| dc.description.abstract | Background
Eight billion kroner from the Norwegian state budget is used on employment interventions every year. These interventions primarily target and assist individuals with reduced work capacity (in need of assistance). The main reasons for reduced work capacity are musculoskeletal disorders and mental illness. Traditional employment interventions have been oriented towards sheltered work and training, with a goal of competitive employment after some time, “train then place”. The interventions Supported Employment (SE) and Individual Placement and Support (IPS) rather aim to rapidly place participants in competitive work (paid work in an ordinary work environment, full time or part time), with support and follow-up offered to the employee and employer (“place then train”). (IPS is particularly targeted toward individuals with mental illness.) While IPS and SE have been used in the United States and other countries for many years, its effectiveness is still contested in Norway. The Norwegian Directorate of Labour and Welfare therefore commissioned a systematic review of effect. In addition, they wanted an assessment of transferability to the Norwegian context.
Aim
The aim of this review is to identify, assess and synthesise research on the effect of Supported Employment, including Individual Placement and Support, among people in need of support to enter and stay in paid employment.
Methods
We conducted a systematic review in accordance with the Knowledge Centre Handbook. We performed a literature search in relevant databases and a search for grey literature. The search was last updated in March 2016. Systematic reviews of high methodological quality and/or controlled primary studies were to be included if the intervention was SE/IPS and the populations included one or more groups in need of employment support. Relevant outcomes were: competitive employment, income, quality of life, mental and physical health measures, and cost-effectiveness. Two researchers screened all titles and abstracts and then the relevant full texts in order to assess eligibility consistent with the inclusion criteria. We identified both primary studies and systematic reviews. Two researchers critically appraised the included studied for potential risk of bias. We synthesised the results and completed meta-analyses when possible. For each finding we considered the certainty of the evidence by using the GRADE approach. We also performed some analyses regarding transferability.
Results
We read 5010 abstracts and 229 articles in full text. We included 38 unique studies, described in 54 publications. In nine of the studies, the intervention was SE, and in 29 the intervention was IPS. The studies were conducted in the US, UK, Australia, Canada, Hong Kong, Japan, Netherlands, Bulgaria, Italy, Switzerland, Germany, Spain, Sweden and Norway. The studies were published between 1995 and 2015. Follow-up time varied from six months to five years, but the majority followed the samples for 12-24 months. Most studies included people with serious mental illness; other samples included people with severe physical injuries or developmental disorders. Control interventions were predominantly variations of usual services, often sheltered work, training, and sometimes groups or other programs. We performed nine comparisons all together. Below, the most important results are summarised:
IPS vs. control
Employment
The largest comparison comprises 21 studies that examined the effect of IPS versus control among people with serious mental illness. We found that IPS probably lead to more people getting competitive employment, most likely twice as many (RR=2.40, CI 2.08 to 2.77) (moderate certainty). They probably spent more time (days, weeks) in competitive work (SMD 0.90, CI 0.45 to 1.35) and had slightly higher income (SMD 0.30, CI 0.14 to 0.46 or MD US$ 200, CI 54.7 to 347.1) (moderate certainty).
Health
Most of the 21 studies also measured health outcomes. 13 studies measured quality of life (five of these could be included in a meta-analysis). IPS has possibly neither positive nor negative effect on quality of life (low certainty). 8 studies measured psychological symptoms. None of the studies found any effect of IPS on psychological symptoms (very low to moderate certainty). Five studies measured psychiatric hospitalisations. Four of these could be included in a meta-analysis, but the results were uncertain.
Cost-effectiveness
Finally, six of the 21 studies measured cost-effectiveness, that is, the costs of the intervention seen in relation to the results. Altogether, the studies’ results suggest that the costs of IPS were comparable to the costs of the control interventions and that IPS lead to more people in employment. IPS was thus more cost-effective (moderate certainty).
Enhanced IPS vs. control
Three studies compared enhanced IPS with control interventions. IPS enhanced with medication management, social support and cost-coverage, compared with traditional interventions, lead to more people with serious mental illness gaining competitive employment (moderate certainty). The IPS participants also spent longer time in employment and had higher income (moderate certainty). They had slightly improved quality of life, improved mental health (low certainty) and slightly fewer days hospitalised (moderate certainty). IPS enhanced with work-oriented cognitive-behavioural therapy (CBT) compared to traditional interventions also lead to more people with mild to moderate mental illness getting employment (high certainty). This (Norwegian) study also demonstrated a possible positive effect on quality of life and psychological symptoms (low certainty).
SE vs. control
The studies of SE were small and could not be synthesised statistically. Still, the trend is that SE lead to more people (with various illnesses or handicaps) gaining competitive work. These findings, however, have low certainty.
Transferability
In order to assess the transferability of the findings to a Norwegian context we performed a subgroup analysis with the 21 studies of IPS versus control, divided in three groups: the US, Europe and other countries. The studies showed a slightly better effect in the US compared to Europe, but the difference was not significant. We also performed a subgroup analysis based on types of control intervention, categorised as simple or extensive. The effects were approximately the same.
Discussion
The included studies indicate a rather consistent effect of SE in general and IPS in particular, regarding competitive work for people in need of support. This approach to assisting people into employment has consistently better effect compared with other, more traditional interventions. The measurements of quality of life, psychological symptoms and psychiatric hospitalisations show that IPS has neither better nor worse effect than other interventions. Other research has indicated that employment leads to better health. Perhaps the participants in our included studies have been employed too short to experience an effect on health outcomes, perhaps it is reasonable that some get stressed by working, perhaps some get better and some get worse, and the average outcome is zero, or perhaps these participants are so ill (most of them have severe mental illness) that it is unlikely that employment (mostly part time) will have significant impact on other parts of life. The overall results are in accordance with previous reviews.
The results seem consistent for standard IPS, whether we perform analyses on the basis of geographical/cultural context or type of control; IPS has a consistently positive effect. This adds to our confidence in the results as transferable to a Norwegian context.
Conclusion
In this review we have found that SE/IPS lead to more people in need of support getting competitive employment, compared to other interventions. The best evidence is for IPS for people with serious mental illness. SE/IPS also lead to more time spent in competitive work and higher income. Seemingly, SE/IPS do not lead to positive changes regarding quality of life, (psychological) symptoms or hospitalisations. However, there are no negative effects. IPS seems to lead to more people in employment to a fairly similar cost as other interventions. Follow-up time, fidelity scales, geographical/cultural context and type of control intervention do not have significant impact on the effect of IPS. The findings are probably transferable to a Norwegian context.
Even if the results are quite consistent for IPS in particular, research with other groups in need of support will be useful, for instance adolescents with moderate psychological problems and immigrants with poor knowledge of Norwegian. More studies with longer follow-up time will also be valuable. | |
