The effectiveness of supported employment programs: Meta-regression analysis of the global secular trend, 1990 - 2015

Abstract

Importance: Supported employment programs aim to re-integrate unemployed people with mental illness into the competitive labour market. While it is known that short-term economic developments like the 2008 recession may have negative consequences on the supported employment effectiveness rates, research into long-term secular trends has not yet been conducted. Objective: The primary objective of the regression analysis was to assess the effectiveness of supported employment programs for clients with any mental disorder over a longer time period. Further objectives were a comparison to prevocational training and the identification of moderating variables.Data sources: We retrieved publications on randomized controlled trials and on routine implementation programs that were included in four recent systematic reviews on supported employment for people with mental illness, plus very recent publications that were identified by a new search. This search was conducted in the following databases: Pubmed, PsycInfo, CINAHL (Cumulative Index on Nursing and Allied Health Literature), and Google Scholar. Study selection: We included any publication that was used in the four meta-analyses plus very recently published studies. The included studies were conducted between 1990 and 2015. Data extraction and synthesis: 80 studies with 14,437 participants were included in the analysis. Data was extracted by one author and cross-checked by the second author. We have conducted univariate and multivariate meta-regressions on the basis of a random-effects meta-analysis of proportions. The best-fitting model was computed by utilizing a statistical model selection procedure. Main Outcome and Measures: The outcome was the competitive employment rate over time. The year of study conduction (1990 to 2015) was regressed on the competitive employment rate. Regression coefficient estimates were computed and graphically displayed.