The Experiences of Adults Experiencing Homelessness When Accessing and Using Psychosocial Interventions: A Systematic Review and Qualitative Evidence Synthesis

Sample

There are a number of definitions of homelessness available, reflecting differences between countries and over time. There are also different forms of homelessness, taking into account the length of time someone has been experiencing homelessness, distinctions between living on the street or in their vehicles, or having a temporary place to stay.

During the scoping work to develop the protocol for this study (O'Leary et al. 2022), we facilitated a workshop with five individuals with lived experience of homelessness to consider definitions, criteria for inclusion and exclusion, and the process of conducting this review. Following recommendations from those involved in this workshop, we used an adapted and widened definition of homelessness developed by Keenan et al. (2021). The revised definition for this review is:

Homelessness is defined as those individuals who are in inadequate accommodation (environments which are unhygienic and/or overcrowded), who are sleeping rough (sometimes defined as street homeless or unsheltered), those in temporary accommodation (such as shelters and hostels), those in insecure accommodation (such as those facing eviction or in abusive or unsafe environments), and people whose accommodation is inappropriate (such as those living in tents or vehicles, or ‘sofa surfing’).

The definition and inclusion criteria for homelessness shared great parity with that of the EGM, that is, including both the literal forms of homelessness and those at risk of homelessness, as such, it is unlikely to have resulted in significant divergences in search strategies/results. Our focus was on adults (men and women aged 18 years and over), undertaken in any high-income country and published in English. Studies of families or children were excluded from the review, as in many countries (particularly the United Kingdom), there are different legal frameworks that apply to families and children experiencing homelessness, and thereby their access to different types of services, and different outcomes expected.

Phenomenon of Interest

The review focused on formal psychosocial interventions used with adults experiencing homelessness. Interventions based solely or mainly on pharmacological approaches or approaches other than psychosocial were excluded. In developing the protocol for this review (O'Leary et al. 2022), we set out our definition of psychosocial interventions and a typology of interventions that we considered to meet this definition. This typology is presented again in the Appendix.

Design

Eligible studies were those that used individual and group interviews, focus groups, observations, or other qualitative-related methods focused on the experiences, views, or opinions of adults experiencing homelessness.

Evaluation

We included studies where empirical data presenting the experiences, views, perspectives or opinions of people who are experiencing homelessness or at risk of homelessness when accessing or using psychosocial interventions, and are directly presented either as direct quotes or summaries, or as reports of participant experiences by researchers.

Research Type

Eligible studies included data reported either as part of a mixed methods or qualitative study about the use of psychosocial interventions focused on people who are experiencing homelessness.

The Homelessness Implementation Studies EGM

The main source of studies included in this review was from the Implementation EGM third edition, published by The Campbell Collaboration (Singh and White 2022). The searches for this updated EGM were conducted by a team from the Campbell Collaboration between March and September 2021. The EGM included 401 primary studies and systematic reviews. The process for undertaking these searches (including search strings) is set out in the published protocol for the Implementation EGM (Singh et al. 2023). This states that the review included a broad range of qualitative studies, mixed method studies, and quantitative studies that reported on implementation issues. The team reported that they searched several databases, websites and registries, conducted backward citation tracking of included studies, contacted researchers and hand-searched a number of journals. The team also reported that, because many relevant evaluations are unlikely to be published in academic journals, an extensive grey literature search was undertaken. The protocol includes details of the websites searched for grey literature.

Two reviewers independently undertook a title and abstract search of each of the studies listed in the EGM. Each title and abstract was reviewed against the inclusion and exclusion criteria for this review. The two reviewers then compared their assessments. One systematic review was identified through this process, and this was unpacked to identify additional primary studies. These were also subject to title and abstract reviews by two reviewers independently. Studies that appeared to meet the inclusion criteria, or where there was disagreement between the reviewers, were subjected to a full-text review, which was again undertaken independently by two reviewers. Any disagreements at this stage were referred to a third reviewer for decision.

Appraisal of the Methodological Limitations of the Included Studies

Eligible studies that are included in the Implementation EGM have already been assessed using Campbell's Critical Appraisal Tool for Primary Studies (White and Narayanan 2021, 60). The appraisal considers the following questions: 1.

Are the research questions clearly stated?

This assessment was undertaken by the Campbell team in their development of the EGM. We did not undertake any further assessment of the methodological limitations of those included studies that were identified from the EGM. For any additional studies identified through either the hand searches of relevant journals, the call for grey literature, or through unpacking of systematic reviews, we conducted an assessment of confidence in the findings using this same tool. For these additional studies, classifications were undertaken by one researcher and judgements (high/medium/low confidence) were verified by a second researcher. These quality assessments were not used as a criteria in our decisions around inclusion of studies in this review. The individual study-level quality assessments are reported in Figure 1.

OPEN IN VIEWER

Figure 1

Individual study-level quality assessment.

Analysis

There are a number of different methods available for undertaking qualitative evidence syntheses. The review team used thematic synthesis (Thomas and Harden 2008) for a number of reasons. First, this method was deemed appropriate because of the focus of the review on the experiences (Carroll 2017) of people experiencing homelessness as they access and use psychological interventions, although the review team recognises that a number of methods would be appropriate for this objective. More importantly, thematic synthesis seemed the most appropriate approach because of its perceived neutrality in terms of the epistemology of underlying studies (Booth 2017), and because it is appropriate given that the review is aimed at policy makers, practitioners, and those involved in designing interventions (Barnett-Page and Thomas 2009). Finally, for practical reasons, because of time and available resources and the expertise of the review team, thematic synthesis was more appropriate than other methods such as meta-ethnography (Booth 2017).

Thematic synthesis (Thomas and Harden 2008) involves three analytic and inductive steps. The first step involves line-by-line descriptive coding of the data from the findings of the primary studies. This involves examining each sentence individually, and giving it one or more short descriptive codes (McMahon et al. 2022). This coding was undertaken by two reviewers, independently. Each reviewer distinguished between direct quotes from study participants with lived experience of homelessness and study authors' interpretation of participants' views, perceptions, and experiences. Once each study had been coded line by line, the two reviewers met to compare descriptive codes and coded material, discuss any differences, and agree on an overall coding framework (presented in the Appendix).

The second stage of analysis involved identifying possible relationships between the descriptive codes, and thus organising the descriptive codes into a series of descriptive themes. These themes describe the experiences of participants as they engage in psychosocial interventions. The reviewers independently developed descriptive themes; the lead reviewer developed 14 and the second reviewer developed 11 descriptive themes. The reviewers met several times to discuss the relationships that were identified and the development of the descriptive themes, using mind maps and other visual methods, reviewing these against the material coded under each of the 1st stage descriptive codes, until agreement was reached. (There was good agreement between the reviewers in most cases; we have highlighted in the analysis section where major differences arose between the reviewers and where reaching agreement was a more iterative process.) Each of the resulting descriptive themes was then presented, with illustrative extracts from the coded material, at a workshop with a panel of individuals with lived experience of homelessness, and changes were made when appropriate. Through this iterative and inclusive process, a final agreement was reached on the descriptive themes presented in this review.

The third step in thematic synthesis is ‘the most difficult to describe and, potentially, the most controversial’ (Thomas and Harden 2008), and is the development of analytical themes. Whereas the first two stages outlined above are intended to describe and summarise data extracted from the included studies, this third stage of analysis ‘goes beyond’ (Thomas and Harden 2008) the primary studies to directly address the questions set out for the review. However, stages 2 and 3 also overlap with each other, as often review questions can be answered during stage 2 analysis (McMahon et al. 2022; Thomas et al. 2017). It involved iteratively examining the descriptive themes, drawing out and inferring from these themes the experiences of participants, their perceptions of whether and how the intervention worked for them, and whether their experiences varied by type of intervention, location, and context. The analytical themes were also discussed and evaluated in a workshop held with a panel of people with first-hand experience of homelessness, held in early 2023.

Experiences

The core focus of this review was to understand the experiences of, views about, perspectives on, and opinions of adults experiencing homelessness when they (a) access and (b) use psychosocial interventions, and what factors they perceive are important to the effectiveness of psychosocial interventions. The evidence set out in the background section of this review clearly demonstrates that adults experiencing homelessness have significantly higher levels of mental ill-health, problematic substance use, and housing instability compared to the general population, and are also less likely to be able to access services (particularly general healthcare) to reduce the impact of, and negative effects from, these experiences. By focusing this review on the experiences of adults experiencing homelessness, the review aims to elevate the voices of people experiencing homelessness and thereby address a fundamental limitation of many systematic reviews – that of treating service users as passive research participants rather than as citizens who may actively participate in their own recovery.

Reflexivity Statement

The two researchers involved in screening and eligibility decisions both have a background in homelessness research, including evidence synthesis and primary research that has involved working with people with first-hand experience of homelessness. Both researchers have experience working in front-line services aimed at, and accessed by, individuals with first-hand experience of homelessness, including individuals experiencing homelessness at the time.

One of the two researchers involved in screening and eligibility decisions was also involved in data extraction and analysis. The other researcher has a background in social work, both academic and practice-based, and considerable experience in evidence synthesis.

None of the researchers involved have had first-hand experience of homelessness. All researchers are committed to involving individuals with first-hand experience in the design of research and using research to elevate the voice of individuals with first-hand experience of homelessness.

Participant Characteristics

The included studies cover 84 participants with experience of homelessness, of which 81 were accessing or using psychosocial interventions, and 3 in 1 study (Lynch et al. 2017) were accessing a comparator intervention. (For six of the included studies, the number of participants was clearly and explicitly stated. Identifying for number of participants in one study – Santa Maria et al. (2020) – was less straightforward. This was a mixed methods study and stated that the number of participants was n = 39. However, the data extracted from this study related to the qualitative findings only, and we determined that this part of the study involved 10 participants.)

Two studies involved 26 women participants (Kahan et al. 2020; Salem et al. 2013), of whom 12 were female-identified survivors of domestic violence aged under 25 (Kahan et al. 2020) and 14 were women who had been in prison (Salem et al. 2013). One study focused on men experiencing homelessness, covering 10 participants (Rayburn and Wright 2009). Two studies focused on adults (both men and women). Lamanna et al. (2018) included 22 participants experiencing homelessness as they were discharged from hospital, and Parkes et al. (2021) 10 participants with substance use issues. Table 3 provides a breakdown of the number of participants by sex for each study.

Two studies focused on young people experiencing homelessness. Kahan et al. (2020) on 12 young women aged 16–24 escaping domestic violence, and Santa Maria et al. (2020) on 10 participants aged 18–21.

Two of the included studies provided details of the participants' race. Rayburn and Wright (2009) state that of the 10 participants, 8 were Black, 1 Hispanic and 1 White. Salem et al. (2013) state that of the 14 participants, African American (79%), followed by White (14%) and Hispanic/Latino (7%). The other five include studies do not provide information on the race of participants.

Geographical Location of Studies

Three studies took place in Canada (Kahan et al. 2020; Lamanna et al. 2018; Lynch et al. 2017), all in Toronto (Lynch et al. (2017) included shelters in Toronto and Calgary). Three studies took place in the United States: in Florida (Rayburn and Wright 2009); California (Salem et al. 2013); and Texas (Santa Maria et al. 2020). One study took place in Edinburgh, Scotland (Parkes et al. 2021).

Intervention Characteristics

Interventions covered included brief intervention (Lamanna et al. 2018), motivational interviewing (Lynch et al. 2017), 12-step programme (Rayburn and Wright 2009), and residential rehabilitation (Salem et al. 2013). Kahan et al. (2020) and Parkes et al. (2021) both described the interventions as group psychosocial interventions, and Santa Maria et al. (2020) as a mindfulness-based intervention.

Study Designs

Five of the studies were qualitative (Kahan et al. 2020; Lynch et al. 2017; Parkes et al. 2021; Rayburn and Wright 2009; Salem et al. 2013). One study was described as being a subset of a larger mixed methods study (Lamanna et al. 2018); and one study was described as a mixed methods feasibility study (Santa Maria et al. 2020).

All but two studies (Rayburn and Wright 2009; Santa Maria et al. 2020) stated the analytical method used to generate findings. Thematic analysis was used in three Canadian studies (Kahan et al. 2020; Lamanna et al. 2018; Lynch et al. 2017). Parkes et al. (2021) stated that framework analysis was used, and content analysis was used by Salem et al. (2013).

Six studies used interviews as the data collection method, of which three (Kahan et al. 2020; Lynch et al. 2017; Rayburn and Wright 2009) used this as the only method and three used it in conjunction with other approaches. Lamanna et al. (2018) used interviews and focus groups, Parkes et al. (2021) used interviews and analysis of policy/service documents, and Santa Maria et al. (2020) utilised interviews and baseline surveys. One study (Salem et al. 2013) used focus groups as the only method for collecting data.

Use of Theory in the Included Studies

The fourth research aim was to describe what explicit theories of change underpinned the psychosocial interventions being evaluated in the included studies. This objective was an important part of the review process because: (1) of the lack of any single agreed definition of what a psychosocial intervention is; (2) the lack of a single explanation of how these interventions work; and (3) policy makers and practitioners need to understand how these interventions work when they design, implement, and deliver psychosocial interventions.

For each of the seven included studies, we examined whether and how the authors explained how the intervention might work. Each study was coded by the lead reviewer. This coding covered: whether an explanation was provided for how the intervention would work; whether a justification was provided for the intervention; and whether and how this related to the three meta-theories identified in the ‘how the intervention might work’ section of this report. Three relevant theories were identified in the protocol for this review (O'Leary et al. 2022) and were: 1.

Interpersonal Relationships – relationships as a driver of homelessness;

None of the included studies reference any of the three meta-theories identified in the protocol for this study.

Of the seven included studies, two include an explicit theory of change detailing the intervention design and how it is intended to work. Lynch et al. (2017) evaluate a multi-component intervention aimed at young people experiencing street homelessness, which combined resilience-based approaches and motivational interviewing. The authors describe the intervention and provide details of the two theoretical models underpinning its design, the Seven Cs Model of Resilience (Ginsburg and Jablow 2006, as cited in Lynch et al. 2017) and motivational interviewing (Miller and Rollnick 2002, as cited in Lynch et al. 2017). The authors state that: ‘The intervention was designed to enhance the capacity of street-involved youth by encouraging their engagement in health-promoting relationships, reducing risk-oriented behaviors, and promoting the acquisition of skills and healthy behaviors to support transition to young adulthood’. Santa Maria et al. (2020) provide a more detailed insight into the theory of change underpinning the intervention, a mindfulness-based intervention for young people experiencing homelessness. The authors reference two theoretical approaches, risk amplification model (Meyer 2003, as cited in Santa Maria et al. 2020) and minority stress model (Whitbeck et al. 1999, as cited in Santa Maria et al. 2020). The authors provided detailed justification for the use of these models, and illustrate the key components of the intervention and the intended outcomes on page 264.

Three of the included studies reference previous research to justify the design and intent of the evaluated intervention. Kahan et al. (2020) provide details on the design of the intervention, locating this design in trauma-informed practices, and providing brief details on why these principles are used. Lamanna et al. (2018) provide a detailed explanation of the level of need within their population of interest (in this case, adults experiencing homelessness as they are discharged from hospital), and describe the structure of the evaluated intervention. The study references previous research by several of the study authors on brief interventions. Rayburn and Wright (2009) provide a detailed literature review around 12-step programmes (in this case, specifically Alcoholics Anonymous) in justification of the design of the evaluated intervention. The authors also provide details of the specific structure and components of the First Steps programme that is the focus of their research.

Two of the included studies did not provide any justification for the design or intent of the evaluated intervention. In a study of service change because of Covid-19 at a homelessness centre in Edinburgh, Parkes et al. (2021) provide details of the services provided, and states that the centre adopts a harm reduction approach and includes a ‘psychosocial’ programme, though do not reference the extant literature or provide any details to justify these approaches. The authors also state that their research was informed by two theoretical approaches, namely psychologically informed environments (PIEs) and enabling environments. Salem et al. (2013), in an evaluation of a residential rehabilitation programme targeting women experiencing homelessness who had been in prison, provide details of the model underpinning the design and conduct of the research. The authors do not provide any insight to justify the intervention design.

Finding 1: An Individual's Goals, Motivation, and Sense of Hope Are Important to Enabling Change

This finding is developed in all seven of the included studies, and talks directly to the review's second research question about what makes for effective interventions. The finding recognises that experiencing homelessness can have a negative impact on an individual's sense of self worth and sense of hope, and that motivation is an important ingredient in an individual's change journey.

Finding 2: An Individual's Negative Thoughts, Feelings, and Behaviours Can Adversely Impact Their Change Journey

This finding speaks to the role that negative feelings and thoughts can play, of powerlessness and being out of control, and how this can adversely affect an individual's change journey. It is about the role that the intervention – and particularly relationships with peers and staff during the intervention – can play in challenging negative thoughts and feelings, and augmenting self-awareness. This finding was developed from six studies (all but Salem et al. 2013), although none of the study-level descriptive themes relates to negative feelings, thoughts, and behaviours.

Finding 3: Recovery Is Not Linear

This finding speaks to recovery or the process of change as a journey, with many bumps and different turns along the route. In particular, an individual's change journey is not a straight line of progression from homelessness to being stably housed, and it is important the both the intervention design and how it is delivered recognises this. During the workshop with a panel of individuals with first-hand experience of homelessness to discuss the descriptive and analytical themes, one panel member used the analogy of ‘a game of snakes and ladders’ to summarise this descriptive theme. In this game, an individual's progress from start to finish often involves episodes of rapid positive progress (ladders) and also the reverse when the individual lands on a snake. In both cases, this is outside the control of the individual player. This finding develops from four of the included studies (Lamanna et al. 2018; Lynch et al. 2017; Parkes et al. 2021; Salem et al. 2013) and addresses the first of the review's research questions.

Finding 4: People Experiencing Homelessness Face Several Challenges When Accessing and Using Mainstream Services

Participants in four of the included studies (Lamanna et al. 2018; Lynch et al. 2017; Parkes et al. 2021; Salem et al. 2013) discuss the challenges they faced when trying to access mainstream services, and the consequences they experienced because of these challenges. These challenges arise both because of lack of signposting to, or knowledge of services and other interventions, but also because of the discrimination and stigma they experience when trying to access services. Although this finding does not directly relate to the first research question, which is specifically about access to psychosocial interventions, it nevertheless does speak to issues faced by people experiencing homelessness more generally about accessing services.

Finding 5: A Non-Judgemental Approach Is an Important Part of Accessibility

This finding addresses the second research question, as it speaks to what participants perceive to be important to making the intervention work for them. This was found in each of the seven included studies. Alongside the physical aspects, accessibility – such as its location, layout, and available facilities – participants identified that it was important to have a warm and inviting approach in which relationships between staff and service users were based on a non-judgemental, genuine and open approach.

Finding 6: Relationships and Trust Are Important Ingredients for Successful Psychosocial Interventions

The importance of relationships, with peers and support workers, and the role that trust plays in these relationships, is found in six of the seven included studies. (It is not found in Rayburn and Wright [2009].) Throughout the included studies, relationships and trust are seen as an important ingredient in how the intervention might work, and therefore addresses the second research question for this review. References to relationships and trust are also found in study-level descriptive themes, identified by study authors through their analysis.

The review team was able to address two of the four research questions raised through this analysis, namely: (1) what are the experiences of study participants when accessing or using psychosocial interventions; and (2) whether and how adults experiencing homelessness perceive the interventions that work for them? The small number of included studies, the limited number of interventions covered by the included studies (and the lack of more than one study covering each intervention), and the fact that six of the seven studies were undertaken in North America, meant that we could not adequately address the third research question. The fourth research question is discussed in the section ‘use of theory in included studies’.