Policy Challenges in Improving Mental Health Services: Some Lessons From the Past

When I was asked to give one of the opening plenary addresses at the national convention of the National Alliance for the Mentally Ill (NAMI), I began to reflect on what I had learned over the past 45 years in my various involvements in the mental health field. Here I briefly review some of those lessons.

Three broad observations define how I see the larger mental health arena. First, in every endeavor I've been involved in, I repeatedly have been impressed by the differences caring makes. Too often I've seen excellent scientists and well-informed clinicians undermine the value of their knowledge and their positive influence with patients through inattention to the power of empathic communication. Second, I've learned that popular ideas that seem intuitively correct and seductive are often wrong and sometimes are harmful. Perhaps the finest hours in my research endeavors have been when ideas I thought obviously true turned out to be wrong and I was required to reformulate my beliefs (1). Third, as someone who works both in mental health and more broadly in health policy, I am impressed that mental health advocates too often talk primarily to one another and to those already committed to their positions. Most policies that have the largest effects on the welfare of persons with mental illness are made outside the mental health sector and even outside of health care.

I elaborate on these points through discussion of five central areas: the importance of a broad research agenda encompassing basic biological and behavioral processes as well as health services research; managed behavioral health care; health insurance and parity; service system availability and linkage; and dignified employment for persons with mental illness. Over the years I have learned that there are no easy answers; that trade-offs, however painful, are needed; and that public interest and support run in cycles. Thus extended efforts are needed to protect the gains that we take for granted. Building a realistic and durable sense of hope depends on a sound foundation and not on exaggerated promises.

A broad research attack

We have repeatedly been told over the past 45 years that we are on the threshold of extraordinary advances in mental health understanding (2). Claims of revolutionary developments were made in the areas of psychoanalysis, psychopharmacology, preventive community psychiatry, the neurosciences, and, more recently, genetics. We have no doubt made some advances, learned a great deal, and developed impressive new investigational tools. But too often, enthusiastic hype far exceeds real advances or the likelihood of tangible help in the foreseeable future, a problem that characterizes advocacy for biomedical science—and indeed all science—in the American political context (3). We would be well served by being more humble in our claims and accepting that we still know relatively little about causes, processes, cures, or even good management of the major mental illnesses. A broad and sustained research effort is needed, from basic biological and behavioral processes to investigation of how to better organize and provide services.

In mental health we can't even assume as factual many ideas that most of us take for granted. For example, I have never met a clinician who did not believe that early treatment of a serious mental disorder, as opposed to later treatment, prevents subsequent disability. The idea is both intuitively appealing and theoretically plausible and appears to be a reasonably basic concept. Although this is not my area of research, and I have no reason to doubt the consensus, there has been little investigation in this area. The research that has been done provides only the most modest support for this consensus (4). The value and cost-effectiveness of early identification and treatment, like many other things in mental health, is worth knowing, because it can be important in the design of treatment strategies.

About ten years ago the U.S. Congress asked the Institute of Medicine to examine what was known about the prevention of mental illness and to develop recommendations for a prevention research agenda. Although the committee, of which I was a member, did a thorough job in defining important opportunities for research, we had to conclude that we did not have the knowledge to prevent any of the major mental disorders (5). However, it was clear that appropriate treatment and rehabilitation could limit some secondary disabilities that substantially impaired peoples' lives. Yet we know from various studies of adherence to practice standards that the major proven successful psychosocial modalities for ameliorating disability are rarely provided (6). Thus we have problems not only in extending our knowledge but also in using what we know.

Management of behavioral health care

Managed care has a bad reputation, and it is not difficult to understand the hostility it engenders in the mental health community. Mental health professionals and advocates commonly bash managed care when their efforts could better be directed toward trying to shape it in more positive directions (7). Professionals are especially hostile, because managed care has resulted in reduced fees and challenges to the clinician's autonomy and sense of professionalism. Clients and their advocates are sometimes enraged because they never before have confronted explicit rationing as exemplified by utilization review and management at the point of service (8).

Management of behavioral health care was introduced into an undisciplined environment characterized by uneven practices, a lack of interest in careful diagnosis, and poorly specified treatment standards. In the private employment sector, managed behavioral health care organizations demonstrated that they could substantially reduce costs without discernible negative impact on outcomes (7,9). Moreover, studies of the employment sector indicated that managed behavioral health care improved access to specialty mental health and substance abuse care.

Managed behavioral health care soon expanded beyond the employment sector to state Medicaid programs and to disabled populations. Here the evidence, although sparse, is more mixed. Overall, researchers have found few differences between managed care and more traditional care, but there are indications that persons with the most serious and persistent illnesses, as well as those who have been high-intensity service recipients, may have done less well under managed behavioral health care (10). There appears to be a time lag before negative effects are discerned, so short-term studies find few differences, but those extending at least two years sometimes find some decrement in outcomes among the most disabled persons.

The reasons behind such findings are not clear. We have proposed the hypothesis that in capitation systems there is a "democratization" effect, so that the intensity of services is more even among clients (11). Those who depend on a very high intensity of care may get reduced services, resulting in some poorer outcomes. Nevertheless, the dominant finding in the managed care literature both in general and in terms of behavioral health care is little difference in outcomes between traditional and managed care services (10,12).

A major inducement for government to manage behavioral health care is the potential to contain expenditures. Public programs facing fiscal pressures aspire to hold capitation levels in check or even reduce them. For-profit managed behavioral health organizations used by some states have been aggressive in constraining payments to hospitals and provider groups beyond what many believe are minimally reasonable rates. Experience differs among states and among managed behavioral health organizations, with some providing more adequate capitation and subcapitation payments. States differ as well in their sophistication in monitoring performance and in ensuring accountability.

Mental health professionals are extremely unhappy with managed behavioral health care, but most systematic studies do not support their negative appraisals (10). Some states have had serious implementation problems—for example, Tennessee's TennCare, which introduced fully capitated carve-outs without adequate preparation and support (13). In other states the transition has gone more smoothly. Managed care approaches differ in structure and strategies—and in many details—and each must be evaluated with attention to these details.

A recent careful evaluation of a two-year post-managed care period in Colorado's capitated Medicaid Demonstration Program for persons with serious mental illness is illustrative of experience in a state in which implementation was careful. Capitation contracts were given to entities in various geographic areas, including community mental health centers that were responsible for care before the demonstration. Two models of capitation were implemented and evaluated, one administered by nonprofit community mental health centers and another by a for-profit managed behavioral health organization. Three community mental health centers continued on a fee-for-service basis. An extensive evaluation carried out by investigators at the University of California, Berkeley, compared randomly selected samples of persons with serious mental illness served by these alternative approaches. The data allow comparisons of experience in the year before and the two years after the implementation of managed care.

Both capitation models reduced per-person expenditures in each of the follow-up years relative to the fee-for-service approach, although the reductions associated with the managed behavioral health organization were much larger (14). The study included a number of measures of access, and these suggest complicated patterns of change. Overall, however, access was reduced in both capitation plans, but it appeared to affect utilization in the managed behavioral health organization only.

The Colorado evaluation is notable because investigators monitored mental health measures, including the Brief Psychiatric Rating Scale (BPRS) to assess symptom severity, the CAGE to assess alcohol problems, a newly developed Drug-CAGE to assess substance abuse, the GAF to evaluate functioning, and components of Lehman's Quality of Life Interview (15). Outcomes were comparable between the fee-for-service model and the two capitated models. In those limited instances in which differences were observed, they favored the capitation models.

The issue is not managed care but how care is managed and the adequacy of capitation rates. The concept of care management had its early development in mental health, because public agencies that were largely responsible for financing and providing mental health services had to work with limited budgets. Some of the approaches we now most admire in community care for persons with serious and persistent illness, such as assertive community treatment, supported housing and employment, and psychosocial rehabilitation, were first developed and implemented in contexts in which care was managed. Capitation was seen as providing the flexibility to use services where they could do most good, break down lines of division among mental health professionals, and expand the range of reimbursable services (16).

The early innovators did not foresee the massive for-profit managed behavioral health care industry, but the challenge now is not so much to damn it as to prod it in constructive directions. We need to better understand how to ensure that clients get the needed services—not too much and not too little. This objective requires clear treatment standards based on evidence and a better understanding of what works. It also requires careful monitoring of patients over time and maintenance of continuity of care. Given that much remains unknown, easily accessible procedures for complaints and reviews must be in place so that fairness prevails and clients, clinicians, and managers can learn from experience. Norman Daniels and James Sabin have given us a good start in their model of procedural justice and how it might be implemented (17).

Much improvement is required, but it would be more productive if such improvement came through fixing managed care rather than by attempting to destroy it. Benefits must be managed if we are to achieve a viable state of parity.

Mental health parity

Parity has been an important objective of mental health advocates for decades. Parity is valued because it is fair, it provides an opportunity to improve care, and, by treating mental illness like any other, it symbolically helps undermine stigma. Parity also has value from the standpoint of health policy, because it reduces the pernicious tendency for plans and providers to seek low-risk, low-cost patients and avoid those who most need care (18).

Insurance plans and capitated providers have strong incentives to avoid enrolling people with serious mental illness. Such patients typically have comorbid mental and physical disorders and are high-cost patients. Even when payment is risk adjusted, such adjustments rarely compensate accurately for actual costs, and high-cost patients threaten the financial viability of plans and providers. In a broad sense, this is a disincentive to developing quality coverage and care, because reputational success attracts even more high-cost patients. Insurers can discourage high-cost patients with a need for mental health care from enrolling in their plans by offering less mental health coverage and poorer services. Although parity does not solve this problem, it puts all plans on a more equal footing by requiring that they provide the same benefit arrangements for mental health as they offer for other services.

Parity, however desirable, is not enough for persons with chronic and persistent illness. Most private health insurers exclude many of the services that are important for effective treatment, such as day care, psychosocial rehabilitation, case management services, housing assistance, residential care, and the like. Moreover, health insurance does not ensure that the needed services of sufficient quality are available and accessible. Studies of states that have and have not passed parity legislation inspire little confidence that parity by itself will significantly improve the scope and quality of care (19).

Finally, in focusing on parity, it is essential not to forget persons who have neither private nor public health insurance. In a recent analysis we found that the 20 percent of respondents in a national community sample of persons with serious mental illness who did not have health insurance were six to seven times less likely to receive any specialty mental health services during the year (20). The evidence from a substantial literature review shows that persons who are uninsured receive less care, receive care later, and have worse health outcomes than those who do have insurance (21).

Parity is an important part of needed care arrangements. Whenever we get close to a federal parity requirement that has any meaning, it is typically sidetracked or diluted by the contention that it will contribute to large cost increases and reduce the willingness of employers to offer health insurance to their employees. Such resistance is rationalized by the economic concept of moral hazard—the notion that increases in insurance coverage induce demand regardless of need. Mental health services are particularly vulnerable to this contention, because some such services are used for self-improvement. Before managed care, mental health services were found to be more responsive to increases in insurance coverage than other medical services (18). Without a clear distinction between serious and less serious illness, clear treatment standards, evidence-based care, and limits on duration of treatment, moral hazard is real.

There have been advances that make it more possible to implement parity. Many treatments for mental disorders are now comparable in effectiveness to other medical treatments. Diagnosis is more careful than in the past, and treatment standards have been to some extent clarified. Nevertheless, it is difficult to overcome the political opposition's moral hazard claims. NAMI has argued that parity legislation should apply only to disorders of the brain and to a limited set of diagnoses. Although this may offer a partially pragmatic solution, it makes little sense from a scientific perspective, is imprecise, and would certainly exclude persons from coverage who should be included.

One solution to the parity challenge is to provide parity in the context of a carefully managed benefit. Doing so is likely to require only modest additional cost (22). Efforts are needed to ensure that the benefit is managed in an evidence-based way, that persons who make these judgments have appropriate qualifications and training, that there is competent supervision, and that appropriate grievance procedures are in place and accessible. Such a managed care benefit would make professional mental health evaluation highly accessible but would ensure that a structure is in place to carefully and professionally manage extended treatment plans. Parity needs managed care.

Service system availability and linkage

This spectrum of needed care covers various sectors with professionals and bureaucrats with different goals, values, and career and reward structures. We have long agonized over the need to better coordinate and integrate services over time and across care domains in order to provide appropriate chronic disease care and long-term care. Although "integration of care" has been a popular buzz phrase in health systems over the past decade, few systems have been able to successfully integrate care at the clinical level. The task is inherently difficult and is even more difficult in our highly decentralized and disorganized medical care and social services systems.

I have worked over the years with several foundations that have sought to improve the integration of services for chronic illness. One such major initiative was the effort of the Robert Wood Johnson Foundation and the U.S. Department of Housing and Urban Development to improve care for persons with serious mental illness in large cities. This was a $125 million effort to build stronger mental health authorities in nine medium to large cities so as to coordinate care more effectively and improve patient outcomes. There was much enthusiasm and anticipation of its value.

One unique aspect of this effort, during a period of prevalent homelessness, was to help authorities develop housing strategies as part of their programs. It was soon evident that the Section 8 housing vouchers, available through the program, were a major incentive for cities to participate. It is also noteworthy that the largest and most complex cities in the country could not demonstrate enough coherence in their programs to compete successfully, although some reasonably large cities, such as Philadelphia and Baltimore, were included.

Looking back, we were extraordinarily naive about the magnitude of the challenge or the difficulty of achieving our objectives. Although Howard Goldman, who was responsible for the evaluation, managed to put a good spin on it (23), the outcome was in fact disappointing. From the start we felt strongly that the only real evidence of success would be to show improvement in the lives of the people involved, but no such effect could be demonstrated. There was some evidence of improved systems and increased continuity of care, but nothing that made a significant difference for clients. The optimistic view that made it appear less of a failure was that the program helped keep mental health services on the national agenda when there was little political interest or attention. The program also made clear the importance of housing as a component of a good mental health care. But for those of us who invested heavily in this effort, it was a significant letdown.

Efforts for better coordination and integration continue throughout our medical system. Although there are model programs that work well in particular instances, success in implementation on a broad scale continues to elude us. Barriers exist at all levels and are extraordinarily difficult to eliminate. Reimbursement often distorts behavior; the incentives are often wrong and counterproductive; professionals fight over turf; teams rarely work effectively together; and the relevant organizational structures are often bureaucratic and not accountable. Our chaotic overall health insurance framework simply adds to the confusion and dysfunctions.

Ironically, many of the things that need to be done are not technically difficult. For example, in a study of inpatient care, we found that linking patients who had no community provider, before discharge, with a clinician with whom the patient had some initial contact was associated with improved continuity of care (24). These patients were less likely to become homeless in the subsequent three months than patients who were not linked (4 percent compared with 14 percent) and reported less difficulty in controlling their symptoms (12 percent compared with 31 percent). Unfortunately, there are no incentives to make this effort, reimbursement is difficult, and professionals persist with the same failed practices. Supported employment provides another impressive example.

Employment of persons with mental illness

In our culture people want to work, because they gain self-respect and dignity through work. Persons with mental illness are no different. Young people with mental illness who are not in school or the workforce are doubly stigmatized. Some ten years ago we did a survey of clients for NAMI and questioned clients about their experience in 15 areas of concern. Sixty-one percent reported a need for help in getting or keeping a job, the second highest area of need (the first was in the medications area). However, although most clients believed that they were getting the help they needed with medications, only 29 percent felt they were getting comparable help in getting or keeping a job (25).

Traditional practice in the vocational area has been to take clients who need assistance in stages toward a competitive job, starting with various types of sheltered activity. Thanks to the excellent efforts of Bob Drake, Gary Bond, and others, we now know that the traditional approach to vocational rehabilitation for persons with mental illness is less effective than placing such persons directly in competitive employment and providing supports as needed (26,27). The evidence is now clear on this point, but word has yet to go out to many professionals in the work disability area who make important decisions affecting the rehabilitation opportunities for persons with mental illness.

As part of our effort to better understand the experience of persons with mental illness in the workplace, we have mined a variety of surveys. The various surveys yield different estimates, depending on the definitions and measures used, but the overall picture is reasonably clear. Some of our findings depart considerably from public stereotypes.

From 75 percent to 83 percent of persons in the general population aged 18 to 65 years are employed. The comparable percentages for persons with mental illness, depending on the survey, are 48 percent to 73 percent. Roughly a third of the population of persons with mental illness who might be expected to work if they had no such condition are lost to the workforce. Persons with severe mental illness are the least likely to work, but even in this group a third to a half manage to work. Persons with schizophrenia and schizoaffective disorders are the least likely to be employed, but even a fifth of such persons have some employment (28).

Persons with mental illness are less likely to work full-time than others. About three-fifths of the working population without mental illness works full-time, but only a third of persons with mental illness and a quarter with severe illness work full-time. Only 12 percent of persons with schizophrenia report working full-time. Many people find it surprising to learn that persons with mental illness who work are distributed across the entire occupational hierarchy, from executive, managerial, and professional positions to partially skilled or unskilled ones, much like the general population. Although persons with mental illness are more likely to work in some jobs than in others, many of the jobs held by these individuals are cognitively and interpersonally demanding and difficult.

The surveys we have worked with are insufficiently refined to enable us to fully understand employment success among persons with mental illness. Adequate analysis has to control carefully for the extent of illness and disability, but it is difficult to do so with complete faith that our adjustments capture all the relevant differences. Persons with mood disorders do better than those with other diagnoses, and the absence of comorbid psychiatric conditions has positive prognostic value as well.

By far the most important factor affecting participation and level of activity is the amount of education completed. We cannot ascertain convincingly whether and to what extent educational preparation preceded illness and the extent to which it is confounded with the seriousness and complexity of the illness. However, to the extent that educational preparation independent of illness predicts employment success, it could usefully direct efforts in early treatment and in encouraging and facilitating return to school and completion of as much schooling as is feasible.

Conclusions

Forty-five years is but a speck of time in the larger scheme of things, but mental health services look very different now than they did a half century ago. The good news is that we know more, have better treatment tools, and provide useful services to more people. The bad news is that many signs of neglect remain, particularly for the most poor and most disenfranchised individuals. As the history of mental illness attests, a decent mental health system depends as much on the interest and compassion of the larger public as it does on the concerned professionals who make mental health their work, and this area requires improvement. It is necessary to face the uncertainties of our knowledge and the uncertainties of our tasks but also the insensitivities of our practices. Through caring engagement and intellectual openness, we have significant opportunities not only to enhance the science but to improve the practice and provide more dignity and hope for those who need it.

Dr. Mechanic is affiliated with the Institute for Health, Health Care Policy, and Aging Research at Rutgers University, 30 College Avenue, New Brunswick, New Jersey 08901-1293 (e-mail, [email protected]). A version of this paper was presented at the national convention of the National Alliance for the Mentally Ill held June 27 to 30, 2003, in Cincinnati, Ohio.