Predicting rehabilitation outcome for supplemental security income and social security disability income recipients: implications for consideration with the ticket to work program.

The United States government provides income replacement assistance for individuals with disabilities through two programs: Social Security Disability Income (SSDI) and Supplemental Security Income (SSI). To be considered disabled under either of these programs, the individual must have a severe disability that will result in death or will result in an inability to perform substantial gainful activity for at least 12 months.

Although an identical definition of disability is used for the two programs, differences do exist in qualifying for each. To be qualified for SSDI one must have worked and made contributions to Social Security. Conversely, only those individuals who have not made enough work contributions to qualify for SSDI and who meet income and resource limits are eligible for SSI (Wheeler, Kearney & Harrison, 2001/2002).

The number of individuals receiving SSDI and SSI dramatically increased during the 1990s. As a result, the Social Security Administration (SSA) is attempting to understand this trend and to find methods to help individuals with disabilities remain in the labor force or return to work (Wheeler, Kearney, & Harrison, 2001/2002). One method of assistance is through the provision of vocational rehabilitation services. Vocational rehabilitation services have been found to have a positive effect on return to work rates for people receiving disability benefits (Hennessey & Muller, 1995). These services, however, appear to be underutilized.

Shrey and Bangs (1991) found that 44% of beneficiaries participating in a Social Security return-to-work projects reported no contact with state rehabilitation agencies or other vocational rehabilitation providers. A 1987 study by the United States General Accounting Office (GAO) (as cited in Shrey & Bans, 1991) revealed 12% of beneficiaries were evaluated for rehabilitation services and only .3% of the beneficiaries returned to work alter receiving rehabilitation services. Similarly, Hennessey & Muller (1994) found that few vocational rehabilitation (VR) services were being provided to beneficiaries of Social Security disability programs, and that those beneficiaries, who did receive services, portrayed VR as not influencing their decision to work.

Vocational rehabilitation services can be provided by public vocational rehabilitation agencies, or, under the new Ticket to Work program implemented by the Social Security Administration, by private rehabilitation providers. The success of the Ticket to Work program will depend significantly upon establishing methods of identifying those individuals who will benefit from vocational rehabilitation services and identifying the most effective types or combinations of rehabilitation services for these individuals.

Numerous studies have been conducted to identify individuals who would benefit from rehabilitation services by examining the factors that may predict employment of these individuals. The majority of studies appear to focus upon demographic characteristics of the individuals. Berry (2000) found that white males were more successful in employment outcomes than nonwhite and female subjects. Other significant factors impacting employment outcomes included education, family income, activity limitations, hospitalization, and health status. Swett (2000) found educational level, age, type of disability, and type of Social Security benefit were significant in predicting employment. Kearney (1997) conducted a survey to determine the medical and non-medical factors that were most successful in predicting return to work for individuals who were disabled by a back condition. He observed that individuals with higher levels of education, who were seeking less physically demanding and more skilled occupations, were more likely to return to work. The type of medical provider (family doctor, company doctor, or specialist) and the existence of other chronic diseases had minimal to no impact on return to work.

The impact of the type of vocational rehabilitation service provided to SSI and SSDI beneficiaries has not been thoroughly examined. Hennessey and Muller (1995) examined the effectiveness of vocational rehabilitation services, i.e., physical therapy, vocational training, job counseling, general education, and job placement, upon actual work outcomes by using data from the New Beneficiary Follow-up survey conducted by the SSA. The effect of job counseling was not significant. Job placement services were found to have a dramatic effect on the tendency to work. Beneficiaries who were provided with this service were over four times more likely to return to work than beneficiaries who did not receive the service. However, job placement was only offered to approximately two percent of the beneficiaries. Vocational training (offered to 5% of beneficiaries) doubled the rate of return to work. Physical therapy (20% of beneficiaries) and general education (approximately 2.5% of beneficiaries) also had positive significant effects on return to work.

Other studies, not specifically restricted to beneficiaries of SSI or SSDI, have also found that individuals who receive job placement services are significantly more likely to obtain competitive employment. Moore, Feist-Price and Alston (2002) found that persons with severe/profound mental retardation were significantly more likely to obtain employment if they were provided job placement services. Hayward and Schmidt-Davis (2003) conducted a longitudinal study for the Rehabilitation Services Administration to "assess the performance of the state-federal VR program in assisting eligible individuals with disabilities to achieve positive, sustained economic and non-economic outcomes as a result of their receipt of VR services" (p. 1). This study was implemented in 1992 and tracked a representative sample of VR applicants and consumers for tip to three years after exit from the VR program. One of the factors investigated was the extent to which specific VR services predicted or contributed to successful employment outcomes. They found that several services increased the likelihood of competitive employment. These services included job placement, job development, and on-the-job training. Postsecondary education and provision of tools, uniforms, and equipment contributed to a lesser degree to competitive employment. Individuals who received job placement services achieved competitive employment more than twice as often as those who did not receive such services. Although job placement services contributed to successful employment outcomes, only 11% of all individuals received this service.

Attempts have been made to determine if there is a particular technique for providing rehabilitation services that is more effective in returning individuals to work. Shrey and Bangs (1991) conducted a demonstration project that used a multidisciplinary disability. management model to assist in returning beneficiaries to work. Services provided included multidisciplinary rehabilitation evaluations, vocational evaluations, vocational counseling, job-seeking skills training and informational programs on Social Security work incentives. In addition, job development and placement networking activities with employers and information on obtaining assistive devices and other resources were provided. The project had an estimated 25% rate of success. This represents a substantial increase in return to work from the GAO study conducted in 1987 and may indicate a vocational rehabilitation service approach that should be considered in providing return to work assistance to beneficiaries.

Additional research is needed to determine the effectiveness of particular vocational rehabilitation services in predicting employment outcomes for SSI and SSDI recipients. The purpose of the current study was to explore the relationship between provision of various services and rehabilitation outcome. The services include: assessment, mental or physical restoration, college or university training, business or vocational training, adjustment training, on-the-job training, miscellaneous training, counseling or guidance that is substantial and directly related to the rehabilitation plan, job finding services, job placement, transportation, maintenance, and other services not classified above. The category of "other services" includes occupational tools and equipment, initial stocks and licenses, services to family members for the benefit of the individual, and medical care for acute conditions arising during rehabilitation and constituting a hazard to the determination of rehabilitation potential or to the achievement of the vocational objective.

Such understanding will help to maximize rehabilitation assistance to individuals by minimizing unnecessary or non-productive service expenditures. Identification of the most effective services may help to direct rehabilitation vendors (under the Ticket to work program) in planning and providing productive services. Specifically, this study examined which services resulted in more successful closures, i.e., return to work.

Method

The data for this study came from the national RSA database for federal fiscal year 2001 (Rehabilitation Services Administration, 2001; data covered October 1, 2000 through September 30, 2001). The national database is compiled from data submitted by the state VR agencies each year and includes only cases that were closed during the federal fiscal year. In 2001 this number totaled 639,823. For the purpose of the present analysis, we were interested only in cases that were (a) closed into either status 26 ("closed rehabilitated") or status 28 ("closed, not rehabilitated, after individualized written rehabilitation program initiated"); (b) recipients at intake of either Social Security Disability Insurance (SSDI) or Supplemental Security Income (SSI). Alter cases not meeting these criteria were removed, we took a random sample of 5% of the remaining cases. This final sample consisted of 6,264 cases.

The selection of cases was guided by the research question, which concerned the effectiveness of services and the relationship of different services to closure status among persons receiving either SSI or SSDI. According to Wheaton and Hertzfeld (2002) the examination of cases in which a rehabilitation plan has been developed, whether the outcome is successful or not, best represents the effectiveness of the services provided because the development and implementation of this plan indicates that (a) a thorough study has determined an appropriate vocational goal, (b) a strategy to achieve the goal has been developed, and (c) the client has agreed to the plan. The remaining (excluded) cases represent persons closed at various stages prior to rehabilitation plan development.

Statistical Analysis

The prediction of closure status based on the type of services received was computed using binary logistic regression analysis. Logistic regression analysis is appropriate when the dependent variable, in this case closure status, is dichotomous (Wright, 1995). The predictor variables in a logistic regression analysis can be continuous or noncontinuous. Like linear regression, the logistic regression model relates one or more predictor variables to a dependent variable, and the logistic model yields regression coefficients, predicted values, and residuals. In this type of analysis, one attempts to predict the probability that an observation belongs to each of two groups. In other words, if a dependent variable is coded as 0 (e.g., "closed, not rehabilitated") and 1 (e.g., "closed, rehabilitated"), the logistic regression analysis predicts a probability that an observation belongs to the group designated as 0 and a separate probability value that the observation belongs to the group designated as 1. The observation is assigned to the group having the higher predicted probability.

Due to the exploratory nature of the analysis, the forward stepwise method of entry was used, with the Wald statistic as the criterion for significance. Stepwise regression is frequently used in the exploratory phase of research to identify the best predictors from a larger set of predictor variables. Because of the relatively large sample size, the significance level for entry into the regression model was set at .01.

The Wald test statistic was used to determine if the probability of a successful closure (Status 26) was significant. The D (predictor coefficient) represents the change in the natural logarithm of the odds ratio (Exp (B)). A positive predictor coefficient means that the predicted odds increase as the predictor variable associated with the predictor coefficient increases. A negative predictor coefficient means that the predicted odds decrease as the predictor variable increases (Wright, 1995). The odds ratio (Exp (B)) estimates the change in the odds of membership in the target group (i.e., "closed, rehabilitated") based on a one-unit increase in the predictor variable (Wright, 1995).

Variables

Criterion Variable

The criterion variable in this study was VR service outcome, defined as VR closure status, including either status 26 or status 28. Cases closed in Status 26 are classified as "closed, rehabilitated", with "rehabilitated" being defined by the RSA as meaning that the person received at least one service and had been employed at closure for a period of at least 90 consecutive days. Cases closed in Status 28 represent those cases in which an individualized written rehabilitation program had been initiated, but the person did not find employment. For the purpose of this study case outcome was a nominal, dichotomous variable.

Predictor Variables

Service variables. The variable of primary importance in this analysis was type of services received. That is, we were interested primarily in understanding whether the type of services received predicted VR closure status. There are 13 categories of services identified by the RSA and coded in the national database. In the present analysis the receipt of services was coded as either 0, did not receive this service, or 1, did receive this service. The services include: assessment, mental or physical restoration, college or university training, business or vocational training, adjustment training, on-the-job training, miscellaneous training, counseling or guidance that is substantial and directly related to the rehabilitation plan, job finding services, job placement, transportation, maintenance, and other services not classified above.

Number of services. In addition to examining the contribution of each service individually to the regression model using these dichotomous variables, we summed the number of services received by each client, creating a separate, continuous variable. The median number of services received was 4 (Mean = 3.84, SD = 2.63).

Cost of services. Cost of case services is defined as the total amount of money (to the nearest dollar) spent by the state rehabilitation agency in providing or arranging for services for the client. This includes expenditures over the "life of the case" and does not refer to a single fiscal year (RSA, 1995). The mean cost of services was $4,246.62 (SD 8147.63) with a range of costs from $0 to $99,999.

Demographic Variables. In addition to the number and type of services received, we explored a number of demographic variables for their possible contribution to VR closure status prediction. These included gender, age, race, and education.

Gender. Gender was coded as a dichotomous variable. The final sample was 53.4% male, and 46.6% female.

Race. The predictor variable race was a polytomous variable with four levels, as defined by the RSA in the national database (White, Black, American Indian or Alaskan Native, and Asian or Pacific Islander). Race was operationalized as the race reported by clients on their application for VR services (RSA, 1995). This categorical variable was coded using the "simple" contrast method (Norusis, 1999) such that the group identified as Whites served as the reference category for this variable and was contrasted with each of the other three race groups. The sample was composed as follows: Whites comprised 74.1%, Blacks 23.1%, American Indians or Alaskan Natives 1%, and Asian or Pacific Islanders 1.7%. Note that the RSA reporting manual makes a distinction between race and Hispanic origin and does not identify a separate racial category for the Hispanic population. For the purpose of the analysis, therefore, the race variable only included the four categories identified above.

Education. The predictor variable education was a continuous variable. This variable was defined and coded as the highest grade completed at the time of application for VR services. The mean highest grade completed was 11.9 (SD = 2.37). The minimum was zero and the maximum was the 21st grade.

Age. Age at application was computed based on the date of birth and date of application. This was a continuous variable. The mean age of the sample was 40.9 years (SD = 13.13). The minimum age was 19 and the maximum was 95 years.

Severity of Disability. According to the RSA definition, a person has a severe disability if he or she has a mental or physical disability, or both, and this disability (a) severely limits the individual in functional capacities, (b) is expected to require multiple rehabilitation services, and (c) causes multiple substantial functional limitations. This variable is coded in the national database as a dichotomous, nominal variable, and was entered into the regression as such.

Results

Nine variables significantly contributed to the explanatory power of the model (2 (8, N=6264) = 24.391, p = .002). Together. these variables accounted for approximately 14% of the variance (Nagelkerke [R.sup.2] = .135). The results of the final stage of the analysis are summarized in Table 1. This model correctly predicted group membership 64.1% of the time. Specifically, the model accurately classified 58.8% of those who were closed "rehabilitated" (Status 26) and 70.1% of those who were closed "not rehabilitated" (Status 28).

Six VR services, including restoration, college or university training, on-the-job training, job placement, miscellaneous services, and other services were found to contribute to the explanatory value of the model. The percentages of consumers receiving each service are displayed in Table 3. The receipt of job placement services was found to be significantly related to closure status (Wald = 300.090, p = .000). The Exp (B) for job placement services (3.187) is interpreted as meaning that compared to people who did not receive job placement services, those who did were over three times more likely to be closed in status 26. People who received on-the-job training were over one and a half times more likely to be closed in Status 26 than those who did not (Wald 14.901, p = .000, Exp (B) = 1.628).

The remaining services that contributed to the explanatory power of the model, in order of odds ratio (Exp (B)) were (1) other services (Wald = 25.476. p = .000, Exp (B) = 1.403), (2) restoration services (Wald = 12.765, p = .000, Exp (B) 1.293), and (3) miscellaneous services (Wald = 10.990, p = .000, Exp (B) = 1.289). While the model suggests that all these services contribute to varying degrees to the likelihood of a successful outcome, college or university training was negatively associated with successful closure (Wald 26.766. p = .000, Exp (B) = .629). The contribution of the variable "cost of services" was negligible.

In terms of the demographic variables, race and education were the only variables that were found to contribute to the model. Race was significant as a variable, however only White race was significant individually (Wald = 3.837, p = .000, Exp (B) = 1.526). White clients were approximately 1.5 times more likely to be closed in Status 26 than non-whites. The model also suggests that education, defined in terms of highest grade completed at application, tends to increase the likelihood of successful case closure (Wald = 7.720, p = .000, Exp (B) = 1.035).

Discussion

The purpose of this study was to determine the effectiveness of particular rehabilitation services in predicting employment outcomes for SSI and SSDI recipients. Results indicate that six services contributed to the explanatory value of the model. Three demographic variables also contributed to the explanatory value of the model.

White race individuals were approximately 1.5 times more likely to be closed rehabilitated. Education, i.e., highest grade completed at application, also contributed to prediction of successful closure. These demographic results are consistent with those found in many previous studies and are not the main focus of this study.

The primary focus of this study was on effectiveness of rehabilitation service variables. The six variables that contributed to the explanatory value of the model included restoration, college or university training, on-the-job training, job placement, miscellaneous services, and other services. A discussion of miscellaneous and other services is difficult as these services are ill-defined and multi-dimensional. Restoration services and training services (college, university, or on-the-job) are often costly. While rehabilitation counselors in the public sector may be able to assist with the costs of restoration and training services, private sector counselors generally do not have budgets to absorb these costs. Therefore, private sector counselors must coordinate restoration and training services through other funding sources, e.g. public rehabilitation agencies. The primary service that private sector rehabilitation counselors, as well as public rehabilitation counselors, can provide which would substantially increase the likelihood of a successful employment outcome for SSI and SSDI beneficiaries based upon this study is job placement.

Results indicate that those who received job placement services are over three times more likely to secure employment than those individuals who did not receive placement services. These results are consistent with other studies (Hennessey & Muller, 1995, Hayward & Schmidt-Davis, 2003) and support the need to emphasize such services in rehabilitation delivery.

We recognize and emphasize that the combination and types of services provided to VR consumers must be, first and foremost, based upon the individual consumer's needs. With this in mind, it must also be recognized that vocational rehabilitation has traditionally had as its primary focus a goal-directed objective of competitive employment for individuals with disabilities. Therefore, it would appear that the provision of job placement services would be appropriate in the majority of cases. The findings of this and other studies (Hayward & Schmidt-Davis. 2003; Hennessey & Mullaer, 1995), however, reveal that the vast majority of consumers do not receive job placement services.

Failure to provide job placement services could be a result of several factors. First, the consumer may not have a vocational goal of competitive employment, e.g. homemaker or unpaid family worker. Second, large counselor caseloads may preclude or impair the ability to provide job placement services that can be quite time consuming. Third, counselors may lack appropriate training in the provision of job placement services. Fourth, if counselors contract for job placement services, budgetary constraints may prevent referrals to contractors.

Job placement services may be provided to consumers by the counselor, an in-house job-placement counselor, another public agency, or a private contractor. In the case of SSI or SSDI recipients, job placement services can also be obtained through private rehabilitation agencies that are registered with the Ticket to Work program. Based upon the current study's findings, rehabilitation agencies that are participating in the Ticket to Work program may wish to consider a proactive approach to job placement to maximize successful employment outcomes.

Limitations of the Study

Nine variables, including six services accounted for approximately 14% of the variance. Fourteen percent is a relatively low figure and consideration must be given to other confounding variables not addressed. The variables considered were restricted to those variables outlined in the RSA data and did not include possible contributing factors of VR office characteristics, counselor-client relationships, urban versus rural consumer locales, community resources, or psychosocial characteristics of consumers. These factors could impact the likelihood of successful employment and are beyond the scope of the current study.

Additionally, the current study was limited to individuals who receive SSI or SSDI benefits and may not be representative of rehabilitation clients in general. In other words, individuals who receive disability benefits may be more or less inclined to actively pursue competitive employment than individuals who do not receive disability income benefits.

Implications for Future Research

Predicting successful outcome of VR services is, at best, an inexact science. A large number of confounding variables can contribute to the ultimate equation. The present study increases current understanding of the relationship between provision of various services and rehabilitation outcome. However, additional research is needed. Important areas for future study include an analysis of the effectiveness of different services based upon such variables as type of disability, vocational goal, and geographical location of consumer.

References

Berry, H. G. (2000). The supplemental security income program and employment for young adults with disabilities: An analysis of the national health interview survey on disability. Focus on Autism and Other Developmental Disabilities, 15 (3), 176-181.

Hayward, B. J., & Schmidt-Davis, H. (2003) Longitudinal study of the vocational rehabilitation services program. (Final report). Research Triangle Park, NC: Research Triangle Institute International.

Hennessey, J. C. & Muller, L. S. (1994). Work efforts of disabled-worker beneficiaries: Preliminary findings from the new beneficiary followup survey. Social Security Bulletin, 57 (3), 42-51.

Hennessey, J. C., & Muller, L. S. (1995). The effect of vocational rehabilitation and work incentives on helping the disabled-worker. Social Security Bulletin, 58 (1), 15.

Kearney, J. R. (1997). The work incapacity and reintegration study: Results of the initial survey conducted in the United States. Social Security Bulletin, 60 (3), 21.

Moore, C. L., Feist-Price, S. & Alston, R.J. (2002). VR services for persons with severe/profound mental retardation: Does race matter? Rehabilitation Counseling Bulletin, 45 (3), 162-167.

Norusis, M. J. (1999). SPSS Regression Models 10.0. Chicago: SPSS Inc.

Shrey, D. E., & Bangs, S.A. (1991). Returning social security beneficiaries to the work force: A proactive disability management model. Rehabilitation Counseling Bulletin, 34 (3), 257.

Swett, E. A. (2000). Correlates of workforce participation for people receiving Social Security benefits: An investigation of selected demographic variables. Dissertation Abstracts International Section: The Sciences and Engineering, 61 (5-B).

Wheeler, P. M., Kearney, J. R., & Harrison, C. A. (2001/2002). The U. S. study of work incapacity and reintegration. Social Security Bulletin, 64 (1), 32.

Wright, R. E. (1995). Logistic Regression. In L. G. Grimm & Yarnold (Eds.). Reading and understanding multivariate statistics (pp. 65-97). Washington, D.C.: American Psychological Association.

Jackie B. Rogers

University of Kentucky

Malachy Bishop

University of Kentucky

Ralph M. Crystal

University of Kentucky

Jackie B. Rogers, Department of Special Education and Rehabilitation Counseling, 224 Taylor Education Building, Lexington Kentucky 40506-0001. Email:[e-mail]

Table 1

Final Model for the Logistic Regression Analysis

Predictor              B      SE     Wald              df

Race                                          15.411
        White         .423   .216     3.837             1
Education             .035   .012     7.720             1
Cost of Service       .000   .000    26.959             1
Restoration           .257   .072    12.765             1
College/University   -.463   .090    26.766             1
OJT                   .488   .126    14.901             1
Misc                  .254   .077    10.990             1
Placement            1.159   .067   300.090             1
Other                 .338   .067    25.476             1

Predictor            ExpB     p

Race                   3
        White        1.526   .000
Education            1.035   .000
Cost of Service      1.000   .000
Restoration          1.293   .000
College/University   0.629   .000
OJT                  1.628   .000
Misc                 1.289   .000
Placement            3.187   .000
Other                1.403   .000

Table 2

Frequency and Percentage of Consumers Receiving Each Service

Service Received               Frequency   Percentage
                                           Receiving
                                            Service

Assessment
       Yes                       4820         76.9
       No                        1444         23.1

Restoration
       Yes                       1250         20
       No                        5014         80

College/University Training
Yes                               744         11.9
No                               5520         88.1

Business/Vocational training
Yes                               613          9.8
No                               5651         90.2

Adjustment training
Yes                              1430         22.8
No                               4834         77.2

On-the-job training
Yes                               496          7.9
No                               5768         92.1

Miscellaneous training
Yes                              1265         20.2
No                               4999         79.8

Substantial Counseling
Yes                              4453         71.1
       No                        1811         28.9

Job-finding
       Yes                       2459         39.3
       No                        3805         60.7

Job placement
Yes                              2059         32.9
No                               4205         67.1

Transportation
       Yes                       1890         30.2
       No                        4374         69.8

Maintenance
       Yes                        837         13.4
       No                        5427         86.6

Other services
       Yes                       1715         27.4
       No                        4549         72.6