Effect of a Rent Subsidy and Mentoring on Youth Transitioning Out of Homelessness

The overarching aim of this mixed methods study is to assess whether and how rent subsidies and mentorship influence social integration outcomes for formerly homeless young people living in market rent housing in three urban settings.

Specifically, the objectives of this study are to:

1. Determine whether rent subsidies plus mentorship results in better social integration outcomes than only receiving rent subsidies with respect to: a) community integration (psychological and physical); and b) self-esteem at the primary endpoint of 18 months.
2. Determine whether rent subsidies plus mentorship results in better social integration outcomes than only receiving rent subsidies with respect to: a) social connectedness; b) hope; and c) sustained academic and vocational participation at 18 months.
3. Explore whether rent subsidies plus mentorship results in better social integration outcomes than only receiving rent subsidies with respect to: a) income; b) perceived housing quality; c) psychiatric symptoms; and d) sense of engulfment at 18 months.
4. Integrate qualitative data to facilitate a fuller understanding of the quantitative data and deepen understanding of what the study participants (young people and mentors) found most beneficial about the intervention and how it could be improved.

This study will employ a convergent mixed methods design (i.e., quantitative and qualitative data are collected concurrently, and the findings combined) embedded within a randomized controlled trial (RCT) and a community-based participatory action research (CBPAR) framework. A mixed methods RCT is appropriate given the complex explanatory pathways (i.e., social and behavioral processes that may act independently and interdependently) of this intervention. In addition, the qualitative data will provide insights on contextual factors that may impact the external validity of the findings. Most importantly, this design provides a crucial (and under utilized) youth-informed perspective on social integration.

The study will be conducted in three Canadian cities: Toronto, Ontario (pop. 2.8 million); Hamilton, Ontario (pop. 552,000); and St. Catharines, Ontario (pop. 133,000). All of the study participants (n = 24) will receive rent subsidies (ranging from $400 - $500/month) for 24 months as a part of the intervention. This study includes funding for the rent subsidies and will be paid directly to the landlords by the community partners. St. Michael's Hospital will establish a service provider agreement with each of the community partners for this purpose. Participants in the intervention group (n = 12) will be matched with an adult mentor recruited by one of the community partners.

The mentors will be encouraged to incorporate the key relationship-based components of natural mentors (e.g., a 'coach' or 'cheerleader' role) to assist with mainstream integration. To facilitate more of an organic, natural mentor-mentee relationship, the mentors will have more flexibility than a typical formal mentorship program in the types of activities they pursue with their mentees. For example, they will not be mandated to attend shelter-based social events. Instead, mentors will be encouraged to initiate activities that direct their mentees away from the shelter system (and their old identities as homeless youth) and toward the mainstream (e.g., meeting for coffee at a local university campus, touring a local library, or visiting the mentor's place of employment during business hours). All of the mentors will meet monthly with their mentees for two years. In addition, the mentor will be encouraged to touch base with their mentee via phone or text message every week. If a mentor is unable to continue their role and there are at least six months left in the study, the study participant will be matched with a new mentor.

Community partners will match all participants with an outreach worker (already employed by each agency) who will communicate regularly with the research team, help ensure the rent subsidies are being distributed appropriately, maintain an ongoing relationship with the study participants, and monitor for 'red flags' in participants matched in mentor-mentee relationships (e.g., mentee reluctant to meet with their mentor). Matching all of the study participants with a worker will also help ensure that everyone is receiving a fairly equal level of social support from community partners, making it easier for the research team to discern whether the outcomes of interest are more likely attributable to mentorship rather than to varied levels of agency-based support. Moreover, a review of services and interventions designed to reduce "problem behaviors" (e.g., substance use and risky sexual practices) among street-involved and homeless young people (ages 12 - 24) found that researchers who had strong relationships with outreach workers and the community had more effective interventions and lower attrition rates than those who did not.

Following the baseline interview, participants at each of the three study sites (Toronto (n=12), Hamilton (n=6), and St. Catharine's (n=6)) will be randomized using block randomization to either the intervention (rent subsidies plus mentoring) or control (rent subsidies only) group. Randomization will be balanced by site based on random block sizes of two and four. The advantage of using block randomization is to uniformly distribute participants into treatment groups within each site. Because small block sizes may increase the risk of guessing the allocation procedure and subsequently introducing bias into the enrolment procedure, random block sizes will be used to avoid this potential selection bias. A unique randomization schedule will be produced for each site using SAS. A research coordinator not affiliated with the study will be the only person with access to the randomization schedule. The research coordinator will prepare sealed, opaque and sequentially numbered envelopes with the randomization results of participants. After assessing for eligibility and obtaining consent of each participant, research personnel responsible for enrolling participants will open the next randomization envelope from the sequentially ordered randomization envelope file to obtain the participant's randomized group assignment.

Quantitative data will be collected at six points in time over the course of 30 months: baseline, month six, month 12, month 18, month 24, and month 30. Qualitative measures are an important feature of this study and will consist of: 1) semi-structured individual interviews (study participants) and 2) focus groups (mentors). At baseline, twelve participants (six from each arm of the study) will be invited to participate in six semi-structured individual interviews, which will take place at the same time as the quantitative data collection: baseline, month six, month 12, month 18, month 24, and month 30. Participants will be purposively selected with a goal of having input from each of the three communities and a fairly equal gender and ethno-racial representation. All of the mentors (n = 12) will be invited to participate in two focus groups, which will take place at month 12 and month 24.

All analyses will be performed using the intention-to-treat principle; that is, all participants will be included and analyzed in the groups they were originally randomized. Baseline characteristics of the intervention and control groups will be summarized using descriptive statistics (i.e., mean, standard deviation, median and interquartile range for continuous variables, and frequencies and proportions for categorical variables). The descriptive statistics for outcomes at each study time point will be calculated, and differences in trajectories from baseline to 30 months follow-up between intervention and control groups using scatterplots and box-plots will be explored. Differences with 95% confidence intervals in continuous outcomes at 18 months (psychological community integration, self-esteem, social connectedness, hope, perceived housing quality, psychiatric symptoms, and sense of engulfment) between participants who received rent subsidies plus mentorship and participants who only received rent subsidies will be estimated using Analysis of Covariance (i.e., linear regression models), including an indicator of intervention group and the baseline value of the outcome. Regression diagnostics will be performed and analyses using the non-parametric Wilcoxon rank-sum test will be repeated if there are extreme outliers or influential observations. Groups will be compared with respect to count outcomes at 18 months (physical community integration) using graphical tools and the non-parametric Wilcoxon rank-sum test. For binary outcomes at 18 months (sustained academic and vocational participation, and income above low income cut-off ), differences in proportions with 95% confidence intervals will be estimated and tested using the chi-square or Fisher's exact test. Given the small sample size of this pilot randomized trial, all results will be interpreted with caution and with the intention of generating data and hypotheses for conducting a larger trial.

Given the emergent, iterative nature of research using a qualitative design, data analysis and interpretation will begin immediately after the first qualitative data generation session (at baseline). The semi-structured individual interviews and focus groups will be audio recorded and transcribed verbatim. In order to conduct a more nuanced analysis of the data, the transcriptionist will be instructed to note short responses, uncooperative tones, and literal silence. Prior to each subsequent qualitative data generation session, members of the research team will conduct a preliminary data analysis, reading the interview transcripts multiple times, separating the data into coded segments, making analytic memos beside sections of the transcripts, identifying emerging themes (and comparing/contrasting these between respondents), and compiling new questions. Those participating in the individual interviews and the focus groups will be asked for their perspectives on the emerging interpretations at each visit and these perspectives will play a key role in helping shape the data analysis and help ensure the trustworthiness of the data. The web-based application Dedoose will be utilized to assist with sorting and coding the qualitative data.

In keeping with CBPAR methodology, disseminating evidence with community partners is key in building community capacity and improving the lives of the young people participating in this study. Moreover, given the use of Critical Social Theory, the study findings will not only be presented, but will also be used to expose and explicate the relational processes (e.g., subjective experience of low socioeconomic position and low social class) that may be preventing formerly homeless young people from achieving meaningful social integration. With an emphasis on 'actionable' data, the findings will be disseminated broadly to both academic and community-based audiences in a variety of formats ranging from scientific journal papers to oral presentations.