There were totally 8 peer support workers and 46 persons with schizophrenia ("service users") participating in the pilot program from April 2017 to August 2018 at the Taipei Veterans General Hospital Yuli Branch (TVGH-YL). The hospital provides treatment and community care for patients with mental illness who reside in the rural area of eastern Taiwan. The TVGH-YL administers a half-way house, a community rehabilitation center, and a supported housing program. The hospital also provides sheltered and supported employment as part of its community care services for persons with mental illness .
The inclusion criteria for both phases of the program were: (i) being certified with psychiatric disability of schizophrenia, or with catastrophic illness of schizophrenia in the health insurance system, (ii) living in half-way houses or receiving services from the supported housing program, (iii) participating in a sheltered or supported employment program, and (iv) showing interest in the job of care attendant. The exclusion criteria were: (i) comorbidity of severe physical illnesses which could lead to hospitalization, (ii) acute exacerbation of psychosis, and (iii) a reading ability below the age of 6 years. For phase I, there was one additional inclusion criterion which required experience of caring for elderly persons in the community for at least 1 year.
Training for peer support workers
The training program for peer support workers involves 8 individuals recruited and 7 trained professionals specializing in 6 medical disciplines, respectively. Table 2 details the themes of curriculum at various intervention levels.
Vocational rehabilitation co-led & assisted by peer support workers
Before each vocational rehabilitation session, peer support workers discussed with 2 occupational therapists ("stakeholders") to decide on the content and processes of the session. Peer support workers’ involvement should account for at least 50% of the session time to ensure the intensity of support. More specifics about this phase of the program can be found in Table 1.
We used the Social Support Scale (SSS)—initially designed by Liu and later modified by Sung and Yeh —to measure the social support received by service users. The Kaiser–Meyer–Olkin (KMO) value of 0.82 and the Bartlett test of sphericity (BT) of 815.37 (P < 0.001) in factor analysis confirm the construct validity of the scale. The internal consistency is also confirmed based on the Cronbach’s α of 0.86 . Three dimensions of social support were measured: relatives or family (SSS-R), staff or professionals (SSS-S), and friends or peers (SSS-F). There are 12 5-point questions for each dimension, with a possible total score ranging from 12 to 60.
Mental health and psychiatric symptoms
The Chinese Health Questionnaire-12 (CHQ-12) was employed to measure the mental health of service users. There are 2 questions on positive symptoms of mental health and 10 on psychological and physical discomfort, with a possible total score ranging from 0 to 12. The value of the area under the Relative Operating Characteristic curve (AUC) is 0.85, and the cutoff value is 3/4 . The sensitivity of the questionnaire is 78%, and the specificity, 77% . The lower the score, the better the mental health.
We also measured the positive, negative and general psychiatric symptoms of the service users by using the Brief Psychiatric Rating Scale-18 (BPRS-18) which contains 18 7-point questions. According to Bell et al., the α values which measure internal consistency for positive, negative, and general symptoms in the BPRS are 0.69, 0.68, and 0.46, respectively, which are deemed satisfactory to acceptable . The interrater reliability (r = 0.87) is also deemed satisfactory .
We measured the social functions of service users objectively by using the Global Assessment of Functioning (GAF) and subjectively by using the Chinese version of the Social Functioning Scale (C-SFS). The GAF score—which was used alongside the DSM-IV-R manual—ranges from 0 to 100, with 100 representing superior functioning. The scale comprises 10 levels (with an internal of 10 points) of psychiatric disturbance, accompanied by descriptions of impaired psychological, occupational, and social functions associated with the disturbance. Jones et al. reported a reliability coefficient of 0.72 and a significant negative association between the GAF score and patients’ medication/support needs .
The Chinese version of the Social Functioning Scale (C-SFS) was adapted by Song  from the Social Functioning Scale (SFS) developed by Birchwood et al., by factoring in distinct cultural characteristics in Taiwan. There are 7 dimensions in this scale: social engagement/withdrawal (5 items), interpersonal communication (4 items), independence-performance (13 items), recreation (15 items), pro-social (22 items), independence-competence (13 items), and employment/occupation (5 items). The internal consistencies are deemed acceptable to good, with a Cronbach’s α value of 0.86 for the scale as a whole and Cronbach’s α values ranging from 0.48 to 0.88 for the subscales . The higher the score, the better the social function.
Earned income from employment
Service users' weekly wages from sheltered or supported employment were collected from users' records on file for the 3 months before and after the intervention of vocational rehabilitation. The averaged weekly income during the 3 months was used as the measure of occupational outcome.
As all indicators examined in this study were continuous variables, we used paired t-test to compare the measurements before and after the intervention. IBM SPSS Statistics 16.0 was employed for statistical analysis.