Trajectories of recovery among homeless adults with mental illness who participated in a randomised controlled trial of Housing First: a longitudinal, narrative analysis.

OBJECTIVES: This study used longitudinal, narrative data to identify trajectories of recovery among homeless adults with mental illness alongside the factors that contribute to positive, negative, mixed or neutral trajectories over time. We expected that participants who received Housing First (HF) would describe more positive trajectories of recovery than those who were assigned to Treatment as Usual (TAU; no housing or support provided through the study).
DESIGN: Narrative interview data were collected from participants at baseline and 18 months after random assignment to HF or TAU.
SETTING: Participants were sampled from the community in Vancouver, British Columbia. PARTICIPANTS: Fifty-four participants were randomly and purposively selected from the larger trial; 52 were interviewed at baseline and 43 were reinterviewed 18 months after randomisation.
METHOD: Semistructured interviews were conducted at both time points. For each participant, paired baseline and follow-up narratives were classified as positive, negative, mixed or neutral trajectories of recovery, and thematic analysis was used to identify the factors underlying different trajectories.
RESULTS: Participants assigned to HF (n=28) were generally classified as positive or mixed trajectories; those assigned to TAU (n=15) were generally classified as neutral or negative trajectories. Positive trajectories were characterised by a range of benefits associated with good-quality, stable housing (eg, reduced substance use, greater social support), positive expressions of identity and the willingness to self-reflect. Negative, neutral and mixed trajectories were characterised by hopelessness ('things will never get better') related to continued hardship (eg, eviction, substance use problems), perceived failures and loss.
CONCLUSIONS: HF is associated with positive trajectories of recovery among homeless adults with mental illness. Those who did not receive housing or support continued to struggle across a wide range of life domains. Findings are discussed with implications for addressing services and broader social change in order to benefit this marginalised population.

RCT Entities:   Population Interventions Outcomes

Trajectories of recovery among homeless adults with mental illness over 18 months of participation in a randomised controlled trial where participants received Housing First (HF) with intensive support services or treatment as usual (TAU; no housing or supports through the study).

Narrative interviews (n=43) were conducted at baseline and at 18 months follow-up and were classified as positive, negative, mixed or neutral trajectories of recovery.

Thematic analysis was used to examine the factors that contribute to different trajectories of recovery.

Narratives from participants assigned to HF were predominantly classified as positive trajectories; no HF participants’ narratives were classified as negative trajectories. Narratives from participants assigned to TAU were predominantly classified as negative or neutral trajectories.

Positive trajectories were characterised by the ‘ontological security’ of obtaining good-quality housing, and included reduced substance use, efforts to expand social supports, positive expressions of identity and the willingness to self-reflect.

Negative, neutral and mixed trajectories were characterised by hopelessness related to continued hardship including recent eviction, substance use problems, perceived failures, loss and social isolation.

This study is the first to use longitudinal, narrative data from adults with mental illness who were randomly assigned to HF or TAU. Additional strengths include specific measures to enhance trustworthiness of the data.

Despite using multiple, independent coders, classifying trajectories among such a marginalised population is challenging given the volatility in people's lives.

Introduction

The co-occurrence of homelessness and mental illness is associated with a range of health, social and systemic challenges that make goals of residential stability and recovery difficult.1–3 As a result, many homeless individuals with mental illness appear to be caught in a ‘revolving door’ of institutional care, shelter use, substandard accommodation and living on the streets.4–6 In an effort to interrupt this cycle, Housing First (HF) was developed to reach the ‘hardest to house.’3 HF provides immediate access to independent, market-lease apartments without requirements around sobriety or engagement in treatment. HF has also been offered in congregate settings where all or a majority of tenants are programme participants.7 In both settings, HF participants are provided access to treatment and services, but choose their level of participation. A growing body of research on HF has documented positive outcomes including residential stability,8 9 quality of life,10 community integration11 and client satisfaction.12

Over the past 20 years, qualitative studies have examined homeless people's subjective experience of life on the streets and in shelters,13 mental illness,14 substance use,15 16 various health and social services17 18 and different types of housing.8 19 20 However, few qualitative studies have followed homeless adults over time, before and after receiving supported housing20 or have examined what factors influence different trajectories of recovery.21 Prior research on supported housing has emphasised the importance of personal and external resources resulting from safe and secure housing20 and the challenges around social isolation, substance use and stigma.14 19 Several studies have reported that substance use disorders are predictive of lower housing stability regardless of residence type.22 23 For the most part, prior research on trajectories among a variety of marginalised populations has followed variable-centred (eg, cluster analysis and other quantitative techniques) rather than person-centred strategies, thereby losing the focus on how multiple events, conditions and experiences contribute to an individual's perception and experience.24 Moreover, few qualitative studies have been conducted alongside randomised controlled trials.25 26

The Vancouver At Home Study is part of a multisite, mixed-methods randomised controlled trial to examine the effectiveness of HF interventions compared with existing services (Treatment as Usual; TAU) among homeless adults with mental illness.26–28 The current study focuses on narrative data from the Vancouver site, which includes a high proportion of participants who met criteria for substance dependence and is the only site that implemented HF in a congregate setting as well as in independent apartments. The current study uses thematic analysis to examine participant narratives before and 18 months after random assignment to HF or TAU (no housing or supports provided by the study) and seeks to identify positive, negative, mixed or neutral trajectories of recovery and the factors related to different trajectories.

Methods

Participants and sampling

Eligibility criteria included legal adult status (over 19 years of age), presence of a current mental disorder based on a semistructured interview, and being absolutely homeless or precariously housed.27 Participants were recruited through referral from a variety of agencies that serve the homeless. All participants met with a trained research interviewer who explained procedures, confirmed study eligibility and obtained informed consent. Eligible participants completed a series of baseline questionnaires and were differentiated into two groups, High Need (HN) or Moderate Need (MN), based on the complexity and severity of their needs.27 HN participants were randomised to one of the three study arms: (1) HF (independent apartments) with Assertive Community Treatment; (2) Congregate Housing with on-site support (CONG) and (3) HNTAU, which provided no additional housing or support services beyond what was available in the community. MN participants were randomised to one of two study arms: (1) HF (independent apartments) with Intensive Case Management; and (2) MNTAU. Participants in the present study were both randomly and purposively selected from the larger study sample, in an effort to represent differences across gender, ethnicity, duration of homelessness and degree of functional impairment. Within 1 month of enrolment in the larger study, selected participants were asked to participate in a ‘personal story interview.’ Participation was voluntary, and 2 of the 54 participants declined. The current study is based on narrative interviews from 43 participants who were interviewed within 1 month of recruitment and reinterviewed 18 months later. Reasons for loss to follow-up include: death (1), refused participation (1), incarcerated (2), moved out of town (1) and unable to locate (4). Baseline interviews (n=52) included 32 HF participants and 20 TAU participants; follow-up interviews (n=43) included 28 HF participants and 15 TAU participants.

Data collection

Four research assistants and one peer interviewer conducted the narrative interviews. Interviews lasted from 1–2 h and were conducted at a setting chosen by the participant, usually a community agency or the project field office. All participants gave informed consent and received $30 upon completion of each interview.

Using a semistructured interview format, participants worked with interviewers at baseline to co-construct a personal story highlighting (1) their pathway into homelessness; (2) experiences of being homeless or inadequately housed; (3) experiences around first learning that they had a mental illness and obtaining help for their illness and (4) key life events. The 18-month follow-up interview focused on changes since the first interview in the areas of typical day, housing, service use, experience of community, social ties, hopes for the future and key life events. Interviews were audio recorded and transcribed verbatim.

Data analysis

Each participant's baseline and 18-month follow-up interviews were compared and the overall trajectory was categorised as ‘positive,’ ‘negative,’ ‘mixed’ or ‘neutral.’ Categorisations were based on 22 domains including housing stability, typical day, mental and physical health, substance use, criminal justice activity, social interactions, hopes for the future, willingness to introspect and interviewer observations. Each transcript was coded for positive, negative, mixed or no change on each domain. If the number of positive domains clearly outnumbered the negative, the trajectory was coded as ‘positive.’ Conversely, if the number of negative domains clearly outnumbered the positive, the trajectory was coded as ‘negative.’ If the number of negative and positive domains was roughly equal, the trajectory was categorised as ‘mixed;’ and if no clear change was observed over time in the majority of domains, the trajectory was categorised as ‘neutral.’ Although we structured coding by using 22 domains, scientific/clinical judgement and discussion among coders and interviewers was a key part of the analytic process.24 Approximately three quarters of the interviews clearly fell into one category, while the remainder were more complex and required collaborative consensus.

Thematic analysis was used to examine the interview transcripts in relation to factors that contribute to positive, negative or neutral trajectories. Our approach reflects ideas brought to the data set from the research questions and existing literature (ie, deductive), as well as themes that emerge from the data (ie, inductive).29 30According to this approach, emphasis is primarily on the content of the text (what is said) rather than on structural or discourse analysis (how it is said); however, we also considered the emotional tone of the interviews and field notes.

The interviewing team met four times during the early phase of thematic analysis to co-code and discuss emergent themes in the narratives. Initial codes and themes were based on interview questions; for example, the question ‘How have your relationships with people changed?’ elicited the code ‘developing trust’ and a preliminary theme of ‘wanting deeper social connections vs isolating.’ Subsequently, the first author independently coded all transcripts line-by-line, classified trajectories and identified repeated or similar codes to build a set of overarching themes;29 30 the second and third authors followed the same process for half of the transcripts. Thus, all transcripts were independently coded by at least two people. After a thorough review of the transcripts and field notes, conceptual impressions were integrated into key thematic areas. At this point, themes and initial interpretations were shared with field interviewers and any coding differences were resolved by group consensus. Five audio files from the baseline and follow-up interviews were reviewed by an external researcher who was familiar with the interview guide. The audio files were checked against their respective transcripts for accuracy and quality.

Results

Sample characteristics

Demographic characteristics at baseline for participants who completed both narrative interviews (baseline and 18 months) are presented in table 1. No significant differences were observed among baseline demographic characteristics for the baseline and the follow-up samples. As noted in table 1, the age of the follow-up sample ranged from 21 to 66 years (mean=43 years) and included 25 men (58%), 16 women (37%) and 2 (5%) transgendered individuals. At baseline, the mean lifetime duration of homelessness was 6 years and 30% had completed high school. The most commonly identified mental disorders among the sample were Psychotic Disorder (49%), Major Depressive Episode (49%) and Substance Dependence (67%).

Table 1

Demographic characteristics at baseline for the original baseline sample (n=52) and those reinterviewed after 18 months (n=43)

	Baseline		18-month follow-up
	Mean (years)		Mean (years)
Age (range, median)	42 (21–66, 42.5)		43 (21–66, 42.1)
Lifetime duration homeless (range, median)	5.5 (0.2–33, 3)		5.9 (0.2–33, 4)
	n	Per cent	n	Per cent
Gender
Male	28	54	25	58
Female	21	40	16	37
Transgender	3	6	2	5
Race/ethnicity
White	31	60	26	61
Aboriginal	12	23	10	23
Mixed/other	9	18	7	16
Housing status
Absolutely homeless	42	81	35	81
Precariously housed	10	19	8	19
Marital status
Single, never married	31	60	27	63
Divorced/separated/widowed	18	35	15	35
Married/common law/other	3	6	1	2
Have children under 18 years	17	33	15	35
Education
Grade 8 or less	10	20	10	23
Incomplete high school	21	40	20	47
Completed high school	21	40	13	30
Mental disorders
Psychotic disorder	25	48	21	49
Major depressive episode	24	46	21	49
Mood disorder with psychotic features	15	29	14	33
Post-traumatic stress disorder	15	29	12	28
Panic disorder	14	27	10	23
Manic or hypomanic episode	10	19	8	19
Substance dependence	33	63	29	67
Alcohol dependence	11	21	10	23

Classification of trajectories

Narratives from HF participants (n=28) were predominantly classified as positive (n=17) or mixed (n=9) trajectories over the 18-month study period; two narratives were classified as neutral and none were negative (see table 2). Narratives from TAU participants (n=15) were typically classified as negative (n=5) or neutral (n=5) trajectories, with the remaining trajectories divided between the positive and mixed categories. Of note, only two TAU participants obtained good-quality housing during the study, and both participants’ narratives were classified as positive.

Table 2

Recovery trajectories over 18 months by study arm

	Trajectory type
Study arm	Positive	Negative	Mixed	Neutral
ACT (n=9)	6	0	3	0
CONG (n=10)	5*	0	3†	2‡
HNTAU (n=7)	0	2	1	4
ICM (n=9)	6	0	3	0
MNTAU (n=8)	3	3	1	1
HF total (n=28)	17	0	9	2
TAU total (n=15)	3	5	2	5

*Two participants left the CONG residence within the first year and were living in supported housing elsewhere.

†Two of these participants left the CONG residence within the first year, one of whom was living in supported housing elsewhere and one of whom was living in an single room occupancy hotel (SRO).

‡One of these participants left the CONG residence within the first year and was living in an SRO.

ACT, Assertive Community Treatment; HF, Housing First; HN, High Need; MN, Moderate Need; TAU, Treatment as Usual.

Key themes: positive trajectories

Two related but distinct themes primarily contributed to the positive trajectories: (1) housing as a stable foundation for change across a variety of domains and (2) the expression of positive aspects of one's identity.

Theme 1: Housing as a secure and stable foundation. Positive trajectories were characterised by the benefits associated with good-quality, stable housing which affected many areas of people's lives (eg, health, substance use, social ties, identity, financial, leisure time). These benefits reflected a sense of ‘ontological security’31 that shifted the way people thought about themselves and others, and allowed them to take steps towards positive change. For example, since running away from home in adolescence, Alice had lived intermittently in shelters, hotels and with acquaintances. She described longstanding fear and hypervigilance that had shifted to a growing sense of freedom and security since receiving stable housing: “I was at the beach, and I was running towards the water, and I was jumping in. ... I was just jumping in the water. I took all my clothes off, and I had my bathing suit on, and I was like, ‘Whoo-hoo!’ And then I walked home in my bare feet, laid on the bed, and I was like, ‘Ahh. Heaven.”’

Several HF participants achieved a period of stable housing (typically for 6–12 months) but had since experienced eviction. The experience of losing housing contributed to different trajectories depending on the timing, context and interpretation of the experience. Among the positive trajectories, eviction was perceived as a turning point and a significant learning experience. Given the project's commitment to rehousing, many HF participants were able to learn from their mistakes, particularly around how to set boundaries, manage substance use and challenge expectations around the inevitability of failure.

Theme 2: Expression of positive identity. Many housed participants struggled to let go of their homeless identity, especially those who had been homeless for long periods. Finding and exploring opportunities that promoted the development of a positive personal and social identity, independent of homelessness, were often present among positive trajectories. For example, one participant joined a community choir; another coached a softball team for homeless people; several others were training to become peer support workers. However, it was also important for many participants to maintain links with their old identities, neighbourhoods and routines. Among the positive trajectories, there appeared to be a gradual process of shifting towards new social roles, networks and routines. The expression of positive identity was related to a willingness to introspect and reflect on one's experience and the lessons learnt, which was often present in positive trajectories. For example, after losing his job and marriage, Tom (positive trajectory) started using drugs and alcohol more heavily and developed intense paranoia and auditory hallucinations. For 2 years prior to the baseline interview, he cycled through shelters, transitional housing and the corrections system. After 18 months of HF, his narrative reflected a willingness to introspect on his recovery to date and a stronger sense of self: “I'm getting more solid in my thinking and in terms of what I want. Like a better relationship with my kids. … I have everything I need right now and the choices are my own. I have to live with them, good or bad. So, there's a level of maturity that's happening.”

Key themes: negative, neutral and mixed trajectories

Three major themes primarily contributed to the negative, neutral and mixed trajectories: (1) feeling like things will never change, including continued poverty, instability and substance use; (2) perceived failure and disappointment, resulting in a sense of learnt helplessness and (3) loss of one's health and loved ones.

Theme 1: Things will never change. Negative, neutral and mixed trajectories were typically characterised by increasing or continued hardship and instability across multiple domains, resulting in feelings of social devaluation, feeling trapped and a profound lack of autonomy (as described in the previous analysis of our baseline narratives).32 In addition, the emotional tone of negative and neutral trajectories was often flat with little elaboration or detail provided around experiences. Negative and neutral trajectories, in particular, were often pervaded by a sense of resignation and hopelessness. For example, at baseline, Alex (negative trajectory) had been living on the streets and in rooming houses for 10 years. For short periods, he was able to obtain low-income housing; however, he described persistent depression and frequent substance use, which made it difficult for him to maintain housing. At follow-up, he described a series of negative interactions with the justice, health and social housing systems, which left him feeling increasingly demoralised and suicidal: “I got beat up at the hospital by two security guards. They treat you like an animal. The nurses assume that you're just high. …They discharged me to a hotel. I left the next day. It was noisy, bug-infested, full of drugs.” Alex admitted that his drug and alcohol use has increased, and he has started selling drugs: “I'm just going in circles. It feels like I'm trapped. Down here, I'll always be an addict.”

Theme 2: Perceived failure and disappointment. Failed attempts at change across multiple life domains were prevalent in the narratives. Participants often made genuine efforts to create change, but encountered personal and/or systemic obstacles such that they did not follow through with the change. Recent or prolonged loss of housing and contact with family, often due to relapse into substance use, were the more common perceived failures. For example, Clara (mixed trajectory), a 54-year-old woman, had been living on the streets and in precarious housing situations for 20 years prior to receiving HF. She was stably housed for 18 months and had reconnected with her daughter. However, she was struggling with the effects of a recent concussion, a long-standing heroin addiction and the loss of past opportunities. “You know, people kept saying, ‘You're so lucky!’ [to receive housing] Well, yeah, I'm lucky. But I'm stupid. I could have had something like this 20 years ago. I put myself where I am. Nobody put me here. You know? I know that. I don't like it, but [shrugs].”

Theme 3: Loss of one's health and loved ones. Many participants experienced significant physical health problems in addition to mental illness. Limited mobility and physical pain were daily experiences for most participants. Deaths and ruptures in key attachment relationships, including trauma and abuse, were also very common and resulted in a profound sense of isolation for many participants. The physical and psychological pain associated with recent loss(es) coloured participants’ perceptions of change and well-being across a range of domains and seemed to confirm an expectation of failure and low self-worth. Many participants, but especially those with negative and neutral trajectories, described being very socially isolated and disconnected from the community. For example, James (neutral trajectory) is a 46-year-old man with a long history of homelessness, severe mental illness and substance use who was assigned to HNTAU. He described a struggle between wanting deeper social ties but also wanting to protect himself: “I haven't really met people that were worth spending that type of energy on. I know that sounds a little cold, but people will—even unknowingly—suck the life out of you, you know? And I have to be really careful with my energy because I get very drained and I've spent a little bit more time understanding what stress can actually do, like mentally, physically and spiritually. And my life has been extremely stressful and traumatic. I don't want to go through that.”

Discussion

This study is the first to use longitudinal, narrative data from homeless adults with mental illness who were randomly assigned to HF or TAU (no housing or supports provided through the study) to examine trajectories of recovery, and is an important contribution to the growing literature on HF. Narratives from participants assigned to HF reflected positive, mixed or neutral trajectories over the 18-month study period; none were classified as negative. By contrast, positive trajectories were rare among TAU participants (3 of 15). In general, TAU narratives reflected negative or neutral trajectories; participants continued to experience numerous challenges related to housing, health, substance use, trauma and marginalisation that culminated in increasingly poor functioning and feelings of hopelessness.

Positive change was primarily related to obtaining good quality housing.33 The sense of security and positive self-worth resulting from good-quality, stable housing allowed individuals to explore new daily routines, reduce substance use and antisocial behaviour, expand social roles and networks, and provided a safe space to reflect on one's experiences. The breadth of domains affected by stable housing supports the construct of ontological security—the psychosocial sense of safety and stability which often accompanies permanent housing.31 34 Past research has shown that homeless adults with mental illness who receive HF manifest greater ontological security than those living in transitional housing.31 Our findings support a life-course approach to recovery from homelessness and mental illness which considers the complex effects of cumulative adversity. The safety, security and ideological construct of ‘home’ is a critical foundation upon which the recovery journey is based.31

Negative, neutral and mixed trajectories were characterised by continued hardship and heavy substance use, perceived failures and disappointments, loss and social isolation. These struggles resulted in a pervasive sense of social devaluation and helplessness, as described in a previous analysis of the baseline interviews.32 Trauma theory and research may provide a useful lens through which to view and understand the experience of and attempts to exit homelessness in at least three respects35: first, becoming homeless (and repeatedly thereafter) may itself produce symptoms of trauma. Second, the ongoing experience of homelessness (loss of safety, predictability, control) may erode coping abilities. Finally, homelessness may exacerbate trauma symptoms among people with pre-existing histories of victimisation. In our narratives, cumulative trauma and adversity were often the common factors underlying barriers to recovery such as psychiatric symptoms, substance abuse and social isolation.

Similar to prior qualitative research, mental illness was not a dominant theme in our narratives despite direct questions on the topic.21 33 Psychiatric symptoms and related stigma were a source of distress and impairment for most participants; however, it was often hard to disentangle the effects of cumulative adversity, substance abuse and mental illness. Most participants had experienced repeated, long-standing trauma and marginalisation, resulting in a level of disorganisation that does not fit traditional diagnostic classification systems. For many participants, heavy substance use started in early adolescence and continued into adulthood as a means of coping with trauma, homelessness, psychological distress and as a way to ‘fit in.’ Participants with long histories of homelessness and substance dependence tended to be more entrenched in the homeless identity and subculture, and had more difficulty adjusting to housing.21 33 A better understanding of the timing, sequence and context for changes in complex trajectories of recovery, including homelessness, mental illness and substance use, is still needed.

Isolation was a barrier to recovery for many participants, including those who received HF. The structural and individual factors that contributed to homelessness (ie, poverty and lack of resources, poor mental and physical health, poor coping skills, lack of meaningful activities) continued to impact all participants’ daily lives. Henwood et al34 suggest that past trauma may lead individuals to view the world as dangerous and unpredictable, thus leading some people to seek the perceived comfort of isolating in an apartment. This isolation could be a necessary consolidation phase before more substantial recovery can occur.21 In our narratives, isolation was often a long-standing way of being; however, participants also had few opportunities to develop new daily routines from which to develop a sense of well-being and identity.15 Some participants alluded to trade-offs between the safety of cohesive social ties and the flexibility of weak ties. It should also be noted that social isolation is a broader societal issue that is widely prevalent. While isolation is clearly heightened among marginalised populations, it is interesting to note that very few participants in our study who received HF left to return to their old neighbourhoods.

Our 18-month follow-up period provided most participants with only initial opportunities for recovery. Many participants who demonstrated positive trajectories had experienced either an eviction or a planned move from their original housing placement. Improvements in domains such as substance use, mental illness and social support were often very fragile. Our findings highlight the importance of supporting homeless people with mental disorders through stages of change that include relapse, eviction and rehousing. Oftentimes, in hindsight, these moves and the support received from service teams were significant turning points and learning experiences for participants. Narratives across all trajectory groups revealed that recovery and reclaiming stability in housing and health, among other domains, constitute very difficult work—slips and relapses are common. Collectively, the narratives highlight the tremendous challenge for people to move indoors and into the community after years and sometimes decades outside, and often longer as marginalised citizens.

For our participants, recovery requires attention to the consequences of cumulative adversity, particularly previous trauma, and the ongoing stressors of poverty and social isolation. Our findings revealed that progress in recovery reflected the stages of change model36: change was gradual and cumulative; the setbacks were often sudden and devastating, and maintaining gains often meant being ‘stuck’ with few options for positive change. Rather than a clinical approach to recovery, a whole-person approach that takes into account experiences of cumulative trauma, marginalisation and individual beliefs and expectations around change is needed.33

Limitations

The narratives in this study were affected by participants’ recent circumstances; however, our analysis relied on two different time points and also incorporated field notes. Classifying trajectories is challenging, given the volatility and multidimensionality of people's lives.21 Neutral and mixed trajectories were inherently more variable than positive and negative trajectories. For example, neutral trajectories could reflect a continued negative path or a continued pattern of positive and negative (mixed) experiences. Mixed trajectories were often reflective of a tumultuous pattern of positive and negative experiences that could not be clearly categorised as positive or negative. However, each category contained exemplar cases, which were easier to code, and other cases (about one-quarter) that were more complex. Coding complex cases requires a collaborative consensus-based approach.24 Finally, given the qualitative and exploratory nature of this research, we caution against conclusions based on the type of HF received (independent vs congregate housing) or comparisons of HF and TAU as the sample sizes within different housing types are small.

Future directions

Future qualitative research should examine the trajectories and themes identified in our study in more detail and over longer periods of time. For example, we found that the expansion or contraction of social roles greatly affected participants’ identities, and that positive shifts in identity were facilitated by HF. However, this shift appeared to be more difficult for those who had invested in a homeless lifestyle. The process of establishing a new sense of self and a sense of belonging requires more investigation.

Our analysis demonstrates that many homeless adults with mental illness who receive good-quality, stable housing with intensive supports (HF) are able to make significant positive change over 18 months. The sense of security and confidence related to stable housing is critical for supporting recovery across a variety of life domains. However, systemic changes at broader sociopolitical levels are also need to address issues related to poverty and homelessness.13 15 Social change is needed to create opportunities for marginalised people to be included in communities and to confront poverty and social inequity.32