How do supported employment programs work? Answers from a systematic literature review

For many people, work means belonging to a society. It is an essential part of their lives and improves their self-esteem, helps establish social relationships, and sets life goals (Knaeps et al., 2012). Unemployment might negatively affect individuals and their families and predict negative societal and economic implications. It can cause increased psychological and physical problems as well as higher mortality and suicide rates (Hedley et al., 2017). Some, especially those with a form of disability or other impairments, are particularly affected, as they often struggle to attain a competitive job. For them, a diverse set of vocational rehabilitation (VR) programs can help them find employment. Most of these programs are categorized as supported employment (SE) programs, which showed their effectiveness in many ways (Drake & Wallach, 2020; Williams et al., 2016). SE is a so-called “place-then-train” model (Corbière et al., 2013; Grigorovich et al., 2017) and is defined as a value-based and person-centered approach to assist and support people with limited access to the labor market in obtaining and retaining paid work in companies in the first labor market. It inverts the traditional VR approach, placing the client in a work environment first and providing support within the work setting rather than training the client before finding a job (Homa & DeLambo, 2015). Some practices, such as the individual placement and support (IPS) model of SE for people with severe mental illness (SMI), are evidence-based and systematic, to help achieve competitive employment (Bond et al., 2012). IPS is based on eight principles: eligibility based on client choice, focus on competitive employment, integration of mental health and employment services, attention to client preferences, work incentives planning, rapid job search, systematic job development, and individualized job supports (Bond et al., 2012). Other programs vary from segregated day programs to competitive employment as well as segregated work settings such as sheltered workshops (Verdugo et al., 2006).

Concerning outcomes, meta-analyses have found that SE, especially high-fidelity IPS, has been effective in helping individuals with SMI attain and maintain competitive employment compared with other VR interventions such as prevocational training, sheltered employment, or psychiatric services only (Hedley et al., 2017; Ng & Shanks, 2020). Program participants holding competitive jobs for a constant period show benefits such as improved self-esteem and better symptom control on the one hand (Hedley et al., 2017). On the other hand, SE enrollment has no systematic effect on nonvocational outcomes, either on undesirable outcomes, such as rehospitalization, or on valued outcomes such as improved quality of life (Bond, 2004). Studies on the cost efficiency of SE conclude that individuals with disabilities fare better financially from working in the community than in sheltered workshops and that SE appears to be more cost effective than sheltered workshops over the entire “employment cycle” and returns a net benefit to taxpayers. Cimera (2000) also points out that individual placements appear to be the most cost-efficient SE methods (Bond et al., 2001; Williams et al., 2016).

Even though different types of programs exist, common features are identifiable. SE is an evidence-based practice with multiple components designed to help adults with mental disorders or cooccurring mental and substance use disorders find and sustain competitive employment (Marshall et al., 2014). Competitive employment is characterized by the following factors: jobs with permanent status, paid minimum wage, and jobs not set aside for people with disabilities (Salyers et al., 2004). SE is usually defined by the following principles: (1) inclusion of all clients who want to work; (2) integration of vocational and clinical services; (3) emphasis on competitive employment; (4) fast job search without prevocational skills training; (5) job development by the employment specialist; (6) attention to the client’s preferences, skills, and experiences about their desired work and disclosure of mental illness to future employers; (7) benefits counseling; and (8) ongoing support services after a job is obtained (Mueser & McGurk, 2014).

Regarding the diverse set of programs, service providers, and target groups, it is not surprising that questions remain about the impact and the comparability of these programs since even though SE programs are mostly specialized for one specific target group, success rates are relatively low (Hedley et al., 2017; Knaeps et al., 2012). There is a need to identify appropriate factors that contribute to the success of these programs. In contrast to traditional systematic literature reviews (SLR) this contribution presents a research synthesis, which is based on the realist approach.¹ It provides an explanatory analysis of complex social interventions or programs, like SE, aimed at discerning what works for whom, in what circumstances, in what respects and how, which at the same time corresponds to the main question. The results of the review combine theoretical understanding and empirical evidence and focus on explaining the relationship between the context in which the SE program is applied, the mechanisms by which it works and the outcomes which are triggered (the so-called CMO configurations, Pawson & Tilley, 1997; Pawson et al., 2005). These are then presented clearly in an SE impact model.

Since the number of available scientific publications on SE increases rapidly each year, it is useful to summarize individual studies on a single topic clearly and concisely. An SLR can provide an overview of the state of research. This method is characterized by a systematic approach to the selection, inclusion, and analysis of literature sources (Machi & Mcevoy, 2016; Pawson et al., 2005). A detailed study protocol and evaluation plan are prepared, and the literature search is carried out in a priori defined literature databases as well as according to a priori defined inclusion and exclusion criteria (Ressing et al., 2009). Finally, the results are summarized qualitatively by contexts, mechanisms, and outcomes, which correspond to a realist review (Pawson et al., 2005). The further structure of the contribution follows the six literature review steps proposed by Machi and Mcevoy (2016), which is briefly explained below.

Step 1 Select a topic and Step 2 Develop the tools for argument: The search inquired which impact factors (i.e., contexts, mechanisms) in a counseling setting influence the effectiveness of SE programs (i.e., outcomes), guaranteeing a broad search approach.

Step 3 Search the literature: Four electronic databases were searched: (1) Web of Science, (2) PsycInfo, (3) WISO, and (4) Social Science Abstracts. The focus was on post-1999 publications because the debate about impact began during that time, and publications have continued to increase since then. Before deciding on this search term, the most common synonyms of “effectiveness” and “supported employment” were combined, including “impact,” “impact factors,” “effects,” “efficacy,” “vocational rehabilitation,” and “individual placement and support.” This procedure allowed for the inclusion of the largest number of relevant articles that have been excluded from a combination of the other search terms. For example, when evaluating the search terms, “impact” was excluded as the results were less satisfactory than for “effectiveness” and yet the relevant “impact” articles occurred in the results. The chosen search term, therefore, used a combination of “effectiveness” AND “supported employment” (German: “Wirkung” UND “Arbeitsintegration”) to include as many publications as possible.² We set the boundaries for the systematic review by defining inclusion and exclusion criteria which can be seen in Table 1. No studies were excluded based on the evaluation of quality. This inclusive approach ensured that the search captured a range of research methods and varied elements of SE programs. Consistent with the approach of realist review (Pawson et al., 2005), there was no reason to exclude literature sources based on the research design.

Step 4 Survey the literature A multistep screening process was executed to identify relevant publications. First, a screening of all 2023 titles eliminated clearly irrelevant articles that did not meet the inclusion criteria. Of the remaining ones, abstracts were checked for relevance. Because of this screening, 1691 publications did not meet the inclusion criteria and were not relevant to the study context. Afterward, full-text availability was proven, excluding 10 unavailable publications. The remaining 322 publications were evaluated by two authors for content based on their full text. After finding no discrepancies, the first author conducted a full review of the identified articles. Only 104 publications that met all the inclusion criteria were analyzed and coded. Those that did not meet the inclusion criteria were flagged for removal. Afterward, two reviewers evaluated those studies. The decision whether the study should be removed or included was based on consensus. Figure 1 provides an overview of the search results at each stage of the process. Of the 104 identified articles, the final review included 1 case study, 21 meta-analyses or SLRs, 16 experimental studies, 19 qualitative and 31 quantitative studies, 9 mixed-method studies, and 7 research overviews.

The publications were coded for further information using MAXQDA, a software package for qualitative data analysis, since it is flexible when subcoding is needed.

The codes were developed by the four-member project team according to the realist review (Pawson et al., 2005). The literature review was guided by the question what works for whom, in what circumstances, in what respects, and how. According to the realist review, the focus was on identifying significant contexts, mechanisms, and outcomes of SE programs. The selected literature was finally coded by a single project member. The project team regularly discussed the inclusion and exclusion of literature as well as the codes and their fit. The basic coding included two parts: Part I program information (i.e., SE program type, target group, professionals, and other stakeholders); and Part II impact factors and impacts separated into context–mechanism–outcome (CMO) regarding RE. This contains the coding to the research question “What components work in supported employment for whom, in what circumstances, in what respect and how?.” We refer to Step 5: Critique the literature—draw conclusions in the following chapter, and to Step 6: Write the review—communicate and evaluate the conclusions in the closing chapter.

As noted at the beginning, in contrast to traditional systematic literature reviews, this contribution shows a research synthesis, which is based on the realist approach. It provides an explanatory analysis of SE programs, aimed at discerning what works for whom, in what circumstances, in what respects and how. The review enables the policy and practice community to reach a deeper understanding of SE programs and how it can be made to work most effectively (Pawson et al., 2005).

Although the implementation of VR programs was shown to be effective, practical implementations lag behind theoretical knowledge. Many studies have focused on evidence-based approaches to analyze factors affecting outcomes. Therein, SE was the most frequently used program type, of which the IPS model provided the best outcome for diverse target groups (e.g., Bond et al., 2012; Knaeps et al., 2012). Regarding the research on these SE programs, principles were developed to measure fidelity. Only when high fidelity to the SE model is guaranteed will program implementation lead to effectiveness and good outcomes for individuals (e.g., Bond et al., 2012). Thus, the greatest individual predictor mentioned for positive outcomes is fidelity (Beimers & Gatlin, 2011).

However, besides fidelity and program implementation, there is evidence that other factors shape the effectiveness of SE programs. As Bond et al., (2001, p. 316) stated, “Further research is needed to clarify critical ingredients of supported employment, which will lead to modifications, refinements, and additions.” There were hints at other criteria (i.e., context and mechanism) that affect the outcome of SE programs. Researchers, using a systematic approach, have shown that, besides a good client–employment fit, key variables include individual characteristics as well as contextual factors, cultural and societal attitudes, and environmental factors (Homa & DeLambo, 2015). The fit between clients’ characteristics and the environment, including cultural, social, and physical aspects, is critical. It is important to understand those factors because of their impact on individual behavior and functioning (Homa & DeLambo, 2015). Furthermore, SE corresponds in many ways to ethical social work practice, such as the zero-exclusion criterion or the basic idea of a strength-based approach (Beimers & Gatlin, 2011). Despite that, the relationship between the professional (e.g., the social worker) and the client comes into focus as an influencing factor.

Furthermore, building the model on RE showed that the interaction between contexts and mechanisms affects the outcomes. This interaction generates the results of the program (Pawson & Tilley, 1997). The review showed that most of the research on SE programs were one-dimensional, looking at either the effects of SE programs, the client or the professional, or the relationship between clients and professionals. The model reveals that impact factors are interconnected (e.g., if employers and the work environment are open to creating working conditions that suit the client’s needs, the satisfaction with these conditions increases and accordingly the possibility of lengthening work tenure). There are cumulative impacts on the client, the professional, the work environment, and so on. By considering more impact factors in the coproduction process with the client, professionals can offer even more individualized SE programs, respond quickly to changing circumstances, and measure the impact of their intervention. This might help in dealing with barriers that otherwise become too challenging and outweigh the support. To plan effective programs, we particularly recommend facilitating organizational change and trusting team relationships (especially the integration of mental health and vocational services within a single service team), clarifying and discussing anticipated gains of integration, and considering continuous support for sustainability (Hillborg et al., 2020).

Thus, this review provides the groundwork for further studies to make SE programs even more effective by building on CMO configurations at the microsystem, mesosystem, and macrosystem levels. A special focus still needs to be on the microsystem level, meaning the relationship between clients and professionals, especially, as actions at the microsystem level are integrated into higher levels. More research is needed to understand how social workers and therapists experience their contribution to IPS (Moen et al., 2020) and to understand better ongoing support and attention to client preferences as well as the correlation between those with SMI and employment outcomes (Bender et al., 2020).

Even though this study provides an expanded impact model for considering SE programs, the limitations of the work must also be presented. First, only German and English publications and only four databases were considered. Notwithstanding the presentation of an expansive SLR, there might still be unlocated articles that meet the inclusion criteria and are relevant to completing the impact model developed. Second, the focus was only on identifying impact factors with some initial evidence of effectiveness; therefore, no quality criteria for the studies were collected and evaluated. Third, certain VR programs were selected based on the selection criteria, which could also contribute to further studies.