Recovery in Supported Accommodations: A Scoping Review and Synthesis of Interventions for People with Severe Mental Illness

With this review, we aim to answer the following questions:

To answer our first research question, we searched the following databases: PubMed, Psycinfo, Embase and Cinahl (January 2018, Update December 2019). These databases were chosen to cover medical (PubMed and Embase) as well as psychological (Psycinfo) and nursing (Cinahl) literature. We formulated and combined search terms concerning: (a) the setting and population (mental disorder/illness, schizophrenia, psychosis, inpatient rehabilitation, supported accommodation, sheltered housing, housing facility, community housing, community facility, supported housing, residential facility and residential care), (b) the scope and outcome of the intervention (psychosocial, societal, recovery, functioning, rehabilitation, health, wellness and cognition), and (c) study type (clinical trial, randomized controlled trial, evaluation study, experimental trial, naturalistic study, follow up study, quasi-experimental and case study).

To select studies that corresponded with our research aims, we formulated inclusion and exclusion criteria. We included peer-reviewed articles that were published in English from January 2000 till December 2019; aimed at adult clients with severe mental illness receiving services from housing services or comparable long-term (> 1 year) supported accommodation; evaluated psychosocial interventions focussing on personal, functional or societal recovery outcomes; evaluated the outcomes of an intervention on the client level; and evaluated outcomes by means of effect evaluation all types of designs except for expert opinions and case studies. As we aimed to give an overview of existing interventions for this group, we also included protocol papers and checked if there were results published already. To be able to provide a clearly defined answer to the research questions and to keep the results manageable, we also formulated exclusion criteria. Studies were excluded if they primarily focussed on substance abuse; intellectual and/or developmental disability, including brain damage; or on homelessness; or if they were executed in developing countries.

In the first and second selection phase, the first two authors each screened a separate part of the titles from the initial search, and of the remaining papers they screened the abstracts on relevance. When there was doubt, the selection was made in consensus. The first and second author determined final inclusion by discussing the interpretation of the inclusion criteria in certain cases. When doubt persisted about an abstract, the article was included so that a more careful decision could be made in the next phase.

In the third phase, the first and second author read the full-text of the remaining articles and made a final selection. In this final phase, both authors each read half of the articles independently. Again, articles about which doubt existed were discussed until consensus was reached. The selected studies then were categorised in a qualitative synthesis, based on the dimensions of recovery: societal, functional and personal, and additional in vivo categories were made when needed.

Our second aim was to evaluate what is already known about the outcomes of these interventions. Therefore, the second phase of the qualitative synthesis was evaluation of each study to understand the status of the available evidence of each intervention found. First, we formulated categories of designs based on Evans’ hierarchy of evidence (2003): randomized (controlled) study, uncontrolled longitudinal study, or other (all other designs except case studies and expert opinions). Next, we evaluated the results of relevant outcomes and (where possible) the effect sizes of these results. Again, three options were possible: Large or medium effects, small effects, or neutral, unclear, unknown or not convincing yet. Based on these criteria, we concluded there was one of three options: (a) added value when a randomized control trial (RCT) resulted in small, large or medium effects, (b) promising first results when other designs than RCTs showed positive results, or (c) no evidence for the effectiveness yet when there were neutral or negative results or no results yet. The first and second author executed this quality assessment independently. Each assessed an equal part and then discussed the results until they reached a consensus. This review is part of a larger research project which received ethical approval from the Medical Research Ethics Committee of the Elisabeth Hospital in Tilburg (NL41169.008.12).

We evaluated the outcomes of all included studies (see Table 2 for a summary). Following is a description of the overall picture for each category.

Of the nine studies that evaluated interventions aimed at goal attainment, seven interventions were totally or partly based on the ‘choose-get-keep’ model (Anthony et al. 2014; Ellison et al. 2011; Fagan-Pryor et al. 2009; Magliano et al. 2016; Pioli et al. 2006; Sanches et al. 2015; Vandevooren et al. 2007). Three of the goal attainment studies were RCTs, and four were uncontrolled/pre-post design. Five of these studies showed (small) positive results (Ellison et al. 2011; Fagan-Pryor et al. 2009; Magliano et al. 2016; Pioli et al. 2006; Vandevooren et al. 2007), among others, concerning functioning and residential status. Bitter et al. (2017) evaluated, by means of a cluster randomized trial, CARe: A rehabilitation approach based on the strengths model and personal recovery in teams of supported accommodation, but did not find any differences in outcomes between the clients of trained and untrained teams.

Of the studies on interventions concerning skills and illness/disability management, two RCT studies evaluated the illness management and recovery (IMR) approach (Beentjes et al. 2018; Levitt et al. 2009). The Levitt study reported significant improvements in illness management, symptoms and psychosocial functioning, while the Beentjes’ e-IMR study did not due to low implementation rates. Lindström et al. (2012) conducted a study on a home-based occupational therapy intervention aiming at daily occupations including remediation and compensatory strategies. The authors observed positive significant results on most outcomes (goal attainment, social interaction, satisfaction with daily occupations, activities of daily living (ADL) and psychiatric symptoms). Anzai et al. (2002) examined an RCT on an training program for illness management skills based on Liberman’s community re-entry module, resulting in positive effects including knowledge and skills and community participation. In a small, pre-post study on a short educational training course on using the internet and touch screen, no effects were found on social isolation, self-esteem and internet use (Loi et al. 2016). Three studies (Park and Han 2018; Seo et al. 2007; Tsang and Pearson 2001) examined societal recovery explicitly focussed on social skills. Tsang and Pearson (2001) evaluated social skills training in the context of vocational rehabilitation. This cluster randomized pilot found positive results for work-related social skills, motivation to seek employment and success in job search. Seo et al. (2007) conducted a quasi-experimental study on social skills group training that included conservational and assertiveness skills based on the Liberman modules. The results showed a difference in improvement of social skills and self-esteem in favour of the intervention group. Park and Han (2018) studied with a quasi-experimental design a 5-week communication program based on communication theory of Walsh and existing of ten sessions. They found improved communication, and relational skills, but no improvement in problem solving, though used an alpha of 0.70.

Two studies evaluated interventions aimed at vocational rehabilitation. Oka et al. (2004) evaluated a hybrid occupational therapy and supported employment intervention by means of a retrospective study. Positive results were achieved concerning social functioning and hospitalisation. Rogers et al. (2006) evaluated the choose-get-keep approach in a vocational context compared with enhanced state vocational rehabilitation and found no differences between the groups. A positive effect on vocational status was found for both interventions, indicating that a rehabilitation approach aiming at work can be effective for this group.

Finally, the remaining studies were concerned with client engagement in activities (Killaspy et al. 2015; Sheridan et al. 2018) and psychoeducation based on cognitive behavioral therapy (McMurran et al. 2011). Killaspy et al. (2015) evaluated a staff-training program designed to increase patients’ engagement in activities. In this cluster-randomized trial, no differences were found between the study groups in engagement in activities. Sheridan et al. (2018) studied the effects of a supported socialisation volunteer partner group to stimulate social and leisure activities. In their qualitative thematic analyses of diary data they found indications for positive effects on involvement in normalising life, connectedness, physical health, social capacity and culture engagement. McMurran et al. (2011) published on a protocol to evaluate a 12-session group intervention aimed at problem solving.

The six studies in this category evaluated interventions aimed at personal recovery (including outcomes on empowerment, hope, confidence, and quality of life or comparable). All studies showed added value or promising first results. Of these studies, one was an RCT and five were semi-controlled or pre-post designs. Two studies were peer-run interventions. One of these peer-run interventions examined confidence and care needs (Boevink et al. 2016) and the other on consumers’ perception of the recovery attitudes on the staff (Mancini et al. 2013).

One study focussed especially on elderly patients and showed a small but positive result concerning life satisfaction (Willemse et al. 2009). Park and Sung (2013) reported results of a study on a 6-week, recovery-oriented nursing intervention. This study also showed positive results on helplessness and recovery, but due to the non-controlled design, these results need further confirmation in replication studies. There were two studies on therapies to enhance personal recovery (Nowak et al. 2018; Randal et al. 2003). Randal et al. (2003) conducted a small, matched-control evaluation study on individual recovery-focused multimodal therapy. Following Evans’ design hierarchy, the results can be interpreted as promising with outcomes showing significantly more improvement of positive and negative symptoms and a decrease of deviant behavior, e.g. verbal aggression and violence. Nowak et al. did a pre-post study on a recovery-oriented cognitive behavioral workshop of 6 weeks. They found no significant change in total recovery, but they did find significant improvements in sub scales including confidence, hope and psycho social functioning.

This category included five studies evaluating interventions focused on improvement of cognitive and executive functions. Four were RCTs, and one had a pre-post design. A study on an integrative program that focused on all basic cognitive functions showed positive results concerning vocational outcomes, family contact and social competence (Sánchez et al. 2013). Lindenmayer et al. (2012) conducted an RCT on an intervention that combined cognitive remediation with social cognition training. The combined intervention resulted in greater improvements in emotion recognition, emotion discrimination, social functioning and neuro-cognition compared with cognitive remediation alone. Another study resulting in interesting results was a cognitive remediation intervention focusing on problem solving skills (Medalia et al. 2001). This study found a significant difference for independent living. Schutt et al. (2017) executed a small pre-post study on a cognitive remediation intervention, but did not find relevant outcomes. Stiekema et al. (2015) published on their protocol to evaluate a cognitive adaptation training (CAT).

We found thirteen studies focusing on lifestyle interventions; all were published after 2010. Seven were RCTs, five were semi-controlled or pre-post studies and one was qualitative. Seven of these studies showed promising first results, four did not show evidence and two were protocol papers. Loh et al. (2016) executed a (pilot) RCT on a structured walking intervention. In this study, the participants of the control group scored slightly better on quality of life, psychiatric symptoms, physical role limitations and physical functioning after 3 months. Hjorth et al. (2016) evaluated an intervention program for improving physical health in staff and its impact on patient’s health. The intervention had a positive effect on the waist circumference and blood pressure for the staff, and there was a statistically significant association between the staff change in each facility and the patients’ change in health parameters.

Looijmans et al. (2017) conducted a cluster RCT on lifestyle intervention that focused on cardio metabolic health. This intervention led to positive results after 3 months on waist circumstance and metabolic syndrome. The same research group studied the use of a web-based tool (Looijmans et al. 2019) in FACT teams and sheltered living teams. Findings indicate no significant improvements on the primary and secondary outcomes and an improvement on the readiness to change. Oertel-Knöchel et al. (2014) conducted a combined cognitive–aerobic/relaxation intervention showing that physical exercise is a valuable addition to cognitive training. Verhaeghe et al. (2013), Cabassa et al. (2015), Forsberg et al. (2010) and O’Hara et al. (2017) also all studied a lifestyle program. Verhaeghe et al. conducted a cluster RCT on a comprehensive lifestyle intervention (psycho-education, supervised exercise and individual support) in sheltered housing services. Although initially small positive results were achieved on weight, body mass index (BMI) and waist circumstances, these results almost all disappeared during follow-up. No differences were found regarding secondary outcomes (i.e., symptoms and quality of life). Cabassa et al. published a study protocol. Forsberg et al. did not find support for the added value. O’Hara et al. studied the results of a peer based group and did this qualitatively, using focus groups and field notes, and added this with structured interviews. The results indicate participants attended on average three quarter of the sessions and perceived them as helpful and satisfactory. Ringen et al. (2018), Hutchison et al. (2016) and Gill et al. (2016) all executed pre-post evaluations on a promotion / motivational program, of which the first did not show improvements and the latter two resulted in positive results on physical activity and physical health. Sweeney and Baker, finally, published two protocol papers on an intervention in which existing peer workers tailor clients to appropriate smoking quitline service (Sweeney et al. 2019).

This category contained seven studies. Two studies (one protocol) evaluated Tai chi (Ho et al. 2012, 2014) of which a pilot RCT showed promising results concerning movement and interpersonal functioning. Three studies (Gold et al. 2013; Hayashi et al. 2002; Kwon et al. 2013) evaluated a form of music therapy. In all studies, positive results were achieved concerning amount others: Negative symptoms (Gold et al. 2013), cognitive function (Hayashi et al. 2002), positive behavior (Kwon et al. 2013), and quality of life (Hayashi et al. 2002). These positive results, however, did not last through the last follow-up.

One study in this category evaluated the effect of watching humorous movies. Watching these movies regularly for 3 months appeared to have a small positive effect on negative symptoms, depression and anxiety, and social competence (Gelkopf et al. 2006). The seventh study was a cluster trial on a ward intervention to improve patient-staff relationships and wellbeing leading to significant differences in depersonalization in staff and criticism experienced by clients (Berry et al. 2016).

This study has several strengths and limitations. A strength was the broad scope. Our aim was to provide an impression of psychosocial interventions that exist for people with SMI who need supported accommodation and to provide first insights into what is known about the effectiveness of these interventions. Therefore, we used a broad search strategy and included a variety of interventions aiming at a broad range of outcomes and executed in different settings and (international) contexts, and included all types of study designs. When developments in the recovery field are a bit further along, a quantitative synthesis would add to our knowledge. At this point the number of studies in the supported accommodation field is too small yet to perform a quantitative review (Chilvers et al. 2006; McPherson et al. 2018). A point of attention is that we used information provided in the included articles only, which sometimes was somewhat poor, for example not all papers published effect sizes. So, it might be that the quality of some papers is displayed more positively if it was based on the p values only. Another note is that when performing a review, a selection of specific search terms is chosen. There is always a risk that not all relevant papers end up in the results due to word use in titles, abstracts and key words. When reading this and other reviews, this should be kept in mind. Nevertheless, this study provides a broad overview of interventions on several dimensions of recovery besides the clinical one that can give supported accommodation an impression of interventions that may be relevant and sufficient to implement, and bring the recovery forward, even in people that cope with severe, and therefore often complex and long-term, mental illnesses.

Research specifically focussing on the recovery dimensions besides clinical recovery of people with severe mental illness who live in supported accommodations remains limited but seems to be in development. We also can conclude that a broader vision towards recovery in these settings has gained attention and that, regarding all other dimensions of recovery, hopeful results have been achieved so far.

Four challenges can be appointed concerning the practice and research of interventions for people with severe mental illness who live in supported accommodations. The first challenge is the further development and professionalization of recovery-oriented care and support offered for this specific group of people. Effective and promising interventions should be developed and made available for all people with severe mental illness, despite their place in the care landscape (Couwenbergh and van Weeghel 2014).

The second challenge is to accompany developments in practice with research to gain more insight into what works, for whom and what does not, so that the provided care can be more personalized. Specific knowledge is needed concerning the group of people who are in need of supported accommodation. For example, we were surprised that for some well-known recovery interventions, for example, the wellness recovery action plan (WRAP) (Fukui et al. 2011) or narrative enhancement and cognitive therapy (NECT) (Fukui et al. 2011), no studies were found explicitly focussing on people living in supported accommodation. Here may lay a chance for further development, as it is worthwhile to study interventions that have proved themselves in ambulant contexts to see if they also can help clients with more complex and supported living needs.

The third challenge is the integration of different approaches towards recovery. In several countries, different forms of support are fragmentized (Boevink et al. 2016). For example, in the Netherlands a separation exists between clinical mental health care services and supported accommodation services. The insight is growing that integration of different aspects of recovery may lead to better outcomes (Corrigan et al. 2012). This might lead to improvement of recovery orientation of the care for people living in supported accommodation. Altogether, it is recommended that supported accommodation services reconsider their scope and position in the care landscape and consider broadening and strengthening their recovery-oriented services, as well as stronger collaborations between stakeholders including mental health treatment providers, supported housing organisations and local organizations for community support.

The fourth challenge is the professionals’ interest, knowledge and implementation skills to adapt and use state-of-the-art interventions. Working evidence based asks for an innovative mind set as well as time and support in keeping up-to-date and using new interventions that were proven effective in research.