Interventions for improving employment outcomes for persons with autism spectrum disorders: A systematic review update

Background

According to the Centers for Disease Control and Prevention (CDC), the incidence rate of autism has increased over the past two decades to 1 in 59 children currently being identified with autism spectrum disorder (Baio et al., 2018). Internationally, autism spectrum disorders (ASD) prevalence is most recently reported at one in 160 children; however, this figure likely represents a significant underestimate (World Health Organization, 2019). Although individuals with ASD are capable of achieving academic and vocational success when provided with accommodations and support, outcomes for this group remain dire in terms of adult outcomes in employment and independent living (Roux et al., 2017; Wehman et al., 2018).

ASD refers to a range of neurological disorders that involve some degree of difficulty with communication and interpersonal relationships, as well as obsessive and repetitive behaviors. ASD occurs in all racial, ethnic, and socioeconomic groups. There is a wide range of effects demonstrated across the spectrum. Those at the lower functioning end of the spectrum often demonstrate physical limitations and may not be able to speak or relate socially to others. Those at the higher functioning end of the spectrum are often able to lead relatively independent lives, graduate from academic institutions, but may also be awkward in their social interactions and have difficulty developing friendships. The critical gap represented from the high to the low end of the spectrum dramatically affects employment-related skills, abilities, and behaviors (CDC, 2011).

Given the increase of ASD prevalence and the number of students with ASD exiting public school settings, attention is increasingly focused on potential employment outcomes for individuals with ASD. Individuals without severe disabilities are eight times more likely to be employed than individuals with very severe disabilities (National Organization on Disability, 2000). Individuals with ASD are among those least likely to be employed (Dew & Alan, 2007; Roux et al., 2017); only 14%–15% of individuals diagnosed with ASD in the United States gained employment (Cameto et al., 2003; Roux et al., 2017).

Although economic conditions and employer attitudes affect employment opportunities, employment outcomes for individuals with ASD can be improved by providing social, behavioral, and work supports specific to individuals with ASD (Schaller & Yang, 2005; Wehman et al., 2013). Several studies have indicated that individuals with ASD are more likely to lose their employment for behavioral and social interaction reasons rather than their inability to perform work tasks (Bury et al., 2020; Dew & Alan, 2007; Hurlbutt & Chalmers, 2004; Solomon, 2020; Unger, 1999; Wilczynski et al., 2013). Conversely, specific employer practices—sometimes referred to as demand side strategies—have been linked to improved outcomes for employees with autism such as collegial understanding, positive work environment, and provision of accommodations (Chan et al., 2010; Hayward et al., 2019). Employment support service providers must often provide social and behavioral interventions along with other supports to effectively facilitate obtaining and maintaining employment for individuals with ASD (Ham et al., 2014). However, relatively few employment support service providers including state vocational rehabilitation counselors have an in-depth understanding of services that are associated with developing successful employment outcomes for individuals with ASD (Dew & Alan, 2007). Social skills and behavioral interventions designed specifically for workplace settings have received increased attention in the research literature as a means of addressing this barrier to employment for individuals with ASD (Gorenstein et al., 2020).

Each state in the United States has at least one state agency that is charged with the provision of vocational rehabilitation services to facilitate employment outcomes for eligible individuals with disabilities. Data accumulated through the national network of state vocational rehabilitation agencies suggest that few individuals with autism requested services. Of the individuals who did request services, few were successfully employed as an outcome of the service. However, in its latest report, the National Vocational Rehabilitation Service System indicated that in the 2007 fiscal year, it provided services that successfully placed 1774 individuals with ASD into employment situations that continued for 90 or more days (Rehabilitation Services Administration, 2009).

Title and abstract stage

With Endnote software, two coders independently assessed and selected for advancement to the next stage of inclusion if titles and abstracts met at least one of the two following criteria:

Full-text stage

Full-texts of studies from all citations/abstracts advanced from Stage 1 were retrieved for a final determination of inclusion in the review and analysis. All of the following criteria were required for each study to be included in the review and analysis. Two coders independently assessed study inclusion at this stage.

Electronic searches

We searched within a large number of electronic databases using search terms from domains associated with ASD, employment, and intervention terminology. Due to the combining of the Westbrook et al. (2012a, 2013) systematic reviews, we did not place any age-related terms in our search strategy. In addition, we searched for our articles after 2008, the most recent year the Westbrook et al. reviews concluded their searches. Our electronic searches concluded in August 2017.

Using three domains (population, treatment, and domain), all linked by appropriate Boolean terms, we used the following search terms (* for truncation):

POPULATION

Autis*

Asperger*

ASD

Pervasive developmental disorder*

Pervasive development disorder*

Pervasive child development disorder*

PDD-NOS

Rett*

Childhood disintegrative disorder*

Heller's syndrome*

Heller syndrome*

DOMAIN

employ*

pre-employ*

work*

vocation*

occupation*

trade*

career*

job*

rehabilitat*

skill*

We searched the following databases:

ABI Inform Global

Searching other resources

Gray literature identified through electronic searches were submitted to the same inclusion criteria as other studies. The same time range (2008–2017) for gray literature types of studies was specified as the other studies. References from individual studies were searched for potential studies to consider for inclusion. In addition, unpublished dissertations and theses were identified through the search strategy for review and consideration. Also, recent conference proceedings from relevant associations and conferences were reviewed to identify unpublished studies to include in the review:

Autism Society of America

National Association of Rehabilitation Research and Training Centers

International Society for Autism Research

National Alliance for Autism Research

Autism Research Institute

National Autistic Society (UK)

Autism Research Centre (Cambridge)

Conference proceedings that were reviewed included:

Asia Pacific Autism Conference 2009

PENN Autism Network Conference.

Autism Society National Conference

NARRTC Annual Conference

Annual International Meeting of Autism Research (IMFAR)

National Autistic Society's Professional Conference

Cambridge Autism Research Conference

Selection of studies

Studies were screened for inclusion/exclusion decisions at two stages, Stage 1: citation and abstract and Stage 2: full-text. The same two coders served as independent reviewers at both stages. A third party was not needed to resolve a coding value difference.

Citation and abstract stage

At Stage 1, the decision for advancing the retrieved citations and abstracts in Endnote to the full-text stage retrieval was made independently by both reviewers based on meeting two items from a and b of the following questions or a designation by a reviewer of “unsure” (item c): A.

Are the participants identified, described, and defined under the Autism Spectrum Disorder category?

If the reviewers were “unsure,” the citation/abstract was advanced to the full-text stage for a final inclusion decision. Inter-rater reliability was tested on a sample of 25 studies at this stage and was found to be 95%. Coding differences were resolved through discussion between pairs of reviewers.

Full-text level

At the full-text Stage 2 level, full texts of all citations advanced from Stage 1 were obtained and coded for an inclusion/exclusion decision. Two reviewers for each study independently decided whether to include the retrieved full-text studies in the final analyses. An inclusion decision for advancement to the coding stage of the process required that a study met all the criteria presented earlier. When multiple studies used the same sample or outcome data, the study providing the most complete information focusing on our desired intervention outcome was selected for inclusion. Inter-rater reliability was established before initiating coding activities, minimizing coding disagreements. At the Full-Text Stage 2 level, the two reviewers recorded all excluded studies and the reason for exclusion independently. The two reviewers resolved their differences through discussion until they came to agreement.

Data extraction and management

Data that were extracted and coded from the primary studies included: publication source, subject characteristics, sample source, employment setting, intervention characteristics, type of employment, and outcome measurement. See Supporting Information Appendix A for a copy of the coding form.

Coding was completed by two independent coders with an Excel-based coding sheet. Any discrepancies were resolved by a third-party author.

Assessment of risk of bias in included studies

Included studies were coded by two independent reviewers for methodological quality addressing dimensions that included:

Design type o

RCT individual randomized design

In addition, an evaluation of the potential risk of bias of all included studies was conducted using “Risk of Bias” procedures developed by Higgins and Green (2011) in which studies are evaluated across five sources of potential bias. A report of this analysis is provided in Section 4.

Measures of treatment effect

The magnitude of the intervention effects was calculated as either a Cohen's d or an odds ratio (OR), depending on the nature of the outcome. If the outcome was continuous (i.e., amount of wages earned), the effect size was calculated as a Cohen's d, or standardized mean difference. When the outcome was dichotomous (i.e., employed or not), then we calculated ORs for effect sizes. Effect sizes were calculated directly from reported means and standard deviations or percentages for the experimental and control groups for studies that reported such statistics along with the sample sizes.

Unit of analysis issues

When studies reported multiple outcomes for the same group comparison, such as various timepoints, there are dependency issues that may arise when synthesizing effect sizes together. Because we did not statistically integrate any effects together, but rather discussed narratively the individual effects, issues of dependency do not impact the results of our summary.

Dealing with missing data

After using our coding guide to extract data, there were instances of missing data where studies did not report complete information covered in the coding guide. Because effect size data were complete, and no moderator analyses were conducted, we did not attempt to request information from authors or conduct any imputation techniques.

Assessment of heterogeneity

Because of our narrative approach to describe interventions effects given the small number of included studies, we did not perform a formal assessment of heterogeneity.

Assessment of reporting biases

Due to the small number of studies, we also did not explore any selection or reporting biases.

Data synthesis

Due to the small number of included studies, we opted to describe the findings of the included studies narratively. We conducted no quantitative data synthesis of the effects.

Subgroup analysis and investigation of heterogeneity

We did not conduct any subgroup or investigations of heterogeneity through moderator tests.

Sensitivity analysis

We did not conduct any sensitivity analyses.

Results of the search

The results of the search yielded a total of 12,982 citations published after 2008. Upon double-screening all titles and abstracts, 278 citations remained for further examination of their full-text documents. After double-screening all full-text documents, three studies were included.

Included studies

This section narratively summarizes the included studies and their characteristics such as research design, sample sizes, setting, recruitment, participants, interventions, and outcomes. Note that the studies will be referred to by the last name of the first author and the year of the publication.

Research design

All three included studies used RCT designs (Smith et al., 2015; Wehman et al., 2014, 2017). Two of the three studies used a random number generator to assign individual participants to treatment or control groups (Wehman et al., 2014, 2017). No studies used stratification or wait-listed control groups as part of their design. All three studies included data from follow-up points.

Sample sizes

The unique number of participants from both treatments and comparison groups across the three included studies was 108. The average sample size per study was 36 participants (T: 23.3 participants; C: 12.7 participants).

Setting and recruitment

In two of the three studies, local education agencies (LEAs) assisted in the recruitment of eligible participants meeting inclusion criteria (Wehman et al., 2014, 2017). Recruitment in the remaining study was conducted through advertisements at local universities, community-based service providers and support groups, and online support groups (Smith et al., 2015).

The settings for the treatment groups in Wehman et al. (2014, 2017) were located at community-based employment sites at hospitals within a single region. The control groups for these studies attended high school and continued their normal course of study for the duration of the academic year. The setting in the third study was private offices for the intervention and a clinical space for the assessment phase (Smith et al., 2015).

Participants

Excluded studies

Studies were excluded on the following bases: (1) did not include an employment outcome; (2) did not assess the impact of an intervention; (3) did not include a sample of individuals with ASD.