Do individual placement and support (IPS) service users leave their jobs soon? A subgroup analysis of a long‐term follow‐up study and a comparison with other population groups’ data

During the past two decades, the number of people with mental illness securing competitive employment has increased, but job retention remains a concern. In Japan, nearly half of individuals with mental illness who obtained employment through Public Employment Security Offices (PESOs) in fiscal year 2023 discontinued their jobs within 1 year.1 This one-year turnover rate was higher than that for new employees who graduated from educational institutes in March 2023.2 The Individual Placement and Support (IPS) model is internationally recognized as an evidence-based approach that improves vocational outcomes for people with mental illness.3, 4 IPS is grounded in core principles—most notably zero-exclusion (e.g., no refusal based on symptom severity or functional impairment) and rapid job search—to assist service users in obtaining employment. Indeed, Japanese IPS programs undergo fidelity assessments using the Japanese version of the Individualized Supported Employment Fidelity Scale (JiSEF) to ensure adherence to the principles.5, 6 However, while these IPS principles facilitate employment among service users, a recurring concern in Japan—as previously observed in other countries7—is that, without pre-vocational training, they may leave their jobs soon after placement. In this letter, we use long-term follow-up data to examine job retention among IPS service users in a routine setting and then compare the retention rates with other national data. We conducted a 6-year follow-up study with 123 participants across 15 agencies who initiated IPS services between January 1 and June 30, 2017. The detailed methods and overall results have been reported elsewhere,5, 6 and the study was approved by the ethics committee of the National Center of Neurology and Psychiatry (No. A2023-056). Among the participants, 87 obtained competitive employment during the follow-up period and were analyzed as a subgroup (See Supplementary Table 1 for employment and retention rates by the fidelity scores in the full sample). We computed two Kaplan–Meier survival curves over a maximum of 5 years after participants began employment: one for the first job obtained and one for the longest-held job during follow-up, as IPS adopts a place-then-train approach and supports job changes among individuals already employed. The supplemental Kaplan–Meier survival curves were also computed to compare the low- and high-fidelity programs based on the JiSEF scores.5 We then contextualized one-year retention against PESO statistics1 and the large-scale data of newly graduated employees.2 Approximately 40% of the analytic subgroup had a diagnosis of schizophrenia, and about 30% obtained employment more than once during follow-up (Supplementary Table 2). The average and median days to obtain the first job were 394.41 (SD = 410.82) and 228 (IQR = 404), and those for obtaining the longest-held job were 519.14 (SD = 478.43) and 331 (IQR = 542). Kaplan–Meier estimates are presented in Supplementary Fig. 1 and Supplementary Table 2. For the first job, 77%, 66%, 35%, and 22% of participants remained employed at 6, 12, 36, and 60 months, respectively. For the longest-held job, the corresponding retention estimates were 86%, 83%, 47%, and 30% at 6, 12, 36, and 60 months, respectively; several observations were censored owing to termination of follow-up. A log-rank test indicated a near-significant difference between the two survival functions (χ2 = 3.600, P = 0.058). The supplemental analysis did not find a significant difference in overall retention rates between the low- and high-fidelity programs in this analytic subgroup (Supplementary Fig. 2). Figure 1 illustrates 12-month retention estimated by Kaplan–Meier curves for IPS service users, alongside data for employees placed via PESO and newly graduated employees. Compared with PESO benchmarks for new employees with mental illness—71% retained at 6 months and 56% at 12 months1—participants in our cohort showed higher retention at the same time points. With regards to the newly graduated employees in March 2023, 86% retained employment at 1 year (Supplementary Tables 4 and 5).2 This rate exceeded the cohort's first-job retention but was comparable to retention for the longest-held job. Relative to PESO's and newly graduated employees' data, our findings appear to not support the concern regarding IPS service users leaving their jobs too soon. In particular, more than 80% of participants retained their longest-held job at 1 year, suggesting that some individuals may achieve better person–job fit after an initial job change. Indeed, prior research indicated that longer work tenure is associated with employment aligning with individual preferences (e.g., wages, illness disclosure, weekly hours).8 The participants' first-job retention was relatively lower than that of new graduates. This finding may reflect structural labor-market conditions commonly faced by people with mental illness, including lower wages and employment precarity,9 whereas most new graduates obtained permanent and full-time employment.10 In such contexts, remaining with the same employer is not necessarily a desirable outcome. In other words, the results of this study may be associated with workplace conditions for people with mental illness. This study has four major limitations. First, the analytic subgroup size was relatively small. Second, the PESO database likely includes some IPS service users, as it encompasses all individuals who obtained employment through PESOs. Third, follow-up durations varied according to each participant's job start date, and several observations were censored. Fourth, although the fidelity scale contains relevant items, we could not examine the association between job retention and the frequency of follow-up services owing to the lack of detailed data. Accordingly, the findings are not definitive, but the analysis suggests that IPS service users tend to maintain their employment for at least 1 year while finding jobs suited to them. Furthermore, measuring quality and frequency of follow-up services, developing supports to promote longer job tenure, and improving workplace conditions remain future challenges. SY conceived and designed the study and conducted the analyses. CF provided administrative and technical support. SY contributed to the writing of the draft manuscript. All authors contributed to the interpretation of the results and to writing the manuscript and approved the final version of the manuscript. We used ChatGPT 5 to check the English grammar in the text. This study was funded by grants to SY from the Japan Ministry of Education, Culture, Sports, Science and Technology (No. 20H01611) and from the Japan Ministry of Health, Labour and Welfare (No. 25GC1002). The authors have no conflicts of interest to declare that are relevant to the content of this article. The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. Supplementary Table 1. Job retention rates between the low- and high-fidelity programs in the full sample. Supplementary Table 2. Characteristics of participants (n = 87). Supplementary Figure 1. Job retention among participants: the results from the Kaplan-Meier analysis. Supplementary Table 3. Number of participants who retained their jobs from the Kaplan-Meier analysis. Supplementary Figure 2. Job retention among participants by the fidelity score: the results from the Kaplan-Meier analysis. Supplementary Table 4. Comparison of job retention rates between IPS service users and employees via PESO. Supplementary Table 5. Comparison of job retention rates between IPS service users and newly graduated employees. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.