Role of Social Disadvantage in Crime, Joblessness, and Homelessness Among Persons With Serious Mental Illness

Social problems among persons with mental illness, such as involvement with the criminal justice system, unemployment, and homelessness, are receiving heightened attention in the media and the psychiatric literature. It would surprise many service providers and advocates to learn that mental illness is not as potent an explanatory factor for these problems as the psychiatric literature might lead us to believe. Researchers have sought to improve services that address these problems among persons with mental illness but have frequently failed to recognize that the experience of people with mental illness is often contextualized in disadvantaged social settings.

In other words, persons with mental illness experience social problems more frequently because they live in a world in which these problems are endemic, not just because they are mentally ill. Thus social problems become erroneously simplified as psychiatric problems, resulting in the creation of overly simple interventions and policies to address a complex phenomenon.

Much of the psychiatric services research is based on samples of participants who meet criteria for mental illness. Using data from these samples, researchers frequently make inferences about the social problems of individuals who have mental illness. According to these data, problems such as unemployment, crime, and homelessness are pervasive among people with mental illness. Thus it is inferred that mental illness itself is a prime explanatory factor for these social problems among these individuals.

However, in most cases these inferences are not valid, essentially because of failure to compare the experience of persons who have serious mental illness with that of persons in similar socioeconomic situations who are not ill. In the few studies in which such comparisons have been made, the impact of mental illness was much smaller than that implied in most of the psychiatric literature.

In arguing for this perspective, we do not want to exchange biological reductionism for social reductionism. Cohen (1) recently warned of the dangers of too great a focus on biomedical reductionism in examining treatment and services for persons with serious mental illness. One aspect of this reductionism is a "social amnesia" when it comes to understanding the broader social context for the difficulties encountered by persons with mental illness. At the same time, we do not want to lose the benefit of gains we have made in biological explanations of psychiatric illness. The question is one of achieving a balance that best informs policy and practice toward a better community life for people with mental illness.

In addressing social problems among persons with mental illness, researchers need to give more consideration to social factors that overlap the boundaries of mental illness and mental health. These factors have links to psychiatric illness but are also closely linked to the social and economic characteristics of the communities in which people with mental illness live.

Our position is that poverty is an important moderator of the relationship between serious mental illness and social problems and that this moderating role is not sufficiently accounted for in research, service planning, and policy. Poverty and its associated social disadvantage can also be linked to important mediating factors, such as decreased self-efficacy and coping. If persons with mental illness are not poor to begin with, they are likely to become poor, and poverty factors become salient in explaining common outcomes, such as quality of life, social and occupational functioning, general health, and psychiatric symptoms.

To discuss the refocusing of research on social problems and mental illness, we use three social problems by way of illustration: crime, unemployment, and homelessness. In this article we suggest ways in which research on each of these problems may have suffered from inattentiveness to the general knowledge base—that is, outside of psychiatry—in these areas. We present selected evidence in each of these areas as a way of illustrating our point.

Crime

The prevalence of mental disorders detected in jail populations gives rise to rhetoric about jails replacing hospitals for the mentally ill. The metaphor of the "largest mental hospital" (2) has been used in the media to refer to the jails of Los Angeles, Chicago, and New York. Among mental health advocates, this phenomenon is routinely attributed to the failures of the mental health system. However, such an explanation ignores the increasing incarceration rates in the general U.S. population. The number of prison inmates in the United States quadrupled between 1970 and 1999 and could double again by 2005 (2). There have been similar increases in the number of individuals in jails or on probation and parole.

An alternative explanation for the increase in the number of people in jail with a mental illness over this period is that individuals with mental illness are also members of other groups with a high risk of being arrested. Persons who are substance users, are unemployed, have fewer years of formal education, and have low incomes have a greater risk of incarceration. It would follow from the general increase in incarceration rates that persons with mental illness are disproportionately caught up in the criminal justice system because of their greater risk of arrest as a result of these other factors, independent of mental illness.

This phenomenon was documented in a recent study in Australia showing that an increase in the numbers of persons with schizophrenia who were arrested after deinstitutionalization was accounted for by an increase in arrest rates in the general population, not by psychiatric status (3). This finding confounds the conclusion that mental illness has been criminalized or that the institutionalization of persons with mental illness has shifted from state hospitals to jails (4).

Studies of persons with serious mental illness who are living in the community have consistently found lifetime arrest rates in the range of 42 to 50 percent (5,6,7). Although these rates initially appear remarkable, they are similar to rates observed in general community samples. Studies of crime indicate that 25 to 45 percent of urban males have been arrested by the age of 18 years, and about half of those who are ever arrested during their lifetime are arrested before the age of 18 (8). African Americans are almost twice as likely as whites to have been arrested (8). The lifetime arrest rates we have noted above were taken from adult populations with an average age well beyond 18 years. Two of the three samples were from communities in which African Americans predominated. Thus when the demographic and community contexts are considered, the apparent disparity in arrest rates for persons with mental illness is no longer so obvious.

Some research has linked symptomatic behavior to a greater likelihood of arrest (9,10), but even this research has not indicated that a substantial portion of persons who show evidence of mental illness are arrested (8,9). Thus an empirical link between the extent of symptomatic behavior encountered by police and the bulk of the jailed population with mental illness is tenuous. It is likely that the majority of the persons in jail who have a diagnosis of a psychiatric disorder have risk factors other than their psychiatric symptoms. However, the empirical weakness of the criminalization argument has not prevented some advocates from using indirect evidence to support such a conclusion.

For example, a recent study by the National Institute of Justice estimated that 16 percent of persons in jails and prisons had a mental illness (11). Mental illness was defined as a self-reported "mental or emotional condition" or any overnight stay in a "mental hospital, unit, or treatment program." For most mental health researchers, this definition is far from sufficient as a valid operationalization of mental illness status. However, even if this operationalization were valid, the study found no data that make a case for a link between change over time and mental illness or psychiatric hospitalization patterns.

Nevertheless, the results of the study were used extensively by advocates and by federal agencies to support the proposition that persons with mental illness are being "criminalized." Multiple studies have shown a disproportionately large number of persons with mental illness in jails and prisons, but few studies are designed to allow investigators to attribute this incarceration to mental illness (12) rather than to other risk factors or to the historical increase in incarceration generally.

As the idea of the criminalization of the mentally ill has become part of the policy context, advocates have pushed for more mental health treatment resources as a means of preventing the incarceration of persons with mental illness. Model mental health interventions are being funded and implemented more extensively as a way of addressing this problem. However, few of these interventions use crime theory and research to conceptualize criminal behavior among persons with mental illness. Instead, their conceptualization tends to focus on shifting people out of the criminal justice system and into the mental health system.

Few mental health interventions are conceptualized around the idea that many individuals could reasonably be treated as both criminal offenders and recipients of mental health services. Such a conceptualization would seem to tolerate the criminalization of mental illness. However, regarding an individual as both an offender and a service recipient has important implications for the treatment of the individual as a citizen, not just as a psychiatric patient. When a person is shifted from the criminal justice system to the mental health system, there is also a significant shift in due process procedures, which the individual may not perceive as being in his or her best interests.

Behavior that is linked to acute symptoms could be addressed by increasing police access to effective crisis services (13). There is no evidence that exempting persons who are not acutely symptomatic at the time of arrest from accountability to the criminal justice system is beneficial to the individuals or to the community (14). This group of individuals probably constitutes the bulk of persons with mental illness in jails. Thus it could be more beneficial to develop a range of interventions that integrate mental health services throughout the criminal justice system. Individuals could maintain their rights as criminal defendants while receiving effective mental health treatment.

Research has shown that psychiatric status is a poor predictor of criminal recidivism (15) and may not be a strong factor in explaining involvement with the criminal justice system (3) when other factors strongly associated with crime are also considered (15). This does not mean that mental illness among persons who are involved with the criminal justice system should be ignored. Interventions that integrate criminal justice services and mental health services may be more effective for many individuals than a shift "back" to the mental health system. Such interventions would consider the complicated context of criminal behavior, including risk factors common to both mental health and criminal behavior—for example, substance abuse, problems with employment and housing, and a low likelihood of having prosocial attachments (15,16).

Thus research and intervention could more realistically accept the overlap between mental health and criminal justice populations as an expected and unsurprising phenomenon. Such acceptance may provide a foundation for a greater understanding of criminal behavior among persons with mental illness and for more services that are effective in the criminal justice system.

Employment

Employment of persons with severe mental illness has historically been related to prognosis (17). Various explanations have been postulated of the impact of mental illness on work, including inability to cope with the stresses of work, poor social skills, and diminished cognitive abilities. One result of these perceived deficits is that persons with serious mental illness may not know how to apply for jobs and may do poorly in job interviews (18).

The data pertaining to the work lives of persons with serious mental illness are grim. Among persons with a disability due to mental illness, the labor force participation rate—that is, whether a person has a job, is on temporary layoff, or is actively looking for work—ranged from a low of 21.5 percent in 1985 to 27.2 percent in 1994 (19). This rate has been relatively stable and is similar to the 25 percent employment rate for persons with serious mental illness computed from surveys conducted in the early 1980s among participants in community support programs (20).

Research on the lifetime work experiences of persons with serious mental illnesses is limited, possibly because of the expectation that the work experiences of persons with mental illnesses are restricted. In a study of 100 young adults with a diagnosis of schizophrenia, 85 had maintained a competitive job for more than three months during their lives (21). In a study involving almost 300 people with a wide range of diagnoses, nearly all had been competitively employed at some point (22).

A small study of persons with serious mental illness who were seeking vocational rehabilitation services showed that the mean±SD number of full-time jobs held was 2.70±2.38 and the mean number of part-time jobs held was 1.89±1.79 (23). The median employment tenure among employed persons with a psychiatric diagnosis was seven months (22), and the mean±SD employment tenure across all jobs held was 18.57 ±26.81 months (23). The temptation to conclude that mental illness is the primary factor in shaping vocational careers is strong.

However, a more complicated picture of the relationship between mental illness and employment status emerges when these data are considered in the context of various sociodemographic factors. For example, there is a strong link between educational level and mental illness, and educational level is also an important factor in employment. Data from the national comorbidity study (24) indicate that the early onset of a psychiatric disorder may have an adverse impact on a person's educational attainment.

Similar observations can be made from data obtained as part of the Epidemiologic Catchment Area study. For example, although 57.4 percent of persons with a diagnosis of schizophrenia graduated from high school—a rate that is slightly, but not significantly, lower than that of the general U.S. population (65.7 percent)—only 4.8 percent were college graduates, compared with 17.2 percent of the U.S. population (25). Thus there is some complexity in the relationship between mental illness and employment. It appears that symptoms over the course of a person's life may have an impact on educational attainment, which also affects employment.

In 1992 the employment rate for persons with a college degree was 2.7 times higher than the rate for those with only a high school diploma (26). High school graduates worked 8 percent fewer weeks from 1991 through 1995 than college graduates and experienced significantly more episodes of unemployment than college graduates (27). Education also dictates the types of jobs a person can seek. Lower educational attainment is associated with entry-level jobs known for rapid turnover—the same types of jobs that persons with serious mental illness frequently take.

For example, McCrohan and associates (22) interviewed 279 persons with schizophrenia (68 percent), affective disorder (21 percent), or another mental disorder (11 percent). The current or last job for 51 percent of respondents was in the service industry—for example, food service, janitorial work, and personal services, such as child care worker or nurse's aide. Thirty-one percent of respondents held benchwork jobs, such as light assembly or employment in a sheltered workshop. Fewer than 5 percent held clerical jobs or sales positions. These types of positions are generally associated with a short job tenure—for example, 1.8 years for retail trade positions, 1.2 years for sales positions, 1.3 years for food service, 1.6 years for construction laborers, and 1.9 years for equipment cleaners, helpers, or laborers (28). Not surprisingly, education also significantly influences earnings and wages (26).

There has been a promising shift in orientation from work—or lack thereof—as a prognostic indicator of illness toward rehabilitation and recovery models that emphasize ameliorative approaches to overcoming the impairment and disability associated with mental illness. Tremendous strides have been made in the development of effective vocational rehabilitation programs. For example, the individual placement and support model has been found to dramatically increase the total numbers of persons with serious mental illness who obtain employment in competitive work settings (29).

However, vocational rehabilitation providers and researchers have been somewhat stymied in the degree to which they can bring about more satisfying and stable employment for persons with serious mental illness. One study found that the maximum employment rate of persons receiving individual placement and support was only 34 percent over an 18-month period, with an average job tenure of 16.5 weeks (29). Other studies have found similar employment rates for other vocational rehabilitation programs (30,31,32).

Race, sex, age, presence of a disability, and residence in a metropolitan area also play important roles in employment status, independent of mental illness. Results from one analysis found that sociodemographic factors play a significantly larger role in the employment of persons with a disability due to a mental illness than the mental disability itself (33). In fact, these data suggest that the impact of programs aimed at addressing the mental illness and resulting disability pales in comparison with what might be achieved if factors associated with various sociodemographic characteristics that limit employment opportunities could be adequately addressed.

For example, expanding supported education programs would go a long way toward improving the work lives of persons with serious mental illness. Other factors that limit employment opportunities include racial factors, such as prejudice in hiring, and factors related to a person's place of residence—for example, the fact that fewer jobs are available in poorer neighborhoods. Addressing such factors could maximize the potential of mental health interventions such as individual placement and support to improve vocational outcomes for persons with mental illness.

Homelessness

The casting of homelessness as "primarily a mental health problem" began during the early 1980s, when a resurgence in homelessness was identified by advocates, the news media, and researchers. Mental health researchers were among the first to study the problem (34,35). The apparent "disturbed" nature of the street homeless population created an expectation that the public mental health system would assume responsibility for this vulnerable group.

The first set of studies argued that as many as 90 percent of homeless persons had a "mental disorder" (34,35), and such estimates were widely reported in the national news media. Rates for more serious mental disorders, such as schizophrenia, were reported to be around 30 percent. Nevertheless, the early research confirmed and even promoted the view that homelessness was a problem that stemmed from the deinstitutionalization of persons with mental illness (36).

Reviews of this research have not been favorable. The many shortcomings to which critics have pointed in questioning the generalizability of research findings include single shelter sites, in one case a shelter attached to a state-owned hospital; screening criteria for admission to shelters; and unsystematic diagnostic procedures (37). Nevertheless, the dictum that "one-third of the homeless have a severe mental illness" can ultimately be traced to this relatively minimal body of evidence (37).

Research conducted in the latter half of the 1980s was based on larger, more representative samples (38,39). These studies also involved structured, standardized diagnostic interviews. Not surprisingly, they yielded far more convergent results (40). In a meta-analysis, Lehman and Cordray (41) reported that the rate of severe mental disorders among the homeless samples in these more rigorous studies was about 18 to 22 percent, or about a third less than that reported in the earlier literature. However, even these larger studies could not be fairly generalized to "the homeless," because they excluded homeless families, a population that has lower rates of mental illness. In some cities, the number of people in homeless families has outpaced the number of homeless adults. Thus generalizations of research findings to "the homeless" need to be qualified.

A study of the rates of severe mental disorders among homeless persons in Philadelphia, which used longitudinal administrative treatment and shelter records, found that 18 percent of homeless adults without accompanying children had received services for a severe mental disorder (42). However, the estimate dropped to a range of 12 to 15 percent when mothers of the children were included. Furthermore, when the children were included, the rate of treatment for severe mental disorders among the homeless population dropped to a range of 7 to 9 percent. In this context, it is clear that mental illness is not the primary cause of homelessness. Excluding data on families has inflated the estimated rate of mental disorders among homeless persons.

Administrative data on use of homeless shelters have provided an opportunity to assess population-adjusted risks of homelessness among persons with mental illness and in other groups. This research has shown that while the annual rate of shelter admission in the general population was about 1 percent in New York and Philadelphia throughout the 1990s (43), the annual rate among persons with serious mental illness in the Medicaid population, at least in Philadelphia, was about 3.6 percent (44).

It is surely significant that the shelter admission rate of persons with serious mental illness is more than three times that of the general population. However, the rate appears to be no lower than that for other poor people. In Philadelphia, about 6 percent of the poor population stays in a shelter each year, including about 9 percent of poor children and 12 percent of poor adults in their 30s and 40s, the primary age groups of homeless adults.

The annual rate of shelter admission among poor black children in New York City was 16 percent in 1995, and the rate among poor black men in their 30s and 40s is as high as 20 percent per year. Given that the comparable rate among persons with serious mental illness is 3.6 percent (43), is it possible that severe mental illness actually confers some protection against homelessness?

More carefully controlled studies that compared the housed and homeless poor populations have found that although serious mental illness might not offer protection from homelessness, it does not represent a distinguishing risk factor for becoming homeless. Multivariate studies of families in New York City (45) and Worcester, Massachusetts (46), showed no effect of mental illness or substance use disorders on the risk of shelter use.

Multivariate studies among single adults in Chicago (47) and Buffalo (48) likewise found no effect of mental illness or substance abuse on becoming homeless or on having been homeless in the past. The evidence more strongly suggests that, among persons with mental disorders, substance abuse is a common risk factor for homelessness (49,50), but, again, mental illness per se does not distinguish between the vulnerable poor population who remain housed and those who become homeless.

Furthermore, and perhaps most interestingly, study results show that although substance abuse does consistently predict long-term homelessness or difficulty in exiting homelessness, no such consistent evidence exists of an association between homelessness and mental illness. In some cities, such as New York and Buffalo, people with mental disorders have been found to stay homeless (48,51); in others, mental illness has no effect (52,47), and in still others it actually appears to have a mitigating effect on the duration of homelessness (51).

It is likely that differences in the local treatment system explain why the findings of these studies have been so disparate. In cities that are better organized to conduct outreach and move people with serious mental illness into housing, mental illness may indeed be a protective factor against long-term homelessness. Availability of Supplemental Security Income, greater access to inpatient services, access to a public system of outpatient and rehabilitation programs, and even special housing outreach programs specifically for persons with mental illness who are homeless should, if developed appropriately and expanded sufficiently, act to reduce the risk of long-term homelessness among persons with mental illness relative to the risk for poor persons in general.

Clearly, homelessness is more than a mental health problem. That homeless persons have much higher rates of mental disorders, including severe mental disorders, than the general population may seem shocking. However, in the context of the poor populations from which they come, such rates are not as noteworthy. When dramatically adverse shifts in housing affordability and crowding occurred during the 1980s and 1990s, people with mental illness were among those who suffered their effects. They may even have been the first to make the problem visible. However, homelessness affects a much broader segment of the poor in general and is not unique to people with severe mental illness.

From a mental health policy perspective, what is disturbing is that severe mental disorders are so prevalent among the poor in general and that having a mental disorder does not convey adequate social protection from impoverishment. Such protection should include effective outreach to improve access to treatment, housing combined with services to prevent homelessness, and adequate income. Thus the problem is one of relating social forces to treatment effectiveness in explaining outcome.

Implications

It has long been understood that persons with mental illness are largely among society's poor (1). The precipitating factors of many socially adverse events are strongly associated with complicated issues such as crime policy, the war on drugs, race, poverty, and economic development—not with psychiatric symptoms. Acknowledging this reality should help keep these social problems among persons with mental illness in perspective.

This is not to imply that having a psychiatric disorder is not a risk factor for social problems or for exacerbation of the problems. It is important to note that much of the current mental health services literature leaves an impression of a greater impact of mental illness in explaining individuals' experience of these problems than is probably warranted. Our understanding of social problems among persons with mental illness could be more effectively anchored by theoretically valid comparisons with persons from the general population who are in similar socioeconomic strata.

Implications for services

Mental illness is nevertheless an important variable to consider in addressing these social problems. In fact, using the proper comparisons to contextualize social problems among persons with mental illness is a step toward providing more effective services. Such an approach may provide data to enable more thorough integration of effective mental health services with the service systems that address these problems in the general population. It also may provide an additional social benefit for people with mental illness. When we conceptualize the problems these individuals face in the appropriate social context, our policy ideas begin with the premise of integration with society.

Specialized services for persons with mental illness may produce additional layers of stigma. This potential should not be treated lightly. It is possible to develop more effective policies and programs to address social problems among persons with mental illness that integrate these individuals into service systems that address such problems. If we view each of these social problems solely as a mental health treatment problem, we may further stigmatize persons with mental illness while providing insufficient interventions.

Implications for research

Current thinking holds that the etiology of major mental illnesses is largely biological. In light of the impact of past thinking, this is fortunate in many ways. The development of scientific knowledge used in the treatment of these disorders has improved the lives of many people. It has also provided ammunition to advocates as they fight the stigma associated with mental illness and mental health treatment. Thus the medical treatment of serious mental disorders is widely accepted and has provided progressively greater advantages to many individuals who experience such disorders. This acceptance has influenced social research on serious mental illness, and many social scientists now focus on the study of service delivery and the development of social interventions.

Mental health services research has necessarily focused on diagnostically homogeneous groups. Designs of the most influential research include controls that heighten internal validity to establish the efficacy of interventions. These studies will affect policy as "best practices" become established on the basis of this research, which in turn will leverage reimbursement mechanisms from Medicaid, Medicare, and managed care organizations. Thus the prestige of the mental health professions will be enhanced by a process that focuses on discrete diagnostic categories.

Over the past 30 years, a meta-diagnostic category—serious and persistent mental illness—has dominated much of public mental health services research. This designation is based on a three-pronged definition that includes diagnosis (generally schizophrenia and major affective disorders), disability, and duration of the disorder. Studies have used this designation to link mental illness to broader social problems and costs. Certainly this effort is well intentioned. It seeks to call attention to the need to develop effective intervention strategies and public health policy for the people most seriously affected by psychiatric illness. However, by isolating this population for study, we subvert our capacity to accurately assess the relative risk mental illness poses for other problems.

In addition to comparing individuals on the basis of their current status—for example, homeless, employed, or incarcerated—examining the role of these states during the life course of individuals with and without mental illness can be informative. Longitudinal research on the relative risk of multiple experiences and their relation to mental illness may increase our understanding of social forces and of treatment effectiveness.

We should expect researchers to do more work that compares the experiences of persons with serious mental illness with the experiences of persons who live in similar circumstances but who do not have mental illness. Theoretically, this work should be conducted among persons who are at risk of homelessness, involvement with the criminal justice system, and unemployment. The impact of this strategy is highlighted in the homelessness research example above. When use of homeless shelters by persons with mental illness was compared with use in the general population, mental illness appeared to be a strong factor in explaining homelessness. However, when use of shelters by persons with mental illness was compared with use in the population of persons living in poverty, the implications were nearly the opposite—mental illness could possibly be seen as a protective factor against homelessness, particularly long-term homelessness.

Here we have provided similar examples for employment and involvement with the criminal justice system. Other factors that might be shown to mediate the impact of mental illness include substance abuse, physical health, educational attainment, child abuse and neglect, and juvenile delinquency.

Making comparisons of this sort will require specialized sampling strategies, which need to be both targeted on serious mental illness and generalizable to larger populations. Much of the epidemiological research that has focused on mental illness and social status has used samples from the general population. As a result it has focused on a much broader conception of mental illness and thus a broader range of illness than the serious mental illness designation (24,53,54). Because persons with serious mental illness account for less than 5 percent of the general population, generalizations of this research to the population of individuals with serious mental illness are tenuous. For example, in an analysis that linked poverty indicators to development of a mental disorder by using data from the Epidemiologic Catchment Area study, only four new cases of schizophrenia were found during the longitudinal follow-up period (53).

Although it initially appears that a great deal of research has been conducted on social mobility, social problems, and mental illness, it quickly becomes apparent that this research is not readily applicable to populations of individuals with serious mental illness. In future studies linking serious mental illness to social problems, sampling designs could include stratified sampling to ensure sufficient numbers of persons with serious mental illness as well as a generalizable comparison with persons who do not have serious mental illness in the same sociodemographic contexts.

Implications for policy

A greater focus on social context has mixed policy implications for mental health service advocacy. The use of "disease" as the explanation for social problems has permitted advocates to acquire additional resources for treatment and to create primary mental health services for people who are homeless, involved with the criminal justice system, or unemployed. Disease-based interventions can be more focused, and outcomes can be expressed more readily on an individual level. Because it is easier to generate interest and support for such interventions than for broader social welfare-oriented interventions, it is easier to fund these interventions as medical interventions.

However, because the interventions are based on the disease, their effectiveness is theoretically limited to the extent to which the disease affects the problem. If psychiatric disease is only a small part of the problem, services focused on psychiatric treatment are limited in the extent to which they can illuminate effects for social outcomes. Such limitations do not suggest that the role of mental illness in these problems should be neglected, only that policies aimed at mental illness will be only partially effective.

If explanatory frameworks highlight a smaller role for mental illness, mental health professionals—and policy makers in general—gain information from the overall framework to develop appropriate interventions. Grounding intervention research further in the existing research from a greater variety of theoretical frameworks will expand the range of intervention activities. This approach should provoke greater innovation in practice and policy, including innovations that are not necessarily linked to the conventional roles of mental health professionals. Such roles may include community-level interventions, social welfare programs, conventional and alternative law-enforcement programs, and development of greater infrastructure for supporting self-help and mutual-aid associations among persons whose lives are affected by mental illness.

In considering a role for more broad-based policies, two sets of goals might be considered. First, public policies could reduce the risk of social problems in general: joblessness, poverty, undereducation, housing crowding, and lack of access to adequate, affordable housing. Interventions focused on these problems will benefit persons with serious mental illness insomuch as these individuals themselves are disproportionately affected by them.

Alternatively, policies could focus on reducing the prevalence of these risk factors among persons with serious mental illness. These persons should be more adequately protected through social insurance programs so that they do not become poor, destitute, and homeless. They should be provided with more employment accommodations to enable them to work and with the services and supports that would help them avoid the risks of crime and incarceration.

The authors are with the department of psychiatry of the Center for Mental Health Policy and Services Research at the University of Pennsylvania, 3535 Market Street, Third Floor, Philadelphia, Pennsylvania 19104 (e-mail, [email protected]).