The term "managed care" may be used to describe a wide variety of arrangements that have different structures, functions, and effects on the care of people who have behavioral health disorders. The evaluation of public-sector managed care plans has been hindered by a lack of a systematic vocabulary for describing them and a lack of instruments to operationalize this vocabulary into a set of measurement procedures.
We developed and pilot tested an instrument to be used in categorizing public-sector managed care arrangements (1). The instrument was used to collect descriptive data on managed care plans in the Managed Behavioral Health Care in the Public Sector Study conducted by the Substance Abuse and Mental Health Services Administration (SAMHSA). In this multisite study, a competitive process was used to fund 21 sites in order to evaluate managed behavioral health services for four target populations. Here we report preliminary descriptive data from five SAMHSA sites at which the impact of managed care on adults with severe mental illness—one of the most vulnerable and important public-sector target populations—was studied.
We began by conducting a review of the literature on managed care in the public sector. Next, working with a template developed by investigators at George Washington University (2), we developed an initial set of critical domains by asking, "Based on current knowledge, what aspects of managed care arrangements are the most likely to have an effect on service use patterns and, therefore, consumer outcomes, broadly conceived?"
A national panel of experts composed of services researchers, managed care industry consultants, representatives of state Medicaid and mental health authorities, national mental health organizations, mental health consumers, and SAMHSA investigators reached consensus on six domains to be included in the instrument: a general description of the Medicaid managed care plan and vendor, the enrolled population, benefit design, payment and risk arrangements, composition of provider networks, and accountability. Closed-ended items were constructed to elicit information in each of these domains. Copies of the instruments, manuals, and the glossary are available from the first author.
The instrument was pilot tested at the SAMHSA site in Florida; more detailed descriptive data from the Florida study have been published elsewhere (3). Data collection at four other SAMHSA sites in four different states was conducted or coordinated by the second author and staff at the Human Services Research Institute, the coordinating center for the project. SAMHSA principal investigators at each site were responsible for verifying the accuracy of the data gathered from key informants at their site. In this column we compare and contrast the features of public-sector managed care arrangements that affect people with severe mental illness in these five states.
The data provide an overview of the "demographics" of managed care, such as waiver type, geography, number of lives covered, and case mix; the managed care structures in place in the five states; financing and risk arrangements; and information about the mental health benefit available to people who have severe mental illness. As we discuss below, there is more heterogeneity that is relevant to policy than would be suggested by simple characterizations such as "managed care" or "carve out."
"Carved out" and "integrated" mental health benefits
In four of the five Medicaid managed care programs being evaluated by SAMHSA-funded investigators (in Hawaii, in rural Oregon, in Philadelphia, and the Florida carve-out), the behavioral health premium was carved out from the general health premium, and the Medicaid agency contracted directly with behavioral health organizations for management of the behavioral health benefit.
In the Tidewater region of Virginia and in eight health plans in the Tampa Bay area of Florida, the premium for the mental health benefit was integrated with that for the health benefit. However, most of the health maintenance organizations (HMOs) in Florida carved out the mental health benefit and used risk-bearing contracts with behavioral health organizations for the provision of mental health services; six of the eight HMOs used behavioral health carve-outs. In Virginia, only one HMO used a behavioral health organization, but that HMO enrolled about 70 percent of the total Medicaid managed care enrollees. As these examples illustrate, premium integration does not necessarily result in integrated management and certainly does not result in integrated general and mental health care.
Substance abuse and pharmacy benefit carve-outs
In Hawaii and Philadelphia the mental health and substance abuse premiums were carved out together. These were the only SAMHSA sites at which there was an integrated benefit that was not extremely limited—for example, available only to pregnant women or to persons experiencing severe withdrawal from substance use.
The mental health premium and the pharmacy premium were integrated at four SAMHSA sites—the Florida HMOs, Hawaii, Philadelphia, and Virginia. However, in the Florida HMOs and in Virginia, the pharmacy benefit was administered within the health plan rather than the behavioral health plan. At the other two sites the pharmacy benefit for psychotropic medication was administered by the carve-out organization; in Philadelphia, management by the carve-out organization was limited to atypical antipsychotic medications. Florida presents an interesting contrast—the carve-out organization was not at risk for the pharmacy benefit, whereas the HMOs were. Preliminary data from Florida suggest that these risk arrangements might reduce the prescription of atypical antipsychotic agents for persons with schizophrenia by as much as 50 percent (unpublished data, Shern D, Robinson P, Boothroyd R, et al, 1999).
A similar concern about medication management was noted by Popkin and colleagues (4) in their 1998 report on the Utah Prepaid Mental Health Plan. They found that by the third year of implementation of managed care, the probability of at least a month's treatment with a suboptimal dosage of an antipsychotic medication was greater in the community mental health centers participating in the plan than in community mental health centers that still operated under fee-for-service reimbursement arrangements.
Is there such a thing as for-profit behavior?
A variety of managed care organizational types were represented at these public-sector sites, including for-profit HMOs (in Virginia and Florida), a partnership between a for-profit behavioral health organization and nonprofit community mental health centers (the Florida carve-out), a quasi-public corporation operating as an arm of city government (in Philadelphia), and nonprofit organizations (in rural Oregon). There is much concern among advocacy organizations that for-profit organizations will respond to financial incentives by emphasizing profit at the expense of clients' needs.
Whether there is such a thing as predictable for-profit and nonprofit organizational behavior is unclear from the literature. There is evidence that organizations do not always behave in a predictable way in response to financial incentives, especially in the face of countervailing incentives. For example, in the Massachusetts Medicaid managed mental health carve-out, Ma and McGuire (5) observed that when the state retained nearly all financial risk, the managed care organization consistently came close to reaching a state-specified cost target in the absence of significant financial incentives to do so. The authors' specific knowledge of the managed care organization and its arrangements with the state enabled them to provide an insightful interpretation of this unexpected finding: the managed care organization attempted to meet payers' expectations about costs to enhance its competitive advantage in subsequent bids; furthermore, it had the flexibility and clinical management tools that enabled it to do so.
Effects of managed care on "safety net" providers
At all five SAMHSA sites, "safety net" providers—in this case, traditional community mental health centers—were included in the managed care networks. Although inclusion in the network is not always a predictor of whether individual provider agencies will receive a significant portion of the referrals, these safety net providers were legal partners in the managed care organizational structures at at least two SAMHSA sites (the Florida carve-out organization and in rural Oregon). In addition, Philadelphia reported using a quasi-governmental agency as an administrative services organization on behalf of the city, increasing the likelihood that public-sector providers would play a significant role in service delivery.
What does "at risk" mean, and who is at risk?
Plans with small base rates of enrollees may be at risk of financial difficulties if costs outpace capitation rates. With the exception of Philadelphia, which reported 380,000 enrollees, none of the sites reported a substantial number of enrollees, especially given that the number reported by the Tidewater HMOs (70,000 enrollees) was spread across four plans and that reported by the Florida HMOs (68,000 enrollees) was spread across eight plans. The Hawaii site, which is a specialty carve-out for people with severe mental illness, has only 475 enrollees. Each of the Medicaid agencies risk-adjusted capitation rates, although typically only by age and eligibility—which are, at best, crude predictors of need.
All of the managed care organizations across the five states were at full risk in their contracts with the state Medicaid agencies, although few of the managed care organizations put their providers at full risk. The exceptions were community mental health centers in both Florida and rural Oregon. However, the community mental health centers in Florida that assumed full risk were partners in the carve-out organization. The fact that they were part of the legal management structure may explain the willingness of both parties to move the risk downstream. Recent data from Florida indicate that these same community mental health centers are beginning to enter into risk-bearing arrangements with HMOs (unpublished data, Shern D, Robinson P, Boothroyd R, et al, 1999).
At all five SAMHSA sites, there appeared to be an inverse relationship between assumption of risk and the application of strict utilization review. In Florida, the community mental health centers have formed a partnership with a for-profit behavioral health organization, which distributes risk to each of the participating centers. These providers are not subject to externally imposed utilization management controls.
On the other hand, in their contracts with HMOs or behavioral health organizations, these same community mental health centers have the administrative burden associated with needing external prior authorization for inpatient and outpatient services. Hawaii's managed care organization also requires prior authorization for both inpatient and outpatient services, even though there is some shifting of risk—albeit through "softer" mechanisms, such as per diem payment.
Whether cost-containment pressures in public-sector managed care programs lead to more efficient and appropriate use of services or to underutilization is probably at least partly determined by whether the public system has sufficient funding to provide an adequate "floor" of care. Large variations across states in per capita spending on public mental health suggest that the overall availability of resources could account for substantial variations in system performance. Thus it is critical to understand the context in which these systems operate as well as historical patterns of spending on behavioral health services when interpreting information gleaned from any study of managed care arrangements and their impact on outcomes.
One of the public policy concerns associated with public-sector managed care is that states are successfully shifting virtually all of the risk of Medicaid-funded behavioral health costs to private organizations that have been licensed by the state to bear risk. However, those organizations have often shifted that risk downstream to other parties—either to more poorly capitalized behavioral health organizations or to providers, especially small nonprofit providers that do not have capital reserves. Along with risk comes the incentive to deny or restrict care in order to stay solvent or to make a profit.
Whether states will pay as much attention to quality in the future as they have to cost containment in the past remains to be seen. Understanding the form and function of the structures that have been put in place by state Medicaid programs and managed care organizations to handle these public responsibilities is as critical to drawing lessons from their experience as are data on access, service use, and costs.
This work was supported in part by grant 5-UR7-TI-11278-02 from the Substance Abuse and Mental Health Services Administration, by the University of South Florida, and by the University of California, Los Angeles-Rand Research Center on Managed Care and Psychiatric Disorders. The authors thank the principal investigators at the other sites—Bentson McFarland, M.D., Ph.D., Joseph Morrissey, Ph.D., Aileen Rothbard, Sc.D., and Michael Wylie, Ph.D.—for their assistance in the collection and verification of data.
Ms. Ridgely is with the Rand Corporation in Santa Monica, California. Dr. Mulkern is with the Human Services Research Institute in Cambridge, Massachusetts. Dr. Giard and Dr. Shern are with the Louis de la Parte Florida Mental Health Institute of the University of South Florida in Tampa. Send correspondence to Ms. Ridgely at the Rand Corporation, 1700 Main Street, P.O. Box 2138, Santa Monica, California 90407-2138 (e-mail, firstname.lastname@example.org). Howard H. Goldman, M.D., Ph.D., and Colette Croze, M.S.W., are editors of this column.