Better Data for Better Mental Health Services
Evidence-based policies for improving care and treatment of those with serious mental illness are urgently needed—but good evidence is hard to find.
In 1948, Mary Jane Ward’s best-selling semi-autobiographical novel, The Snake Pit, brought widespread attention to the deplorable conditions in state psychiatric hospitals. Subsequently made into an Academy Award-winning movie, the novel’s vivid descriptions of understaffing, overcrowding, and inhumane treatment profoundly affected the general perception of treatment for individuals with serious mental illness (SMI) and prompted many states to begin making significant reforms. Widespread recognition of the need to improve the care of this vulnerable population, which had been so shockingly neglected, served as a major impetus to the development of a policy known as “deinstitutionalization.”
Deinstitutionalization shifted much of mental health care for individuals with SMI (schizophrenia, bipolar disorder, major depression, and other disorders that can result in significant functional impairments) from inpatient state psychiatric hospitals to outpatient community settings. The guiding principle of deinstitutionalization was that individuals with SMI should receive treatment in the “least restrictive setting.” This view emerged from the confluence of many factors, including the history of abuses in state hospitals, the development and widespread availability of new psychotropic medications, and an increasing societal concern for civil liberties. In particular, the advent of new antipsychotic medications in the 1950s and 1960s allowed, for the first time, limited control of delusions and hallucinations, and therefore made life in the community a possibility for persons with serious and chronic mental disorders.
The dramatic changes in the mental health system that have taken place over the past 50 years had their origins in the Community Mental Health Systems Act, signed in October 1963 by President John F. Kennedy and conceived with the noblest of intentions. But in policy makers’ haste to correct the abuses revealed in state hospitals, deinstitutionalization was carried out, in the words of psychiatrists H. Richard Lamb and John Talbott, writing in the Journal of the American Medical Association (JAMA) in 1986, with “much naïveté and many simplistic notions.” In his recent book, American Psychosis, E. Fuller Torrey, a former National Institutes of Mental Health psychiatrist, traced such notions to the Interagency Committee on Mental Health, whose 1962 report influenced the subsequent law: “Because no committee member really understood what the hospitals were doing, there was nobody who could explain to the committee that large numbers of the patients in these hospitals had no families to go to if they were released; that large numbers of the patients had a brain impairment that precluded their understanding of their illness and need for medication; and that a small number of the patients had a history of dangerousness and required confinement and treatment.”
Torrey, founder of the Treatment Advocacy Center, a national nonprofit organization dedicated to eliminating barriers to the treatment of severe mental illness, argues that the 1963 law was fatally flawed because it encouraged the closing of state mental hospitals without any realistic plan as to what would happen to the discharged patients, especially those who refused to take medication they needed to remain well. It did not include a plan for the future funding of mental health centers, and it focused on prevention when no one understood enough about mental illnesses to know how to prevent them.
Discharging long-term patients from institutions was a way for states to cut their expenses, since outpatient therapy and drug treatment were less expensive than inpatient care. Increasing attention to the civil liberties of those involuntarily hospitalized also brought the enactment of laws in many states that made it much more difficult to hospitalize the mentally ill against their will.
However, although the number of patients discharged from state hospitals increased and the number of inpatient psychiatric beds declined precipitously after 1960, the planned network of 1,500 community mental health centers, which was intended to assume responsibility for the care of those with SMI, failed to fully materialize because of a chronic lack of funding and shifts in political priorities. Only half of the proposed centers were ever built.
For years following the initial wave of deinstitutionalization, many individuals with SMI—either newly discharged from state hospitals, or in psychiatric crisis—were left to fend for themselves in ”board-and-care” homes or group homes with little or no supervision or treatment other than psychotropic medication. These homes were often clustered in certain communities; one of the best-known locations in the 1970s was Long Beach on New York’s Long Island, which housed hundreds of former patients discharged from several very large state hospitals located nearby (some of these hospitals, such as Creedmoor and Pilgrim State, had more than 10,000 beds in the mid-1960s). At the time, the concerns of mental health professionals and advocates focused on the potential for residents of these board-and-care homes to be victims of crime and on quality of life issues raised by a lack of appropriate treatment, lack of daily structure or employment, and isolation and lack of social support.
Today, deinstitutionalization is viewed by most experts as a policy failure, and the mental health system more broadly is recognized as unable to meet the needs of persons with SMI. Many experts also believe that these failures are the cause of increases in homelessness among seriously mentally ill persons, as well as a dramatic increase in the number of persons with SMI seen in hospital emergency departments, which increased from 5.4% of total visits in 2000 to 12.5% in 2007. This increase has also led to the need for “boarding” when no psychiatric beds are available; the average wait time for a psychiatric admission in general hospitals is currently more than 18 hours, compared with just under six hours for non-psychiatric admissions.
One consequence of these problems has especially come to the fore. In the past few years, tragic and violent events such as the mass shootings in Newtown, CT, and Aurora, CO, have given a new impetus to ongoing concerns about the adequacy of mental health treatment. Following the shooting at the Navy Yard in Washington, DC, in September 2013, Jeffrey Lieberman, president of the American Psychiatric Association, issued a statement titled “The U.S. Mental Health Care System is Broken” in which he noted that there had been 21 mass shootings in the country since 2009, and the perpetrators in over half of these were suffering from or suspected to have an SMI. He emphasized that the system of mental health care in the United States is “inadequate, where individuals with mental illness too often fall through the cracks.”
The dramatic and continuing reduction in the number of inpatient state psychiatric beds in recent decades has been a source of concern and alarm among many observers in the field. According to “No Room at the Inn,” a 2011 report by the Treatment Advocacy Center, the number of public psychiatric beds in the United States per 100,000 population fell from 340 in 1955 to 17 in 2005.
Some observers have suggested that the decrease in public beds has been at least partially offset by increases in beds in private psychiatric and general hospitals. But as psychiatrists Benjamin Liptzin, Garry Gottlieb, and Paul Summergrad pointed out in a 2007 commentary in the American Journal of Psychiatry, although it is true that there were very modest increases in both types of facilities in the 1980s and 1990s, the numbers have subsequently decreased to near their previous levels. Among the likely reasons for the decline are poor reimbursement for psychiatric hospitalizations from all payer sources and conversion of these beds to medical-surgical beds, which were needed and also contribute much more to hospital margins.
Similarly, the Subcommittee on Acute Care of the New Freedom Commission appointed by President George W. Bush reported in 2004 that from 1990 to 2000, the number of inpatient beds per capita declined 44 percent in state and county mental hospitals, 43 percent in private psychiatric hospitals, and 32 percent in nonfederal general hospitals. And more than three-fourths of psychiatric beds in general hospitals are in private facilities that are often reluctant to admit uninsured individuals or those who are deemed to be “disruptive” or “too violent.” The American Medical Association describes the problem of access to psychiatric beds and overcrowding of emergency departments as “an urgent crisis and a national disgrace.”
The sharp decline in the number of beds and the changing philosophy regarding hospitalization have led to a decrease in the median length of stay (LOS) in state facilities. Historically, the presumed purposes of state mental hospitals were to monitor the course of illness and provide psychiatric treatment, medical care, rehabilitation, short- and long-term asylum, residential care, crisis intervention, and social structure. With a relatively short LOS, however, many of these goals are not attainable, leading to a qualitative change in the type of care these facilities are able to provide, from long-term treatment to acute care with relatively quick discharge. When patients’ conditions are only partially stabilized at discharge, and if they are discharged without adequate attention to transition planning, then outpatient treatment regimens are more likely to fail, resulting in frequent hospital readmissions.
But there has been another alarming, if predictable, consequence of the reduction in national capacity to treat people with SMI. Seventy-five years ago, in a seminal article called “Mental Disease and Crime: Outline of a Comparative Study of European Statistics,” Lionel Penrose, a British psychiatrist, medical geneticist, and mathematician, found an inverse relationship between prison and mental health populations, and theorized that if one of these forms of confinement is reduced, the other will increase. It seems reasonable to assume that many of the individuals with SMI who are seen today in jails and prisons in the United States, particularly those who committed minor crimes, could just have easily been hospitalized if psychiatric beds had been available. As H. Richard Lamb points out, “Unfortunately, the inadequate and underfunded community treatment of persons who are the most difficult to treat, and the insufficient number of hospital beds (acute, intermediate, and long term) for those who need them, are some of the realities of deinstitutionalization that have set the stage for criminalization.”
A special report by the Bureau of Justice Statistics in 2006 found that at mid-year 2005, more than half of all prison and jail inmates had some type of mental health problem; other studies have found that between 1984 and 2002, the estimated prevalence of SMI among male jail inmates tripled, from 6.4 percent to 17.5. percent. In November 2014, the Treatment Advocacy Center reported that approximately 20 percent of inmates in jails and 15 percent of inmates in state prisons had an SMI. Based on the total number of inmates, this would translate into approximately 356,000 inmates with SMI in jails and state prisons—10 times more than the approximately 35,000 individuals with SMI remaining in state hospitals.
Ethicists Dominic Sisti, Andrea Segal, and Ezekiel Emanuel point out that care for an inmate with mental illness in a correctional institution ranges from $30,000 to $50,000 per year, compared with $22,000 per year for an inmate without mental illness. Moreover, they describe the environment for inmates with mental illness as “anathema to the goals of psychiatric recovery”—often unsafe, violent, and designed to control and punish. They maintain that long-term inpatient settings are a “necessary but not sufficient component of a reformed spectrum of psychiatric services” that will continue to be essential to mental health patients who cannot live alone, cannot care for themselves, or are a danger to themselves or others.
“The New Asylums,” a segment of PBS’s Frontline series that first aired in 2006, provided an in-depth look at Ohio’s state prison system and the complex and growing issue of caring for mentally ill prisoners. The filmmakers had access to prison therapy sessions, mental health treatment meetings, crisis wards, and prison disciplinary tribunals; and the film provides a graphic and disturbing portrait of the new reality for the prison system and for its mentally ill inmates. Scenes of group therapy sessions conducted while the participants are confined to metal enclosures—essentially small cages—are particularly unnerving, as is the sequence when the camera follows a group of prison guards dressed in black riot gear with facemasks and helmets as they enter the cell of an agitated psychotic prisoner who needs to be transported to the prison’s infirmary.
A dearth of data
In a world where both policy and medicine are increasingly expected to be “evidence-based,” the evidentiary basis for addressing SMI in the United States is disturbingly weak. From assessing hospital and residential care capacity, to developing consensus diagnosis and treatment regimes, many important questions remain unanswered.
Making progress on helping people with serious mental illness will depend not just on new drugs but on good information on which effective policies and treatment regimens can be based.
A recent search of the Department of Health and Human Services’ (HHS) National Registry of Evidence-Based Practices, which focuses exclusively on mental health and substance abuse services, found that only 30 of the 355 entries in the database mentioned “serious mental illness,” “schizophrenia,” or “psychosis,” suggesting the need for increased efforts at consensus development regarding evidence-based practices for people with SMI. A July 2015 report by the National Academy of Medicine points out that “a considerable gap exists in mental health and substance abuse treatments known as psychosocial interventions between what is known to be effective and those interventions that are commonly delivered . . . [This gap] is due to problems of access, insurance coverage, and fragmentation of care—which include different systems of providers, separation of primary and specialty care, and different entities sponsoring and paying for care.”
Although the prevalence rate of SMI is relatively similar across states, the number of state psychiatric hospital beds per 100,000 civilian population currently varies widely from state to state, from 3.9 beds per 100,000 population in Arizona, to 30.1 beds per 100,000 in Wyoming. The reasons for these differences are not well understood, but they seem to reflect the lack of consensus on the purpose of these beds; many are used for forensic purposes (for example, for defendants found not guilty by reason of insanity) and thus are not available to other persons with SMI living in the community. In 2010, 38.2 percent of state psychiatric hospital budgets nationwide were accounted for by forensics and sex offender services, up from 25.6 percent in 2001. A report by the Treatment Advocacy Center found that nationwide in 2010 about one-third of public psychiatric beds were occupied by forensic patients, but the situation differed dramatically among the states, ranging from 66 percent in Ohio and 57 percent in Oregon, to less than 5 percent in in Idaho, Iowa, Mississippi, New Hampshire, North Carolina, North Dakota, and South Dakota.
One possible reason for the apparent differences among the states is that there is neither consensus on the nature of an inpatient psychiatric bed, nor on the number of inpatient psychiatric beds needed in the United States; nor is there a generally accepted or agreed-upon method among policy makers or researchers for projecting or estimating how many beds are needed. Although planners in many states have devoted serious effort to grappling with this issue, as evidenced by recent reports issued by such organizations as the Washington State Institute for Public Policy, the California Hospital Association, and the North Carolina Department of Health and Human Services, a consistent and effective strategy remains elusive. A 2012 survey by the National Association of State Mental Health Program Directors found that only 16 states reported having any method for making such projections; most of those states that did report a method indicated that they relied on previous use data or benchmarking against other states. My own search of the literature did not uncover other documented methods, with the exception of a commercial simulation model for mental health planning called Planning by the Numbers (PBN), which was initially developed over 30 years ago and may still be in use, although I could not find any information on its potential users or their experiences. As with any model, projections using PBN will vary widely, depending on assumptions about the availability of resources in the community and attitudes toward hospitalization. Its value for informing policy is thus unclear, all the more so given the lack of uniformity around definitions of key variables such as “inpatient bed,” as well as other data problems discussed here.
Complexities in gathering and analyzing data related to psychiatric beds do not end there. Forensic patients may have longer lengths of stay than other patients, which complicates state-by-state comparisons. Many states contract with private psychiatric hospitals or community hospitals for the use of beds, complicating the definition of a “state” bed. Obtaining consistent information regarding the number of psychiatric beds in private psychiatric hospitals and community and general hospitals is also difficult, since beds in designated psychiatric units in general and community hospitals often do not generate as much income as beds for other purposes, and psychiatric patients may be housed instead in “swing” or “scatter” beds in medical and surgical units.
The federal government has no oversight or regulatory role in relation to the number of psychiatric beds or the appropriate ratio to total beds, and experts and stakeholders alike disagree about how many beds there should be—or even if they are needed at all. For example, the “consumer/survivor” movement, which has gained widespread attention over the past two decades, is predicated on the idea that SMI is best dealt with through mutual support from peers with mental illness who have “survived” the interventions of psychiatry. Some individuals within that movement believe that encouraging adherence to medication regimes is “paternalistic,” that inpatient hospitalization has no place in the mental health system, and that “recovery” should be entirely self-directed. Many professionals, however, believe that state hospitals play a crucial role in the continuum of care, and that there will always be some individuals whose disorders cannot be treated solely in the community and who need the structure of a more protected setting. In the words of Howard Goldman, a well-known expert in mental health policy, “Many [researchers] have foundered on the shoals of trying to address and answer the question of how many psychiatric beds are needed.”
If good data on psychiatric hospital capacities are scarce and ambiguous, data on residential care—another important component of the treatment system for people with SMI—is even more problematic. Detailed and reliable information on the number of beds in residential settings is very limited and difficult to interpret. Definitions vary widely across states, leading to inconsistencies in reported numbers. In a 2004 article on this problem, psychiatrist Martin Fleishman observed that residential care facilities (RCFs) are also known as board-and-care homes, adult residential facilities, community care homes, and sheltered care facilities, among many other names, which, in turn, has discouraged national statistical categorizations. Fleishman further notes that “Data collection for RCF patients is complicated by the fact that it is difficult to distinguish the RCF population from the population of other community-based domiciles for long-term patients, such as nursing homes.” Based on data from California, he estimates that almost 160,000 persons who are mentally ill occupy RCF beds in the United States, although he suggests it is likely that the real number is considerably higher.
The lack of good data on psychiatric residential facilities is hardly surprising. A 2007 national survey of regulation and certification for these facilities found the regulatory environment to be very complex: in most states, several separate state agencies with differing missions and functions (state mental health authorities, departments of health, departments of social services, and so on) are involved in the licensing, funding, and oversight of the facilities. According to Fleishman, “Because of the difficulties in obtaining reliable statistics, little research has been done on the population of persons with mental illness who require long-term care, and the most effective modalities of treatment have yet to be determined.”
Comprehensive national data on residential psychiatric facilities is also critical to a complete understanding of treatment for persons with SMI. Such data are not available, however. The most recent national survey of psychiatric residential facilities for adults in the United States was conducted in 1987. And although the U.S. Department of Housing and Urban Development conducted a new national survey of residential facilities in 2010, that survey description states that “residences licensed to serve exclusively persons with mental illness, mental retardation, or developmental disabilities are ineligible,” although no explanation is given. An extensive, multi-stage national survey of psychiatric residential facilities currently being conducted in Italy might serve as a useful model for such an effort in the United States. The Progetto Rezidenze (PROGRES) residential care project, funded by the Italian Institute of Health, is described as “the first systematic attempt in Italy to fill the gap between psychiatric services planning and evaluation, by setting up a network of investigators throughout the country and evaluating an entire typology of services in a consistent fashion.” Similar studies of residential facilities have recently been conducted in Australia and Denmark.
Needed: policies for evidence
Researchers working to understand the prevalence of behavioral health disorders currently depend on large-scale, federally funded household surveys, such as the National Survey on Drug Use and Health (NSDUH). Such computer-assisted data collection efforts excel in providing self-reported data on trends such as the use of illegal drugs, but are limited in the data they can provide on mental disorders, particularly those of a more serious nature. For example, as household surveys, they capture the “civilian, non-institutionalized population,” but they do not collect data from individuals in correctional and psychiatric institutions, or from the homeless. Although they do provide some information on self-reported SMI, the methodology of these surveys limits their ability to capture data on individuals with the most severe conditions.
How, then, might data collection on SMI be improved? One possibility, suggests William Eaton of Johns Hopkins University, would be to re-examine the potential for the type of “psychosis registries” that existed in certain locations, such as the states of Maryland, Hawaii, and North Carolina; Rochester County, New York; and Washington Heights, New York, in the 1950s through 1970s. There are comprehensive psychiatric registries in Denmark, Sweden, Finland, Israel, and Taiwan, among other countries. Comparable comprehensive longitudinal databases do exist today in the United States for other types of illnesses; for example the Surveillance, Epidemiology, and End Results (SEER) Cancer Registries have been run by the National Cancer Institute since 1973. It is possible that health maintenance organizations such as Kaiser Permanente, Geisinger, and Group Health Cooperative will push their data collection toward these types of comprehensive longitudinal registries. Although confidentiality concerns may present a considerable challenge, the breadth of information that such registries could provide would potentially be extremely valuable to researchers, stakeholders, and policy makers seeking to improve services to the SMI population. Another approach might be to consider altering existing household surveys to inquire specifically about siblings and other household members with SMI who are not currently residing in the home, assuming that this could be done within the limitations of confidentiality and privacy laws.
Improved data from modified survey methods or psychosis registries might help researchers to develop algorithms for estimating the frequency and intensity of episodes that are likely to require crisis intervention, short-term, and longer-term hospitalization among persons with SMI. Such estimates are a necessary foundation for planning new facilities. Increased attention to the collection of data on location and availability of mental health resources in communities, and improved identification of areas with shortages of mental health facilities and providers, is also important; new mapping technologies may prove to be valuable tools for the assessment and redistribution of such resources.
National-level data are also needed on the availability and effectiveness of other services. In addition to a lack of community mental health centers, many communities are unable to provide the “wraparound” services that persons with SMI often need, such as supported housing, vocational education, social and peer support, crisis management teams, and interventions such as “assertive community treatment.” Such services are often costly and not reimbursable, although they are widely believed to be important for individuals with SMI.
Making progress on helping people with SMI will depend not just on new drugs but on good information on which effective policies and treatment regimens can be based. To start with, we need to understand how the current policy regime, with its focus on deinstitutionalization, has influenced the delivery of mental health services over the past five decades. As Richard Frank and Sherry Glied observed in their 2006 book Better But Not Well, a comprehensive, longitudinal database would provide the best foundation for such an assessment. Given that no such single database exists, Frank and Glied instead combined information from multiple sources—administrative data, epidemiological surveys, general health and medical surveys, and research studies on the effectiveness of specific therapies. They concluded that “improved treatment for mental illness, a growing supply of mental health professionals, and enhanced private insurance coverage have contributed to greater use of services by those with less serious conditions,” although it is not clear that people with SMI have benefitted from these improvements to the same degree.
Lack of coordination among federal programs also contributes to the challenge of good data, sound analysis, and effective policies. A 2014 U.S. Government Accountability Office (GAO) report entitled “Mental Health: HHS Leadership Needed to Coordinate Federal Efforts Related to Serious Mental Illness” found that the 112 federal programs that generally supported individuals with SMI were spread across eight federal agencies, and that only 30 of the 112 programs were specifically targeted toward persons with SMI.
The report also found that “agencies completed few evaluations of the programs specifically targeting individuals with serious mental illness.” GAO recommended that HHS establish a mechanism to facilitate interagency coordination across all programs that support individuals with SMI, and also that a coordinated approach to program evaluation should be implemented. In its written comments on the report, HHS disagreed with both recommendations.
The development of a comprehensive and coordinated research agenda for improving delivery of services to persons with SMI is crucial if the situation is to be improved. This agenda, on which federal, state, and local governments should collaborate, must include a focus on the identification and dissemination of evidence-based practices, and should emphasize the development of financial and regulatory incentives, such as pay-for-performance approaches, to encourage high quality care.
Awareness of the urgent need for such efforts is growing. The Helping Families in Mental Health Crisis Act of 2013 (H.R. 3717), re-introduced in the House in June 2015 by Rep. Timothy Murphy (R-PA) and Rep. Eddie Bernice Johnson (D-TX), would create an Assistant Secretary for Mental Health and Substance Use Disorders as well as a National Mental Health Policy Laboratory and an interagency Serious Mental Illness Coordinating Committee. The legislation has strong endorsements by organizations such as the American Psychiatric Association and the National Alliance for the Mentally Ill, and momentum for mental health reform appears to be building. In a bipartisan vote on November 5, 2015, the Energy and Commerce Health Subcommittee voted to advance the legislation, which would increase funding for additional outpatient and inpatient treatment slots, add new enforcement provisions to the mental health parity law, and ease some privacy restrictions to help parents obtain more information about their adult children’s treatment. The bill would also allocate more money for research into the causes and treatment of mental illness and remove a rule that bars Medicaid from paying for mental health treatment and physical health treatment on the same day. It would also establish a new office at HHS devoted to providing oversight of the federal government’s role in mental health care, headed by the Assistant Secretary for Mental Health and Substance Abuse Disorders.
Better data are critically needed in order to inform these proposed policy changes. Steps to achieve this goal should include:
- Creating a comprehensive national research agenda on treatment of persons with SMI, through focus groups and consensus panels that include providers, payers, family members, mental health advocates, and consumers
- Improving available data on the prevalence of SMI through integrated and innovative data collection efforts aimed at including people who are incarcerated, homeless, and otherwise not captured in current household surveys
- Convening a national panel of experts, including researchers, representatives of state mental health agencies, and policy makers, to use their expertise to develop a rigorous scientific methodology for estimating the need for inpatient psychiatric beds
- Conducting targeted surveys of psychiatrists and other treatment providers to improve our understanding of their experiences and to leverage their knowledge in treating patients with SMI
- Conducting comprehensive research efforts on the effectiveness of innovative approaches such as court-ordered assisted outpatient treatment and assertive community treatment
- Leveraging the increasing availability of detailed claims data from public and private sources to improve our understanding of treatment trajectories for people with SMI
As the Affordable Care Act brings positive changes in the health care system as a whole, it will be vitally important to ensure that substantial and widespread improvements in the care of persons with SMI, and increases in appropriate and adequate facilities, are included. More comprehensive data are crucial to assist policy makers in focusing on those parts of the mental health system most in need of attention, and to aid in developing solutions for this most vulnerable population.
M.L. Berk, C.L. Schur, and J. Feldman, “Twenty-Five Years of Health Surveys: Does More Data Mean Better Data?” Health Affairs 26, no. 6 (2007): 1599-1611.
William W. Eaton, ed., Public Mental Health (New York, NY: Oxford University Press, 2012).
Richard G. Frank and Shelly Glied, Better But Not Well: Mental Health Policy in the United States Since 1950 (Baltimore, MD: The Johns Hopkins University Press, 2006).
Gerald N. Grob and Howard H. Goldman, The Dilemma of Federal Mental Health Policy: Radical Reform or Incremental Change? (New Brunswick, NJ: Rutgers University Press, 2006).
M. Horvitz-Lennon, J.M. Donohue, M.E. Domino, and S.T. Normand, “Improving Quality and Diffusing Best Practices: The Case of Schizophrenia,” Health Affairs 28, no. 3 (2009): 701-712.
B.A. Nicks and D.M. Manthey, “The Impact of Psychiatric Patient Boarding in Emergency Departments,” Emergency Medicine International (2012): doi:10.1155/2012/360308.
Mark Olfson, Harold A. Pincus, and Herbert Pardes, “Investing in Evidence-Based Care for the Severely Mentally Ill,” JAMA 310, no. 13 (2013): 1345-1346.
E. Fuller Torrey, American Psychosis: How the Federal Government Destroyed the Mental Illness Treatment System (New York, NY: Oxford University Press, 2014).