid or State Insurance	26.0	26.9	27.0	26.5	25.6	24.7	*
Employer-Sponsored Insurance	29.0	30.1	31.8	32.2	32.3	32.7	*
Other Private Insurance	7.1	6.9	6.3	5.8	5.7	5.7	*
Other Insurance	6.2	4.9	3.9	3.7	3.7	3.7	*
Uninsured	31.7	31.2	31.1	31.8	32.7	33.3	*
Tennessee
Medicaid or State Insurance	29.7	31.1	330.4	331.7	335.5	337.5	*, #
Employer-Sponsored Insurance	32.0	33.5	334.6	330.4	330.3	331.9	—
Other Private Insurance	5.0	6.1	36.8	35.5	34.6	35.4	—
Other Insurance	5.9	9.2	38.4	33.9	33.3	33.8	—
Uninsured	27.5	20.1	319.7	328.5	326.3	321.4	*, #
Hawaii
Medicaid or State Insurance	24.3	18.2	20.9	327.5	328.9	325.6	—
Employer-Sponsored Insurance	35.2	45.3	44.2	332.9	332.7	339.2	—
Other Private Insurance	6.8	6.6	6.3	37.4	38.9	37.9	—
Other Insurance	15.5	10.9	11.3	313.8	312.8	39.3	*
Uninsured	18.2	19.0	17.2	318.4	316.7	318.0	—
Rhode Island
Medicaid or State Insurance	34.7	29.8	23.2	328.1	335.1	333.5	—
Employer-Sponsored Insurance	34.4	35.1	35.1	334.8	334.3	331.6	—
Other Private Insurance	6.7	7.3	8.6	36.5	34.0	37.7	—
Other Insurance	5.1	4.0	4.6	34.5	34.2	34.4	—
Uninsured	19.0	23.7	28.5	326.1	322.5	322.9	—
Oklahoma
Medicaid or State Insurance	17.2	18.4	18.3	15.8	15.2	15.4	—
Employer-Sponsored Insurance	25.7	28.0	28.4	31.1	35.0	34.8	*
Other Private Insurance	5.4	5.5	7.6	6.9	5.6	5.8	—
Other Insurance	9.4	6.4	5.7	7.3	9.7	10.1	—
Uninsured	42.2	41.7	40.0	38.9	34.6	33.9	*, #
Maryland
Medicaid or State Insurance	23.9	23.0	24.0	27.0	22.4	13.5	*, #
Employer-Sponsored Insurance	31.2	37.3	37.9	33.7	32.8	30.5	—
Other Private Insurance	7.0	7.9	5.3	6.1	10.3	9.4	—
Other Insurance	10.0	4.7	2.4	1.7	1.6	2.3	*, #
Uninsured	28.0	27.1	30.4	31.5	32.9	44.5	*, #

2-year rolling averages.

In this column, * indicates that the 1992-1993 versus 1997-1998 difference (within a State or the United States) is significant with 95-percent confidence; # indicates that the State 1992-1993 versus 1997-1998 difference, less the U.S. 1992-1993 versus 1997-1998 difference, is significant with 95-percent confidence.

States with Medicaid expansions over the whole 2-year period.

NOTES: Standard errors are available from the lead author upon request. Insurance status is presented as a hierarchy, so that those who get Medicaid and employer-sponsored coverage are shown as Medicaid and so on. Excludes data on those in active military duty and those institutionalized.

SOURCE: Current Population Surveys, March 1993-1999, as tabulated by The Urban Institute.

In Table 2, we indicate whether insurance levels for the years 1992-1993 and 1997-1998 are significantly different from each other and whether the State-specific trend is different from the overall national trend. First we test the 1992-1993 versus 1997-1998 difference within each State (or the United States). We also examine how these State differences compare with overall trends, by examining the State difference less the national difference. This difference-in-differences approach may better portray unique State trends, controlling for the national trend. These years are selected because 1992-1993 is the last period that is clearly before the implementation of the eligibility expansions and 1997-1998 is the most recent 2-year period. Any differences across time periods should be interpreted with caution, because we do not control for other forces that might cause differences in insurance trends. Other methodological problems with the use of CPS data may also affect interpretation.

One reason to focus on the last period is that the CPS data suggest that there was a reduction in self-reports of Medicaid or other State insurance participation in the 1 or 2 years following each State's participation, possibly because the Census Bureau did not list the new programs' names or because participants may have simply been confused about the names of their insurance programs. After the initial period, the participation levels appeared more reasonable and more consistent with administrative participation estimates. In the case of Tennessee, there was no reduction in 1994-1995, but underreporting still appeared to be a problem, because the CPS did not show the large increase in case-loads that actually occurred.

Between health insurance year 1993 and 1994, there were methodological changes in the CPS questions. Swartz (1997) discusses these changes and suggests that the CPS modification did not significantly affect differences in the Medicaid or uninsurance levels but appears to be associated with increases in the reporting of employer-sponsored group coverage. Levels of Medicaid participation reported in the CPS are much lower than administrative estimates (Lewis, Ellwood, and Czajka, 1998). Ku and Bruen (1999) noted that CPS underreporting of Medicaid enrollment increased in recent years.

In Tennessee, Medicaid coverage of people with incomes below 200 percent of the FPL rose from 30 percent in 1992-1993 to 38 percent in 1997-1998. During this period, the percentage of low-income uninsured persons in Tennessee fell from 28 to 21 percent. Telephone surveys conducted by the University of Tennessee at Knoxville also show a reduction in the level of uninsurance for State residents from 8.9 percent in 1993 to 5.9 percent in 1997 (Fox and Lyons, 1998). Although the point estimates for coverage differ from the CPS, the trends of falling uninsurance are comparable. Tennessee now has one of the lowest uninsurance rates in the Nation. Given the national trend of falling Medicaid coverage and rising uninsurance rates, Tennessee's success in expanding coverage is even more noteworthy.

There were no significant changes in employer-sponsored coverage or other private insurance (e.g., non-group coverage), suggesting that there was relatively little substitution of Medicaid for private coverage, also known as “crowd out,” in the low-income group. (There was also no significant difference for employer-sponsored or other private coverage comparing 1992-1993 with the 1995-1996 or 1996-1997 periods.) The fact that private coverage did not fall during this period of Medicaid growth might be partly because of the State's requirement that expansion groups had to be either uninsurable (based on denial of a policy) or uninsured (based on lack of private coverage on some date before application).

For those with incomes from 200 to 400 percent of the FPL, the share of persons covered by Medicaid in Tennessee grew from 3.2 percent in 1992-1993 to 8.3 percent by 1997-1998. (Only Tennessee had significant changes in this higher income range; neither Hawaii nor Rhode Island had significant changes, and changes in Oklahoma or Maryland were unrelated to Medicaid policy changes.) However, there was a corresponding decline in the level of other private coverage from 9.5 to 5.2 percent, suggesting that there might have been some substitution of Medicaid for non-group coverage in this middle-income range. This is consistent with State concern about the increase in the number of uninsurable enrollees and belief that private insurers may have become more restrictive in underwriting non-group insurance, knowing that TennCare could provide coverage (Snyder and Klausnitzer, 2000). However, when we control for the national trends, the difference in other private coverage in Tennessee is not significant. During this period, there was an overall national trend in reduced other private coverage, probably caused by escalating premium costs. Thus, the reduction in other private coverage in Tennessee may have been part of a broader secular trend and not a direct result of TennCare expansions in the middle-income group.

Both Hawaii and Rhode Island also had Medicaid eligibility expansions, but there were no significant changes in percent of low- or middle-income people under Medicaid or uninsured. For Hawaii, at least part of the explanation might be that the QUEST expansion largely substituted for a pre-existing State program. There were no significant differences when comparing 1992-1993 with the 1995-1996 or 1996-1997 periods, either.) Further, Hawaii has had among the lowest uninsurance rates in the Nation for many years now, in part because it is the only State in which employers are required to provide health insurance to full-time workers. Rhode Island's expansion was relatively narrow initially and the number of newly covered people was small, so it is not surprising that there are no significant differences.

Oklahoma and Maryland had no significant expansions of Medicaid eligibility under the section 1115 projects, so no changes in Medicaid coverage were expected. Maryland appears to have had a large decline in Medicaid participation among low-income persons from 1992-1993 to 1997-1998 and a steep increase in uninsurance rates, but we believe that these CPS estimates are not valid. Administrative data show that Maryland's Medicaid caseload changed very little in this period, in marked contrast to the CPS data. The State implemented HealthChoice in 1997, but the CPS did not list this alternative program name, so many people may not have reported participation. Also, participants no longer received Medicaid cards but rather health plan-specific cards (e.g., Free State or Prudential), so they might have been unaware of the connection to Medicaid. (In separate analyses, the Urban Institute used its microsimulation model, TRIM, to adjust CPS estimates based on administrative Medicaid counts, which increased Medicaid enrollment rates and lowered the uninsurance rate estimate.)

Evolution of Managed Care Systems

Structural Issues

All five States embraced mandatory Medicaid managed care systems under which: (1) enrollees must join a capitated managed care plan, (2) the managed care plans must agree to the State's contractual and price requirements, and (3) enrollees must have a choice of health plans. Within these requirements, the programs differed across the States, and each has changed over time. (Oklahoma uses a partial capitation approach in rural areas. Primary care physicians are paid a capitated rate for a bundle of primary care services, but other services are fee-for-service. The primary care physicians contract directly with the State and recipients choose their doctors from the State's list of those in their area.)

In the beginning, States were somewhat looser in setting plan requirements, hoping to attract existing plans and foster new ones. Over time, contractual requirements stiffened. For example, Tennessee initially included plans that were not health maintenance organizations (HMOs) (i.e., they did not require primary care providers) but by 1997 required that all TennCare plans be HMOs. Although plans in all five States were supposed to submit quality assurance or encounter data from the first year, States did not impose stronger contractual requirements or sanctions until much later.

The basic managed care program designs did not change much after the first year, allowing administrators to concentrate on important administrative tasks, such as ensuring that provider networks were adequate, refining operational policies, and coordinating with plans and provider representatives. Because of continuous turnover in participating doctors, States and plans must maintain efforts to ensure adequate geographic and specialty distribution of physicians.

Two structural issues have been problematic in the States: safety net providers and behavioral health. Safety net providers include non-profit community health centers, certain public and non-profit hospitals, and public health agencies (e.g., maternal and child health, family planning, or developmental disability clinics) that serve low-income Medicaid or uninsured patients. Because these facilities had few privately insured patients, they depended upon Medicaid funding and, before managed care, often received supplemental payments, such as cost-based reimbursement for federally qualified health centers or Medicaid disproportionate-share hospital payments.

Medicaid managed care threatened these providers by either lowering their share of the Medicaid patients (with the balance going to other private providers) or reducing their payment rates. Although the evidence is mixed on whether Medicaid managed care has financially harmed the safety net providers (Hoag, Norton, and Rajan, 1999; Lewin and Altman, 2000), the treatment of safety net providers has been the subject of great debate in each State. Federally qualified health centers and similar agencies argued that if they lost money on Medicaid, then they could not afford to serve uninsured patients. All the States made some special accommodations for safety net providers, such as requiring managed care plans to include them in provider panels or providing special financial or technical assistance. Certain types of clinics, such as family planning or maternal and child health clinics, may have had problems because they were not viewed by managed care plans as being either primary care providers or specialists, but in a middle ground of specialized primary care services.

Behavioral health care has been another thorny issue in these projects. There is no clear consensus among the States on whether or not to “carve out” behavioral health using separate behavioral health managed care plans, fee-for-service, or some mixture. Even where mental health is provided under the medical managed care plan, there are typically subcontracts with behavioral health organizations, blur-ring the issue of whether to include mental health within the medical managed care plans.

All the States but Maryland originally included mental health services as part of the medical managed care plan, except for persons with serious mental illness or serious emotional disturbances (SMI/SED). People with SMI/SED diagnoses were to be identified by the managed care plans and, if determined after a State review to be sufficiently impaired, were placed back in the fee-for-service system (Tennessee, Oklahoma, and Rhode Island) or in separate behavioral health plans (Hawaii). In Maryland, all specialty mental health services were provided through a new State program, the Specialty Mental Health System, that blended Medicaid and other public funding sources under the control of the State mental health agency.

The SMI/SED carve-out process has not always worked well. The process of referral by the plans and determination by the State has been slow, and there have been disagreements about the standards for severity. Maryland permits self-referral, which greatly lessened but did not eliminate such disagreements. In Rhode Island, mental health expenses beyond the stop-loss level were not the plan's responsibility, so there was little incentive for plans to refer SMI/SED patients for specialty mental health care.

An additional factor that complicated behavioral health arrangements has been that many, perhaps most, of the core Medicaid providers, particularly for SMI/SED patients, are State-operated or State-funded community mental health centers and facilities (Hogan, 1999). State mental health agencies were concerned that their mental health safety nets be maintained because they served uninsured and other needy non-Medicaid patients. This concern was strongest in Maryland and was the impetus for creation of its Specialty Mental Health System. But other States had similar concerns and ensured that State-operated or funded facilities were included in managed care plans and that SMI/SED cases were carved out. Many of these mental health facilities have limited management capacity for managed care because they are primarily grant-funded, pool funds from a variety of sources, and have little experience with private insurance or managed care. For example, one large community mental health center in Oklahoma had funding from more than 50 sources and cited examples of how it supplemented care for Medicaid managed care patients (beyond that authorized by the behavioral health organization) using funds from other grants. Therefore, bringing them into managed care has often been difficult.

Some of the problems have been most evident in Tennessee, which in 1996 began a special behavioral health carve-out program, TennCare Partners, taking away mental health and substance abuse services from the medical plans and shifting them to separate behavioral health organizations. It also provided behavioral health services to some uninsured people not in TennCare. The Partners program was instituted at least partly to reduce State Medicaid expenditures through increased control of mental health expenses. It has been controversial, with serious complaints of poor quality care, underfunding, and other problems. In each year, restructuring the Partners program has been seriously debated, and its future is still uncertain.

One final structural issue is worth mentioning: mandatory managed care for the disabled. Medicaid programs have been more hesitant to require that disabled people, with high health care needs, shift to managed care. Until recently, only TennCare and Maryland's HealthChoice included the disabled. Hill et al. (1999) found that, in Tennessee, disabled people receiving Supplemental Security Income with higher risks were more likely to join certain plans than others, so that some plans experienced adverse selection, which could undermine their finances. In Maryland, there were concerns that disabled peoples' medical care patterns were sometimes disrupted because plans did not always include all the usual providers for disabled patients, although the State tried to minimize this problem through collaborations with providers and advocates for the disabled. Oklahoma began requiring that disabled enrollees join managed care in mid-1999, and Hawaii was planning to include disabled persons, so there is not much experience in those two States.

Capitation Rates

One of the most important functions of the State is to establish initial capitation rates and to periodically update them. Hoag, Wooldridge, and Thornton (2000) discuss the types of information needed to set capitation rates for behavioral health organizations. The capitation rates directly affect the overall project budgets, as well as the amount that plans can pay providers, and also determine the extent to which plans and providers can afford to participate.

Three States (Rhode Island, Hawaii, and Oklahoma) used a form of managed competition to set the capitation rates, while Tennessee and Maryland set uniform rates for all plans. (In managed competition, plans bid to participate in Medicaid, under rules and within price ranges established by the State agency. Multiple plans are awarded contracts and there may be small rate differences.) Except for Maryland, the States used relatively simple methods of risk adjustment: The capitation rates were based on members' age, gender, and other basic characteristics (e.g., Tennessee had a different rate for the disabled; Hawaii paid more for those receiving General Assistance; Rhode Island and Oklahoma provided supplemental payments for pregnant women). The rates were based on actuarial estimates of average costs, using historical Medicaid data with other adjustments. Maryland developed a more sophisticated system, setting capitation rates for each member based on his or her prior Medicaid services, using an adaptation of the Johns Hopkins University Ambulatory Care Groups risk-adjustment system, with separate schedules for disabled and non-disabled enrollees; members without enough prior Medicaid experience had rates set on demographic categories.

With time, the States increased their control over the capitation rates and essentially set them administratively, even in the three States using a competitive model. In these three States, the rates paid to managed care plans have tended to converge: the States have narrow bands of allowable rates and force all plans into those ranges. After the first year, the plans know their competitors' rates, which effectively keeps them in a tight range. For the sake of equity, the States often felt obliged to treat plans consistently when they changed bidding procedures or capitation rates. For example, in its 1999 bid, Oklahoma told plans the top rates it would accept, so plans generally bid the top rate.

The changes in capitation rates paid depended on both the plans' prior experiences and the States' financial needs. Tennessee and Rhode Island increased rates somewhat faster than expected when some plans showed evidence of underpayment. In contrast, Hawaii reduced rates because of State fiscal problems. Maryland had to reconsider payment rates after the first year, when it was discovered that the risk-adjustment algorithm resulted in payments, particularly for the disabled population, that exceeded the intended rates. The State was not able to fully lower rates in the next year to compensate for this initial error because of concerns that large capitation payment reductions might destabilize some plans.

Plan Participation

The participation of managed care plans has emerged as one of the most important issues in the long-term viability of Medicaid managed programs. As of August 2000, there were at least three managed care plans in each of these five States, but the number of plans had fallen since the programs began. In each program, there is a mix of commercial (i.e., serving mostly non-Medicaid members in that State) and Medicaid-dominant plans (i.e., those with only or mostly Medicaid members in that State) (Table 3). The Medicaid-dominant plans were often created by safety net health care providers (e.g., hospitals or community health centers) to participate in the section 1115 projects.

Table 3

Number of Participating Managed Care Plans Through Late 1999

Characteristic of Plans	Tennessee1	Hawaii	Rhode Island2	Oklahoma3	Maryland4
At Start
Total	12	5	5	5	9
Commercial	6	4	4	2	3
Medicaid-Dominant	6	1	1	3	6
By Late 1999
Total	9	6	4	4	8
Commercial	5	6	3	2	3
Medicaid-Dominant	4	1	1	2	5
Changes
New Plans
Commercial	0	1	0	0	1
Medicaid-Dominant	0	0	0	0	0
Reduced Through Consolidation
Commercial	1	0	1	0	0
Medicaid-Dominant	1	0	0	0	0
Exiting Plans
Commercial	1	0	0	0	1
Medicaid-Dominant	0	0	0	1	1

One Tennessee plan was Medicaid-dominant in 1994 but became a majority commercial by 1999, so its status changes in the table. Further, one commercial firm exited the State in 2000.

One commercial plan in Rhode Island exited in early 2000 and terminated all its product lines, not just Medicaid.

The plan that left was Medicaid-dominant in Oklahoma but was part of a multistate commercial chain.

One commercial plan left Maryland's Medicaid program, but its Medicaid product line was purchased by the new commercial plan that was entering the State. That is, the entering and exiting plans involved the same network and membership.

NOTE: This table includes medical plans only and does not include dental or behavioral health plans.

SOURCES: The Urban Institute and Mathematica Policy Research, Inc.

As of 1997, a majority of plans were profitable and relatively few had dropped out of the program, although there were some with substantial losses (Ku and Hoag, 1998). The number of participating managed care plans across the five States fell from 36 at the beginning to 31 as of late 1999. Two new plans joined, three were consolidated through acquisition, and four dropped out. Typically the exiting plans were relatively small. The plans that were acquired or dropped out completely were equally mixed between commercial and Medicaid-dominant, in contrast to the apparent national norm that commercial plans are more likely to leave Medicaid (Felt-Lisk, 1999).

A certain degree of plan consolidation is not necessarily detrimental to the program. The departure of one plan effectively increases member volume for the remaining plans, which may improve the overall financial health. When plans exit the program, the Medicaid agency and managed care plans typically make transition arrangements to ensure that members of closing plans can, to the maximum extent possible, keep the same providers under a different plan arrangement.

In mid-1999, TennCare's largest plan (operated by Blue Cross Blue Shield and serving about one-half of the enrollees) announced that it would leave the program by June 2000 because of financial losses, although that date was later extended. The announcement helped trigger a major debate about the future of TennCare. Since then, the State legislature has substantially increased funding for TennCare, enabling payment rates to be increased. The State is working to restructure the managed care system so that plans are selected on a regional basis, with no more statewide plans, and there are alternative business models, including some that have risk-sharing between the plan and the State so that straight capitation is not the only option. As of August 2000, the State was negotiating with four new managed care plans (Snyder, 2000).

Since 1999, other plans (e.g., Queens and Kapiolani in Hawaii and Harvard Community Health Plan in Rhode Island) have withdrawn or announced their withdrawal, and some plans in other States might also exit. Although not as dramatic as the changes in Tennessee, these actions continue to reduce the pool of plans from which consumers may choose. In Rhode Island, for example, consumers only have three plans available.

In addition to actual plan terminations, there are subtler actions that may also limit participants' health care choices. Many plans are not available statewide and, in a few areas, the number of plans fell to two, the minimum permitted under Federal rules. Further, participating plans often capped the number of Medicaid enrollees they would take and, in a couple of instances, only one plan in some areas was accepting new members.

Evolution of Program Management

The new programs required that State Medicaid agencies and the managed care plans make major investments in new administrative resources. The State agencies needed to upgrade staff skills, procedures, data systems, and other capacities to develop the resources for plan contracting, enrollment/assignment of members into managed care plans, consumer and provider relations, quality assurance monitoring, and collection and analysis of encounter data. States took on these new responsibilities while continuing to administer the remaining fee-for-service systems. To help them, the State agencies often used contractors for a host of functions, including general management consulting, external quality review, data systems, managed care enrollment, and actuarial services. In all the States, Medicaid agencies formed advisory councils or workgroups, sometimes spurred by the governor or State legislature, to get external advice and monitoring of the new programs. The advisory groups typically included representatives of the managed care plans, heath care providers, consumers, and other State agencies.

We can view the States' new roles as fitting into three categories: (1) program planning and design (e.g., determining plan responsibilities and contracting with them), (2) direct program management (e.g., member enrollment and consumer/provider relations), and (3) program monitoring, which principally involves assessing the quality of care provided by the plans. These first two roles were preeminent in States' minds in the first 2 years of the programs.

The monitoring efforts were not fully operational until the third year of operation and continue to need improvement (Wooldridge and Hoag, 1999). It is practically impossible for new programs to monitor care adequately in their first year, because of the heavy demands during initial implementation, as well as because of the relative lack of experience of the States and the plans in that first year. Over time, the States worked with the managed care plans to develop quality assurance/quality improvement systems, using guidelines such as the Health Plan Employer Data and Information Set (HEDIS), Quality Assurance Reform Initiative (QARI), and Quality Improvement System for Managed Care (QISMC). In addition, all the States contracted with external quality review organizations to help develop quality improvement procedures and to measure the quality of care provided by health plans, using both process and outcome assessment methods. Some of the States have been more successful than others, but all made substantial progress as the programs evolved. States continue to face challenges in getting complete and reasonably valid encounter data submissions from the managed care plans (who must get them from the providers) and developing strong systems for ongoing performance measurement and feedback to the plans.

A gratifying sign of the improvement in monitoring systems is that States have begun to issue a number of reports about the quality of care, consumer satisfaction, and health outcomes. In addition to serving research and monitoring functions, these data have been used by States for other management purposes. In some States, quality reviews are used to help select and approve plans for participation. Oklahoma disseminates comparative data from its managed care consumer satisfaction surveys to help new enrollees select their managed care plans.

Conclusions

Perspectives on the section 1115 projects have evolved over time. During the design and initial implementation, these initiatives were controversial and chaotic. Several years later, the waters have calmed in most of the States. We conclude this article by summarizing the key lessons learned, including issues that continue to challenge program administrators.

Although technically designed as time-limited demonstration projects, section 1115 projects initiated long-term changes in State Medicaid programs and as platforms for other policy changes in the States. It seems clear that the State initiatives are not just experiments but fundamental policy shifts. All the States have had major changes in political and/or administrative leadership but retained the basic original designs. Despite recent controversies, the basic structure of TennCare remains intact.

One sign of the support for these programs is exemplified in their SCHIP expansions. Across the Nation, the majority of States either established separate, non-Medicaid SCHIP programs, sometimes combined with Medicaid expansions (Ullman, Hill, and Almeida, 1999). The creation of separate programs could be interpreted as signs of States' discontent with their Medicaid programs. By contrast, these five States all used their section 1115 programs as the only platform for SCHIP expansion, because these States perceived that they had already “reinvented” their Medicaid programs.

Large eligibility expansions are feasible, can reduce the number of uninsured people, and can improve people's access to health services. TennCare greatly expanded eligibility for low- and middle-income people. Between 1992-1993 and 1997-1998, the percent of Tennessee residents with incomes below 200 percent of the FPL who reported getting Medicaid in the CPS rose from 30 to 38 percent, while the share of uninsured people fell from 28 to 21 percent. No significant changes in employer-sponsored or other private coverage occurred for low-income persons, suggesting that crowd-out was not much of a problem in this income range. Those with middle incomes (200 to 400 percent of the FPL) also had an increase in Medicaid coverage from 3 to 8 percent, but there is mixed evidence of an offsetting reduction in non-group private insurance coverage.

These results are particularly important when coupled with other recent research that indicates that people who were covered under the TennCare expansions had better access to care, greater use of preventive services, lower out-of-pocket medical costs, and fewer unmet medical needs than similar uninsured people (Moreno and Hoag, 2001). Other evidence indicates that TennCare was associated with higher physician participation, better immunization rates, and lower emergency room use, although there are also signs of lower prenatal care use (Conover and Davies, 1999). On balance, evidence indicates that TennCare expanded insurance coverage, and this led to beneficial increases in access to health care by low-income people.

States may have to modify the Medicaid eligibility expansions in section 1115 projects, as fiscal resources permit. The two States with the largest initial expansions, Tennessee and Hawaii, trimmed their eligibility expansions because of fiscal problems. Though TennCare was able to partially restore eligibility between 1997 and 1999, the recent financial crisis has brought about new proposals that may lead to eligibility restrictions. Rhode Island began expansions modestly but found that it had both fiscal capacity and political support to expand eligibility in later years. Nonetheless, these three States still have among the most generous Medicaid eligibility standards in the Nation.

In retrospect, it is not surprising that eligibility criteria were fluid and affected by budgetary concerns. Reducing the flow of expansion enrollees is a feasible way to control spending. As section 1115 projects, the eligibility criteria were established at the State's option, unlike mandatory rules for other Medicaid eligibility criteria. In comparison, States have more difficulty reducing capitation rates substantially in times of budget need, because large reductions in the capitation rates would jeopardize the viability of the managed care plans that had become necessary players in the system.

No consensus has been reached on the best way to address either the protection of safety net providers or treatment of behavioral health under Medicaid managed care. Policies about safety net providers and behavioral health have been problematic in all the States. Part of the problem is that the States have conflicting objectives in these areas. On one hand, the States want to reduce expenditures and let private managed care plans manage the providers. On the other hand, the States have an interest in protecting safety net providers, including many of the mental health providers, because of their roles in serving the uninsured and other needy people, supported in large measure by other government grants. Efforts to extract substantial managed care savings from these providers might undercut their other missions to serve the uninsured. Most of the States have provided extra financial or technical assistance to federally qualified health centers participating in Medicaid managed care plans. Some providers, particularly mental health facilities, had little experience with insurance billing or managed care and had great difficulties adjusting their management systems.

There is little consensus on whether it is better to include behavioral health with other medical services in a medical managed care plan or to carve out behavioral health in separate plans. All States carved out behavioral health (as separate plans or fee-for-service) for the people with the most severe mental illness or emotional difficulties, but this also had difficulties.

Establishing and updating capitation payment rates is complicated and involves assessments of actuarial data, plan profitability, and program budgets. States may need to develop better risk-adjustment methods to help ensure plans' financial stability. Three of the States established capitation rates through managed competition and bidding, while two States unilaterally set rates. Although all the States had plans about how they would update capitation rates, circumstances caused them to modify their methods. Tennessee and Rhode Island increased capitation rates faster than planned when faced with evidence that plans were losing money and that some would terminate. Hawaii reduced rates because of State budget problems but also reduced the scope of benefits. Maryland originally overpaid capitation rates relative to its plan but was not able to fully lower rates in the next year to offset this initial error because of concerns about plan stability. Although some States used bidding for capitation rates and others set them by fiat, there were no simple, mechanistic systems for ratesetting. Even in the States using managed competition, the States administratively set the range of bidding and exercised substantial control over the final payment rates. To determine changes in the capitation rates, States needed to factor in not only actuarial estimates of medical costs, but information about profitability of the managed care plans and the current State budget environment.

Four of the five States had relatively simple risk-adjustment methods. Improving risk-adjustment could help ensure that plans are paid more fairly, based on the severity of the health care needs of their members. On a longer term basis, more sophisticated payment systems can aid health plans' fiscal stability. However, Maryland initially had flaws in its risk-adjustment method, which indicates the technical challenges that may be faced.

Relatively few plans dropped out of Medicaid managed care in the first several years, but reductions in plan participation may still be a problem in the future. States have had to remain vigilant in assuring that they have enough participating plans. In the first several years that we observed, the number of participating managed care plans dropped slightly. As of 1997, we found that, on average, plans made slight profits, though some had substantial losses. Typically, the exiting plans were smaller and, in some cases, plans consolidated. Modest reductions in the number of participating plans did not cause grave problems in the State systems, although there were disruptions for patients and providers in the exiting plans.

Since then, some other plans have announced they are withdrawing from the program. Most notably, TennCare's dominant Blue Cross Blue Shield plan announced that it would withdraw during the year 2000, but the State now has plans to attract more managed care plans to TennCare and to shift all plans to a regional basis.

States have a delicate balancing act: They can save money through capitation, but if the capitation rates are too low (or the burdens on plans too high), then plans may drop out. Although some plan loss may be acceptable, if too many plans leave, then the managed care system may become unstable. Similarly, within the plans, many health care providers continue to believe that their own payment rates are too low, and they, too, threaten to leave the programs.

Monitoring the quality of Medicaid managed care requires a long-term investment of State and plan resources. Because of the risk that capitated managed care may result in underservice to patients, most health care analysts believe it is important to monitor the quality and adequacy of health services delivered in managed care. In the first couple of years, States and plans were too busy with initial implementation to make much progress in quality-monitoring efforts. By the third and fourth years, States were more successful in getting HEDIS reports, in completing QARI reviews, and receiving and checking encounter data. Even so, problems and gaps in the monitoring data still exist, and no State has a fully functional system, even after 4 to 5 years. No State fully anticipated the level or type of resources that were required for a rapid startup of quality assurance and monitoring systems.

Some delays were inevitable, related to the startup of any major program and also because national guidance for Medicaid managed care quality assurance was still under development in the mid-1990s. Now that more quality measurement tools for Medicaid have been developed, the challenge for future States implementing large-scale Medicaid managed care is to shorten the learning curve.

The long-term viability of projects like these depends as much on broader economic forces as on the efforts of State Medicaid agencies. The five programs appear to have both structural and political stability, although questions remain in certain areas such as behavioral health and managed care for the disabled. The slow reduction in the number of participating plans and financial losses for some plans indicate that financial stability is somewhat more elusive. The ultimate determinants of financial stability are outside the control of Medicaid agencies. Overall health care cost pressures, including pressure by commercial payers, shape the underlying costs of plans and providers and affect their ability or willingness to accept lower prices from Medicaid programs. The late 1990s was a period of relatively slow increases in health care costs, but this is changing in the future. Equally important, the State's economic strength affects the number of participants and the budget resources available. Many of the challenges faced by Hawaii's QUEST were caused by the weak State economy; how the other States might have fared if they also had an economic downturn is an open question. Much of the concern about the recent TennCare fiscal crisis was related to an overall projected State budget deficit. The governor linked the TennCare shortfall to broader State fiscal policy, including the lack of a State income tax (Snyder, 1999). The Medicaid managed care programs, if adequately funded, can be stable—from structural, political and financial perspectives—but they do not exist in isolation from broader economic forces.

Provided with the opportunity for greater State flexibility, these five States designed, started, and fostered major changes in their State Medicaid programs. There were many challenges and occasional pitfalls, but the State agencies, together with the managed care plans and other groups, worked hard to solve problems as they arose. When first implemented, the State programs were often castigated as chaotic or poorly managed. The confusion became less serious as management systems improved and as beneficiaries and providers became more familiar with the new systems. Although all the States have gained substantial experience in operating managed care systems, the challenges of program monitoring and achieving long-term financial stability remain to be more fully addressed.