Predictors of community-based employment for young adults with intellectual disability and co-occurring mental health conditions

Abstract

Purpose:

To identify predictors of community-based employment and employment quality for young adults ages 23–30 with intellectual disability and co-occurring mental health conditions (YA-ID-MH).

Methods:

We conducted secondary analysis of the 2017–2018 National Core Indicators^® (NCI^®) In-Person Survey. The NCI^® survey was conducted in 35 states and Washington DC. Participants: YA with ID, ages 23–30 who had complete data. We conducted multiple regression analyses to examine demographic and environmental predictors of community-based employment, in addition to employment quality indicators: hourly wages, hours worked, and job duration. We also descriptively examined job satisfaction.

Results:

YA-ID-MH were somewhat less likely to be employed per record review and self-report than YA with ID only, but these findings did not reach statistical significance. On average, YA with ID only had higher hourly wages and worked more hours than those with ID-MH, but there were no significant differences in job duration. For YA-ID-MH, predictors of employment included gender, race, level of ID, and residential setting. Multiple demographic and environmental factors predicted employment quality.

Conclusions:

YA-ID-MH experience employment disparities compared to YA with ID only. Service providers should specifically attend to those at the highest risk of unemployment/low quality employment.

Introduction

Employment is a valued component of community and social participation for all people. Individuals with intellectual disability (ID) are significantly underemployed, with recent estimates of community-based employment in the United States at 20% for the population of people with intellectual disability and/or a developmental disability¹ [3,4]. Employment is associated with increased life satisfaction and social and community integration, demonstrating the importance of meaningful work for overall health and quality of life [5,6]. While barriers to employment and employment disparities for the general population of individuals with intellectual disability are well documented [e.g., 4,7–9], little is known about predictors of employment for the estimated 23–33% [10–12] of adults with intellectual disability who have co-occurring mental health conditions (ID-MH), such as anxiety, depression, and PTSD. Given the additional functional challenges of individuals with ID-MH and the low rates of employment for individuals with MH conditions without intellectual disability [13–16], it is possible that individuals with ID-MH are at even greater risk for disparities in community-based employment than individuals with intellectual disability, without mental health conditions.

While we are not aware of previous studies with a primary aim of exploring employment for individuals with ID-MH, data from the National Core Indicators In-Person Survey suggest a need for more focused inquiry on employment for this population. The National Core Indicators (https://www.nationalcoreindicators.org/) is a portfolio of surveys aimed at measuring outcomes and performance of state developmental disabilities services [4]. Findings from the survey in 2011–2012 and 2012–2013 suggest taking medication for behavioral health is associated with lower rates of paid community employment [3]. Additionally, other studies have documented lower rates of employment with individuals with “behavioral problems” when compared with to those without diagnosed or parent-reported behavioral problems [8]. Yet, data generated from the 2017–2018 National Core Indicators Survey data show few differences in rates of community-based employment for adults receiving services from state developmental disability agencies with and without mental health conditions [17]. These contradictory findings and inconsistent definitions and characterizations of mental health conditions (e.g., “behavioral challenges,” categorizing based on medication, etc.) demonstrate the need for a more focused exploration of factors associated with employment for adults with ID-MH.

A life course approach to health suggests that transactions between individuals and their environment greatly influence outcomes, including participation [18]; thus employment outcomes may be also influenced by environmental factors. Importantly, a life course perspective highlights the importance of environmental resources and supports, especially during critical transition periods. Evidence supporting this framework documents how environmental factors such as city size [19] and availability of transportation are associated with employment for individuals with disabilities [e.g., 20,21]. Accordingly, it is critical to identify both the individual characteristics and modifiable (i.e., environmental) factors associated with employment for individuals with ID-MH.

Job quality

Additionally, not all jobs are created equally, and understanding the factors that predict employment is insufficient. Having “a job,” does not necessarily confer health and wellness benefits or financial stability. The United Nations recognized the importance of job quality in their definition of “decent work” which provides opportunities for fair income, security in the workplace, better prospects for personal development, and social integration among other criteria [22]. Individuals with intellectual disability are significantly underpaid, often not receiving minimum wage and benefits such as paid sick and vacation time [4,9,23]. Therefore, in addition to employment rates, it is also important to examine predictors of employment quality. Legislation, such as the Workforce Innovation and Opportunity Act [24] and the Department of Labor’s “Employment First Initiative” [25] establishes the value of “competitive employment” for individuals with intellectual disability, defined as community-based, with wages a least equal to minimum wage. In this study, we also examined employment quality indicators, including job duration, hourly wages, and number of hours worked for individuals with community-based jobs.

Study purpose

Understanding employment for young adults with ID-MH is particularly important, as early employment is associated with participation in the workforce later in life [8,26]. Many school-based transition programs are designed to establish employment for young adults. Yet, research shows that maintaining a job post-graduation can developmental/developmental disabilities [9,19,27] and that increased age can be associated with lower employment for individuals with intellectual disability [8] and people with schizophrenia, without intellectual disability [16,28]. Therefore, we sought to describe employment for young adults at least one year distant from intensive school-based supports by focusing on young adults with ID-MH ages 23–30. Using data from the NCI^® project, we aimed to examine the following hypotheses:

H1) Young adults ages 23–30 with ID-MH will have lower rates of community-based employment than those without mental health conditions.

H2) For young adults ages 23–30 with mild/moderate intellectual disability and co-occurring mental health conditions, living in non-institutional settings, and who have access to transportation,

H2a) Younger age, guardianship, and less restrictive residential setting will predict employment and employment quality, as measured by wages, hours worked.

H2b) Guardianship, and less restrictive residential setting will predict employment satisfaction.

H3) Young adults ages 23–30 with ID-MH will have lower quality employment than those with ID without mental health conditions

Methods

We conducted secondary analysis of data from the 2017–2018 National Core Indicators^® (NCI^®) surveys to describe employment for young adults with ID-MH. The National Core Indicators are a collaborative effort between Human Services Research Institute, the National Association of State Directors of Developmental Disabilities Services, and participating states. The purpose of the NCI^® project is to collect standardized information across states about system performance through the measures of outcomes for people with intellectual/developmental disabilities. Data collected in the NCI^® surveys include data at the individual level (i.e., service recipient) gathered from existing records, in-person interviews with individuals with intellectual/developmental disabilities, family-member surveys completed using mail, and on-line surveys completed by provider agencies. This study includes data from existing records and face-to-face surveys with people receiving services.

The 2017–2018 In-Person Survey is composed of three parts. The Background Information section collects data from existing records and includes demographic information, such as age and gender. Section I includes subjective information that can only be answered by the individual with intellectual/developmental disability. Section II includes observable information that can be answered by a proxy respondent if the individual with intellectual/developmental disability cannot or chooses not to respond. The survey collects information in four domains: 1) descriptive data; 2) health, welfare, and rights; 3) individual outcomes; 4) system performance [29]. In this study, we drew upon two domains: descriptive data (e.g., demographic factors, such as age, diagnosis, type of residence) and individual outcomes (which includes a work subdomain).

Participants

Currently 46 states and the District of Columbia are members of the NCI^® program. Of those states, 35 states and the District of Columbia chose to participate in the In-Person survey in 2017–2018. Within these states, adults receiving at least one service in addition to case management from state developmental disability systems are eligible for the In-Person survey, regardless of functional abilities. Thus, the sample is limited to individuals receiving state developmental disability system services. Each state is tasked with a goal of administering the survey to 400 randomly selected individuals. In 2017–2018, data were collected from over 25,000 adults with intellectual/developmental disabilities [30].

To examine employment rates (Hypothesis 1), we included respondents ages 23–30, with a diagnosis of intellectual disability, and those who had complete data regarding community-based employment and the presence or absence of a mental health condition (defined as diagnosis of mood disorder, anxiety, psychosis, and/or other mental illness as documented in the participant’s records). To examine predictors of employment quality and satisfaction (Hypotheses 2a-b), we included only those individuals who had a community-based job (per record-review or self-reported) and had complete data on the following variables: quality criterion (hourly wages, weekly hours, job duration), satisfaction, gender, race, diagnosis of autism (yes/no), level of intellectual disability (mild, moderate)², primary language, mobility, guardianship, residential setting (e.g., group home, foster care, relative’s home), perceived availability of transportation for daily needs, and living setting (rural, small town, micropolitan, metropolitan).

After examining the sample, we identified that all but four individuals indicated a preference to communicate using speech. As a result, we added an additional inclusion criterion that participants communicate using speech. For some functional and environmental variables, the sample had limited variability; we thus removed two individuals with “severe” intellectual disability, one individual who lives in an institutional setting, four who live in “other” settings, and one individual who reported they almost never have transportation for daily need to prospectively improve model fit.

Analysis

All analyses were conducted using R 3.5.2 [31]. We recoded variables of interest to remove “I don’t know” responses and missing data. In addition, we created several new variables to increase interpretability of findings; for example, we created a binary white/non-white variable for race and also collapsed the residential setting variable from 14 categories into six residential settings (i.e., institutional setting, group home, own home, with parents or relatives, foster care, and other).

Hypothesis 1: employment rates

To examine employment rates for individuals with ID-MH, we calculated the percent of respondents with a response of “Yes” to the question, “person was in paid individual job in community-based setting during typical 2-week period” (per record review) and self-reported “Yes” to the question, “Do you have a paid job in the community?” We used chi-square tests to compare rates of employment between the two groups of interest (individuals with intellectual disability without a MH condition vs. individuals with ID-MH).

Hypothesis 2a: predictors of employment and employment quality

We utilized logistic regression to examine predictors of employment as a dichotomous outcome (yes/no) and linear regression to examine individual indicators of employment quality (i.e., hourly wages, hours worked within the last two weeks, job duration). While we were most interested in mental health condition, age, guardianship, and residential setting, we included additional covariates that extant literature suggests may impact employment: gender, race, diagnosis of autism, level of intellectual disability, primary language, use of mobility device, perceived availability of transportation for daily needs, and living setting [4,9,13,19]. Prior to running regression models examining predictors, we examined the data for multicolinearity, as several variables were thought to be dependent (e.g., level of intellectual disability and guardianship). No colinear variables were identified (using conservative criteria, correlation coefficient <0.6; [32]). We also evaluated the outcome variables (hours, wages, and duration) for outliers, defined as working >80 h/2 weeks (3 participants removed) and for hourly wages and job duration, ≥3 standard deviations above or below the sample mean [33] (10 participants removed due to outlying wages; 12 participants removed for outlying job duration).

Hypothesis 2b: satisfaction with employment

We recoded responses to “Do you like working at your job in the community?” to create a binary yes/no variables (we used a conservative approach in which the responses from the 23 individuals whose response was “in between” were recoded as “no”). While we intended to examine predictors of satisfaction, given the limited variability in this outcome measure, we were not able to do so. Rather, we descriptively report job satisfaction and describe patterns observed across those individuals who reported dissatisfaction with their job.

Hypothesis 3: differences in employment quality between individuals with and without mental health conditions

To compare average hours worked, job duration, and wages for those with and without mental health conditions, we conducted independent samples t-tests. Due to demographic differences between groups with regards to race, residential setting, and guardianship—all predictors of employment outcomes—we chose to primarily examine group differences by qualitatively describing differences found in the regression models between the subsamples with and without MH conditions.

Results

Hypothesis 1: employment rates

We included 2,149 individuals with intellectual disability in this analysis, 49.2% (n = 1,062) of whom had ID-MH. Compared to individuals with ID-MH, a higher percentage of individuals without mental health conditions (“ID only”) were employed based on data from both record review and self-report, but these differences were not significant when demographic, functional, and environmental differences were not controlled for (Table 1).

Table 1.

Descriptive outcomes.

Outcome	ID-MH	ID only	Test statistic (df)	p Value
Hypothesis 1: Employment Status	n=1062	n=1087
Record review-reported employment (%)	18.46%	21.62%	X2(1)=3.353	0.067
Self-reported employment (%)	28.81%	32.47%	X2(1)=3.387	0.066
Hypothesis 2a: Employment Quality	n=131	n=168
Hourly wages (mean, sd)	$9.46 ($1.89)	$10.27 ($1.89)	t(279.44)=3.705	<0.001
Job duration in months (average, sd)	24.72 (24.43)	28.00 (25.53)	t(284.86)=1.130	0.259
Hours worked in two week period (mean, sd)	27.66 (17.34)	34.43 (22.76)	t(296.86)=2.918	0.004
Hypothesis 2b: Satisfaction	n=125	n=163
Satisfaction (% like job)	86.40%	92.02%	X2(1)=2.398	0.122

In this sample, individuals with “ID only” differed from those with ID-MH on a number of demographic, functional, and environmental variables (Supplementary Table S1), including: race (ID-MH: 38.2% nonwhite; ID only: 50.9% nonwhite); use of mobility devices (ID-MH: 6.2%; ID only: 11.3%); English is not the respondent’s first language (ID-MH: 8.1%; ID only; 11.9%); diagnosis of autism (ID-MH: 25.0%; ID only: 20.3%); geography (ID-MH: 76.7% live in metropolitan region; ID only: 82.6% live in metropolitan region); residential setting (those with ID-MH were most likely to live in group homes: 29.7% or family members’ homes: 43.1%, whereas those with ID only were most likely to live in a family member’s home: 73.1%); and guardianship (individuals with ID-MH were more likely to be under full guardianship: 30.9% compared to those with ID only: 23.2%). Therefore, we further explored differences in employment using logistic regression, with employment (yes/no) as a dichotomous outcome, mental health condition as a predictor, and the following covariates: all variables with differences between groups (race, use of mobility device, English as a first language, autism diagnosis, geography, residential setting, and guardianship status) and two additional predictors, hypothesized to influence employment: access to transportation and level of intellectual disability. As described above, prior to running the models, we searched for and found no collinearity among variables.

When controlling for differences in demographic, functional, and environmental factors between groups, having a mental health condition was a significant predictor of self-reported and record review-reported employment (with mental health condition associated with unemployment). Additionally, the following variables were associated with unemployment: non-white race, being female, more significant intellectual disability (odds of having employment were increasingly lower for more significant intellectual disability), and speaking a language other than English as a first language. Compared to those living in a group home, those living in their own home or in foster care were more likely to be employed, as were those with consistent access to transportation. Individuals living in a family member’s home were also more likely to be employed (per record review data only). Additionally, individuals living in a small town were least likely to be employed compared to individuals living in rural, micropolitan, or metropolitan settings. See Table 2 for a summary of odds ratios. Supplementary Table S2 reports full model information, including differences in significant predictors across the subsamples of individuals with and without mental health conditions. According to the Hosmer and Lemeshow goodness of fit test, the models predicting self-reported and record review-reported employment yielded good fit (self-reported: X²(8)= 6.57, p = 0.58; record review: X²(8)= 4.70, p = 0.79).

Table 2.

Odds ratios for factors that were significantly associated with employment.

	Record review-reported employment (n = 2149)	Self-reported employment (n = 2149)
Predictor
Mental health condition	0.760*	0.790*
Demographic variables
Age
Gender (female)		0.703***
Race (not white)	0.675**	0.774*
Functional/diagnostic variables
ID level
Moderate ID (vs. mild)	0.622***	0.683***
Severe ID (vs. mild)	0.118***	0.375**
Profound ID (vs. mild)
Mobility device (vs. w/o aids)	0.356***	0.367***
English not primary language	0.349***	0.680*
Autism diagnosis
Environmental variables
Transportation
No transportation (compared to sometimes)
Almost always have transportation (compared to sometimes)		1.486*
Residential setting
Institution (compared to group home)
Own home (compared to group home)	2.320***	1.672**
Parent or relative’s home (compared to group home)	1.568**
Foster care (compared to group home)	1.766*	1.735*
Other (compared to group home)
Type of town/city
Rural (compared to small town)	3.245**	3.189**
Micropolitan (compared to small town)		2.195**
Metropolitan (compared to small town)	1.888*	1.973*
Guardianship
Limited guardianship (compared to no guardianship)
Full guardianship (compared to no guardianship)

^*p < 0.05

^**p < 0.01

^***p < 0.001.

Note: Odds ratios reported for each group. Results for significant predictors only are reported. See Supplementary Tables S2a and S2b for full list of model parameters.

Hypothesis 2a: predictors of employment and employment quality

Predictors of employment

Analyses did not support hypothesized predictors, as within this sample (Table 3), mental health condition, age, guardianship, and living environment did not consistently predict employment quality. As seen below and in Supplementary Tables S3–S5, the investigated predictors explained a small proportion of the observed variance, particularly with regards to wages and job duration, suggesting that other, unexamined factors may be more salient.

Table 3.

Demographic characteristics of participants for hypothesis 2a-b^a.

	Full Sample (n=299)	ID-MH (n=131)	ID only (n=168)	Test statistic
Mental health condition	43.81%	-	-	-
Demographic variables
Age (mean, sd)	26.47(2.17)	26.47 (2.05)	26.47(2.27)	t(290.4)=−0.012, p = 0.990
Gender	38.80%	38.17%	39.29%	X2(1)= 0.039, p=0.844
Non-white	37.12%	29.77%	42.86%	X2(1)= 5.400, p=0.020*
Functional/diagnostic variables
ID level
Mild	78.60%	74.81%	81.55%	X2(1)=1.987, p=0.159
Moderate	21.40%	25.19%	18.45%
Mobility
Moves without aids	96.99%	97.71%	96.43%	Fisher’s exact test: p=0.736
Uses mobility device(s)	3.01%	2.29%	3.57%
English is not primary language	4.68%	3.05%	5.95%	X2(1)= 1.386, p=0.239
Autism diagnosis	23.75%	25.95%	22.02%	X2(1)= 0.628, p=0.428
Environmental variables
Availability of transportation
Sometimes have	7.02%	6.87%	7.14%	X2(1)=0.008, p=0.927
Yes, almost always have	92.98%	93.13%	92.86%
Residential setting
Group home	14.38%	24.43%	6.55%	X2(3)= 30.83, p<0.001***
Own home	22.74%	25.95%	20.24%
Parent or relative’s home	57.19%	41.22%	69.64%
Foster care	5.69%	8.40%	3.57%
Type of town/city
Small town	3.34%	6.11%	1.19%	Fisher’s exact test: p=0.051
Rural	5.02%	6.87%	3.57%
Micropolitan	11.71%	12.21%	11.31%
Metropolitan	79.93%	74.81%	83.93%
Guardianship
No guardian	60.87%	51.91%	67.86%	X2(2)= 7.979, p=0.019*
Limited guardianship	11.71%	13.74%	10.12%
Full guardianship	27.42%	34.35%	22.02%

^aThe final sample for hypothesis 2b included 125 participants with ID-MH and 163 with ID only, as some participants did not have data about satisfaction.

^*p<0.05

^***p<0.001.

Predictors of employment quality

We provide a summary of significant predictors in Table 4. See Supplementary Tables S3–S5 for full output from the regression models.

Table 4.

Summary of significant predictors for employment quality.

Predictor	Full Sample (n=299)	ID-MH (n=131)	ID only (n=168)
Mental health condition	↓ Hourly wages	-	-
Demographic variables
Age	↑ Job duration	↑ Job duration	↑ Job duration
Gender (female)	↓ Job duration
	↓ Hours		↓ Hours
Race (not white)	↓ Job duration	↓ Job duration
	↑ Hours	↑ Hours	↑ Hours
Functional/diagnostic variables
Moderate intellectual disability (vs. mild)	↓ Hours	↓ Hours	↓ Hours
Mobility device use (vs. without aids)			↓ Hours
English not primary language
Autism diagnosis
Environmental variables
Almost always have transportation (compared to sometimes have transportation)
Residential setting
Own home (compared to group home)		↑ Job duration
Parent or relative (compared to group home)	↑ Job duration	↑ Job duration
Foster care (compared to group home)
Type of town/city
Rural (compared to small town)
Micropolitan (compared to small town)
Metropolitan (compared to small town)
Guardianship
Limited guardianship (compared to no guardian)		↑ Hours	↓ Hours
Full guardianship (compared to no guardian)	↓ Hours		↓ Hours

Hours worked.

For young adults with ID-MH, having moderate intellectual disability (compared to those with mild intellectual disability) significantly predicted a lower number of hours worked in the last two weeks (β= −7.136, p = 0.038). Individuals who were non-white (β = 8.810, p = 0.013) and were under limited guardianship (β = 10.227, p = 0.026) (compared to no guardianship) worked more hours per week (model adjusted R²=0.137).

Hourly wages.

We identified no significant predictors of hourly wages for young adults with ID-MH. The model, inclusive of all predictors, explained only 1.4% of observed variance in hourly wages, suggesting additional, unevaluated variables may be necessary to consider (model adjusted R²= −0.014).

Job duration.

For young adults with ID-MH, age, race, and residential setting significantly predicted job duration. Individuals who are non-white (β= −10.423, p = 0.041), had shorter job durations. When controlling for all other variables, for each year of age, on average, individuals worked 2 months longer (β = 2.248, p = 0.045). All residential settings, with the exception of foster care, were more favorable than a group home for increased job duration, though only living in a family’s member’s home significantly predicted longer job duration (β_foster= −0.263, p = 0.977; β_ownhome= 17.664, p = 0.006; β_familymember= 12.938, p = 0.027) (model adjusted R²=0.088).

Hypothesis 2b: satisfaction with employment

Most individuals reported liking their job (Table 1). Only 7 of 288 individuals said they did not like their job, and 23 individuals reported “in between” for the question, “Do you like your job?” There was no significant association between mental health condition and job satisfaction (X²(1)=2.398, p = 0.122). We found that the individuals who did not endorse liking their job (“no” or “in between” response) were similar across demographic and functional characteristics compared to those who did like their job (all statistical tests examining the associations between these characteristics and satisfaction were nonsignificant; results not shown). Notably, all individuals who did not like their job had mild intellectual disability.

We conducted t-tests to examine differences in employment quality criterion (hours, wages, job duration) between those who report they were and were not satisfied with their jobs and found no significant differences. Individuals with ID-MH who did not like their job worked slightly fewer hours (mean = 25.35, sd = 19.00) than the full sample of individuals with ID-MH (mean = 27.66, sd = 17.34) and had very similar earnings.

Hypothesis 3: differences in employment quality between individuals with and without MH conditions

Compared to individuals without a mental health condition, individuals with ID-MH had statistically significant lower hourly wages and worked fewer hours. While, on average, individuals with ID only had a longer job duration, this difference was not statistically significant (Table 1). Because participants with and without mental health conditions differed on key variables associated with employment quality, we explored factors associated with employment quality by examining regression models in which these differences were controlled for (Table 4; Supplementary Tables S3–S5).

Hours worked

While in the full sample, we observed that females and those who have full guardianship, on average, work significantly fewer hours per week, this held true only within the subpopulation of people with ID-only. Interestingly, when all other predictors were controlled for, individuals with ID-MH who are under limited guardianship, were employed on average, approximately 10 hours more per week (β_{limitedguardian}= 10.223, p = 0.026) than those without a guardian, but this relationship was the opposite for individuals without mental health conditions (β_{limitedguardian}= −13.575, p = 0.018) and for the full sample (β_{limitedguardian}= −2.595, p = 0.485).

Hourly wages

We observed that the only predictor of hourly wages was the presence of a mental health condition. Within the subpopulations of individuals with and without mental health conditions, no significant predictors of hourly wages were identified. Notably, all individuals earned $16.00/hour or less (maximum hourly earnings for individuals with ID-MH = $14.00/hour; maximum for individuals with ID only = $16.00).

Job duration

For individuals with ID-MH, being non-white, on average, was significantly associated with shorter job duration (β=−10.423, p =.041); this was not true for individuals without mental health conditions (β= −5.927, p = .174). Additionally, we observed that individuals with ID-MH living in their own home or a family member’s home (compared to a group home) were significantly more likely to have a longer job duration, but residential setting was not significantly associated with job duration for individuals without mental health conditions. Of note, many more individuals with ID-MH lived in group homes (n = 32, 24.43%) compared to individuals with ID only (n = 11, 14.38%), raising the possibility that there may have been more variability of job duration within the subgroup of individuals with ID-MH in this living setting than the smaller subgroup of individuals with ID only (n = 11). Post-hoc analyses suggested that individuals who live in group homes with ID-MH and ID had similar median duration (median_IDMH= 13 months; median_ID-only = 15 months), but that a small subset of individuals with long job tenure may have influenced these findings: mean_IDMH=18.09 months, interquartile range_IDMH=22.25 months; mean_ID-only= 28.45 months, interquartile range _ID-only= 39.5 months.

Discussion

We described demographic, functional, and environmental factors associated with community-based employment and employment quality for individuals with ID-MH ages 23–34. We had expected that individuals with ID-MH would have lower rates of participation in community-based jobs and experience lower employment quality. However, we observed that, in our sample, while having a mental health condition did significantly predict employment and hourly wages, a diagnosis of a mental health condition does not appear to drive all observed employment disparities. Rather, even after controlling for mental health condition, several personal and environmental factors appear to be highly associated with employment rates and quality. These findings suggest an additive effect of personal and environmental factors, when overlaid with a mental health condition. When considering intervention approaches to increase workforce participation for individuals with ID-MH, it is critical to examine the specific predictors of employment and employment quality within this underemployed subgroup; as such, the remainder of this discussion focuses on findings related specifically to individuals with ID-MH.

Similar to extant literature in the mental health field, we observed that race and gender were associated with outcomes, with both non-white individuals and females employed at lower rates. In addition, individuals with ID-MH who were not white had shorter job tenure, yet worked more hours than white individuals with ID-MH. Our observation of lower employment rates and duration is consistent with research within the broader population [34–37], and may be attributed to systemic racism that influences hiring decisions, availability of supports on the job, vocational and educational training [36], and the ability of familial social networks to help individuals with intellectual disability attain jobs [38]. Additional research is needed to interpret the finding that non-white individuals worked more hours than white individuals. One possibility may be that due to structural inequality non-white individuals have fewer financial resources, and thus may be more likely to take on additional hours, when provided the opportunity to do so.

A recent review of research regarding the relationship between gender and employment for individuals with disabilities suggests that further research is needed to understand gender-based differences in both employment status and quality [39]. However, employment of females at a lower rate is consistent with population-level disparities [40], likely attributable to unconscious bias and social norms/expectations, etc.

Case managers, educators, and vocational specialists should be aware of the additional barriers faced by people of color [41,42] and women [43] in the labor market, potential gaps in skills due to a lifetime of reduced opportunities, and how these challenges may be compounded by mental health conditions [44]. Accordingly, professionals providing support to employees with ID-MH should recognize how bias may influence all aspects of acquiring and maintaining employment (e.g., job applications, interviews, performance reviews) [45,46] and advocate that employers provide additional job-based supports, training, and feedback, to help increase job tenure. Austin and colleagues [35] explored service-level predictors of employment outcomes for vocational rehabilitation enrollees with ID-MH of all ages. They identified that services such as on-the-job supports, job placement, job training, transportation, and additional financial supports for basic needs (e.g., shelter, clothing) significantly predicted attainment of a competitive job. This is aligned with research indicating that employers may be more interested in hiring employees with disabilities who have job coaches and other on-the-job supports [47]. Such supports can foster person-environment transactions that support job performance [18,48]. Accordingly, ensuring that individuals with ID-MH who are female and/or non-white receive such services to help reduce observed disparities is critical. It follows that societal-level interventions addressing systemic racism may have a positive impact on employees with ID-MH who are not white by influencing service equitable provision. These services must be provided early in transition, as early employment is associated with employment later in life [8,26,49]; such services could be the focus of future policy.

While functional variables did not strongly predict outcomes in these analyses, several environmental variables were associated with outcomes, as hypothesized and found in previous research [19–21]. Though residential setting and guardianship are often perceived as proxies for function, in this sample, these environmental variables had small and non-significant correlations with diagnostic and functional variables (we note that different relationships may have been observed had we included individuals with severe intellectual disability). We observed that individuals who conceivably have greater independence due to less restrictive housing, on average, had longer job tenure. While in general, guardianship was associated with fewer weekly hours, individuals with ID-MH who had limited guardianship, on average, worked more hours than those with no guardian. Given this surprising result, understanding how guardians of individuals with ID-MH may facilitate acquisition and maintenance of employment merits further research.

Aligned with previous research, we found that self-reported employment was significantly associated with availability of transportation [7], yet transportation was not associated with other outcomes. While research suggests that transportation can limit participation for individuals in more restrictive living settings [50], in this sample, lack of transportation in more restrictive settings cannot explain these disparities, as most participants reported that they always have the transportation they need, and availability of transportation was not associated with residential setting. One explanation for the high proportion of individuals who have needed transportation may be that the majority of the sample resided in metropolitan regions in which transportation may be more readily available and accessible. Furthermore, individuals with limited transportation may not be accessing state developmental disabilities service systems, and thus, not included in this sample. Importantly, we observed that many individuals with access to transportation still do not have community-based jobs, suggesting that while transportation is of great importance, it is just one piece of the puzzle for improving employment outcomes in this population.

Similar to the findings of Kocman & Weber [51] we found that most individuals reported job satisfaction. This is an important finding, as job satisfaction is associated with wellbeing and mental health [52]. No individuals with moderate intellectual disability reported that they did not like their jobs. Individuals with higher functional support needs may have had fewer opportunities to make choices and participate in the community [53], perhaps leading them to be more satisfied by “any” employment opportunity. Additionally, they may be more beholden to service providers and feel restricted from providing “negative” feedback about services [54]. Another alternative explanation is that individuals who did not like their jobs were no longer working, and thus not in the analyzed subsample. Together, these factors may have contributed to skewed reporting of job satisfaction. Future work should address the factors that contribute to job satisfaction for individuals with ID-MH. Research may explore how job accommodations specific to mental health related needs influence job satisfaction, in addition to employment quality and job tenure.

Limitations and future research

This study provides a description of predictors of employment and employment quality indicators for young adults with ID-MH by drawing upon an existing dataset. There are several limitations that should be addressed in future research. First, employment policy and the resulting support available to assist individuals to pursue work differs greatly by state in the United States [55]; the level of employment supports is likely an important predictor of employment and employment quality, and we did not include employment services as a covariate in our analyses. We aimed to include state-level information about employment policy as a covariate. However, identifying current state-level information not only about their policies, but assessing implementation of those policies in a valid way proved almost impossible. Future research may include information about the specific employment programs and supports individuals receive; importantly, eligibility may be impacted by functional factors, including mental health support needs. Second, the difficulty of diagnosing mental health conditions in individuals with intellectual disability has been well documented [56,57]. Information about mental health conditions was reported in the NCI^® survey based on record review. There may be significant variability across individuals, and perhaps states, as to how mental health conditions are diagnosed and documented. Additionally, we do not know if individuals without documented mental health conditions have ever been evaluated for mental health conditions—it is possible that several of these individuals did in fact, have unidentified mental health conditions. For this reason, we did not conduct subgroup analyses by mental health condition. Interpretability would be enhanced by a clear and consistent definition of mental health conditions for individuals with intellectual disability. Third, in order to increase interpretability of findings, we included only individuals who had no missing data across all outcomes of interest. This reduced our sample size, reducing confidence in the generalizability of findings. Importantly, individuals with more significant cognitive impairment, those living in institutional settings, and without access to transportation were not represented in analyses of employment quality and satisfaction (Aims 2a-b); more research is needed to understand employment for these individuals. Additional limitations of this secondary analysis include the cross-sectional nature of the dataset, lack of information about individuals’ employment history, and lack more precise functional data (e.g., functional abilities related to work skills rather than diagnostic variables), in addition to limited variability across diagnostic variables (e.g., level of intellectual disability), function (e.g., communication preferences and mobility), and discrepancies between record review and self-reported employment that we were not able to further investigate. For example, it seems likely that some individuals reporting having a community-based jobwho were not categorized as having a community-based job per record review may have had a different type of vocational setting/position, given the low average wages. Prospectively planned projects are necessary to address these limitations and to also capture data about employment for individuals who are not served by state developmental disability service systems. Finally, as relatively little observed variance in wages and job duration was accounted for by the predictors, qualitative research may be an important preliminary step in determining other salient predictors of employment outcomes for study in future research.

With regards to job satisfaction, there was limited variability within the data set. To learn more about factors related to job satisfaction, qualitative research may be needed to deeply explore the experiences of individuals who are not satisfied with their jobs. We had hoped to explore job satisfaction in relation to job industry and individuals’ role in choosing their job. However, there was also limited data for this variable and limited variability, and future research in this area is necessary.

Conclusions

Individuals with ID-MH experience employment disparities compared to individuals with intellectual disability only. We observed that community-based employment was significantly predicted by gender, race, level of intellectual disability, first language, and residential setting. In addition, employment quality was significantly predicted by age, race, level of intellectual disability, residential setting, and guardianship status. Service providers should specifically attend to those at the highest risk of unemployment/low quality employment (e.g., females, people who are non-white, and those living in group homes). Policies that fund societal-level interventions to address racial and gender-based bias may support individuals with ID-MH to acquire and maintain jobs by addressing disparities in social networks/social capital and ensuring equitable service provision and supports for those at the highest risk for unemployment, especially during transition to adulthood, when environmental supports may be particularly useful. Additionally, given the high proportion of people with intellectual and/or developmental disabilities who have co-occurring mental health conditions, it is critical that educators and developmental disability service providers be trained to support individuals who have co-occurring mental health conditions. This may take the form of early identification of symptoms so appropriate supports can be established prior to youths’ transition to adult service systems, identifying reasonable accommodations that mitigate the impact of mental health symptoms/challenges, and/or teaching self-management strategies that can be implemented in the workplace.

IMPLICATIONS FOR REHABILITATION.

• Young adults with intellectual/developmental disabilities and co-occurring mental health conditions (ID-MH) experience employment disparities.

• Young adults with ID-MH who are non-white and female may have particularly low employment rates and employment quality.

• Societal-level interventions to address racial and gender-based bias may support individuals with ID-MH to acquire and maintain jobs by addressing disparities in social networks/social capital and ensuring equitable service provision and supports for those at the highest risk for unemployment.

• Policy makers should consider additional funding for employment services for transition-age youth with ID-MH, particularly those from marginalized populations.