Reverse Integration Initiatives for Individuals With Serious Mental Illness

Abstract

Medical progress has greatly extended the life span of individuals living in the United States, yet certain groups have lagged behind in achieving wellness and longevity. Prominent among these are individuals with serious mental illness. Because of this, various initiatives have been launched at the community, state, and national level to improve the medical care of those with serious mental illness. Many of these initiatives promote “reverse integration,” or the provision of collaborative care services in behavioral health locations. Despite significant barriers to implementation, these initiatives have shown moderate success in improving medical outcomes for those with serious mental illness, in both research and “real-life” settings. Additionally, the role of psychiatrists in addressing physical health has been explored, and there is a need for educational opportunities to optimize competency in this area. Overall, work still needs to be done before the mortality gap for those with serious mental illness dissipates.

Scope of the Problem

The early mortality of individuals with serious mental illness was documented as early as the 1800s (1). Over the subsequent centuries, treatment advances have greatly increased the life span of individuals in the general population. However, those with serious mental illness have lagged behind, which has resulted in a widening disparity in mortality. A U.S. study in 2003 that gathered data from patients in the public mental health sector found that people with serious mental illness were dying, on average, 25 years earlier than the general population. More recent studies confirm the persistence of the mortality gap (2, 3).

Although increased mortality of those with serious mental illness was initially thought to be due to violent death (accidents, homicide, and suicide), increasing evidence revealed that natural causes were largely to blame. In 1932, Malzberg published a landmark longitudinal study of individuals institutionalized in New York state psychiatric hospitals, demonstrating an average of 15-year premature mortality compared with the general population. He reported that this mortality gap was largely due to “natural” rather than “unnatural” causes (4). Subsequent research has shown that death from natural causes contributes approximately 60% to early mortality of people with serious mental illness (5); a recent meta-analysis demonstrated that approximately 67% of deaths among people with mental illness are due to natural causes (6).

Medical comorbidity contributes heavily to the premature medical mortality of those with serious mental illness. Medical illness is highly prevalent among individuals with psychiatric diagnoses: More than 68% of adults with mental illness were found to have at least one medical disorder in the 2001–2003 National Comorbidity Survey Replication (7). Rates of disease among those with serious mental illness exceed those of the general population in every disease category (8), and individuals with serious mental illness have higher standardized mortality ratios compared with the general population for cardiovascular, respiratory, and infectious diseases (8).

The etiology of increased medical morbidity and mortality among those with serious mental illness is complex and multi-factorial. Modifiable risk factors, including adverse health behaviors, contribute significantly. Prominent among these are poor diet and low levels of physical activity, which are highly prevalent among people with serious mental illness (9–11).

Perhaps even more impactful is substance use, including heavy use of tobacco; up to 44% of all cigarettes smoked in the United States are consumed by individuals with mental illness (12). Additionally, side effects of medications prescribed for patients with serious mental illness also contribute to medical morbidity. Adverse metabolic effects, such as weight gain, have been associated with many psychotropic medications but are most prominent with second-generation antipsychotic use (13). Moreover, adverse social determinants of health contribute to medical comorbidity of those with serious mental illness. Persons with mental illness have increased rates of poverty, addiction, lack of access to healthy food choices, unsafe living conditions, exposure to early trauma, chronic psychological stress, and poor social networks (14). The impact of these factors on physical health is well documented.

A final factor contributing to the mortality gap among people with serious mental illness is poor quality of medical care. Individuals with serious mental illness receive fewer routine preventative services, including immunizations, age-appropriate cancer screening, and disease-specific standard-of-care screening (e.g., routine eye examination for people with diabetes), than members of the general population (2, 8). Moreover, common chronic illnesses frequently go undiagnosed and untreated among persons with mental illness (15, 16). Even when diseases are recognized, quality of care is inferior. Individuals with serious mental illness and cardiac disease are less likely to receive angioplasty or coronary artery bypass grafts. After myocardial infarction, persons with serious mental illness are less likely to receive drug therapies of proven benefit, such as beta-blockers, angiotensin-converting enzyme inhibitors, and aspirin (17).

The poor quality of care of individuals with serious mental illness is partially explained by patterns of health care utilization: because of poor access to continuity of quality medical care, persons with serious mental illness underuse primary care services and overuse emergency and medical inpatient care (18). Many factors contribute to this pattern of health care utilization, including fragmentation of the health care system and resulting poor coordination of care between medical and mental health (8); finances, including lack of adequate health care coverage (19); symptoms of mental illness that lead to difficulty with keeping appointments and following recommended treatment plans (15); and the stigma of mental illness, which leads to poor therapeutic alliance between patients with serious mental illness and health care providers (15).

Conceptual Framework for Behavioral Health Integration

To eliminate the mortality gap among people with serious mental illness, the health care system likely needs to address all of the contributing factors. Yet perhaps the most intensive effort thus far has focused on improving access to quality medical care for those with serious mental illness. As described above, systematic barriers to care experienced by those with serious mental illness result in a lack of utilization of primary care; mental health clinics are thus likely to be the principal connection that such patients have to the health care system (20). Because of this, initiatives to improve access to medical care for those with serious mental illness have largely focused on behavioral health integration located in mental health clinics, or so-called reverse integration.

Reverse integration initiatives have generally followed the principles originally outlined by Wagner for the treatment of complex individuals with chronic medical illness (21). These principles are a key component of other collaborative chronic care models, including the patient-centered medical home outlined in the Affordable Care Act of 2010 (22), as well as traditional behavioral health integration initiatives based in primary care settings (23). Core components of Wagner’s chronic care model include patient self-management support, such as coaching for behavioral change, engaging patients in shared decision making, and providing problem-solving therapies; use of clinical information systems, such as patient registries; redesign of delivery systems such that preventative rather than reactive care may be provided; support of generalist providers to allow informed decision making, via the use of expert consultation and care algorithms; linkage to community resources, such as housing and support groups; and support from leadership in the health care organization, including adequate clinical staff and resources (Table 1) (24). All randomized controlled trials (RCTs) to date have included at least three of these chronic care model principles. The principle that was most commonly excluded was support from the health care organization (25), which is essential for real-world initiatives to succeed.

TABLE 1.

Core Components of Wagner’s Chronic Care Model^a

Component	Focus	Example
Support of patient self-management	Allowing patients to recognize and manage their own symptoms, increase their sense of control, and participate more actively in their clinical care	Use of care managers to implement education programs, problem solving, and goal setting
Use of clinical information systems	Facilitation of the flow of information to improve patient care	Use of patient registries, pharmacy refill monitoring to assess medication adherence, shared electronic medical records for medical and behavioral health care
Delivery system redesign	Redefinition of roles for physicians and staff to promote preventative care, and creation of new positions as needed to support care model	Use of case managers and health educators, designated staff to monitor registries, increased screening
Decision-making support	Provision of expert-level input without need for real-time specialty consultation	Use of algorithms and practice guidelines, facilitated consultation with specialist as needed
Linkage to community resources	Support for nonclinical needs	Housing resources, peer support groups
Support from the health care organization	Leadership and resources to support practices	Provision of key staff and services, commitment to sustainability

^aSources: Wagner et al. (21), Woltmann et al. (24), and Gerrity et al. (25)

The majority of studies employed a care manager to apply Wagner’s principles. Care managers interfaced with patients to encourage and educate them on self-management, monitored registries and managed clinical information systems, communicated with outside consultants concerning patient care, and linked patients with community resources. Care managers in these RCTs consisted of nurses, master’s-level health specialists, or midlevel providers (e.g., nurse practitioners). Most studies also provided additional training for the care managers, related specifically to the intervention. In the majority of studies, the care managers’ activities were overseen by a physician or the study’s principal investigators (25). Caseloads for care managers varied according to the intensity of the intervention and ranged from approximately 20 to 125 patients (26–28).

Although reverse integration projects generally apply Wagner’s principles of chronic disease management in their conceptual framework, the projects vary in their model of operation and level of collaboration. Collaboration in behavioral health integration projects may be viewed, theoretically, as operating across a continuum of integration, from none to full integration. On one end of the spectrum are initiatives that employ off-site collaboration. In these circumstances, there is generally minimal discussion among various off-site providers; however, facilitated referral may occur. Moving toward greater integration, initiatives may employ colocated care. Although these circumstances allow a greater opportunity for direct communication among providers, different cultures or different record-keeping systems may prevent optimum collaboration. Finally, care models may provide fully integrated care, with teams of providers creating collaborative care plans (29). Rarely, a single provider can provide fully integrated care (usually a physician dually trained in internal medicine or family medicine and psychiatry), with support from a multidisciplinary team (see Table 2).

TABLE 2.

Continuum of Integration of Care in Behavioral Health Settings^a

Type of Care	Location of Medical Care	Potential Medical Services Offered On Site	Structure of Collaboration
Coordinated care	Off site	Medical care managed by psychiatrists:judicious prescribing; screening for metabolic conditions; counseling on lifestyle changes; management of common medical conditions; health navigators; care/case managers	Minimal collaboration; Patients are referred to a primary care provider at another site, often with limited communication
Colocated care	On site	Facilitated referrals; communication about care plans among providers; sharing of medical records	Providers are located in the same physical site but may or may not share medical records and treatment plans
Integrated care	On site	Fully integrated primary care services;specialty medical services	Providers plan and implement care plans for all patients; in some cases, a dually boarded physician manages medical and psychiatric conditions

^aAdapted from Wise and Reynolds (29)

Much of the work done thus far in reverse integration has been via colocation or integrated systems (25), likely because of the complexity of patient care needs and challenges with access. Although reverse integration initiatives with colocated primary care offer the possibility of increasing access to medical care for those with serious mental illness, the majority of patients with serious mental illness receive care in behavioral health settings that do not offer colocated primary care services. Thus, psychiatrists may be the only physicians persons with serious mental illness encounter. Care for the physical health of these patients may, by default, fall in the hands of psychiatrists. Accordingly, the role of the psychiatrist in providing fully integrated care, by addressing both medical and psychiatric comorbidities, has been explored.

In 2015, the American Psychiatric Association released a position statement advocating for the expansion of the psychiatrist’s scope of practice to include general health conditions (30). In line with these recommendations, psychiatrists may employ various models, within a continuum of level of involvement, to meet the medical needs of individuals with serious mental illness. On one end of this continuum, psychiatrists may manage chronic medical conditions themselves. They may achieve optimum competency in practicing integrated care by completing combined training in internal medicine or family medicine and psychiatry. However, management of medical comorbidity does not require completion of a combined residency; there is a growing emphasis in general psychiatry residency programs on educating trainees on the assessment and treatment of common medical problems (14). Practicing psychiatrists may gain more medical knowledge through continuing medical education or self-learning. Continuing medical education opportunities increasingly have arisen to prepare practicing psychiatrists to provide medical care for their patients who are most in need (31).

However, many psychiatrists are not comfortable with managing the physical health of their patients unassisted. In such cases, psychiatrists may still play an active role in attending to their patients’ physical health, by screening for common medical conditions, making facilitated referrals for medical care, monitoring the quality of medical care that patients may be receiving, and communicating directly with primary care physicians to express concerns about their patients’ health. Such interactions also allow an opportunity for psychiatrists to employ their liaison skills to educate primary care providers about mental illness, thereby improving the relationship between individuals with serious mental illness and primary care physicians.

At a minimum, psychiatrists may reduce harm for persons with serious mental illness through judicious prescribing of antipsychotics. This includes practices such as prescribing the minimum effective dose of medication, regularly assessing for clinical indication for dose reduction, and avoiding polypharmacy. The use of medications with lower potential for metabolic disturbances, particularly for patients with high baseline cardiovascular risk, also has the potential to improve medical outcomes for those with serious mental illness. Additionally, the 2004 American Diabetes Association-American Psychiatric Association guidelines recommend that psychiatrists routinely screen for metabolic abnormalities among patients prescribed second-generation antipsychotics (32). More frequent monitoring is indicated for patients at higher risk, including those with a personal or family history of type 2 diabetes, hypertension, and cardiovascular diseases; those taking antipsychotic drugs for the first time; younger patients; and those with substantial weight gain (33).

When metabolic abnormalities arise, psychiatrists may counsel their patients on therapeutic lifestyle behaviors to reduce cardiovascular risk. Robust data indicate that lifestyle interventions are efficacious for individuals with serious mental illness. RCTs of lifestyle interventions targeting obesity among people with serious mental illness have indicated that up to 40% of patients can achieve clinically significant weight loss (defined as >5% of initial body weight) (34). Expert consensus reflects these data, and the 2014 National Institute for Health and Care Excellence Guidelines for the Treatment of Adults With Schizophrenia state that mental health care providers should offer a combined healthy-eating and exercise program to persons with psychosis, particularly those taking antipsychotic medications (35). Additionally, for patients who have gained more than 7% of their pretreatment weight or have developed hyperglycemia, hyperlipidemia, or hypertension, physical benefits of switching drugs may be considered (36).

Despite expert panel recommendations, further research is necessary to examine the optimal role for psychiatrists in addressing medical care for persons with serious mental illness, as well as the best way to educate psychiatrists on assessment and treatment of medical illness.

State and Federal Initiatives

A number of initiatives at the state and federal level have supported integration of behavioral health and medical care, largely through funding opportunities. In 2009, the Substance Abuse and Mental Health Services Administration (SAMHSA) began its Primary Behavioral Healthcare Initiative (PBHCI) program. The PBHCI provides grant funding to community mental health centers (CMHCs) for the purpose of improving the physical health of patients with serious mental illness by making available an array of coordinated primary care services. Through this initiative, CMHC grant recipients have generally linked with primary care organizations, providing facilitated coordination of care through the use of patient registries and evidence-based interventions. Currently, SAMHSA has supported the creation of more than 200 projects (37).

The Home Health Initiative (Section 2703) of the 2010 Affordable Care Act included funding for state-led Medicaid demonstration projects to improve care for complex patients with chronic comorbidities, including serious mental illness, through the creation of health homes. As of November 2016, 29 health home projects have been created, with the majority focusing on patients with serious mental illness (38).

As part of the Protecting Access to Medicare Act of 2014 (H.R. 4302), the U.S. Department of Health and Human Services selected eight states to participate in a demonstration project providing collaborative care in Certified Community Behavioral Health Clinic (CCBHCs). Medical directors of participating CCBHCs oversee the integration of medical and psychiatric care in each facility, promoting several quality metrics, including appropriate preventative health screenings. In CCBHCs in which primary care services are not colocated, there is a requirement for care coordination with a partner Federally Qualified Health Center or Rural Health Clinic (39–41). However, the changing political landscape may drastically impact current federal funding initiatives aimed at improving medical care for persons with serious mental illness, given the repeal and replacement of the Affordable Care Act.

Research Initiatives in Reverse Integration

A wide body of literature supports collaborative care for individuals with co-occurring medical and behavioral comorbidities. However, the majority of research has examined models of care that are based in primary care settings, providing behavioral health interventions for a single diagnosis, such as depression or anxiety (42). Far fewer studies have examined reverse integration for individuals with serious mental illness, and the existing literature has several limitations. First, many studies examined mental health rather than medical outcomes, limiting our ability to draw conclusions about the ability of reverse integration initiatives to impact physical health (43, 44). Next, few studies included pediatric and adolescent populations. Additionally, few studies examined individuals with co-occurring serious mental illness and substance use disorder (25). Substance use disorders are common among persons with serious mental illness; in some urban centers, as many as 50% of all persons with mental illness have drug and alcohol addiction (45), and conclusions drawn from the existing literature may not be applicable to a large percentage of the real-world population. Finally, few studies reported outcomes related to health care costs, and overall conclusions on the cost benefit of reverse integration initiatives cannot be drawn from the existing literature (25).

Despite these limitations, the existing body of literature includes several well-designed RCTs that provide evidence for positive medical outcomes in reverse integration initiatives (see Table 3). In 2001, Druss et al. published the results of a trial examining a cohort of 120 individuals with serious mental illness treated at a Veterans Affairs Medical Center, who were randomized to receive medical care at an integrated medical clinic, located at the mental health clinic, or to medical care in the Veterans Affairs general clinic. In addition to access to on-site primary care, study participants were also assigned case managers, who provided education and collaboration with mental health providers. When compared with patients who received care in the general medical clinic, patients who were treated in the integrated care clinic attended more primary care visits; received a greater number of preventative services; and had a significantly greater improvement in health, as measured by the Short-Form Health Survey. Additionally, participants had fewer emergency department visits. There were no significant differences in cost between the two groups (46).

TABLE 3.

Randomized Controlled Trials Examining Reverse Integration for Individuals With Serious Mental Illness

Study	Sample Size and Follow-Up	Medical Outcomes	Health Care Utilization and Cost
Druss et al, 2001 (46)	N=120, 12 months	Study participants had a significantly greater improvement in health, as measured by the Short-Form Health Survey.	Study participants attended more primary care visits, received a greater number of preventative services, and had fewer emergency department visits. There were no significant differences in cost between the two groups.
Druss et al, 2010 (47)	N=407, 12 months	Study participants had greater receipt of evidence-based cardiometabolic treatments and had lower Framingham Risk Scores at 12-month follow-up.	Study participants had greater receipt of preventative services. Individuals assigned care managers were also more likely to have a primary care provider. Cost was not studied.
Kilbourne et al, 2013 (28)	N=136, 12 months	Study participants had significant decreases in blood pressure but not in cholesterol. There were no statistically significant differences between groups in physical health-related quality of life.	Health care utilization and cost were not studied.
Druss et al, 2017 (48)	N=447, 12 months	Study participants showed significant improvements in quality of cardiometabolic care and a significant decrease in systolic blood pressure. There were no other significant differences in change in other cardiometabolic parameters.	Study participants had greater receipt of preventive services. Cost was not studied.

In the Primary Care Access, Referral, and Evaluation Study, Druss et al randomized 407 individuals with serious mental illness treated in a community mental health center to usual care or medical care management. Individuals in the treatment arm of the study were assigned care managers who provided health education and logistical support to overcome barriers to accessing medical care. The care managers also communicated with primary care providers and advocated for their patients. When compared with participants receiving usual care, participants assigned care managers experienced many statistically significant positive health outcomes, including greater receipt of preventative services and evidence-based cardiometabolic treatments. Individuals assigned care managers were also more likely to have a primary care provider and experienced lower Framingham risk scores at 12-month follow up (47).

In a study published in 2013, Kilbourne et al. randomized 136 individuals with bipolar disorder, and at least one cardiovascular risk factor, to enhanced usual care or to a Life Goals Collaborative Care program. Enhanced usual care consisted of quarterly wellness newsletters sent over a 12-month period in addition to usual care; the Life Goals Collaborative Care program included four weekly self-management sessions, followed by up to monthly contacts for 12 months with health specialists promoting health behavior change. The health specialists also tracked outcomes using a registry, communicated with mental health and medical care providers about patient outcomes, and provided evidence-based guidelines to providers. At the 24-month analysis, individuals enrolled in the Life Goals Collaborative Care program showed significant decreases in systolic and diastolic blood pressure, but not in cholesterol, compared with those enrolled in the enhanced usual care arm. There were also no statistically significant differences between groups in physical health-related quality of life (28).

In the Health Outcomes Management and Evaluation study, Druss et al randomized 447 individuals with serious mental illness, and at least on cardiometabolic risk factor, to usual care or treatment in a behavioral health home. The behavioral health home consisted of a partnership between a CMHC and a Federally Qualified Health Center, whereby patients received medical care at the CMHC from an on-site nurse practitioner and a nurse case manager subcontracted from the health center. Compared with patients receiving usual care, those enrolled in the behavioral health home experienced significant improvements in quality of cardiometabolic care and receipt of preventative services. Although both groups experienced improvements in cardiometabolic parameters, there were no statistically significant differences, with the exception of a decrease in systolic blood pressure in the group enrolled in the behavioral health home (48).

Real-World Initiatives in Reverse Integration

Limited data have been published on reverse integration initiatives based in “real-world” community settings. In 2010, the RAND Corporation conducted an early evaluation of SAMHSA’s PBHCI program following its first year of implementation (49). The evaluation included an assessment of the structure of integration at each grantee site. Sites varied in staff level of training, services offered, and infrastructure (including variability in sharing of information systems between behavioral and physical health providers, as well as framework for collaboration). Evaluation of medical outcomes was limited to a study of three matched PBHCI and CMHC control clinic pairs. Results indicated an improvement in medical outcomes for patients seen in the PBHCI when compared with controls, with statistically significant mean decreases in diastolic blood pressure, total cholesterol, LDL cholesterol, and fasting plasma glucose. However, no significant difference between groups was detected in other health measures, including systolic blood pressure, body mass index, HDL cholesterol, hemoglobin A1c, triglycerides, and self-reported smoking.

Barriers to Implementation

Although reverse integration initiatives offer the possibility of improving medical outcomes for those with serious mental illness, significant barriers exist to their widespread implementation. Perhaps foremost among these are financial barriers, such as funding and reimbursement. Start-up costs for research initiatives are generally grant funded, and sources for initial costs may be difficult to obtain in real-world settings. Even for those initiatives supported by state and federal funding opportunities (and other grants), maintenance costs may limit sustainability. Typical fee-for-service management plans do not reimburse for some of the most essential and common practices of collaborative care, including conversations between providers, expert consultations, and care management (50).

Additionally, integration of care may require overcoming organizational norms and structures, including the creation of new processes and specialized training for team members. Because of competing responsibilities and lack of resources, health care providers and support staff may be resistant to practice changes. Finally, confidentiality rules for behavioral health treatment, including barriers placed on accessing health records associated with behavioral health treatment, may complicate coordination of care (25).

Several of these challenges were experienced by the PBHCI grantees. These included lower-than-expected rates of consumer enrollment, as well as difficulties with financial sustainability, communication within teams, and creation of an integrated clinic culture. Although most enrolled patients accessed some degree of behavioral health and primary care services during the program’s first year, wellness programs (e.g., those targeting smoking cessation and weight loss) were underutilized. As a result of these findings, SAMHSA has added participation requirements for PCBHI grantees, including greater standardization of infrastructure and practices.

Lessons Learned From Existing Initiatives in Reverse Integration

Although limited literature exists on reverse integration, in both research and real-word settings, several conclusions can be drawn from the existing data. Implementation of the principles of Wagner’s chronic care model appears to be associated with greater success. All published RCTs on reverse integration employed at least three of these principles (25), and analysis of SAMHSA’s PCBHI projects showed that the more successful sites developed registries to track patients (37). Shared information systems, such as the electronic medical record, were also cited as an essential component to success in at least one RCT (46).

Additionally, careful selection of staff appears to be essential. The change in clinic culture required to implement collaborative care can be slow and difficult, and resistance by staff members can make a large difference in success. In order to change the culture of the clinic, it may be necessary to train clinic staff, including case managers and psychiatrists, to include the medical care of their patients as part of their core mission. In addition, reverse integration initiatives may benefit from hiring individuals who specifically desire to work with individuals with serious mental illness and who have the flexibility to tolerate the challenges of addressing adverse social determinants of health among this population. In the SAMHSA PCBHI projects, those grantees that hired staff (including primary care physicians) with prior experience with individuals with serious mental illness, homelessness, or other related groups were more likely to have greater engagement with patients, as well as decreased staff turnover (37).

Moreover, collaborative care models that employ greater levels of integration are more likely to find success. Three of the four RCTs of reverse integration for serious mental illness employed colocation of primary care and behavioral health services (28, 46, 48). PCBHI projects with colocated primary care and behavioral health services were associated with greater enrollee access to care (37). Finally, the modest outcomes seen in the existing literature highlight the difficulty of making meaningful improvements in the physical health of persons with serious mental illness. Innovative practices must be developed and tested to address the role of adverse social determinants of health in poor medical outcomes among persons with serious mental illness; technological innovations may play a role in improving service delivery. Additionally, optimal practices may be based in the community rather than in the clinic setting and will likely include a multidisciplinary approach and engage patients' available support networks.

Conclusion

The increased medical morbidity and mortality of individuals with serious mental illness is a public health crisis. Although the reasons for poor physical health among this population are multifactorial, decreased access to quality medical care plays a large role. Recent initiatives in reverse integration have aimed to address this problem and have achieved variable, although significant, results. Data from these initiatives indicate that changing the clinic culture is a significant barrier to success in reverse integration and may require the training of a new workforce in systems of collaborative care at the community level (e.g., CMHCs). Additionally, existing studies indicate several factors that may improve outcomes, including the application of the principles of Wagner’s chronic care model, greater levels of integration, and shared information systems. However, further large, longer duration studies are needed to inform best practices for integrated care initiatives that are based in behavioral health settings.

Moreover, the modest gains achieved in existing reverse integration studies indicate that reverse integration may not be the sole answer to improving care for individuals with serious mental illness. Access to medical care for those with serious mental illness may require psychiatrists to address the overall health of the patients they treat, including physical comorbidities. Innovative and multifactorial interventions may be necessary to address the adverse social determinants of health and other barriers to care experienced by persons with serious mental illness.