Characteristics of Enrollees in the Supported Employment Demonstration

The Social Security Administration (SSA) funds two disability benefits programs. Social Security Disability Insurance (SSDI) is an insurance program for disabled workers who accumulated a sufficient number of work credits. Supplemental Security Income (SSI) is a means-tested program for disabled, low-income individuals who have never worked or have not accumulated enough work credits to qualify for SSDI. In December 2018, almost 12.5 million adults between the ages of 18 and 64 received SSDI, SSI, or both. Nearly 30% of those in the SSDI program and about 28% of those in the SSI program were awarded benefits based on a mental impairment (1, 2). Self-report data from the 2015 National Beneficiary Survey agreed with administrative records: large proportions of beneficiaries in both programs reported mental health impairments (3). Although 65% of initial disability applicants are denied (4), many denied applicants alleging mental impairment, particularly those of lower socioeconomic status, subsequently reapply and receive approval for benefits (5).

SSA has funded several demonstration programs with the goal of increasing employment among beneficiaries and enabling some beneficiaries to leave the disability rolls. To date, these demonstrations have often increased employment without resulting in participants’ leaving disability programs (6–9). Noting that these previous interventions relied heavily on populations already receiving disability benefits, SSA posited that earlier intervention in the disability process might achieve greater success. In 2016, SSA contracted with Westat to conduct the Supported Employment Demonstration (SED) and examine whether offering employment, health care, and other supports to disability applicants who had been initially denied could enhance employment and reduce future dependency on benefits.

The SED is a randomized controlled trial in which participants in the experimental group receive a multicomponent intervention aimed at increasing employment. The study provides supported employment services integrated with medical treatment and other supports, including help obtaining health insurance, reimbursement for out-of-pocket health care, and payments for other expenses, to enrollees over a 3-year period of individual participation. The primary question that SSA seeks to answer with the SED is whether offering individual placement and support (IPS) employment services (10) that are fully integrated with clinical and other services will foster employment and, consequently, lead to clinical improvement and reduced demand for disability benefits.

The goal of this analysis was to describe the characteristics of SED enrollees (N=2,960), including their demographic characteristics, work history, diagnoses, health status, medical conditions, health behaviors, and mental and physical functional impairments. Analyses included comparisons of functional assessments by employment status at enrollment.

Methods

We analyzed data from the SED baseline survey data and employed descriptive statistics to characterize the SED sample.

Design

Riley et al. (4) provided a summary of the overall study and evaluation design. In brief, the SED evaluation uses a randomized controlled trial design to compare the outcomes of two treatment groups against a control group. Both treatment groups receive employment services following the evidence-based IPS model, as well as behavioral health services and employment-related support. In addition, one treatment group also receives medication management services and liaison with medical providers from a nurse care coordinator. The control group receives a comprehensive resource manual providing resource information for people with mental illness, and individuals in this group may seek services on their own (4).

Setting

The interventions take place at 30 demonstration sites in 20 states across the United States. Twenty-three of the demonstration sites are community mental health centers, five sites are social service agencies that provide services to low-income populations, and two sites are employment agencies (linked with mental health clinics) primarily delivering services to people with disabilities.

The three primary factors in site selection included a state office recommendation that the local site had a good track record of offering IPS services, the necessary infrastructure to continue providing services and a sufficient pool of study-eligible individuals within the geographic area served by the site. Secondary factors included racial-ethnic and geographic diversity among populations served. Each site selected had an extensive history of serving people with mental illness and low-income individuals.

For 29 of the 30 demonstration sites, the Westat Institutional Review Board (IRB) served as the IRB of record and approved the study on June 12, 2017. For the remaining site, the Los Angeles County Department of Mental Health Human Subjects Research Committee approved the study on October 31, 2017. All participants gave written informed consent, and the study followed principles of the Declaration of Helsinki.

Participants

The Westat team used SSA administrative data to identify potential study participants in geographical catchment areas served by sites. Applicants denied disability benefits were eligible for the study if they were not currently receiving disability benefits, had a self-reported or documented mental health impairment, indicated that they wanted to work (or wanted a better job), were not living in a nursing home or other custodial institution, and were able to provide informed consent. Recruitment ended after 15 months with 2,960 eligible enrollees, after 40 enrollees were determined ineligible postenrollment because they were already receiving benefits.

Data Collection

Westat recruiters collected baseline survey data during in-person interviews with all SED enrollees between December 2017 and March 2019. To reduce survey burden on enrollees, Westat staff administered the World Health Organization Composite International Diagnostic Interview (CIDI) in person or by telephone to a subset of enrollees (N=1,842) in a separate interview shortly after the baseline interview, which was administered to all enrollees (N=2,960).

Measures

The baseline data contained several validated measures to assess enrollees’ mental and physical health and the influence of health conditions on their functioning. CIDI versions 19 and 21.1.3 (11) assessed the presence of selected mental disorders, including anxiety disorders, personality disorders, mood disorders, indicators (not diagnoses) of psychosis, and eating disorders. The CIDI is a structured, modular interview used by lay interviewers to assess mental health conditions according to definitions and criteria of the ICD-10 and DSM-IV (11). Most of the 1,842 enrollees assessed with the CIDI were administered the CIDI over the telephone; however, some received in-person administration. In this study, we took a conservative approach to determining the likely presence of personality disorder by including positive scores for probable personality disorder and excluding positive scores for possible personality disorder.

The 14-item Colorado Symptom Index (CSI) assessed psychiatric symptoms on the basis of self-report. The CSI measures the frequency of selected psychiatric symptoms by using a 5-point Likert scale—1, not at all; 2, once during the month; 3, several times during the month; 4, several times a week; and 5, at least every day. Aggregate scores range from 14 to 70, with higher scores indicating more frequent symptoms (12). Previous research associated a score of 30 with clinically relevant symptoms in a Medicaid population (12). Interviewers administered the CSI in person to all 2,960 enrollees.

Mental composite summary (MCS) scale scores and physical composite summary (PCS) scale scores from the 12-item Short-Form Health Survey (SF-12) assessed mental health– and physical health–related quality of life (13, 14). Higher scale scores indicate better health-related quality of life. For the U.S. adult population, 50 is the normed mean score for both the MCS and the PCS; one standard deviation for each is 10 points. Interviewers administered the SF-12 in person.

Scale scores from the Work Disability Functional Assessment Battery (WD-FAB) assessed work-related functioning (15–19). The WD-FAB, which incorporates the International Classification of Functioning, Disability, and Health conceptual framework, assessed work-related functional limitations in eight subdomains: communication and cognition, resilience and sociability, self-regulation, mood and emotions, basic mobility, upper-body function, upper-extremity fine-motor function, and community mobility (including both driving and riding) (20, 21). Responses to survey items with 5-point scales determine WD-FAB scores. Higher subdomain scores indicate better work-related functioning in the area assessed. The WD-FAB uses computerized adaptive testing methodology, where an item is initially presented from the midrange of a defined list of items and subsequent items are selected at an appropriate level on the basis of the respondent’s previous answers. Responses are standardized on a national normative sample. The normed mean scale score for the working-age population in each subdomain is 50; one standard deviation is 10 points.

Analysis

We used descriptive statistics, including proportions, means, and medians, to describe the SED sample. To compare subsamples, we used two-tailed t tests. We set statistical significance at p<0.05 and used Bonferroni correction for multiple tests within single instruments.

Results

As shown in Table 1, most SED enrollees self-identified as female (57%), White (56%), and non-Hispanic (87%). Over half (58%) were 35 years or older when they enrolled in the study. Half of enrollees (51%) reported some education beyond high school, although few were college graduates (17%, including those with associates degrees). More than two-thirds (69%) of enrollees lived with relatives, and more than half (55%) had never married. At the time of the baseline interview, slightly less than two-thirds (64%) had been employed in the past 2 years, and nearly one-fifth (19%) were currently employed.

SED enrollees typically lived in low-income households. Nearly three-quarters (72%) reported total household income of $2,000 or less in the 30 days prior to the baseline interview. Median total household income for the 30 days prior to the baseline interview was $1,200; mean±SD total household income for the same period was $1,837±$2,753. Among those employed in the past 2 years and who reported wages and hours worked, the mean of estimated hourly wages for the most recent job held was $12.84±$8.30 (N=1,795). The mean number of hours worked per week for the most recent job among those employed in the past 2 years was 33.4±15.5 (N=1,777). Among the subset employed at the baseline interview and who reported wages and hours worked, estimated average hourly wages for the current job were $12.99±$7.60 (N=532), and the average number of hours worked per week was 28.5±14.5 (N=525).

Examination of supports revealed that most enrollees (80%) had health care coverage (Table 2). More than half (55%) indicated that they (or someone in their household) were receiving Supplemental Nutrition Assistance Program benefits at the time of enrollment, and over two-thirds (69%) indicated that they (or someone in their household) had received such benefits in the past 12 months. About 7% of enrollees indicated that they (or someone in their household) were receiving Temporary Assistance for Needy Families at the time of enrollment. Over half of enrollees (54%) indicated that they had been arrested at some time in their life. About 14% of enrollees indicated that they had been arrested (and booked) at least once by law enforcement in the past 12 months.

Table 3 presents a summary of results from scored CIDI responses (N=1,842). Most enrollees scored positive for anxiety disorders (71%), personality disorders (65%), and mood disorders (61%). More than a third (38%) reported at least one symptom of psychosis. Nearly all (91%, N=1,677) of those who completed the CIDI scored positive on at least one CIDI diagnostic module administered. Nine percent of enrollees (N=165) did not score positive on any of the CIDI diagnostic modules administered. There are several reasons why enrollees may not have scored positive for any mental health condition assessed, including respondent denial of symptoms, selected diagnostic modules fielded, and the conservative approach to assessing the likely presence of personality disorder. On average, enrollees who took the CIDI scored positive for 2.5±1.3 mental health conditions.

Table 3 also includes mean CSI scores for enrollees who scored positive for the condition or symptom assessed. In general, mean scores were at least 40, although the range was from 34.9 for the few cases who scored positive for agoraphobia to 45.5 for those who indicated that they felt that others were plotting to harm them.

In addition to mental illnesses, many enrollees reported multiple severe general medical conditions. The baseline survey probed 18 specific conditions, including obesity (Table 4). Enrollees reported an average of 3.5±2.1 of these conditions (median 3.0). Back pain was the most frequently reported condition (63%), and nearly half (47%) reported heights and weights indicating obesity. Participants also frequently reported lung conditions (33%). Six percent reported cancer. The great majority (91%, N=2,701) reported at least one of the 18 comorbid conditions. The list did not include “other” unspecified conditions that enrollees reported.

Among those who completed the CIDI (N=1,842), nearly all (99%, N=1,824) scored positive for at least one of the mental health conditions assessed, or they reported at least one of the 18 general medical conditions probed. The great majority (84%, N=1,556) scored positive for at least one mental health condition assessed and at least one of the 18 general medical conditions, not including unspecified “other” conditions that enrollees reported and reflect our conservative approach to assessing the likely presence of a personality disorder.

Mean SF-12 MCS scores (32.6) and PCS scores (38.3) indicated that enrollees’ health-related quality of life was low, compared with the general U.S. population (Table 5). Work-related functioning, as assessed by the WD-FAB, was lower than population norms in the areas assessing control of mood and emotions, upper-body functioning, and ability to drive a car.

Assessments of work-related functionality differed significantly by baseline work status (employed at baseline versus unemployed at baseline but worked in the past 2 years) (Table 5). In the behavioral health domain, magnitudes of mean differences in scores were less than 2 points in each subdomain. Two-tailed t tests with Bonferroni correction indicated that the mean difference in the mood and emotions subdomain was statistically significant. In the physical health domain, absolute mean differences were also small. Two-tailed t tests with Bonferroni correction indicated that the mean differences in subdomains of basic mobility and upper-body function were statistically significant. Significantly different SF-12 MCS and PCS scores for the two groups echoed differences in functionality.

Discussion

Most SED enrollees self-identified as female, White, and non-Hispanic. Many were 35 years or older, reported more than a high school education, lived with relatives, had never married, were unemployed, and were poor. Anxiety disorders, personality disorders, and mood disorders were prevalent, and mean CSI scores indicated that self-reported symptoms were clinically relevant. Enrollees reported multiple mental health and general medical conditions, and health-related quality of life, as assessed by the SF-12, was low, compared with national norms. Scores on the WD-FAB subscales were more than a standard deviation below norms in the areas assessing control of mood and emotions, upper-body functioning, and the ability to drive a car.

Compared with the U.S. civilian labor force ages 16 years and older, SED enrollees are somewhat more likely to be female (57% versus 51%), more likely to be African American (31% versus 14%), somewhat less likely to identify as White only (56% versus 60%), and less likely to be Hispanic (13% versus 19%) (22). They were much more likely to report never having married, compared with adults in the general population (55% versus 29%) (23). Enrollees are also much more likely to be poor: in 2018 dollars, the typical household of those in the civilian labor force reported more than four times the median monthly income as that reported by SED enrollees ($5,024 versus $1,200) (22).

In addition to self-reported or documented mental health conditions, many in the sample reported serious physical conditions. Compared with U.S. adults ages 20 and older, SED enrollees were much more likely to report back pain, obesity, lung conditions, and cancer (24). As evidenced by SF-12 and WD-FAB scale scores, the contribution of comorbid general medical conditions to the overall picture of diminished health is nontrivial, although a validated instrument that provides a single measure of overall diminishment from multiple impairments, as opposed to a single severe impairment, has not been established.

The SED will determine whether enrollees are good candidates for employment. We will also explore the relationship between WD-FAB scores, including subdomain scores, and employment, as well as associations between changes in subdomain scores and employment outcomes.

Findings from the CIDI distinguish the SED sample from those previously studied in IPS supported employment research. CIDI scores suggest that this sample of denied applicants is not uniformly a population of individuals with severe mental illness (25). Most enrollees reported symptoms of anxiety, a personality disorder, or major depression or dysthymia, rather than symptoms of mental illnesses typically described as severe mental illness, for example, schizophrenia, bipolar I disorder, and recurrent major depression. With the exception of posttraumatic stress disorder (26), research on IPS supported employment has typically involved a majority of individuals with schizophrenia spectrum disorders (27–31). The most striking clinical difference between SED enrollees and the typical mental health center population of individuals with severe mental illness is the high rate of probable antisocial personality disorder. People with antisocial personality disorder have difficulty fulfilling adult role responsibilities, such as sustaining work behavior. Consequently, this subgroup of enrollees may well prove to be one of the most challenging subgroups of individuals served in this study.

We acknowledge several limitations. SED enrollees may differ from the entire population of individuals with an observed or alleged mental health impairment who have been denied disability benefits, particularly because eligible enrollees expressed an interest in employment. In addition, study sites are located in a selected number of states, and few sites are in rural areas.

Conclusions

By examining the characteristics of individuals who enrolled in the SED, this study sought to fill the gap in knowledge regarding characteristics of the population of applicants who are denied disability benefits and who have a self-reported or observed mental health impairment and an expressed interest in employment. Baseline data for SED enrollees indicate that this is a low-income population with multiple mental health and general medical conditions that substantially diminish both health-related quality of life and aspects of work-related functioning. SED enrollees do not uniformly exhibit severe mental illness but rather have a high prevalence of anxiety and personality disorders, combined with serious general medical disorders. Nonetheless, study eligibility criteria included an expressed desire to work, and the intervention arms of the SED offer IPS services intended to facilitate competitive employment.