Effect and Cost-effectiveness of the Everyday Life Rehabilitation Intervention

The investigators intend to study the extent to which the intervention under study; the Everyday Life Rehabilitation (ELR), is an effective, and cost-effective model for co-planned rehabilitative services in sheltered and supported housing facilities, aiming at supporting residents´ personalized recovery pathway. The project is of high relevance with its permeating perspectives of severe psychiatric disability (SPD), equity, and quality improvement, in the intersection of health and social care in accommodations.

The ELR project is both user- and practice-oriented. The main focus is to study whether the ELR-model is effective, cost-effective, and relevant for users, based on pre-/post-intervention variables from a resident perspective, such as self-perceived recovery, quality of life, everyday functioning, and goal fulfilment of residents. The project is also practice-oriented with a web-based training, methods and tools for staff. The project is pragmatic in the sense of implementing and studying the ELR model in as natural context as possible. Throughout the investigators will collaborate with user- and relative associations, municipalities, and R\&D units, in order to develop an intervention that is relevant and useful.

It has been shown that people with SPD and impaired autonomy have significantly poorer health than the general population, while at the same time they do not have access to equal health care, despite that they belong to the highest priority group according to the Swedish Parliament's principles for prioritization of healthcare. The high priority is based on the fact that the target group has low autonomy and does not speak out loudly for themselves, or can argue for their right to health care, further that they do not actively seek and demand health care, along with the magnitude of suffering, and the impact on life quality. There is a gap in knowledge, regulations, and lack of interventions suiting this context. Both rehabilitative healthcare interventions in practice, and RCT´s, are sparse regarding interventions for this target group living in sheltered or supported housing facilities. Thus, it is important to develop and test new potentially useful and effective interventions.

Therefore, a manualized but individually flexible model for integrated healthcare-rehabilitation in collaboration with housing staff in supported or sheltered housing facilities; the Everyday Life Rehabilitation (ELR), has been developed by the project manager (Lindström, 2007), and tested in feasibility studies (Lindström et al, 2011; 2011b; 2012; 2013; 2017), aiming at personal recovery and meaningful everyday activities for persons with SPD. The ELR development is thus based on five years of previous research with feasibility studies, and has shown significant impact on residents' everyday functioning and health, as well as the practices of professionals. Based on these feasibility studies, the original ELR has been slightly revised, adding a monthly follow-up by housing managers, and clarified tools for co-planning, in order to promote early involvement in enhancing strategies. The Medical Research Council (MRC) guidelines for complex interventions have been thoroughly applied in the development process, and next, evidence is needed for implementation, requiring randomized controlled trial (RCT)-studies. Therefore, the investigators want to expand the design, and go further with a cluster RCT built on a slightly revised manual of the ELR intervention, adding clarified focus on management, the tools for collaboration, and a cost-effectiveness perspective.

Overall aim and research questions:

The overall aim of the project is to investigate the effectiveness, and cost-effectiveness of a co-planning, person-centered, motivational, recovery-, and activity-based intervention package for people with SPD living in sheltered or supported housing facilities.

The specific research questions (RQs) are as follows 1-2:

RQ 1. What is the effectiveness of ELR intervention on recovery, quality of life, everyday functioning, and goal attainment, compared to Treatment as usual (TAU)? RQ 2. What is the estimated cost-effectiveness based on participant outcomes following ELR plus TAU, and TAU alone, according to baseline differences of recovering quality of life (ReQoL) transformed into QALY´s? For RQ 1 and 2, data will be collected pre- and post intervention. Allocation will be concealed for the independent blinded testers, and partly concealed for the people with SPD giving consent to participate in the study, being informed that a waiting list of up to six months may become relevant. The treating occupational therapists, housing staff, and housing managers, will due to practical reasons not be masked to the intervention. Data will also be blinded to researchers until analysis have been conducted. A person at Umeå university, who is not involved in neither the interventions nor analysis, will administrate, collect and store coded data in a safety-box.

An initial power-analysis was conducted, dimensioning the study to detect a difference of 5 points on the ReQoL-scale with 80% power. Assuming a standard deviation of 10 (Keetharuth et al, 2018), an average cluster size of 2 participants per hosting facility and an intraclass correlation of 0.1, a total of 35 housing facilities in each group is required to reach a power of 80% when using a significance level of 5%.

Oral explanation of the study and a leaflet describing the methods will be given prior to the participant consent. Considering the target population of the intervention is a vulnerable group, there are particular ethical reasons to minimize the number of exposed participants, and thus an internal pilot from year one and two will form the basis for the determination of the sample size for the full scale RCT. This is in line with the recommendation of MRC guidelines for complex intervention.

Statistical analyses:

Regarding analysis for RQ 1, mixed effects models will be used to analyse the primary outcome, including a random intercept for the clustering factor housing facility. The post-intervention measurements will be used as outcome variables, while adjusting for corresponding baseline measurements. The adjustment for baseline variables will performed on both individual level and on aggregated cluster level using the average of the baseline measurement within the housing facility (Hooper et al, 2018). Details about the model, and the adjustment for baseline covariates will be prespecified in thein the statistical analysis plan.

The primary analysis will use an intention-to-treat (ITT) approach and include all allocated participants with valid data, whether they did or did not receive the complete intervention. Multiple imputation will be used for variables where missing-at-random can be assumed. A complementary per protocol analysis will also be made.

For the intervention group only, goal attainment scaling (GAS) will be calculated according to the specific statistical formula of GAS. The secondary outcomes (RAS-DS and GAS) will be analysed using mixed effects models or semi-parametric ordinal cumulative link models, including housing facility as a random effect. This is pre-specified in detail in the statistical analysis plan.

Subgroup analysis will also be conducted, stratifying on gender in all outcomes.

Specifically, for RQ2 on cost-effectiveness, the following analyses will be conducted: For univariate testing of significance of cost and ReQoL, Student´s test will initially be used. For cost-effectiveness analysis, the costs and effects of the ELR and the TAU alternative, will be calculated and presented in a ratio of incremental cost to incremental effect. Effects are health outcomes, such as quality-adjusted-life-years (QALY). For this matter, ReQoL will be transformed into QALY, accounting for baseline differences.

The Swedish Board for Health and Welfare has developed a model for priorities in health care, which consists of the following criteria: severity of the condition, effectiveness in treatment, cost-effectiveness in treatment, and evidence base. The model assumes a specific condition linked to a specific treatment.

In this study the conditions are X and Y, and the treatments are X and Y. This study is designed to provide this model with proper data. A high degree of severity in these conditions have repeatedly been reported.

Thus, the trial is designed to measure effectiveness, and cost-effectiveness. The latter is defined as the cost per QALY gained. The resources needed for the interventions will be measured in physical units (mainly time) and transformed to monetary values.

When all sub-studies have been finalized, the investigators plan to suggest a priority rank for the interventions under study. The Board for Health and Welfare use a scale in 10 steps; 1 is the highest possible rank and 10 is the lowest. This ranking is based on the four criteria described above.

Protocol amendments:

After the first year, the internal pilot will be analyzed. A non-comparative interim analysis will be performed to assess variability in outcome variables. This will form the basis of an updated sample size calculation.

The internal pilot will also be used as a basis for possible minor adjustments of the study protocol. In addition, feasibility of the full scale study will be evaluated. Should it be concluded that ethical problems arise due to unexpected problems in the study process and it is unlikely that further knowledge will be gained from continuing the full study with recruitment of more housing facilities and participants, the study will be interrupted. Results and gained insights about feasibility from the interval pilot alone will be published.

Results and conclusions from the internal pilot will be presented at the trial's site at clinicaltrials.gov. Any adjustments of the protocol and the statistical analysis plan following from the evaluation of the internal pilot will be published as amendments.

3-year update: the recruitment was somewhat slower than expected, and sample size recalculation based on data from the internal pilot showed that the original target sample size would render in an expected statistical power of 53% for detecting a group difference of 5 points in ReQoL. Therefore, the investigators decided to redimension the study to detect a group difference of 10 points, which has been pointed out as a minimum clinically important difference for ReQoL by the instrument developers. Given that the expected number of recruited housing units will be 45, revised calculations based on Monte Carlo simulations showed that this should be sufficient to reach 80% for detecting a group difference of 10 points. Descriptions of the sample size calculations are available in the statistical analysis plan.