Comparative Analysis of Independent Reviewer-Rated and Self-Rated Fidelity Scores in Individual Placement and Support Programs: Repeated Cross-Sectional Surveys

Fidelity assessments can contribute to maintaining the adherence to the individual placement and support (IPS) model, which enhances vocational outcomes for individuals with mental illness worldwide. While independent reviews are standard, self-assessments could broaden the implementation of IPS. This study aimed to evaluate reviewer-rated and self-rated fidelity assessments using the Japanese version of the Individualized Supported Employment Fidelity Scale (JiSEF), and to compare the two assessment methods in terms of their correlations with employment outcomes in Japan. Over the 3-year research period, fidelity assessments were conducted by independent reviewers and trained program staff members across 26 programs, totaling 58 assessments. Analyses involved kappa statistics for item-level comparison, the intra-class correlation coefficient (ICC) and paired t-test for the overall fidelity scores, and Pearson's correlations to examine the relationship between the fidelity scores and program-level employment outcomes. Most individual JiSEF items demonstrated fair to good reliability between reviewer-rated and self-rated assessments. The ICC for the overall JiSEF scores between the two assessment methods was 0.756, yet the distribution of self-rated scores was more scattered compared with that of reviewer-rated scores. The mean total scores from self-assessments were significantly lower than those from reviewer assessments (t = 2.072, P = 0.043). While both sets of scores correlated significantly with employment rates (r = 0.640, P < 0.001 for reviewer assessments; r = 0.325, P = 0.013 for self-assessments), the correlation was stronger for reviewer ratings (z = 2.207, P = 0.027). Self-rated fidelity assessments offer several benefits. However, since independent reviews had a more normal distribution and higher correlation with employment outcome, they should remain the priority in fidelity assessments within the Japanese IPS framework.