- Psychological Issues
schizophrenia exacts a heavy toll on a person’s capacity to work, and a variety of vocational rehabilitation interventions have been developed over the past few decades to enhance the vocational capacities of persons with this disorder. The research literature on outcomes of vocational rehabilitation during the era of deinstitutionalization for persons with schizophrenia is reviewed. Most vocational rehabilitation programs have a positive influence on work-related activities, but most have failed to show substantial and enduring impacts on independent, competitive employment. Recent advances in supported employment suggest that vocational rehabilitation offers greater promise than do transitional and sheltered employment approaches. Vocational rehabilitation intervention may also exert positive influences on such clinical outcomes as medication compliance, symptom reduction, and relapse.
Schizophrenia Bulletin, 21(4): 645-656, 1995.
Vocational rehabilitation has assumed increasing importance as part of the array of services available for persons with schizophrenia. Work not only provides financial remuneration but is a normalizing experience, allowing individuals to participate in society, and may promote self-esteem and quality of life. Furthermore, the vast majority of persons with severe mental illnesses identify paid employment as one of their goals (Rogers et al. 1991). Despite this, rates of competitive employment for persons with schizophrenia remain dismally low — below 25 percent (Anthony et al. 1978).
Many different types of vocational rehabilitation programs have been developed and implemented. These are classified as follows (Bond and Boyer 1988): (1) hospital-based programs; (2) sheltered work; (3) assertive case management; (4) psychosocial rehabilitation, including prevocational training, transitional employment, and volunteer placements; (5) supported employment; and (6) counseling and education. The outcomes targeted by vocational rehabilitation interventions are divided into two broad categories. Vocational outcomes include full-time competitive employment, acquisition of job-related skills, acquisition of any job (paid or volunteer), percentage of time in paid employment (full-time or part-time, competitive or sheltered), total job earnings, level of job (unskilled, skilled, etc.), job satisfaction, and job performance. Vocational rehabilitation also may enhance outcomes other than work. These therapeutic outcomes include treatment compliance and symptom reduction, functional status in other areas (activities of daily living, maintenance of living situation, etc.), self-esteem, and subjective quality of life.
This review addresses the following questions:
As best as can be determined, all the studies reviewed examine these questions in the context of ongoing clinical treatment. Therefore, these questions are posed with the assumption that patients are receiving adequate clinical care, including antipsychotic medications.
This review focuses on the research on vocational rehabilitation interventions in the era of deinstitutionalization; research on vocational rehabilitation for patients who were living in long-term hospitals before deinstitutionalization has questionable relevance to today’s community-based care. The computerized bibliographic data bases, PSYCLIT and MEDLINE, were searched for 1966 to 1993. The key words used for these searches were vocational rehabilitation, sheltered workshops, employment or employment supported, and rehabilitation counseling. A total of 497 articles were identified.
This search yielded two excellent reviews (Bond 1986, 1992) that met the review criteria of Beaman (1991). However, neither review focused exclusively on studies of persons with schizophrenia. This reflects the fact that in vocational rehabilitation research, psychiatric diagnoses often are not specified, either because they are not available to the researchers or are viewed as irrelevant. These previous reviews cover the broader category of “psychiatrically disabled” or “severe or chronic mentally ill.” To deal with this problem, the primary studies covered by the reviews were examined to obtain information about the number and proportion of the subjects in the studies who had a diagnosis of schizophrenia and any information about the relationship of diagnosis to intervention impacts. This allowed an estimation of whether conclusions drawn in the reviews about the more heterogeneous category of persons with psychiatric disabilities need to be modified when restricted only to persons with schizophrenia.
Additional primary studies identified through the search were included in this review if they met the following criteria: (1) a design that at a minimum includes some comparison group — either a control or comparison condition or the same intervention offered to patients with schizophrenia compared with some other diagnostic group; (2) explicit inclusion of patients diagnosed with schizophrenia and at a minimum documentation of the proportion of the study sample that had this diagnosis; (3) evaluation of an explicit vocational rehabilitation intervention; and (4) systematic evaluation of defined and operationalized outcomes.
For the review questions posed, this article summarizes the findings from Bond’s previous reviews, reexamines the primary studies that he reviewed for findings specific to persons with schizophrenia, and adds results from studies that have been published or are “in the pipeline” since his 1992 review.
Table 1 summarizes the characteristics of the studies covered by Bond’s reviews and additional studies since 1992 that were identified in this search and not reviewed by Bond.
|Table 1. Characteristics and findings of vocational rehabilitation studies|
(% with schizophrenia)
|Vocational outcome results|
|Field et al. (not dated)||RA; followup period = 18 mo; VR = job developer and ACT; comparison condition = ACT and sheltered workshop||36 (unknown)||No group differences|
|Beard et al. (1963)||Rotating assignment; followup period = 3-12 no; VR = community-based PSR with transitional employment; comparison condition = waiting list||352 (75)||No group differences|
|Walker and McCourt (1965)||Non-RA; followup period = 6 mo; VR = active in inpatient VR program at VA; comparison condition = not active in VA VR program||211 (100)||No group differences|
|Briggs and Yater (1966)||RA; followup period = 13 mo; VR = vocational counseling; comparison condition = no VR||134 (unknown)||No group differences|
|Meltzoff and Blumenthal (1966)||RA; followup period = 18 mo; VR = day treatment with workshop; comparison condition = psychotherapy||69 (unknown)||VR > comparison condition for sheltered employment; no differences in competitive employment|
|Becker (1967)||RA; followup period = 8 mo; VR = hospital-based, intensive social/vocational program; comparison condition = custodial care||50 (78)||VR > comparison conditin for sheltered paid employment jobs; no group differences in employment after leaving program; results did not replicate when program was expanded|
|Weinberg and Lustig (1968)||RA; followup period = 2 yr; VR = sheltered workshop; comparison condition = vocational counseling||38 (unknown)||No group differences|
|Fairweather et al. (1969)||Randomized block assignment; followup period = 12 mo; VR = Fairweather Lodge; comparison condition = usual aftercare||150 (unknown)||VR > comparison condition for transitional employment; no differences in employment after program|
|Walker et al. (1969)||RA; followup period = 6 mo; VR = community transitional employment; comparison condition = hospital-based job placement||28 (50)||VR > comparison condition for transitional employment; no differnces in competitive employment|
|Purvis and Miskimins (1970)||RA; followup period = unknown; VR = group and individual counseling; comparison condition = no counseling||152 (unknown)||No group differences|
|Wolkon et al (1971)||RA; followup period = 12 mo; VR = PSR; comparison condition = usual aftercare||315 (78)||No group differences|
|Lamb and Goertzel (1972)||RA; followup period = 2 yr; VR = high-expectation halfway house; comparison condition = low-expectation halfway house||93 (unknown)||VR > comparison condition for paid employment and duration of employment; no differences in competitive employment|
|Marx et al. (1973)||RA; followup period = 5 mo; VR = ACT; comparison condition = continued inpatient care||61 (79)||VR > comparison condition for paid employment; comparison condition > VR for sheltered employment; no differences in competitive employment|
|Griffiths (1974)||RA; followup period = 18 mo; VR = sheltered workshop; comparison condition = usual aftercare||28 (unknown)||No group differences|
|Kuldau and Dirks (1977)||RA; followup period = 18 mo; VR = day hospital and sheltered work; comparison condition = rapid hospital discharge||89 (18)||VR > comparison condition for job starts after discharge and full-time employment|
|Azrin and Philip (1979)||RA; followup period = 4 mo; VR = job club; comparison condition = vocational lecture||25 (unknown)||VR > comparison condition for job starts|
|Stein and Test (1980)||RA; followup period = 12 mo; VR = ACT; comparison condition = short-term hospital stay and usual aftercare||116 (50)||VR > comparison condition for full-time employment and job starts|
|Velasquez and McCubbin (1980)||RA; followup period = 6 mo; VR = community residence; comparison condition = waiting list||94 (unknown)||VR > comparison condition for hours employed|
|Dincin and Witheridge (1982)||RA; followup period = 9 mo; VR = PSR and transitional employment program; comparison condition = social club||93 (75)||No group differences|
|Bell and Ryan (1984)||Non-RA; followup period = 6 mo; VR = in-hospital and community job placements; comparison condition = traditional inpatient||87 (71)||VR >comparison condition for employment after discharge earnings, and duration of jobs|
|Bond and Dincin (186)||RA; followup period = 15 mo; VR = PSR and rapid placement; comparison condition = PSR and gradual placements||125(55)||VR > comparison condition for job starts and job duration|
|Bell et al. (1993)||RA; followup period = 6 mo; VR = hospital-based sheltered job with pay; comparison condition = hospital-based sheltered job without pay||100 (100)||VR > comparison condition for job retention,, hours worked, and earnings|
|Drake et al. (1994)||Non-RA; followup period = 12 mo; VR = supported employment and continuous care team; comparison condition = partial hospitalization program||183 (44)||VR > comparison condition for competitive employment, hours worked, and earnings|
|Bon et al. (in press)||RA; followup period = 12 mo; VR = accelerated supported employment placement; comparison condition = gradual supported employment placement||86 (66)||VR > comparison condition for any competitive employment, fulltime employment, and earnings during followup|
|Note — ACT = assertive community treatment; PSR = psychosocial rehabilitation; RA = random assignment; VA = Department of veterans Affairs; VR = vocational rehabilitation condition.|
Do Vocational Rehabilitation Interventions Enhance the Vocational Outcomes of Persons With Schizophrenia? Bond (1986) conducted a formal meta-analysis on 19 studies spanning from 1955 to 1985, comparing a vocational rehabilitation intervention with some control condition. Eighteen of the studies used random assignment and one used a rotating assignment, which was judged to be a good approximation of randomization. The pooled paid employment rates among the experimental and control groups were 51 and 29 percent, respectively. The difference (22%) was judged significant by using Stouffer’s method for pooling results (z = 5.81, p < 0.0001). None of the nine studies that examined competitive employment found advantages to the vocational rehabilitation interventions compared with controls. However, for the seven studies that reported actual rates, the average rates were 31 percent for experimental subjects and 18 percent for controls, yielding a statistically significant difference by using Stouffer’s method (z = 2.26, p < 0.05). In 1992, Bond updated the results from his preceding review, adding four studies and extending some of the analysis of the literature. This second review upheld the previous conclusion that vocational rehabilitation interventions enhance vocational outcomes but, again, no studies found advantages for competitive employment. A number of studies found that vocational interventions were advantageous for a variety of vocational outcomes, including paid employment (5/12), full-time employment (4/8), job starts (5/8), duration of employment (6/12), and earnings (3/7). No vocational interventions showed statistically significant advantages on competitive employment. However, a formal meta-analysis found a modestly significant impact of vocational interventions on competitive employment. Pooling the results from 13 studies, the intervention effect on competitive employment was significant (Stouffer’s z = 2.9, p < 0.01).
These programs have succeeded in placing clients in jobs and in helping clients retain these jobs. The consistent deficiency of these programs has been their failure to prepare clients for future competitive employment outside the support provided by the rehabilitation program.
Thus, the success of vocational programs has been in helping clients adjust to a specific vocational environment, which has sometimes been sheltered or transitional employment. [p. 250.]
We reexamined the studies reviewed by Bond (1986, 1992) and the additional studies identified in our search with regard to their relevance for persons with schizophrenia. Fourteen of these studies (58%) provided adequate diagnostic information to assess this issue (see table 1). The proportions of patients with schizophrenia in these studies ranged from 18 to 100 percent (median = 75%). Three of four studies that examined sheltered employment as an outcome found advantages for vocational interventions. Two of three studies found advantages for vocational rehabilitation interventions in achieving employment after discharge, but only one of six found any advantage for vocational rehabilitation interventions on competitive employment. There does not appear to be any relationship between the proportion of the sample with schizophrenia and the vocational outcomes. Only one study (Walker et al. 1969) presented vocational outcome data according to diagnosis. The sample numbers are quite small, but they found that the same numbers of patients with and without a diagnosis of schizophrenia (5/7 for each) had regular employment. Hence diagnosis did not relate to vocational outcome in that study. Three other studies (Wolkon et al. 1971; Bell and Ryan 1984; Bond and Dincin 1986) report no differences in outcomes related to diagnosis but do not provide specific results. In sum, Bond’s conclusion that vocational rehabilitation interventions enhance employment, but not competitive employment, appears to hold for the subset of 10 studies in which the proportion of persons with schizophrenia could be ascertained.
Two studies not previously reviewed by Bond provide additional information about the effectiveness of vocational rehabilitation interventions. A quasi-experimental study in New Hampshire (Drake et al. 1994) compared vocational outcomes of two programs: one, a traditional partial hospitalization program, and the other, a supported employment program. Patients in one town, which converted its partial hospitalization program to a supportive employment program, coordinated closely with a clinical service team, had significantly better vocational outcomes compared with patients in another town that offered partial hospitalization. During the followup year the supportive employment patients increased their competitive employment rate from 25 to 39 percent (X2 = 5, p = 0.04) compared with no change among the partial hospitalization group (13.4% to 12.5%, not significant). The supportive employment patients also worked more hours (p < 0.001). Preliminary data from a more recent replication of this intensive supported employment approach conducted by Drake and colleagues (R.E. Drake, personal communication 1994) is yielding similarly promising results. The 12-month employment outcomes among 95 chronically mentally ill subjects randomly assigned to either the intensive supported employment approach or to standard “vendor” private vocational rehabilitation agencies consistently favor the intensive supported employment program. These outcomes include weeks worked (19.1 vs. 5.6, p < 0.0001). In summary, these additional studies support the previous conclusion by Bond (1986, 1992) that vocational rehabilitation enhances employment rates among persons with psychiatric disabilities. The Drake et al. (1994) studies suggest that supportive employment closely coordinated with clinical care can enhance the rate of competitive employment among these patients, an encouraging finding in light of the less favorable overall results with regard to competitive employment outcomes in previous vocational rehabilitation studies.
Do Vocational Rehabilitation Interventions Enhance Outcomes Other Than Vocational Functioning? In his reviews, Bond (1986, 1992) concluded that vocational interventions are associated with reduced hospital admissions. Nine of seventeen studies that used rehospitalization as an outcome found an advantage for the patients receiving vocational rehabilitation intervention (meta-analysis Stouffer’s z score = 6.2, p < 0.01). Four of the studies reviewed by Bond provide some information about the impact of vocational rehabilitation interventions on outcomes other than hospital recidivism. Kuldau and Dirks (1977) reported that compared with patients in the control group, those receiving vocational rehabilitation were more likely to be taking antipsychotic medications at an 18-month followup (among the patients with schizophrenia: 86% of vocational rehabilitation patients vs. 33% of controls, Fisher’s exact test p = 0.02). Although not presented according to diagnosis, the vocational rehabilitation patients at the 18-month followup were also more likely to have activities with friends (82% of vocational rehabilitation patients vs. 61% of controls, X2 = 4.02, p < 0.05 2 = 4.2, p < 0.05). 2 = 462, p < 0.05). Although not covered elsewhere in this review because of its limited scientific design and lack of a comparison group, a recent naturalistic follow-up study of 88 patients (about 88% with schizophrenia) in three psychosocial rehabilitation programs in South Carolina (Ares and Linney 1993) examined the longitudinal relationship between vocational status, self-efficacy/self-esteem, and quality of life. Concerned that other studies have failed to find a clear relationship between employment and quality of life among persons with psychiatric disabilities, the researchers hypothesize that it is change in vocational status (e.g., getting a job) that mediates impacts on self-esteem and quality of life. Their analyses tend to support this. The relevance of this finding here is that future research on the relationship of vocational interventions to such outcomes as self-efficacy and quality of life should focus not only on employment status but also on changes in employment status.
Are There Differences in the Effectiveness of the Various Types of Vocational Rehabilitation Interventions for Persons With Schizophrenia? Bell et al. (1993) found that adding pay to hospital-based sheltered jobs significantly increased the likelihood that patients would start and remain in the jobs. During the first week the job start rates were 97 percent for those assigned to the paid group and only 37 percent for those assigned to the unpaid group. The rates at week 13 were 65 and 14 percent, respectively, and at week 26, 35 and 5 percent, respectively. The patients who were paid worked more hours during the 26 week intervention period (275 vs. 112 hours, t = 4.83, p < 0.001. A study by Bond et al. (in press) is exceptional in regard to its exploration of the differential effectiveness of two vocational rehabilitation interventions: accelerated supported employment and gradual entry into supported employment. The findings consistently favored the accelerated condition. The accelerated-entry patients showed 1-year outcomes superior to the gradual entry patients on the following vocational measures: (1) any competitive employment during the 1 year (56% vs. 29%, F = 6.23, p < 0.05); What Patient Characteristics Predict Response to Vocational Rehabilitation? The literature does not permit an adequate assessment of the impacts of patient characteristics on response to vocational interventions. The relationships of demographic characteristics (age, gender, race, education) of persons with schizophrenia to their response to vocational interventions are largely unknown. In Fabian’s (1992) time series study of supported employment, she found no gender differences, but race was a significant predictor, with job survival rates of 52 percent among whites and 27 percent among minorities (p < 0.05). A major controversy has existed with regard to the relationship of clinical characteristics, especially diagnosis and symptoms, to vocational capacity. In a highly influential review, Anthony and Jansen (1984) argued that psychiatric symptomatology and diagnosis are poor predictors of future work performance. This conclusion has generated much debate because of its counter-intuitive implications to clinicians, namely that the manifestations of psychiatric illnesses do not relate to vocational disability. The impetus for the Anthony and Jansen review was a position taken by the Social Security Administration (SSA) under the disability Amendment of 1980 to the Social Security Act that mentally disabled recipients of benefits who no longer exhibit psychiatric symptoms are no longer disabled and therefore no longer eligible for benefits. Their review challenged this position, arguing instead that SSA’s position was unfounded because of the lack of evidence that symptoms and even diagnosis correlate with the ability to work. They asserted that SSA must independently assess a person’s capacity to work. Anthony and Jansen pointed out that better predictors of future work performance are past employment history, ratings of a person’s work adjustment skills made in a workshop setting or sheltered job site, a person’s ability to “get along” or function socially with others, and pencil-and-paper tests that measure a person’s ego strengths or self-concept in the role of worker. Unfortunately, this conclusion has been cited at times as evidence that clinical factors are irrelevant to the process of vocational rehabilitation for persons with psychiatric disabilities and has contributed to rifts between clinicians and rehabilitation specialists.
Taken in its original context, most clinicians would endorse Anthony and Jansen’s conclusion that psychiatrically disabled persons should not lose their disability benefits because their symptoms have abated in response to treatment. This is particularly important for the positive symptoms of schizophrenia. There is in fact general agreement that past work history is the best predictor of future work (Strauss and Carpenter 1974; Anthony and Jansen 1984; Bond and Boyer 1988). However, beyond this general level of agreement, controversy remains as to the relevance of clinical symptoms, diagnosis, and treatment to vocational performance and response to vocational interventions.
Milstein et al. (1991) have noted several potential problems with the research, leading to the conclusion that symptoms and work performance are unrelated. First, procedures for assessing symptoms and diagnosis have changed significantly over the years and many of the earlier studies on which the Anthony and Jansen (1984) review was based used imprecise or currently outdated assessment procedures. Second, most studies use diagnostically heterogeneous samples, which will obscure the relationships of symptoms to work within diagnostic categories. Studies have primarily focused on the ability of symptoms to predict future work performance. The relationship between symptoms and work function may be more immediate because symptoms and capacity to work can vary considerably over time. Finally, current conceptualizations of schizophrenia emphasize the multiple domains of impairment that define this disorder: hallucinations and delusions, cognitive and attentional impairments, and primary deficit symptoms (Carpenter et al. 1993). The relationships of these separate domains of psychopathology to the capacity to work and potential for vocational rehabilitation have not been assessed.
Certainly there is evidence in the literature that diagnosis and symptoms are related to work performance. Employment rates among persons with psychiatric disabilities (and schizophrenia in particular) are low relative to the general population. Overall competitive employment rates below 20 percent are the norm for these persons (Anthony et al. 1972). Among the studies reviewed here, Fabian (1992) and Jacobs et al. (1992) found that the diagnosis of schizophrenia predicted poorer response to a vocational intervention. Fabian (1992) reported significantly higher job dropouts among patients with schizophrenia or schizoaffective disorders compared with other diagnostic groups. Only 19 percent of the patients in the schizophrenia group were skill employed at the end of the 1 year followup compared with 46 percent for the major affective disorder patients (D = 7.3, p = 0.006) and 57 percent for the patients with personality disorders (D = 6.4, p = 0.012). Jacobs et al. (1992) also found that patients with schizophrenia fared more poorly than other patients in their job-finding club program. Among the patients with schizophrenia, only 15 percent had a positive outcome, defined as either finding a job or entering a job training program. The overall rate for all patients was 36 percent, and the rate for patients with schizophrenia and bipolar disorders combined (19%) was significantly lower than for the other diagnostic groups combined (51%) (X2 = 10.5, p = 0.02).
Milstein et al. (1991) and Breier et al. (1991) found significant correlations between symptoms and work performance. Most recently, Anthony and colleagues (1995) found in a longitudinal followup of 275 psychosocial rehabilitation clients (57% with schizophrenia) that current symptoms correlated significantly with current work skills and employability over time (r = -0.23 to -0.37 over five time points, p < 0.0001). Finally, in a more detailed analysis of cost-effectiveness on the data from the Stein and Test (1980) study of assertive community treatment, Weisbrod (1983) examined differential intervention effects for three diagnostic groups: schizophrenia, personality disorder, and nonschizophrenic psychoses. With regard to employment outcomes, he reports that the differences between the experimental and control groups (E-C) in earnings from competitive employment for these three diagnostic groups were $532, -$52, and $2,206, respectively. These E-C differences were statistically significant only for the nonschizophrenic psychoses group. E-C differences in sheltered employment earnings for the three groups were $247, -$20, and $182, respectively. The E-C difference on sheltered employment was significant only for the schizophrenia group. These findings highlight the potential importance of diagnosis for intervention effects and the different patterns of findings that may occur depending on the vocational outcome assessed (competitive vs. sheltered work).
The fact remains that the relationships between the psychopathological processes underlying the schizophrenia syndromes and the ability to work are not well understood. Much needs to be done to clarify which patients with schizophrenia will benefit from the various available models of vocational rehabilitation. These studies suggest that in general patients with schizophrenia are less responsive than patients in other diagnostic groups to vocational interventions and that past level of functioning, age at onset, and duration of unemployment are significant prognostic factors for vocational outcomes.
The literature on vocational rehabilitation is limited in a number of ways. The number of controlled trials of vocational rehabilitation programs is small, especially with regard to any single model of vocational rehabilitation. The controlled studies that exist are generally of limited quality and have examined diagnostically heterogeneous patient groups, often not specifying diagnoses, thus further limiting their utility for this review on schizophrenia. Research methodology for studying vocational rehabilitation is progressing with regard to description of the interventions, standardization of vocational outcomes, and specification of the targeted subgroups, but further work is needed. The changing nature of vocational rehabilitation over the past 40 years is reflected in the types of programs that have been evaluated over the years, ranging from hospital-based workshops to supported employment in the community. The recent rapid transition in the nature of these programs makes the relevance of earlier work questionable.
No definitive conclusion can be offered as to whether vocational rehabilitation interventions enhance the vocational outcomes of persons with schizophrenia. We can state with moderate confidence that vocational rehabilitation programs by definition enhance the vocational activities of persons with psychiatric disabilities while patients are in these programs, but that they do not have significant effects on rates of competitive employment after leaving the programs. The most recent studies on supported employment show more promise with regard to competitive employment, but it is too early to draw conclusions. However, the degree to which these findings apply to persons with the more specific spectrum of schizophrenic disorders is not known.
Nor is there enough information to draw any conclusions about the capacity of vocational rehabilitation interventions to enhance outcomes other than vocational functioning. There is a trend that improvements in vocational functioning are correlated with improvements in other outcomes, such as reduced symptoms and relapse, but it is not clear that there is any causal relationship between this correlation and vocational rehabilitation interventions.
No conclusions can be drawn about differences in the effectiveness of the various types of vocational rehabilitation interventions for persons with schizophrenia. Only two of the studies reviewed addressed this question in any way.
The most consistent finding on what patient characteristics predict response to vocational rehabilitation is that more extensive past employment experience predicts a more favorable response to vocational rehabilitation programs. It can be stated with limited confidence that a diagnosis of schizophrenia is a negative predictor of response to vocational rehabilitation compared with other diagnoses. There is no evidence that other demographic characteristics predict response to vocational rehabilitation.
Anthony F. Lehman, M.D., M.S.P.H., is Professor of Psychiatry, University of Maryland School of Medicine and Co-Director of the Center for Mental Health Services Research, University of Maryland, Baltimore, MD
This article, originally from the Schizophrenia Bulletin, is in the public domain and may be reproduced or copied without requesting the author’s permission.