People with severe psychiatric disorders experience unemployment rates typically above 85 percent; chronic unemployment is common (1). Although numerous studies have investigated the relationship between clinical and demographic indicators and vocational outcomes, most of them focus on outcome and follow-up or on program evaluation rather than on work history trajectories (2,3). Given the longitudinal and episodic nature of employment among people with serious and persistent mental illness (4), it is important to move beyond recording employment status as a single-moment-in-time characteristic and to describe employment history over time.

This study compared two groups of unemployed adults living with serious and persistent mental illness—a group who had held one or more jobs during the past five years and a group who had not. The relationship between long-term unemployment and demographic and clinical characteristics and participation in vocational rehabilitation programming was evaluated.

The sample included 219 participants in the Baltimore sample of the multisite Employment Intervention Demonstration Program. The five-year program (1995-2000) was funded to evaluate new ways of enhancing employment opportunities for mental health care consumers. Prospective participants were identified through a randomized list of all consumers receiving psychiatric care in one of four programs serving persons with serious and persistent mental illness. Written informed consent was obtained from participants; all met criteria for serious and persistent mental illness (5).

Fifty-seven percent of the sample (124 participants) were men, and 165 participants (75 percent) were from minority groups (mostly African Americans). Ages ranged from 21 to 69 years (mean±SD=41±8.5 years). A total of 107 participants (49 percent) did not graduate from high school. A total of 195 (89 percent) received Supplemental Security Income, Social Security Disability Insurance, or both types of Social Security benefits.

Of the 219 participants, 162 (75 percent) had a psychotic disorder as their primary diagnosis, and 54 (25 percent) had a mood disorder—most had bipolar disorder. The mean±SD lifetime number of hospitalizations was 11.4±10.1; only two participants (1 percent) had never been hospitalized. A total of 165 (75 percent) had a lifetime substance use disorder diagnosis. Eighty-seven (40 percent) reported current use of alcohol or an illicit substance; 110 (50 percent) reported some use of alcohol or an illicit substance in the past year.

Enrollment in the Employment Intervention Demonstration Program was restricted to those who were unemployed for at least three months before program entry. Of the 219 participants in the sample, 109 (51 percent) had been continuously unemployed for five or more years, and 105 (49 percent) had held one or more jobs during that time. No data were available for five participants. We obtained this information from responses to an item on a questionnaire that asked about work experience in the past five years. The dichotomous split was also selected to highlight the differences between those with a sporadic recent employment history and those who had not worked in several years. This dichotomous measure of long-term unemployment was the dependent variable.

At enrollment in the Employment Intervention Demonstration Program, participants completed the Structured Clinical Interview for DSM-IV (SCID-IV) Patient Version and the Positive and Negative Symptoms Scale (PANSS). They were interviewed about their clinical history and experiences with vocational programming. During the first 18 months of the demonstration program, all enrollees completed a full neuropsychological battery, which included the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) and four subscales of the Wechsler Adult Intelligence Scale, which were used to estimate full-scale IQ scores. A total of 150 of the participants in our study completed the neuropsychological battery.

Of the 219 participants, 208 (95 percent) reported at least one paid job during their lifetime. However, the tenure of work was relatively brief; the mean±SD number of months employed at the longest-held job was 40.9±46.8; median=24 and mode= 12). Among the 105 participants (49 percent) who held at least one job in the past five years, recent employment experiences were usually part-time. Among these participants, the mean±SD number of hours worked a week was 29.9±17.57. Job tenure was brief—a mean±SD of 20±45.06 months. The jobs held were low paying, with a mean±SD hourly wage of $5.14±$2.62.

Gender, race, and educational status were not significantly related to long-term unemployment. Older people, however, were more likely to have been unemployed; the mean± SD age of those who had been unemployed for more than five years was 43.13±8.7 years. The mean age of those who had held at least one job was 38.72±7.4 years (t=-3.99, df=108, 104, p<.001).

Those with a psychotic disorder were more likely to have been unemployed for five or more years than those with a mood disorder—87 of the participants with a psychotic disorder (55 percent) versus 20 of those with a mood disorder (38 percent) (χ²=4.77, df=1, p<.05).

t1 present the results of logistic regression models used to test relationships between long-term unemployment and a variety of clinical, neurocognitive, and vocational training variables. Diagnosis (psychotic disorder versus mood disorder), age, gender, education level, and race were entered as covariates. More hospitalizations and earlier age at first hospitalization were significantly associated with five-year unemployment. More pronounced negative symptoms were associated with a greater likelihood of long-term unemployment; however, the relationship fell just short of significance. Lower RBANS scores, indicating greater neurocognitive deficits, and lower estimates of global IQ were also both significantly associated with longer-term unemployment.

Current, past year, and lifetime DSM-IV diagnoses of alcohol and drug use disorders were not associated with long-term unemployment, nor was vocational training in the past five years—either job skills training or job-finding skills training.

This study examined longitudinal employment histories rather than considering employment as a single-moment-in-time characteristic. The only demographic characteristic associated with long-term unemployment was age. The lack of relationship of gender, race, and education to long-term unemployment suggests that the varied and deleterious effects of long-term serious mental illness may overshadow the impact of demographic factors on unemployment.

This hypothesis is reinforced by the positive correlation between indicators of illness severity, including more lifetime hospitalizations and younger age at first hospitalization, and the likelihood of long-term unemployment. The positive correlation between neurocognitive deficits, including lower RBANS scores and lower estimated IQ, are consistent with research demonstrating the importance of neurocognitive performance for various functional outcomes (6). Together they highlight the importance of vocational programs' carefully attending and responding to the neurocognitive and clinical deficits associated with serious and persistent mental illness.

The lack of a significant relationship between substance use diagnoses and long-term unemployment was surprising given the robustly documented ways that substance use impairs functioning across a variety of domains (7). This finding may suggest that substance abuse alone is not very useful as a screening measure for employment readiness.

Further, unlike most past studies, which evaluated relationships between specific vocational services and prospective employment outcomes, this investigation considered whether previous rehabilitation experience was related to employment history, and it found no correlation. There are several possible explanations. First, study participants came from a population of consumers with severe and long-standing psychiatric illnesses who may have needs that are not being met by traditional vocational interventions.

Second, because all study participants were drawn from the same treatment system, the vocational programming they received and the employment environment in which they lived may not be representative of other locales and systems, warranting caution in generalizing beyond the study sample. Third, the vocational services available to people with serious and persistent mental illness have long been criticized. Only 27 percent of the sample in this study, for example, reported ever having received such services, despite the fact that most people living with serious and persistent mental illness regard employment as a top priority (8).

In addition, the specific vocational rehabilitation services study participants received in the past simply may not have been effective. In Maryland and elsewhere during the time span of this study, vocational rehabilitation focused on psychosocial support and preemployment transitioning models and generally was not well integrated with clinical services. More recently, supported employment and "place and train" programs have been showing greater success (9,10). Thus the results of this study raise questions about what services would lead to more positive long-term employment outcomes; effective programs may include integrated services and more supported employment models.

Overall, the findings underscore the relevance of clinical impairments and neurocognitive deficits to long-term employment and highlight the need to critically reevaluate the effectiveness of traditional vocational rehabilitation services.

This study was made possible by cooperative agreement number UD7-SM-51824 from the Center for Mental Health Services as part of the Employment Intervention Demonstration Program. Collection of the neurocognitive data was supported by a grant from the Stanley Foundation.

The authors are affiliated with the Center for Mental Health Services Research of the department of psychiatry at the University of Maryland, Baltimore, MSTF Building, Room 300, 685 West Baltimore Street, Baltimore, Maryland 21201 (e-mail, rgoldber@umpsy.umaryland.edu). Dr. Goldberg, Dr. Lucksted, Dr. Dixon, and Dr. Lehman are also affiliated with the Mental Illness Research, Education, and Clinical Center of the Veterans Affairs Capitol Health Care Network in Baltimore.