A Randomized Clinical Trial of Supported Employment for Inner-city Patients With Severe Mental Disorders FREE

COMPETITIVE employment is a primary goal for many people with severe mental disorders (SMD).¹ Moreover, they prefer competitive jobs (ie, regular jobs, supervised by the employer, for regular wages, and in integrated work settings) over prevocational training or sheltered jobs (ie, jobs supervised by mental health staff, at less than minimum wage, in segregated settings, and exclusively for persons with disabilities).^{1 - 2}

The actual rates of competitive employment for people with SMD are quite meager, typically less than 15%.^{3 - 5} One explanation for the low rates is that mental health professionals often discourage patients from seeking employment because they believe that the stress of competitive work produces adverse effects.^{6 - 7} Another is that persons with SMD in the standard vocational service system tend to become stalled in sheltered work settings and do not transition to competitive jobs.^{8 - 9} The supported employment movement represents a promising departure from standard vocational approaches by emphasizing rapid, individualized placement in competitive jobs in integrated work settings with time-unlimited, follow-along supports.^{10 - 12} Promising results have emerged from initial research on supported employment for people with SMD.^{13 - 19} Current research also indicates that supported employment neither causes undue stress nor precipitates exacerbations of illness.^{9 ,20 - 21}

Individual placement and support (IPS) provides a manualized approach to supported employment for patients with SMD, in which vocational services are integrated within mental health teams.^{22 - 23} In a previous clinical trial in New Hampshire, we compared employment services delivered via the IPS model with standard, stepwise vocational services delivered by a rehabilitation agency.¹⁶ During an 18-month period, competitive employment outcomes favored the IPS participants by a ratio of approximately 2:1. For example, the overall rates of obtaining competitive jobs were 78.1% for IPS participants and 40.3% for those in the rehabilitation agency. This study could be criticized because the participants generally had good vocational histories, few came from minority backgrounds, and all members of the control group were assigned to a single rehabilitation vendor rather than to an array of vocational programs.

The purpose of the present study was to replicate the New Hampshire study in Washington, DC, with a more diverse and disadvantaged group of patients with SMD and a typical variety of vocational agencies. In the current experiment, one group of participants received IPS, while the other group received enhanced vocational rehabilitation (EVR) services from several local rehabilitation agencies, all of which used stepwise approaches in which prevocational experiences, such as sheltered workshop trials, preceded competitive employment. Thus, the models offered a clear contrast between the rapid job search method of IPS and the stepwise approach of EVR.

This study tests 3 hypotheses: (1) that for persons with SMD, IPS results in a higher rate of competitive employment, more total hours of competitive work, and more wages than EVR; (2) that the stepwise approach (EVR) leads to high rates of sheltered employment, but, contrary to program goals, does not lead to competitive employment; and (3) that the rapid job search approach of IPS does not result in more hospitalizations or other negative outcomes than the more gradual EVR approach.

All study participants met enrollment criteria: SMD, unemployment, willingness to give informed consent, and lack of memory impairment or medical illness that would preclude working or participating in research interviews. Consistent with standards used in most states,²⁴ adult clients of the District of Columbia Commission on Mental Health Services are considered to have SMD if they have major mental illness (defined as schizophrenia, schizoaffective disorder, bipolar disorder, recurrent major depression, or borderline personality disorder) and at least 2 years of major role dysfunction. Participants were recruited at Community Connections, an agency in southeast Washington, DC, that serves patients with SMD who need intensive case management, usually because their psychiatric disorders are complicated by homelessness, comorbid substance use disorder, or human immunodeficiency virus infection.²⁵

During an 18-month period from January 1994 to July 1995, all clients at Community Connections were encouraged to attend informational groups and were screened for eligibility by the project director. Of the 309 clients at Community Connections, 200 (64.7%) attended at least 1 informational group, and 152 (49.2%) joined the project. Interviews with 43 clients who attended the group but declined to join the project indicated that the majority either decided that they were not interested in competitive employment or planned to move out of the area.²⁶ Community Connections clients who did not join the project (n=159) were similar to study participants (n=152) in terms of sex, minority status, and primary diagnoses, but they were more likely to be elderly (21.7% vs 1.3% older than 60 years, χ²=34.4, P<.01). Random assignment was stratified according to work history (>1 year of employment in a previous job); 76 participants were assigned by computer-generated random numbers to IPS and 76 to EVR.

Individual placement and support was developed to provide supported employment services for people with SMD in community mental health centers.²³ The IPS program integrates mental health and vocational services by having an employment specialist join multidisciplinary case management teams. Rather than providing preemployment assessment and training in sheltered workshops, IPS employment specialists assist clients in searching for jobs rapidly and, after securing employment, provide individualized, follow-along supports (eg, counseling, transportation, or intervening with an employer) as needed without time limits. For this study, 3 employment specialists were hired to implement IPS at Community Connections. Each had a caseload of 25 clients and carried out all phases of the vocational process. The IPS program was implemented according to a manual²² and monitored by the research team. Fidelity ratings, made regularly throughout the project by the research team, indicated that the IPS program consistently scored within the "high fidelity" range on the IPS Fidelity Scale.²⁷

The comparison condition, EVR, consisted of vocational rehabilitation services provided by several well-established agencies recommended by the District of Columbia Rehabilitation Services Administration. The EVR condition was considered "enhanced" because an extra vocational counselor was placed in the Rehabilitation Services Administration office to ensure that participants assigned to this condition were referred to appropriate rehabilitation agencies expeditiously. The vocational counselor monitored participants monthly, and if a client was dissatisfied with the program to which he or she was assigned or dropped out of vocational services, the counselor attempted to link that participant with another agency. All of the EVR agencies endorsed competitive employment as their goal but used stepwise approaches that involved prevocational experiences, primarily paid work adjustment training in a sheltered workshop.

To assess the amount of vocational services and to document that the 2 vocational interventions differed on critical process dimensions, researchers monitored program participants and their service use through bimonthly interviews.

Prior to enrolling in the trial, all participants received a diagnostic interview using the Structured Clinical Interview for DSM-III-R²⁸ to confirm eligibility. Major assessments were conducted at baseline and at 6-, 12-, and 18-month follow-up points. At each major assessment, research interviewers completed the Employment and Income Review,²⁹ the Global Assessment Scale,³⁰ the expanded Brief Psychiatric Rating Scale,³¹ the Rosenberg Self-Esteem Scale,³² and sections of the Quality of Life Interview.³³

The primary outcome, competitive employment, was defined as work in the competitive job market at prevailing wages with supervision provided by personnel employed by the business and in integrated work settings.^{10 - 11} Employment was assessed weekly by employment specialists in both programs. Every 2 months research interviewers asked participants about employment, vocational services received, and satisfaction that the program was helping them to achieve their goals. Employed participants were also assessed every 2 months with the Indiana Job Satisfaction Scale.³⁴ Both interrater and test-retest reliabilities were maintained at high levels (κ and intraclass correlations consistently above 0.7).

Complete vocational data were obtained on 150 participants (98.7%). For research interviews, follow-up rates at 6-, 12-, and 18-month assessments were 98.7%, 96.7%, and 94.7%, respectively. There were too few noncompleters to compare them statistically with completers.

Standard statistics were used to test group differences throughout the analyses. Univariate tests of significance were conducted using χ² tests for discrete data and t tests for continuous data to examine group equivalence at baseline and group differences on cumulative measures of employment. All statistical tests were 2 tailed with α set at .05. For distributions with high skewness or underused categories, the data were log-transformed or recoded into fewer categories. The mixed-effects regression approach (SAS PROC MIXED) for longitudinal data was used to test group differences on nonvocational outcomes. For these analyses, the covariance structure was specified as unstructured, and time was treated as a class variable.

The statistical analysis of employment outcomes was hampered because the group means and associated variances differed so greatly that the assumptions of conventional parametric tests, such as the t test, were not met. Nonparametric analogs, such as the Mann-Whitney test, were conducted to validate the parametric tests reported here and in every case produced equivalent results.

The demographics, Structured Clinical Interview for DSM-III-R diagnoses, recent histories of homelessness and hospitalization, and employment histories of the participants are given in Table 1. The 2 groups were comparable except that the IPS participants were slightly younger and had a higher rate of current drug use disorder (primarily cocaine). Other differences were not significant.

All IPS and EVR participants reported receiving some amount of vocational services. Participants in the EVR program were referred to 10 different rehabilitation programs, but three fourths (76.3%) received services from 1 of 3 main programs. Several (15.8%) attended more than 1 program.

There were no group differences in the proportion of participants who reported receiving vocational services during bimonthly intervals. For IPS participants, the proportion receiving vocational services declined gradually from 95% at 2 months to 61% at 18 months, and for EVR participants, the proportion changed from 84% at 2 months to 57% at 18 months. However, the IPS participants did report receiving more total vocational services during the study (IPS log mean hours=1.37, SEM=.06; EVR log mean hours=1.12, SEM=.05; t₁₄₈=3.22, P=.002). Consistent with the respective models, IPS participants received more help with job searches (IPS log mean hours=0.86, SEM=0.94; EVR log mean hours=−0.77, SEM=1.27; t_137.8=8.97, P<.001), while EVR participants received more job planning services (IPS log mean hours=0.25, SEM=1.03; EVR log mean hours=0.60, SEM=0.72; t_129.8=−2.43, P=.02). The 2 groups received similar amounts of job support. In addition, the same proportion of each group (98.7%) received mental health case management services throughout the study.

Competitive employment outcomes consistently favored IPS (Table 2). Participants in the IPS program were more likely than EVR participants to obtain a competitive job and to be working competitively during any month of the study, and they also had superior outcomes in other dimensions of competitive employment, including various measures of job tenure and earnings. The IPS participants rapidly obtained a rate of competitive employment of 22% by the second month of participation and maintained a level between 17% and 34% throughout the 18-month follow-up, whereas 5% or less of EVR participants were competitively employed throughout the study (Figure 1). Consistent with their educational and work histories, the 45 competitively employed IPS clients occupied a range of mostly entry-level jobs, such as clerical, sales, and service positions. On average, they worked 529.5 hours and earned $3084.11, at an average hourly wage of $5.82, during the 18-month study. Their average job tenure was 16.5 weeks.

Consistent with the stepwise approach, EVR participants were more likely than IPS participants to work in sheltered jobs (Table 2). The 54 EVR participants who worked in sheltered jobs earned on average $1878.57. Most of this work was piece rate, and the average earning rate was typically less than $1 per hour. There were no differences in vocational outcomes among the EVR participants who received services from the 3 major vocational rehabilitation agencies.

In addition to competitive and sheltered jobs, a few individuals in both the IPS and EVR programs worked in federal set-aside jobs through the National Industries for the Severely Handicapped program. These jobs have some characteristics of competitive employment (work for regular wages in existing federal agencies) and some characteristics of sheltered jobs (work in enclaves in positions that are only available to persons with disabilities). Total wages (from competitive, sheltered, and National Industries for the Severely Handicapped jobs) were equivalent for the 2 conditions (Table 2).

Other outcomes directly related to work, such as satisfaction with income, job, and vocational services, were generally high for both groups, but sometimes favored IPS (Table 3). Thus, participants in both groups reported greater financial support adequacy, satisfaction with finances, and satisfaction with vocational services (time effects). Similarly, Indiana Job Satisfaction ratings (averaged over all jobs and all rating periods) were high for both groups and revealed no group difference (IPS mean=3.40, SEM=0.45; EVR mean=3.25, SEM=0.50; t₉₉=1.57, P=.12). On the other hand, IPS participants were more satisfied with their vocational services over time than were EVR participants (interaction effect), were more satisfied that their vocational program was helping them to achieve their goals (group main effect), and reported higher Quality of Life Interview job satisfaction ratings at 1 of the 3 follow-ups.

Neither psychiatric diagnosis nor total psychiatric symptoms was associated with the amount of competitive work, the amount of sheltered work, or the probability of working. Work history, defined as the number of months of paid employment in the community in the past 5 years, did predict total wages earned in competitive work (r=0.27, n=150, P=.001), total wages earned in sheltered work (r=0.20, n=150, P=.02), and total wages earned in the study period (r=0.34, n=150, P<.001).

There were essentially no group or group × time interaction differences for nonvocational outcomes. However, both groups improved over time on global functioning, general quality of life, and self-esteem (Table 3), with post hoc comparisons consistently revealing that improvements occurred during the first 6 months of the study and then remained stable. Consistent with baseline differences, IPS participants had more psychiatric hospital days during the study (Table 3), and this group difference disappeared when the baseline value was entered as a fixed covariate rather than as part of the response vector. Both groups decreased hospital use after starting their vocational programs, but the reductions were not significant. Brief Psychiatric Rating Scale total scores increased significantly, but these increases were clearly in the subclinical range, with no average symptom scores above 2.0 and no increases on subscales.

The major finding of this study was that IPS was more successful than EVR in helping multiply impaired, inner-city patients with SMD to obtain competitive employment. Participants in the IPS program obtained competitive jobs faster and maintained their advantage throughout the 18 months of the study. Further group differences in time or amount of competitive work were consistent with the differential rates of employment. This main outcome accords with the positive results of previous studies of IPS and similar approaches to supported employment²⁰ and the negative outcomes in other studies of stepwise approaches.³⁵

Our study expands earlier results by showing that IPS is superior to EVR for patients who are usually considered poor candidates for vocational services owing to complicating conditions such as poor employment histories, dual diagnosis, and homelessness. This study also addressed another limitation of our earlier New Hampshire study¹⁶ by including several vendors, rather than a single vendor, in the EVR condition. Because competitive employment was negligible for all of the vendors, the failure of the stepwise EVR approach could not be attributed to poor performance by a single vendor. Inefficiencies within the Rehabilitation Services Administration were eliminated as an explanation for the failure of EVR because an extra vocational counselor insured that project participants were rapidly assessed and linked with vocational programs.

Although we believe that the differences in competitive employment in this study are most parsimoniously explained by the direct vs stepwise difference between the IPS and EVR approaches, other possibilities should be considered. One rival explanation concerns the organizational linkage between mental health and vocational services. In the standard service system, a patient receives mental health care from one agency and vocational services from a separate agency, which has the theoretical advantages of allowing the vocational agency to specialize, avoid the stigma of mental illness treatment, and develop a culture of high expectations.³⁶ In practice, however, separate and parallel agencies have difficulty cooperating to individualize services.^{37 - 38} The IPS program overcomes the communication difficulties of parallel services by having employment specialists join mental health teams, which could account for experimental differences.

The IPS participants reported more total vocational services, but these reports did not include time working in sheltered workshops, which is the core of the EVR approach. In fact, EVR programs engaged most of their patients in extensive sheltered work. Failure seems to have occurred at the step of transitioning from sheltered work to competitive jobs. Contrary to the models and stated goals of the EVR programs, their participants only rarely moved on to competitive work. Although rehabilitation agencies usually do not specify the length of time to transition from sheltered to competitive jobs, clients clearly expect the transition to occur before 18 months. In practice, however, participants in stepwise programs became stalled in sheltered jobs, because of either their own inertia or conservative forces within sheltered work programs.⁹

Another finding of the current study was the lack of evidence that entering vocational services in general or the more intensive, high-expectations approach of IPS in particular produced negative outcomes in nonvocational areas. On the contrary, these data suggest that participating in a vocational program is associated with improvements not only in vocationally related outcomes such as income but also in nonvocational outcomes such as quality of life and self-esteem. Because participants in the 2 programs engaged in similar rates but different types of work, our data suggest that type of work is less important than participating in a vocational program for the effect on nonvocational outcomes.

Our hypothesis concerning vocational earnings was not confirmed. Although IPS was clearly superior to EVR in helping patients to attain competitive employment, EVR participants earned as much money as IPS participants (primarily through sheltered workshops). The EVR participants also expressed similar satisfaction with their jobs as IPS participants, and the 2 groups had similar nonvocational outcomes. Given these findings, are there nevertheless reasons to favor competitive employment over sheltered work? The IPS participants were more satisfied than EVR participants that the program was helping them to achieve their goals, consistent with evidence from other studies that people with psychiatric disorders prefer competitive employment^{1 - 2} and do not move on from sheltered work to competitive jobs.⁹ Ultimately, however, the answer to this question may involve values: Do we believe it is better to integrate people with mental illness into mainstream society, or do we want to maintain separate work settings and keep them segregated from society?

In addition to the usual limits to generalization regarding particular patients and the enthusiasm of new programs, one caveat seems particularly warranted for this study. Although the patients were high risk, in the sense that most had histories of homelessness and dual diagnosis and continued to live in high-crime neighborhoods, they had the advantage of receiving intensive case management in a high-quality program that integrated mental health, substance abuse, and housing services. Inner-city patients in a more fragmented and less well-integrated service system may not benefit from IPS to the same extent or at all. Another limitation was the inability to maintain the blindedness of interviewers, though this factor could not account for the major difference in competitive employment.

In conclusion, IPS significantly increased the rate of competitive employment, even for this highly vulnerable population, while standard stepwise EVR approaches involved similar proportions of patients in sheltered work with few transitions to competitive employment. The current study also adds to mounting evidence that vocational services do not destabilize patients with SMD and may instead enhance their outcomes in other, nonvocational areas.

Accepted for publication March 30, 1999.

This work was supported by grant Mh51346 from the Substance Abuse and Mental Health Services Administration and the National Institute of Mental Health, Washington, DC, and grant Mh00839 from the National Institute of Mental Health.

Reprints: Robert E. Drake, MD, PhD, Psychiatric Research Center, 2 Whipple Pl, Lebanon, NH 03766 (e-mail: Robert.E.Drake@Dartmouth.edu).