Employment retention by persons with schizophrenia employed in non-assisted jobs.
Analysis of the employment retention experience for this study population may advance understanding of important issues relating to provision of rehabilitation and treatment services. In particular, examining the risk of employment loss for persons not currently receiving vocational support services, and identifying the groups for which this risk is highest, can be informative about unmet need for vocational services, and help in targeting service expansions. It may also be the case that, in the absence of formal vocational supports, clinical and treatment factors are particularly important determinants of employment retention.
In this study, data from a large observational study of persons under treatment for schizophrenia in six different geographic areas of the U.S. is used to examine employment stability of persons working at study enrollment in jobs where they did not receive formal vocational supports. Using baseline and six-month follow-up data on these individuals, the significance of selected indicators of social, demographic, clinical, functional, and treatment status as predictors of employment status at six months after study enrollment was evaluated.
Data from the Schizophrenia Care and Assessment Program (SCAP), which is an ongoing longitudinal study of adults with schizophrenia recruited from systems of behavioral health care in six regions of the U.S. was analyzed (Mark, Dirani, Russo, & Slade, 200l). The regions are: Baltimore, New Haven, San Diego, selected counties in Florida, selected counties in Colorado, and selected counties in North Carolina. The behavioral health care systems include community mental health centers, public psychiatric hospitals, Veterans Administration mental health providers, and university outpatient clinics. Recruitment began in July 1997 and was completed in January 2001. All participants were over age 18 and had a schizophrenia, schizophreniform, or schizoaffective diagnosis prior to recruitment. Data were obtained from interviews, clinical assessments, and medical records reviews. The SCAP is designed to examine the impacts of medical treatment on "clinical, functional, and economic outcomes" for persons with schizophrenia (Mark et al.).
The subjects for this study were SCAP participants who were classified, based on their interview responses, as employed in non-assisted jobs at baseline. Participants were defined as employed if they reported paid employment at some time in the four weeks preceding the baseline interview. This parallels the standard U.S. Census Bureau definition of employment used in the Current Population Survey and the Survey of Income and Program Participation (Westat & Mathematica Policy Research, 2001). Similar questions for defining employment have also been used in other survey instruments designed specifically for persons with schizophrenia, such as the Survey Utilization and Resources Form for Schizophrenia (SURF). Employed persons in non-assisted jobs were identified by when they did not report 1) that their job was "in a sheltered workshop" or 2) that they had a "job coach or special supervisor". These same survey items were previously used in the Schizophrenia PORT study (Lehman et al., 1998) to determine the presence or absence of formal vocational supports.
Study subjects were drawn from the population of all 2,250 SCAP consumers for whom baseline and six-month follow-up data had been collected between July 1997 and July 2001. Those with incomplete medical records information (522), or missing employment information at baseline (233), were deleted from our analysis. Of the remaining 1,495 consumers, 173 were identified as holding a non-assisted job at baseline. Of these 173 consumers, 14 were dropped from the analysis because of missing employment status data in the six-month follow-up interview.
Information on the characteristics of these non-assisted jobs is not available in our data set. Survey responses concerning days and hours of work suggest that these jobs were a mixture of fulltime and part-time positions. Respondents reported a mean of 12.78 days worked in the four weeks preceding the baseline interview, with 35.4 percent working 20 days or more. The mean usual hours worked on working days was 5.83, with 36.5 percent reporting working eight hours or more. Mean earnings per hour was $6.94.
The outcome variable is self-reported employment status at the six-month follow-up interview, using the prior four weeks as the reference period. For most of our analyses, this outcome was coded as employed vs. not employed. In some regression analyses, we used an ordered trichotomy to describe outcome: not employed, employed in an assisted job, and employed in a non-assisted job. The same survey items used in determining baseline employment status were also used for the six-month outcome data.
As the sample size was small (n = 159), only a limited number of predictors were used in testing for relationships to these employment outcomes. Demographic predictors were gender, age, and race. Socio-economic status was captured by years of education completed.
Baseline clinical/functional status measures were the Global Assessment of Functioning (GAF) score, a binary indicator for a (self-reported) psychiatric hospitalization in the year preceding the baseline date, a binary indicator for living in an independent residential situation (alone or with family), and the consumer's Montgomery-Asberg Depression Rating Scale (MADRS). These measures were intended to capture the consumer's ability to work productively and maintain their regular working schedule.
Two factors relating to the treatment process were included in the study. It has been noted that the occurrence of acute episodes represents a serious obstacle to maintaining employment (Peckham & Muller, 1999; Rutman, 1992). This factor is incorporated into our analysis by including a binary indicator for one or more psychiatric hospitalizations during the 6-month follow-up period. Recent literature also suggests a possible relationship between the type of antipsychotic medication used and employment status at any point in time (Slade & Salkever, 2001). To test for a medication effect on employment retention, we classified each patient's drug therapy for the 6 months preceding the baseline date into one of three modes: l) consumers who only received prescriptions for typical antipsychotics, 2) consumers who received at least one prescription for a second generation antipsychotic (clozapine, olanzapine, quetiapine, risperidone, sertindole), and 3) consumers who received no prescriptions for antipsychotics. The hospitalization and drug treatment variables were based on information abstracted from medical records.
Finally, in multiple regression models, separate regression intercepts for each of the six geographic regions from which patients were recruited were included. These were included to control for regional mental health care system or labor-market effects that might have impacted on employment outcomes.
Three different methods were used to test for relationships between our predictor and outcome measures. First, we used the simple dichotomy of employed vs. not employed as our six-month outcome and tested for the bivariate associations between this outcome and each predictor variable. Second, using this same outcome dichotomy, we estimated a multiple logistic regression with all predictors included. This provided a test for significance of individual predictors while controlling for other predictors in our analysis. We also tested the stability of significant regression results by deleting insignificant predictor variables from our regression and reestimating the coefficients for the remaining variables and the regional intercepts. Third, we estimated an ordered multiple logistic regression model with the following ordering for our outcome categories: not employed, employed in a job with formal vocational support, and employed in a non-assisted job. Unlike the two preceding analyses that treat employment in assisted and non-assisted jobs as equivalent outcomes at follow-up, the third analysis implicitly treats the outcome of employment in a non-assisted job as more positive than employment in an assisted job. The rationale for this was that the non-assisted job represents a greater degree of vocational independence. In supplementary analyses of the 111 study consumers who were employed at 6 months, bivariate and multiple logistic regression comparisons of consumers in assisted versus non-assisted jobs were conducted. The statistical package used for all computations was STATA.
Characteristics of the 159 study subjects are shown in Table 1. Roughly two-thirds are male, just under half are under the age of 40, 42.2 percent are non-Caucasian, and just under half have education beyond high school. In terms of level of functioning, just under one-third had GAF scores over 60 at study entry and 78 percent were living in the community without supports. With regard to treatment history, 28.3 percent were hospitalized for mental or emotional disorders in the 12 months prior to study entry, and 12.6 percent were hospitalized in the six months following study entry. One or more prescriptions for antipsychotic drugs during the six months prior to study entry were reported by 93.1 percent of consumers, including 53.5 percent who received at least one prescription for a second-generation antipsychotic,
We compared the study subjects to SCAP consumers who were excluded from the analysis. Study subjects tended to be relatively high functioning and well educated compared to the 1,322 SCAP consumers who were excluded because they did not have a non-assisted job at baseline. Comparison of the two groups showed the following significant differences (p < 0.1) for our study subjects: a lower fraction reporting psychiatric hospitalization in the year prior, a lower fraction hospitalized in the six-month follow-up, higher GAF scores, higher levels of educational attainment, and a higher percentage in residential situations without support services.
Compared to the 140 consumers excluded because they held an assisted job at baseline, our study subjects had significantly more education and were significantly more likely to have been living in non-supported residential situations, while their higher GAF scores fall just short of a significant difference (p-value =0.104). Finally, no significant differences were found between our study subjects and the 14 consumers who met the baseline employment criterion but were excluded because they did not report any information on employment status at the six-month follow-up.
Table 2 reports on employment status at the six-month follow up for the study subjects. Overall, 69.8 percent continued to report being employed for pay (Column (a)). The data in Columns (b) and (c) show that 80.2 percent of the study subjects who maintained employment remained in non-assisted jobs while 19.8 percent reported formal job supports at six months.
Results of the bivariate chi-square tests in Columns (e) and (f) show that education was strongly associated with continued employment; the risk of joblessness is about half as low for subjects with more than a high school education than for other subjects. Risk of joblessness was also significantly greater for subjects who had been hospitalized in the prior year and for subjects who were hospitalized in the follow-up period. The latter result strongly supports the expectation that acute episodes of schizophrenia have negative consequences for employment stability.
Differences in jobless rates among the three different drug treatment groups were also marginally significant. The highest jobless rate (39.7%) was observed for subjects who were only prescribed typical antipsychotics in the six months prior to study entry. The lowest jobless rate (18.2%) was observed for the few subjects with no prescriptions for antipsychotics during the same period, while subjects receiving atypical antipsychotics reported a jobless rate of 24.7%. Presumably the result for subjects receiving no antipsychotics reflects a lower severity level for their schizophrenia. This conjecture is supported by comparisons of GAF scores; those receiving no antipsychotics have a higher mean score (56 vs. 49) and a higher minimum score (35 vs. 10).
Bivariate chi-square tests were also performed, for the 111 consumers maintaining employment after six months, to compare the percentages remaining in non-assisted vs. assisted jobs at the six-month follow-up, (The data used in these tests are based on columns (b) and (c) of Table 2.) Of the characteristics shown in Table 2, only residential status was significantly related to the probability of remaining in a non-assisted job (p < 0.1), but the corresponding result for the GAF score groups were nearly significant (p=0.116). These results indicate that among those consumers who remained employed, the probability of remaining in a non-assisted job is greater for consumers who live independently or whose overall level of functioning is higher. (Detailed test results are available from the authors.)
Results of multiple logistic regression analyses reported in Tables 3 and 4 were fairly consistent with the bivariate results in columns (e) and (f) of Table 2. Regression models of the risk of joblessness at 6 months are shown in Table 3. Results with all predictor variables (i.e., the left-hand panel) suggest a very strong negative relationship between years of education and risk of joblessness. Coefficients for atypical antipsychotics and for no antipsychotics (versus the reference group of only typical antipsychotics) show reductions in the relative risk of being jobless of 53.4 percent (p=0.087) and 78.7 percent (p=0.119) respectively. Hospitalization during the follow-up period more than triples the relative risk of joblessness, though this result is not quite statistically significant (p=0.103). Age effects on joblessness risk are also significant; older patients are at greater risk of joblessness and the added risk from each additional year of age is greater for older patients than for younger patients. The result from our bivariate comparison concerning prior hospitalization is, however, not replicated in the logistic regression. (This is due to the correlation between the prior and follow-up hospitalization variables.) Results of testing the sensitivity of our coefficient estimates to deletion of insignificant predictors are shown on the right-hand side of Table 3. Findings are quite similar to the results with all predictors included.
Analogous logistic regression models with three ordered employment outcomes at the six months follow-up (jobless, employed in an assisted job, and employed in a non-assisted job) are shown in Table 4. Results for the full regression model are in the left-hand panel and results when the list of regressors is reduced via stepwise elimination are in the right-hand panel. The regression coefficients are similar, in terms of sign, magnitude and significance, to the coefficients reported in Table 3. The main differences are that the coefficients of the age and follow-up hospitalization variables now have associated p-values that just exceed 0.1 while the GAF variable is a stronger predictor of more positive employment outcomes.
Finally, a multiple logistic regression on assisted versus non-assisted jobs for the 111 consumers employed at follow-up only yielded two significant predictors. Lower GAF scores and the binary indicator for African-Americans were both predictive of assisted jobs. (Results are available from the authors.)
Our analysis suggests several predictors that may be important protective factors against the risk of job loss for adults with schizophrenia. The strongest evidence pertains to the influence of education; more highly educated consumers have a much higher probability of maintaining employment. Several interpretations of this effect are possible. One is that consumers with more education are better able to cope with the problems presented by their disorder without seriously compromising their work performance. Another possibility is that jobs held by consumers with more education offer greater flexibility and make it easier to accommodate treatment side effects and fluctuations in symptoms over time. A third possibility is that informal job supports are more substantial in the types of jobs held by consumers with more education.
There is also consistent evidence that consumers receiving atypical antipsychotics were more likely to remain employed than consumers treated only with first-generation drugs. Atypical antipsychotic medications may increase employment retention by improving patient compliance or side-effect profiles. We cannot rule out the possibility that unobserved factors related to receipt of an atypical antipsychotic explain our finding, but this alternative explanation is unlikely, given that patients who are more severely ill, and thus are at greater risk of becoming jobless, are more likely to be receiving an atypical antipsychotic medication (Mark et al., 2001). Thus, selection on unobserved factors may imply that the true impact of atypicals on employment is underestimated in this sample. In the case of the small number of consumers receiving no antipsychotics, however, selection factors are the most likely explanation for our results; the considerable evidence supporting the effectiveness of antipsychotic therapies (United States Department of Health and Human Services, 1999) makes alternative explanations unlikely.
Our findings also provide strong confirmation of the detrimental employment impact of acute episodes that result in hospitalization. This suggests that treatment approaches that reduce hospitalization, such as assertive community treatment (Salkever et al., 1999), should have important payoffs in helping consumers who are employed to retain their jobs.
Finally, the increasing risk of job loss with age is surprising in view of evidence that disorder severity and risk of hospitalization tend to decline with age (USDHHS, 1999; Eaton et al., 1992). Possible explanations include an increasing risk with age of disability due to causes other than schizophrenia and the tendency for older workers to take early retirement or apply for disability benefits (Kouzis & Eaton, 2000).
Several study limitations should be noted. One important limitation is the absence of random assignment. As noted above, this raises the possibility that unobserved factors that influence the choice of drug therapies may explain some reported findings. A second limitation is the absence of detailed information on job characteristics and informal, job supports. The importance of informal supports has been noted as a factor contributing to positive employment outcomes (Blankertz & Keller, 1997); their omission in our analysis clouds the interpretation of results for factors that may be correlated with informal supports (e.g., education). Third, generalizations to all patients with schizophrenia may not be warranted. SCAP subjects were recruited from large public mental health and university-based treatment systems. While most schizophrenia care is provided in such settings, there are some types of provider and patient groups not captured in our data. Moreover, we have focused on a group of consumers that is characterized typically by relatively high levels of functioning and relatively low risk profiles for acute episodes and hospitalizations. Finally, small study group size is a limitation; a larger study group is required to resolve the several observed differences between our bivariate and multiple regression results.
In this study of 159 persons with schizophrenia in non-assisted jobs, less than 70 percent were still employed at six-months follow-up. This high rate of job loss among employed persons receiving no formal vocational supports suggests the possibility that expanding vocational services to this group will yield benefits in terms of greater employment stability. We also find that the rate of employment retention for our study subjects varies with patient and treatment characteristics. Education, treatment with second-generation antipsychotics, and no medications were predictive of a higher rate of employment retention while the risk of job loss was greater for persons hospitalized during the 6-month follow-up and for older workers. These results are of potential interest in targeting expanded vocational support services to those persons at the highest risk of job loss. These results can help to bring greater focus to subsequent, more substantial explorations of the factors that promote or inhibit job retention among persons with schizophrenia who are working in community settings without formal job supports.
Table 1: Characteristics of the Study Sample (N= 159) Gender % Male 67.9 Female 32.1 Age Under 40 yrs. 47.8 Over 40 Yrs. 52.2 Race/Ethnicity Caucasian 57.9 African American 34.0 Other 8.2 Years of Education Completed Less than High School 20.8 High School 32.1 More than High School 47.2 Residential Status Supported 78 Not Supported 22 Prior Yr. Psych. Hospitalization Yes 28.3 No 71.7 Psych. Hospitalization in Follow-Up Yes 12.6 No 87.4 Antiosychotic Rx Prior 6 Mos. First Generation Only 39.6 Second Generation 53.5 None 6.9 MADRS at Baseline (higher score = more depressed) <15 61.6 15-40 38.4 GAF at Baseline (higher score = better functioning) <40 25.2 40-60 42.8 60 32.1 Table 2: Six-month Employment Outcomes and Bi-variate Tests for Predictors of Employment Retention % Employed All Non-Assisted Assisted Jobs Jobs Jobs (a) (b) (c) All Study Consumers 69.8 56 13.8 Gender Male 71.3 57.4 13.9 Female 66.7 52.9 13.7 Age Under 40 yrs. 71.1 54 17.1 Over 40 Yrs. 68.7 57.8 10.8 Race/Ethnicity Caucasian 70.7 59.8 10.9 African American 68.5 50 18.5 Other 69.2 53.9 15.4 Years of Education Completed Less than High School 57.6 45.5 12.1 High School 62.8 45.1 17.7 More than High School 80 68 12 Residential Status Assisted 62.9 37.1 25.7 Not Assisted 71.8 61.3 10.5 Prior Yr. Psych. Hospitalization Yes 57.8 46.7 11.1 No 74.6 59.7 14.9 Psych. Hospitalization in Follow-Up Yes 50 40 10 No 72.7 58.3 14.4 Antiosychotic Rx Prior six-months First Generation Only 60.3 47.6 12.7 Second Generation 75.3 60 15.3 None 81.8 72.7 9.1 MADRS at Baseline < 15 71.4 56.1 15.3 15-40 67.2 55.7 11.5 GAF at Baseline < 40 65 45 20 40-60 67.7 52.9 14.7 60 76.5 68.6 7.8 % Not Test for Employed Employed vs. Not Employed Chi Squared P (d) (e) (f) All Study Consumers 30.2 Gender Male 28.7 0.35 0.553 Female 33.3 Age Under 40 yrs. 29 0.106 0.744 Over 40 Yrs. 31.3 Race/Ethnicity Caucasian 29.4 0.076 0.963 African American 31.5 Other 30.8 Years of Education Completed Less than High School 42.4 7.247 0.027 High School 37.3 More than High School 20 Residential Status Assisted 37.1 1.03 0.31 Not Assisted 28.2 Prior Yr. Psych. Hospitalization Yes 42.2 4.312 0.038 No 25.4 Psych. Hospitalization in Follow-Up Yes 50 4.261 0.039 No 27.3 Antiosychotic Rx Prior six-months First Generation Only 39.7 4.659 0.097 Second Generation 24.7 None 18.2 MADRS at Baseline < 15 28.6 0.317 0.573 15-40 32.8 GAF at Baseline < 40 35 1.664 0.435 40-60 32.4 60 23.5 Table 3: Multiple Logistic Regressions for Jobless vs. Employed at Six Months Full Model P- Odds Predictor Variables ** Coefficient value Ratio Gender (Male = 1) -0.376 0.374 0.687 Age (yrs.) at Baseline -2.061 0.070 0.127 Age Squared 0.265 0.055 1.304 African-American (=1) -0.009 0.987 0.991 Other Non-Caucasian (=1) 0.095 0.900 1.100 Years of Education Completed -0.229 0.007 0.796 No Antipsychotic Drug (=1) -1.548 0.119 0.213 Atypical Antipsychotic Drug (=1) -0.763 0.087 0.466 Supported Residence (=1) 0.473 0.405 1.605 Psych. Hospitalization in Prior Yr. (=1) 0.286 0.579 1.330 Psych. Hospitalization in Follow-Up (=1) 1.158 0.103 3.184 Baseline GAF (<40=1, 40-59=2, >59=3) -0.229 0.479 0.795 Baseline MADRS >14 (=1) 0.016 0.462 1.106 Model Chi squared 27.790 0.065 Pseudo R-squared 0.1427 Percent Correct Predictions 75.500 Reduced Model * Gender (Male - 1) -1.946 0.086 0.143 Age (yrs.) at Baseline 0.249 0.073 1.282 Age Squared African-American (=1) Other Non-Caucasian (=1) Years of Education Completed -0.217 0.007 0.805 No Antipsychotic Drug (=1) -1.718 0.071 0.179 Atypical Antipsychotic Drug (=1) -0.778 0.061 0.459 Supported Residence (=1) Psych. Hospitalization in Prior Yr. (=1) Psych. Hospitalization in Follow-Up (=1) 1.146 0.063 3.147 Baseline GAF (<40=1, 40-59=2, >59=3) Baseline MADRS >14 (=1) 0.023 0.262 1.023 Model Chi squared 25.840 0.011 Pseudo R-squared 0.1327 Percent Correct Predictions 74.800 Full Model 90% Conference Interval for Odds Ratio Predictor Variables ** 0.343 1.376 Gender (Male = 1) 0.020 0.827 Age (yrs.) at Baseline 1.038 1.637 Age Squared 0.413 2.380 African-American (=1) 0.315 3.845 Other Non-Caucasian (=1) 0.692 0.914 Years of Education Completed 0.041 1.091 No Antipsychotic Drug (=1) 0.224 0.970 Atypical Antipsychotic Drug (=1) 0.630 4.087 Supported Residence (=1) 0.571 3.102 Psych. Hospitalization in Prior Yr. (=1) 0.987 10.254 Psych. Hospitalization in Follow-Up (=1) 0.467 1.354 Baseline GAF (<40=1, 40-59=2, >59=3) 0.980 1.054 Baseline MADRS >14 (=1) Model Chi squared Pseudo R-squared Percent Correct Predictions Reduced Model * 0.022 0.924 Gender (Male - 1) 1.021 1.611 Age (yrs.) at Baseline Age Squared African-American (=1) Other Non-Caucasian (=1) 0.706 0.918 Years of Education Completed 0.038 0.859 No Antipsychotic Drug (=1) 0.232 0.908 Atypical Antipsychotic Drug (=1) Supported Residence (=1) Psych. Hospitalization in Prior Yr. (=1) 1.140 8.683 Psych. Hospitalization in Follow-Up (=1) Baseline GAF (<40=1, 40-59=2, >59=3) 0.989 1.058 Baseline MADRS >14 (=1) Model Chi squared Pseudo R-squared Percent Correct Predictions * Reduced model results were obtained via stepwise deletion of insignificant predictors (with replacement) from the full model until P-values for all remaining predictors were <0.3. Both full and reduced models include a separate intercept for each study region. ** The omitted (reference) category for the regressions is female Caucasian consumers in an independent (nonsupported) living situation, with baseline MADRS<15, who received only prescriptions for typical antipsychotics in the six months prior to baseline. Table 4: Ordered Multiple Logistic Regressions for Employment Status at Six Months Full Model Predictor Variables *** Coefficient P-value Gender (Male = 1) -0.363 0.326 Age (yrs.) at Baseline -1.634 0.107 Age Squared 0.208 0.092 African-American (=1) 0.441 0.346 Other Non-Caucasian (=1) 0.305 0.643 Years of Education Completed -0.204 0.006 No Antipsychotic Drug (=1) -1.329 0.124 Atypical Antipsychotic Drug (=1) -0.752 0.056 Assisted Residence (=1) -0.038 0.934 Psych. Hospitalization in Prior Yr. (=1) 0.07 0.881 Psych. Hospitalization in Follow-Up (=1) 0.762 0.247 Baseline GAF(<40=1 ,40-59=2,>59=3) -0.365 0.194 Baseline MADRS >14 (=1) 0.02 0.323 Model Chi-Squared 33.19 0.016 Pseudo R-Squared 0.1087 Reduced Model ** Predictor Variables *** Coefficient P-value Gender (Male = 1) -0.404 0.269 Age (yrs.) at Baseline -1.639 0.101 Age Squared 0.199 0.102 African-American (=1) Other Non-Caucasian (=1) Years of Education Completed 0.205 0.005 No Antipsychotic Drug (=1) -1.448 0.078 Atypical Antipsychotic Drug (=1) -0.811 0.030 Assisted Residence (=1) Psych. Hospitalization in Prior Yr. (=1) Psych. Hospitalization in Follow-Up (=1) 0.942 0.102 Baseline GAF(<40=1 ,40-59=2,>59=3) -0.45 0.070 Baseline MADRS >14 (=1) Model Chi-Squared 31.19 0.003 Pseudo R-Squared 0.1022 * Ordered outcomes are: jobless (high); employed in assisted job (middle); employed in non-assisted job (low). ** Reduced model results were obtained via stepwise deletion of insignificant predictors (with replacement) from the full model until P-values for all remaining predictors were <0.3. Both full and reduced models include a separate intercept for each study region. *** The omitted (reference) category for the regressions is female Caucasian consumers in an independent (nonsupported) living situation, with baseline MADRS<15, who received only prescriptions for typical antipsychotics for the six months prior to baseline.
We gratefully acknowledge financial support from grants MH01647 and MH43703 from the National Institute of Mental Health, and from Eli Lilly and Company through a subcontract between the Medstat Group and the Johns Hopkins University. Thanks are also due to Joe Gibson and the referees for helpful comments on earlier drafts, to James Collins for expert assistance in manuscript preparation, and to Ann Skinner and Maureen Fahey for providing unpublished tabulations from the Schizophrenia PORT data.
Blankertz, L., & Keller, C. (1997). The provision of long-term vocational supports for individuals with severe mental illness. Continuum, 4, 51-63.
Eaton, W. W., Bilker ,W., Hart, J. M., Herrman, H., Mortensen, P., Freeman, H., & Burgess, T. (1992). Long-term course of hospitalization for schizophrenia: Part II, change with passage of time. Schizophrenia Bulletin, 18, 229-24t.
Kouzis, A. C., & Eaton, W. W. (2000). Psychopathology and the initiation of disability payments. Psychiatric Services, 51, 908-913. Lehman, A. F., Steinwachs, D. M. and the Survey Co-Investigators of the PORT Project. (1998). Patterns of usual care for schizophrenia: Initial results from the Schizophrenia Patient Outcomes Research Team (PORT) Client Survey. Schizophrenia Bulletin, 24, 11-20.
Mark, T. L., Dirani. R., Russo, P., & Slade, E. P. (2001). Access to new medications to treat schizophrenia. The Journal of Behavioral Health Services & Research, 29, 15-29.
Peckham, J., and Muller, J. (1999). Employment and schizophrenia: recommendations to improve employability for individuals with schizophrenia. Psychiatric Rehabilitation Journal, 22, 399-402.
Rutman, I. (1992). How psychiatric disability expresses itself as a barrier to employment. In Matrix Research Institute (Ed.), Strategies to Secure and Maintain Employment for People with Long-Term Mental Illness (pp. 1-19). Philadelphia: Editor.
Salkever, D., Domino, M. E., Burns, B. J., Santos, A. B., Deci, P.A., Dias, J. et al. (1999). Assertive community treatment for people with severe mental illness: The effect on hospital use and costs. Health Services Research, 34, 577-601.
Salkever, D. Tickets without takers? Potential economic barriers to the supply of rehabilitation services to beneficiaries with mental disorders. In Rupp, K. and Bell, S.H. (eds.), Paying for Results in Vocational Rehabilitation (pp. 327-354). Washington, D.C.: The Urban Institute, 2002.
Scheid, T. L., and Anderson, C. (1995). Living with chronic mental illness: Understanding the role of work. Community Mental Health Journal, 31, 163-176.
Slade, E. and Salkever, D. (2001). Symptom effects on employment in a structural model of mental illness and treatment: analysis of patients with schizophrenia. The Journal of Mental Health Policy and Economics, 4, 25-34.
United States Department of Health and Human Services (USDHHS). (1999). Mental health: A report of the Surgeon General. Rockville, MD: USDHHS, National Institutes of Health, National Institute of Mental Health, Center for Mental Health Services, Substance Abuse and Mental Health Services Administration.
Westat and Mathematica Policy Research. (2001). Survey of Income and Program Participation Users' Guide (Third Edition) [Report under Contract No. 500-YABC-7-60016]. Washington, DC: United States Department of Commerce, United States Census Bureau, Economics and Statistics Administration.
David S. Salkever, Ph.D., Professor, Dept. of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health Hampton House Room 429, 624 N: Broadway, Baltimore, MD 21205. Email: Salkever@jhu.edu
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Employment Retention and Schizophrenia|
|Publication:||The Journal of Rehabilitation|
|Date:||Oct 1, 2003|
|Previous Article:||Psychiatric disabilities: challenges and training issues for rehabilitation professionals.|
|Next Article:||Job placement: the development of theory-based measures.|