Study design
This prospective observational study was part of a two-year, two-centre, non-blinded, parallel-group, non-inferiority cohort field study on the effectiveness of ISH vs HAU for non-homeless persons with SMI. The study was registered (ClinicalTrials.gov: NCT03815604) and approved by the Swiss Association of Research Ethics Committees (Swissethics; 2018–02381). The study was conducted according to the published protocol [21]. The 12-month data have already been published [19]. The present study reports on the longitudinal two-year results.
Procedure
The study was conducted in Bern, Switzerland. Recruitment took place between April 2019 and December 2020. Sample size calculation was performed with regard to the non-inferiority hypothesis and yielded a sample size of 28 participants for the intervention group. The control group was intended to be two- to three times larger than the intervention group in order to facilitate matching based on propensity scores [21]. However, for reasons explained in the limitations, we could not apply propensity score methods for the two-year data.
After admission to the respective housing rehabilitation service, participants were consecutively recruited by housing rehabilitation staff (see setting and study conditions below). All service users were asked a few days after their admission to HAU or at the first or second meeting with the ISH coach if they were potentially interested in taking part in the study. The study collaborator then contacted interested participants and asked them for written informed consent. Consenting participants were enrolled in the study and were assessed at baseline (T0), after six months (T1), 12 months (T2) and 24 months (T3). Follow-up assessments were intended to be continued after participants have been discharged from or have moved on to other housing rehabilitation settings. Participants were not financially compensated for their participation; however, care was taken to ensure they did not incur any participation costs (travel, phone costs).
Setting and study conditions
Independent Supported Housing (ISH) has been provided since 2012 by the Center of Psychiatric Rehabilitation of the University Psychiatric Hospital Universitäre Psychiatrische Dienste (UPD) in Bern, Switzerland. ISH targets non-homeless adult persons with SMI and a need for housing and related daily life support. ISH follows the ‘Housing First’ paradigm [22] and offers flexible and targeted support according to the service users’ individual needs. Support is provided by an outreach coach in service users’ independent accommodations, which are chosen and rented by the service users’ expenses and are independent of treatment and care services. ISH coaches are mainly non-medical mental healthcare providers with nursing or social work training. ISH is independent of treatment and care, which appropriate external specialists offer. According to the STAX-SA classification [23], ISH corresponds to a type 4 service with no staff on-site providing low to moderate (sometimes also high) support in independent individual accommodations in the area around Bern without emphasis on moving on.
The control condition consisted of different housing rehabilitation as usual (HAU) settings in Bern, Switzerland. These settings provide housing rehabilitation and support according to the traditional continuum rehabilitation approach [9]. Each setting on this continuum aims to help service users stabilise and gain housing skills to enable them to live independently. Some of these settings provide transitional support and seek service users to move along the continuum from higher to lower levels of supported accommodation. In each HAU setting, users have a rental contract with the service that includes housing and support. According to the STAX-SA taxonomy, the control condition included housing rehabilitation services of types 1, 2, and 3, with staff on-site providing moderate to high (sometimes low) levels of support in a congregate setting with limited or strong (sometimes no) emphasis on moving on. In addition, the control condition included host families (not covered in the taxonomy) providing moderate support on-site supported by outreach staff and limited emphasis on moving on.
Participants
The housing rehabilitation settings in both study conditions (ISH, HAU) target similar populations. The inclusion and exclusion criteria of the study were defined following the requirements of the service providers. Eligible participants were aged between 18 and 65 years, had a psychiatric diagnosis, could communicate in German, were able to take their medication if indicated, had a source of income (including social insurance benefits), and could provide written informed consent. Persons who lacked the capacity to consent and had impaired cognitive abilities that affected the feasibility or validity of the assessments, including intoxication, delirium, and dementia, were excluded.
Data collection and outcome measures
Assessments consisted of questionnaires and semi-structured interviews using the German versions of each instrument. Assessments took place in person before the Coronavirus pandemic, were continued by phone call assessments during the pandemic, and were completed according to participants’ wishes after the pandemic (in person, by phone call). In-person meetings occurred according to participants’ wishes in their homes (independent accommodation or HAU setting), in neutral places like parks or restaurants, or in the research office. Questionnaires were completed by the participants or by an interview according to their choice. Following the UN CRPD, which states that service users’ social inclusion and participation are the highest priority goal of rehabilitation [4], social inclusion was the primary outcome variable. Secondary outcome variables included self-reported quality of life, psychiatric symptom severity, capabilities, social support, support needs, observer-rated functioning, service utilisation, and housing rehabilitation service provision and costs.
Demographic and clinical information was also collected through interviews with the participants. Demographic data included participants’ age, gender, nationality, highest education, and the number and duration of previous stays in housing rehabilitation settings (see Table 1). Clinical information included the primary psychiatric diagnosis according to ICD-10 categories [24]. Diagnoses were verified using patient medical records where possible. One HAU participant did not allow diagnosis verification; she reported suffering from PTSD.
Social inclusion was measured using the Social Functioning Scale (SFS) [25, 26]. The 76-item questionnaire with mostly four-point Likert scales asks for participants’ social inclusion and participation among seven subscales (social engagement, interpersonal behaviour, pro-social activities, recreational activities, independence-competence, independence-performance, and employment/occupation). Raw subscale scores were converted into standardised scale scores with m = 100 and SD = 15 [25], and higher scores mean better social inclusion.
Quality of life was assessed using the Manchester Short Assessment of Quality of Life (MANSA) [27]. This questionnaire assesses participants’ satisfaction with twelve life domains on a 7-point Likert scale. The overall quality of life was summarised as total mean scores between 1 and 7, with higher values indicating better quality of life.
The subjective severity of psychiatric symptoms was assessed using the 9-item Symptom Checklist (SCL-K-9) [28, 29]. The severity of nine symptoms could be rated on a 5-point Likert scale and were summarised as total mean scores between 0 and 4, with higher scores indicating more severe symptoms.
The Oxford Capabilities Questionnaire – Mental Health (OxCAP-MH) [30, 31] assessed participants’ capabilities among 16 items on a 5-point Likert scale. Total sum scores were translated into standardised scale scores between 0 and 100, with higher scores indicating better capabilities.
Emotional social support was assessed using the ENRICHD Social Support Inventory (ESSI-D) [32, 33]. This 5-item questionnaire could be answered on a 5-point Likert scale between 1 and 5, which were summarised as total mean scores. Higher scores indicate more social support.
Observer-rated functioning was rated by participants’ key workers using the Health of the Nation Outcome Scales (HoNOS) [34,35,36]. If key workers were non-available due to participants’ discharge from their setting or after they moved on to a non-cooperating service, the first author rated participants’ functioning based on their answers during the assessment interviews.
Service utilisation encompassed participants’ utilisation of inpatient and outpatient psychiatric and somatic treatment, criminal justice contacts and psychotropic medication prescription during the past six (twelve at T3) months. It was assessed by interview using the adapted Client Sociodemographic and Service Receipt Inventory (CSSRI) [37].
Service provision and costs of ISH were assessed using the Swiss medical tariff reimbursement tool (TARMED) or the tariff reimbursement tool of the canton Bern, where the duration of contacts with service users and the corresponding support costs will be assessed. Prices for rent and basic needs of ISH participants were estimated based on the guidelines from the local social insurance using the upper limit of the coverage granted to provide a rather conservative comparison of costs between ISH and HAU [38, 39].
To assess the costs of HAU services, heads of included HAU services were asked about the daily flat rates. The daily flat rates usually include housing rehabilitation support, rent, and basic needs. Therefore, the individual amount of support received cannot be estimated separately.
Statistical methods
Statistical testing of differences in baseline sample characteristics between conditions (ISH, HAU) and between study completers and dropouts were performed using Kruskal–Wallis (numeric) and Chi-square tests (categorical). Scale scores from outcome measures were computed based on averaged available item scores [40, 41] and were descriptively analysed for the total sample and completers only. Descriptive analyses of costs, service utilisation, and move-on rates were conducted based on available data from the total sample. Within-group changes in service utilisation between T1 and T3 were analysed using McNemar’s test.
Because of the considerable amount of study dropouts in the control condition (48.1%), we did not impute outcome data of withdrawn participants. However, we imputed scale scores of missing data at baseline (one single data point) and of completers who missed single assessments (n = 3) or instruments (n = 1) without dropping out from the study (7.4% of study completers). Imputation of missing scale scores was performed by multiple imputation [42] using the R package mice [43], applying the predictor matrix quickpred (mincor = 0.3) and performing m = 5 iterations.
Mixed-effects models for repeated measures were conducted with the multiply imputed datasets to analyse the long-term effects of conditions and between-group differences at each time point. Mixed-effects models were performed on the dependent variables social inclusion (SFS), quality of life (MANSA), symptom severity (SCL-K-9), capabilities (OxCAP), social support (ESSI-D), and functioning (HoNOS). Variance across participants was modelled as random effects. Study conditions (ISH, HAU), assessment time points (T0, T1, T2, T3), and the interaction of condition*time points were modelled as fixed effects. Longitudinal within-group and between-group effect sizes for each dependent variable are presented as the model-derived fixed-effect parameters and 95% Confidence intervals (95% CI). Mixed-effects analyses were conducted using the lmer function of the R package lme4 via Maximum Likelihood estimation [44]. Between-group differences were tested by hypothesising the non-inferiority of ISH to HAU regarding the primary outcome of social inclusion. The non-inferiority margin was pre-defined to be 15, which refers to one standard deviation on the SFS (for details on the non-inferiority margin and sample size calculation, see [21]). All statistical analyses were performed using the statistical software R version 4.0.3 [45]. The significance level was set to α = 0.05 for all analyses.
Add Comment