Housing and Support Interventions for Homeless Youth in High-Income Countries: a Systematic Review — supplementary material

Authors
Affiliation

David Taylor

Monash University

Stephanie Vecchio

Monash University

Susan Baidawi

Monash University

Aron Shlonsky

Monash University

Published

January 15, 2026

1 About

This page contains supplementary material for a systematic review titled Housing and Support Interventions for Homeless Youth in High-Income Countries: a Systematic Review.

The following information is provided here:

Reporting Standards:

  • PRISMA 2020 Checklist — includes details of how this review meets the PRISMA 2020 reporting guidelines

  • SWiM Checklist — includes details of how this review meets the SWiM reporting guidelines

Supplementary methodology:

  • Search methods — includes search strategies and results for each published database and grey sources

  • Data extraction — includes the template and information extracted from each included study used to transform into a common effect size

  • Effect size transformation — includes detailed description of the methods used to transform reported results into a common effect size

Supplementary results:

The code to produce this supplementary material, along with the analysis code for this review is available in a GitHub repository.

2 Reporting Standards

2.1 PRISMA 2020 checklist

The PRISMA 2020 checklist (Page et al., 2021) is included in Table S1.

Table S1: PRISMA 2020 Checklist
Section and Topic Item # Checklist item Location where item is reported
Title
Title 1 Identify the report as a systematic review. In the article title
Abstract
Abstract 2 See the PRISMA 2020 for Abstracts checklist. Abstract is fully compliant
Introduction
Rationale 3 Describe the rationale for the review in the context of existing knowledge. Article section: ‘Why is it important to do this review?’
Objectives 4 Provide an explicit statement of the objective(s) or question(s) the review addresses. Article section: ‘Objectives’
Methods
Eligibility criteria 5 Specify the inclusion and exclusion criteria for the review and how studies were grouped for the syntheses. Article section: ‘Selection criteria’ and Table 1
Information sources 6 Specify all databases, registers, websites, organisations, reference lists and other sources searched or consulted to identify studies. Specify the date when each source was last searched or consulted.

Article section ‘Search methods’

Supplementary material: Section 3.1.
Search strategy 7 Present the full search strategies for all databases, registers and websites, including any filters and limits used. Supplementary material: Table S3 and Table S4.
Selection process 8 Specify the methods used to decide whether a study met the inclusion criteria of the review, including how many reviewers screened each record and each report retrieved, whether they worked independently, and if applicable, details of automation tools used in the process. Article section: ‘Data collection and extraction’
Data collection process 9 Specify the methods used to collect data from reports, including how many reviewers collected data from each report, whether they worked independently, any processes for obtaining or confirming data from study investigators, and if applicable, details of automation tools used in the process. Article section: “Data collection and extraction”
Data items 10a List and define all outcomes for which data were sought. Specify whether all results that were compatible with each outcome domain in each study were sought (e.g. for all measures, time points, analyses), and if not, the methods used to decide which results to collect. Supplementary material: Table S6
Data items 10b List and define all other variables for which data were sought (e.g. participant and intervention characteristics, funding sources). Describe any assumptions made about any missing or unclear information. Supplementary material: Table S5
Study risk of bias assessment 11 Specify the methods used to assess risk of bias in the included studies, including details of the tool(s) used, how many reviewers assessed each study and whether they worked independently, and if applicable, details of automation tools used in the process. Article section: ‘Assessment of risk of bias’
Effect measures 12 Specify for each outcome the effect measure(s) (e.g. risk ratio, mean difference) used in the synthesis or presentation of results. Article section: ‘Measures of treatment effect’
Synthesis methods 13a Describe the processes used to decide which studies were eligible for each synthesis (e.g. tabulating the study intervention characteristics and comparing against the planned groups for each synthesis (item #5)). Article section: “Synthesis of results”
Synthesis methods 13b Describe any methods required to prepare the data for presentation or synthesis, such as handling of missing summary statistics, or data conversions.

Article section: “Measure of treatment effect”

Supplementary material: Section 3.3
Synthesis methods 13c Describe any methods used to tabulate or visually display results of individual studies and syntheses. Article section: ‘Synthesis of results’
Synthesis methods 13d Describe any methods used to synthesize results and provide a rationale for the choice(s). If meta-analysis was performed, describe the model(s), method(s) to identify the presence and extent of statistical heterogeneity, and software package(s) used. Article section: ‘Synthesis of results’
Synthesis methods 13e Describe any methods used to explore possible causes of heterogeneity among study results (e.g. subgroup analysis, meta-regression). Not applicable, quantitative synthesis was not undertaken.
Synthesis methods 13f Describe any sensitivity analyses conducted to assess robustness of the synthesized results. Not applicable, quantitative synthesis was not undertaken.
Reporting bias assessment 14 Describe any methods used to assess risk of bias due to missing results in a synthesis (arising from reporting biases). Article section: ‘Risk of bias assessment’
Certainty assessment 15 Describe any methods used to assess certainty (or confidence) in the body of evidence for an outcome.

Article section: “Assessment of bias and confidence in results”

Supplementary material: Section 4.4
Results
Study selection 16a Describe the results of the search and selection process, from the number of records identified in the search to the number of studies included in the review, ideally using a flow diagram. Article: Figure 1
Study selection 16b Cite studies that might appear to meet the inclusion criteria, but which were excluded, and explain why they were excluded. Supplementary material: Table S7
Study characteristics 17 Cite each included study and present its characteristics. Article: Table 2
Risk of bias in studies 18 Present assessments of risk of bias for each included study.

Article: Figure 2 and Figure 3

Supplementary material: Table S10 & Table S11
Results of individual studies 19 For all outcomes, present, for each study: (a) summary statistics for each group (where appropriate) and (b) an effect estimate and its precision (e.g. confidence/credible interval), ideally using structured tables or plots. Article: Tables 3, 4, 5, 6, 7, 8 & 9
Results of syntheses 20a For each synthesis, briefly summarise the characteristics and risk of bias among contributing studies. Article section: ‘Impact of included interventions’
Results of syntheses 20b Present results of all statistical syntheses conducted. If meta-analysis was done, present for each the summary estimate and its precision (e.g. confidence/credible interval) and measures of statistical heterogeneity. If comparing groups, describe the direction of the effect. Not applicable, quantitative syntheses was not undertaken.
Results of syntheses 20c Present results of all investigations of possible causes of heterogeneity among study results. Not applicable, quantitative syntheses was not undertaken.
Results of syntheses 20d Present results of all sensitivity analyses conducted to assess the robustness of the synthesized results. Not applicable, quantitative syntheses was not undertaken.
Reporting biases 21 Present assessments of risk of bias due to missing results (arising from reporting biases) for each synthesis assessed. Not applicable, quantitative syntheses was not undertaken.
Certainty of evidence 22 Present assessments of certainty (or confidence) in the body of evidence for each outcome assessed. Not applicable
Discussion
Discussion 23a Provide a general interpretation of the results in the context of other evidence. Article section: ‘Discussion and Applications to Practice’
Discussion 23b Discuss any limitations of the evidence included in the review. Article section: ‘Discussion and Applications to Practice’
Discussion 23c Discuss any limitations of the review processes used. Article section: ‘Discussion and Applications to Practice’
Other Information
Registration and protocol 24a Provide registration information for the review, including register name and registration number, or state that the review was not registered. Article section: ‘Method’
Registration and protocol 24b Indicate where the review protocol can be accessed, or state that a protocol was not prepared. Article section: ‘Method’
Registration and protocol 24c Describe and explain any amendments to information provided at registration or in the protocol. Article section: ‘Deviations from the protocol’
Support 25 Describe sources of financial or non-financial support for the review, and the role of the funders or sponsors in the review. Article section: ‘Funding statement’
Competing interests 26 Declare any competing interests of review authors. Article section: ‘Declaration of conflicting interest’
Availability of data, code and other materials 27 Report which of the following are publicly available and where they can be found: template data collection forms; data extracted from included studies; data used for all analyses; analytic code; any other materials used in the review. Data collection forms: available in supplementary material and Github repository Data extracted from included studies: available in Github repository Data used for all analyses: available in Github repository Analytic code: in supplementary material and Github repository Other materials used in the review: all material required to replicate this review is available in the Github repository

2.2 SWiM Checklist

The Synthesis Without Meta-analysis (SWiM) checklist (Campbell et al., 2020) is included in Table S2.

Table S2: SWiM Checklist
SWiM reporting item Item description Location where item is reported
Methods
1. Grouping studies for synthesis 1a) Provide a description of, and rationale for, the groups used in the synthesis (eg, groupings of populations, interventions, outcomes, study design) See sectopn 'Synthesis of results'
1. Grouping studies for synthesis 1b) Detail and provide rationale for any changes made subsequent to the protocol in the groups used in the synthesis See section 'Deviations from the protocol'
2. Describe the standardised metric and transformation methods used Describe the standardised metric for each outcome. Explain why the metric(s) was chosen and describe any methods used to transform the intervention effects, as reported in the study, to the standardised metric, citing any methodological guidance consulted See section 'Measures of treatment effect'
3. Describe the synthesis methods Describe and justify the methods used to synthesise the effects for each outcome when it was not possible to undertake a meta-analysis of effect estimates See section: 'Data analysis and synthesis'
4. Criteria used to prioritise results for summary and synthesis Where applicable, provide the criteria used, with supporting justification, to select the particular studies, or a particular study, for the main synthesis or to draw conclusions from the synthesis (eg, based on study design, risk of bias assessments, directness in relation to the review question) Not applicable
5. Investigation of heterogeneity in reported effects State the method(s) used to examine heterogeneity in reported effects when it was not possible to undertake a meta-analysis of effect estimates and its extensions to investigate heterogeneity Not applicable, heterogeneity was no assessed
6. Certainty of evidence Describe the methods used to assess the certainty of the synthesis findings Not applicable, certainty of findings was not assessed
7.Data presentation methods Describe the graphical and tabular methods used to present the effects (eg, tables, forest plots, harvest plots) Results are presented in Tables 3-9
7.Data presentation methods Specify key study characteristics (eg, study design, risk of bias) used to order the studies, in the text and any tables or graphs, clearly referencing the studies included Study characteristics are presented in Table 2
Results
8. Reporting results For each comparison and outcome, provide a description of the synthesised findings and the certainty of the findings. Describe the result in language that is consistent with the question the synthesis addresses, and indicate which studies contribute to the synthesis See section 'Impact of included interventions'
Discussion
9. Limitations of the synthesis Report the limitations of the synthesis methods used and/or the groupings used in the synthesis and how these affect the conclusions that can be drawn in relation to the original review question See section 'Discussion and Applications to Practice'

3 Supplementary methodology

This section contains additional information on the methodology used in this review.

3.1 Search methods

3.1.1 Search strategy and results for database

The databases and search terms are detailed in Table S3 below.

Table S3: Search terms used for databases
Database Search terms Date searched Record #
Cochrane CENTRAL Registed of Controlled Trials via Ovid

1. exp Homeless Youth/ or exp Homeless Persons/ or Homeless*.mp or Couchsurf*.mp or (Unstab* adj2 hous*).mp or Runaway.mp or Rough sleep*.mp or Insecur* hous*.mp

2. exp Young Adult/ or exp Adolescent/ or (Emerging adj2 (Adult or Youth)).mp or adolescent.mp or Youth.mp or (young adj1 (adult or person or people)).mp

3. 1 AND 2

4. exp Randomized Controlled Trial/ or exp Clinical Trial/ or exp Controlled Clinical Trial/ or exp Comparative Study/

5. (RCT or trial* or randomi* or random* allocat* or random* assign* or (control* adj1 intervention*) or (treatment* adj1 control*) or (control adj1 (group or condition*)) or (comparison adj1 (group* or condition*)) or int* time series or comparative effectiv* or experiment* or difference in difference* or instrumental variable* or (propensity adj1 (score or match*)) or (match* adj2 (control or comparison)) or (control* adj1 treat*) or wait* list or quasi exp*).mp.

6. 4 OR 5

7. 3 AND 6

 7 June 2023 234
CINAHL via EBSCO

1. (MH “Homeless Persons”) OR (MH “Homelessness”) OR TI Homeless* OR AB Homeless* OR TI Couchsurf* OR AB Couchsurf* OR TI(Unstab* n2 hous*) OR AB(Unstab* n2 hous*) OR TI Runaway OR AB Runaway OR TI Rough sleep* OR AB Rough sleep* OR TI Insecur* hous* OR AB Insecur* hous*

2. (MM “Young Adult”) OR (MH “Adolescence”) OR TI (Emerging n2 (Adult OR Youth)) OR AB (Emerging n2 (Adult OR Youth)) OR TI Adolescent OR AB Adolescent OR TI Youth OR AB Youth or TI (Young n1 (Adult or Person or People)) OR AB (Young n1 (Adult or Person or People))

3. 1 AND 2

4. (MH “Randomized Controlled Trials”) OR (MH “Clinical Trials”) OR (MH “Experimental Studies”) OR (MH “Interrupted Time Series Analysis”) OR (MH “Quasi-Experimental Studies”)

5. TI RCT OR AB RCT OR TI Trial* OR AB Trial* OR TI Randomi* OR AB Randomi* OR TI Random* Allocat* OR AB Random* Allocat* OR TI Random* Assign* OR AB Random* Assign* OR TI (Control* n1 Intervention*) OR AB (Control* n1 Intervention*) OR TI (Treatment* n1 Control*) OR AB (Treatment* n1 Control*) OR TI (Control n1 (Group OR Condition*)) OR AB (Control n1 (Group OR Condition*)) OR TI (Comparison n1 (Group* OR Condition*)) OR AB (Comparison n1 (Group* OR Condition*)) OR TI Int* Time Series OR AB Int* Time Series OR TI Comparative Effectiv* OR AB Comparative Effectiv* OR TI Experiment* OR AB Experiment* OR TI Difference in Difference* OR AB Difference in Difference* OR TI Instrumental Variable* OR AB Instrumental Variable* OR TI (Propensity n1 (Score OR Match*)) OR AB (Propensity n1 (Score OR Match*)) OR TI (Match* n2 (Control OR Comparison)) OR AB (Match* n2 (Control OR Comparison)) OR TI (Control* n1 Treat*) OR AB (Control* n1 Treat*) OR TI Wait* list OR AB Wait* list OR TI Quasi exp* OR AB Quasi exp*

6. 4 OR 5

7. 3 AND 6

 7 June 2023 429
Criminal Justice Abstracts via EBSCO

1. SU Homeless OR SU Homeless Persons OR SU Homeless Youth OR SU Runaway Teenagers OR TI Homeless* OR AB Homeless* OR TI Runaway OR AB Runaway

2. SU Youth OR SU Teenagers OR SU Adolescence OR TI (Emerging n2 (Adult OR Youth)) OR AB (Emerging n2 (Adult OR Youth)) OR TI Adolescent OR AB Adolescent OR TI Youth OR AB Youth OR TI (Young n1 (Adult or Person or People)) OR AB (Young n1 (Adult or Person or People))

3. 1 AND 2

4. SU Treatment Effectiveness OR SU Randomized Controlled Trials

5. TI RCT OR AB RCT OR  TI Trial* OR AB Trial* OR TI Randomi* OR AB Randomi* OR TI Random* Allocat* OR AB Random* Allocat* OR TI Random* Assign* OR AB Random* Assign* OR TI (Control* n1 Intervention*) OR AB (Control* n1 Intervention*) OR TI (Treatment* n1 Control*) OR AB (Treatment* n1 Control*) OR TI (Control n1 (Group OR Condition*)) OR AB (Control n1 (Group OR Condition*)) OR TI (Comparison n1 (Group* OR Condition*)) OR AB (Comparison n1 (Group* OR Condition*)) OR TI Int* Time Series OR AB Int* Time Series OR TI Comparative Effectiv* OR AB Comparative Effectiv* OR TI Experiment* OR AB Experiment* OR TI Difference in Difference* OR AB Difference in Difference* OR TI Instrumental Variable* OR AB Instrumental Variable* OR TI (Propensity n1 (Score OR Match*)) OR AB (Propensity n1 (Score OR Match*)) OR TI (Match* n2 (Control OR Comparison)) OR AB (Match* n2 (Control OR Comparison)) OR TI (Control* n1 Treat*) OR AB (Control* n1 Treat*) OR TI Wait* list OR AB Wait* list OR TI Quasi exp* OR AB Quasi exp*

6. 4 OR 5

7. 3 AND 6

 8 June 2023 131
EMBASE via Ovid

1. exp Homeless Person/ or exp Homelessness/ or exp Homeless Youth/ or Homeless*.mp or Couchsurf*.mp or (Unstab* adj2 hous*).mp or Runaway.mp or Rough sleep*.mp or Insecur* hous*.mp

2. exp Young Adult/ or exp Adolescent/ or exp Juvenile/ or (Emerging adj2 (Adult or Youth)).mp or adolescent.mp or Youth.mp or (young adj1 (adult or person or people)).mp

3. 1 AND 2

4. exp Randomized Controlled Trial/ or exp Clinical Trial as Topic/ or exp Controlled Clinical Trial/ or exp Comparative Study/ or exp Quasi Experimental Study or exp Experimental Study/

5. (RCT or trial* or randomi* or random* allocat* or random* assign* or (control* adj1 intervention*) or (treatment* adj1 control*) or (control adj1 (group or condition*)) or (comparison adj1 (group* or condition*)) or int* time series or comparative effectiv* or experiment* or difference in difference* or instrumental variable* or (propensity adj1 (score or match*)) or (match* adj2 (control or comparison)) or (control* adj1 treat*) or wait* list or quasi exp*).mp.

6. 4 OR 5

7. 3 AND 6

 8 June 2023 933
ERIC via Proquest

1. MAINSUBJECT.EXACT(“Homeless People”) OR MAINSUBJECT.EXACT.EXPLODE(“Runaways”) OR noft(Homeless*) OR noft(Couchsurf*) OR noft(Unstab* N/2 hous*) OR noft(Runaway) OR noft(Rough sleep*) OR noft(Insecur* hous*)

2. MAINSUBJECT.EXACT(“Young Adults”) OR MAINSUBJECT.EXACT.EXPLODE(“Late Adolescents”) OR noft(Emerging N/2 (Adult or Youth)) OR noft(Adolescent) OR noft(Youth) OR noft(Young N/1 (Adult OR Person OR people))

3. 1 AND 2

4. MAINSUBJECT.EXACT(“Randomized Controlled Trials”) OR MAINSUBJECT.EXACT(“Quasiexperimental Design”) OR MAINSUBJECT.EXACT(‘Program Evaluation’)

5. noft(RCT) OR noft(Trial*) OR noft(Randomi*) OR noft(Random* allocat*) OR noft(Random* assign*) OR noft(Control* N/1 Intervention*) OR noft(Treatment* N/1 Control*) OR noft(Control N/1 (Group OR Condition*)) OR noft(Comparison N/1 (Group* OR Condition*)) OR noft(Int* Time Series) OR noft(Comparative Effectiv*) OR noft(Experiment*) OR noft(Difference in Difference*) OR noft(Instrumental Variable*) OR noft(Propensity N/1 (Score OR Match*)) OR noft(Match* N/2 (Control OR Comparison)) OR noft(Control* N/1 Treat*) OR noft(Wait* list) OR noft(Quasi exp*)

6. 4 OR 5

7. 3 AND 6

 8 June 2023 304
APA PsycINFO via Ovid

1. exp Homeless Youth/ or exp Homeless/ or exp Homeless Mentally Ill or exp Runaway Behavior/ or exp Shelters/ or Homeless*.mp or Couchsurf*.mp or (Unstab* adj2 hous*).mp or Runaway.mp or Rough sleep*.mp or Insecur* hous*.mp

2. exp Emerging Adulthood/ or exp Late Adolescence/ or (Emerging adj2 (Adult or Youth)).mp or adolescent.mp or Youth.mp or (young adj1 (adult or person or people)).mp

3. 1 AND 2

4. exp Randomized Controlled Trials/ or exp Clinical Trials/ or exp Treatment Effectiveness Evaluation/ or exp Quasi Experimental Methods/ or exp Experimental Methods/

5. (RCT or trial* or randomi* or random* allocat* or random* assign* or (control* adj1 intervention*) or (treatment* adj1 control*) or (control adj1 (group or condition*)) or (comparison adj1 (group* or condition*)) or int* time series or comparative effectiv* or experiment* or difference in difference* or instrumental variable* or (propensity adj1 (score or match*)) or (match* adj2 (control or comparison)) or (control* adj1 treat*) or wait* list or quasi exp*).mp.

6. 4 OR 5

7. 3 AND 6

 8 June 2023 360
Ovid MEDLINE

1. exp Homeless Youth/ or exp Ill-Housed Persons/ or Homeless*.mp or Couchsurf*.mp or (Unstab* adj2 hous*).mp or Runaway.mp or Rough sleep*.mp or Insecur* hous*.mp

2. (limit 1 to (“all child (0 to 18 years)” or “adolescent (13 to 18 years)” or “young adult (19 to 24 years)”)

3. exp Young Adult/ or exp Adolescent/ or (Emerging adj2 (Adult or Youth)).mp or adolescent.mp or Youth.mp or (young adj1 (adult or person or people)).mp

4. 1 AND 3

5. 2 OR 4

6. exp Randomized Controlled Trial/ or exp Clinical Trial/ or exp Controlled Clinical Trial/ or exp Controlled before-after Studies/ or exp Comparative Study/ or exp Evaluation Study/ or exp Feasibility Studies/ or exp Follow-up Studies/ or exp Interrupted Time Series Analysis/

7. (RCT or trial* or randomi* or random* allocat* or random* assign*).mp or (control* adj1 intervention*).mp or (treatment* adj1 control*).mp or (control adj1 (group or condition*)).mp or (comparison adj1 (group* or condition*)).mp or int* time series.mp or comparative effectiv*.mp or experiment*.mp or difference in difference*.mp or instrumental variable*.mp or (propensity adj1 (score or match*)).mp or (match* adj2 (control or comparison)).mp or (control* adj1 treat*).mp or wait* list.mp or quasi exp*.mp

8. 6 OR 7

9. 5 AND 8

 8 June 2023 959
Sociological Abstracts via Proquest

1. MAINSUBJECT.EXACT(“Homeless People”) OR MAINSUBJECT.EXACT.EXPLODE(“Runaways”) OR noft(Homeless*) OR noft(Couchsurf*) OR noft(Unstab* N/2 hous*) OR noft(Runaway) OR noft(Rough sleep*) OR noft(Insecur* hous*)

2. MAINSUBJECT.EXACT(“Young Adults”) OR MAINSUBJECT.EXACT(“Adolescents”) OR noft(Emerging N/2 (Adult or Youth)) OR noft(Adolescent) OR noft(Youth) OR noft(Young N/1 (Adult OR Person OR people))

3. 1 AND 2

4. MAINSUBJECT.EXACT(“Experiments”) OR MAINSUBJECT.EXACT(“Effectiveness”) OR MAINSUBJECT.EXACT(“Evaluation”)

5. noft(RCT) OR noft(Trial*) OR noft(Randomi*) OR noft(Random* allocat*) OR noft(Random* assign*) OR noft(Control* N/1 Intervention*) OR noft(Treatment* N/1 Control*) OR noft(Control N/1 (Group OR Condition*)) OR noft(Comparison N/1 (Group* OR Condition*)) OR noft(Int* Time Series) OR noft(Comparative Effectiv*) OR noft(Experiment*) OR noft(Difference in Difference*) OR noft(Instrumental Variable*) OR noft(Propensity N/1 (Score OR Match*)) OR noft(Match* N/2 (Control OR Comparison)) OR noft(Control* N/1 Treat*) OR noft(Wait* list) OR noft(Quasi exp*)

6. 4 OR 5

7. 3 AND 6

 7 June 2023 327
Social Services Abstracts via Proquest

1. MAINSUBJECT.EXACT(“Homelessness”) OR MAINSUBJECT.EXACT(“Runaways”) OR noft(Homeless*) OR noft(Couchsurf*) OR noft(Unstab* N/2 hous*) OR noft(Runaway) OR noft(Rough sleep*) OR noft(Insecur* hous*)

2. MAINSUBJECT.EXACT(“Young Adults”) OR MAINSUBJECT.EXACT(“Adolescents”) OR noft(Emerging N/2 (Adult or Youth)) OR noft(Adolescent) OR noft(Youth) OR noft(Young N/1 (Adult OR Person OR people))

3. 1 AND 2

4. MAINSUBJECT.EXACT(“Evaluation Research”) OR MAINSUBJECT.EXACT(“Evaluation”)

5. noft(RCT) OR noft(Trial*) OR noft(Randomi*) OR noft(Random* allocat*) OR noft(Random* assign*) OR noft(Control* N/1 Intervention*) OR noft(Treatment* N/1 Control*) OR noft(Control N/1 (Group OR Condition*)) OR noft(Comparison N/1 (Group* OR Condition*)) OR noft(Int* Time Series) OR noft(Comparative Effectiv*) OR noft(Experiment*) OR noft(Difference in Difference*) OR noft(Instrumental Variable*) OR noft(Propensity N/1 (Score OR Match*)) OR noft(Match* N/2 (Control OR Comparison)) OR noft(Control* N/1 Treat*) OR noft(Wait* list) OR noft(Quasi exp*)

6. 4 OR 5

7. 3 AND 6

 8 June 2023 538
Violence and Abuse Abstracts via EBSCO

1. SU Homelessness OR SU Homeless Persons OR SU Homeless Youth OR SU Runaway Teenagers OR TI Homeless* OR AB Homeless* OR TI Runaway OR AB Runaway

2. SU Youth OR SU Teenagers OR SU Adolescence OR TI (Emerging n2 (Adult OR Youth)) OR AB (Emerging n2 (Adult OR Youth)) OR TI Adolescent OR AB Adolescent OR TI Youth OR AB Youth OR TI (Young n1 (Adult or Person or People)) OR AB (Young n1 (Adult or Person or People))

3. 1 AND 2

4. SU Treatment Effectiveness OR SU Randomized Controlled Trials

5. TI RCT OR AB RCT OR  TI Trial* OR AB Trial* OR TI Randomi* OR AB Randomi* OR TI Random* Allocat* OR AB Random* Allocat* OR TI Random* Assign* OR AB Random* Assign* OR TI (Control* n1 Intervention*) OR AB (Control* n1 Intervention*) OR TI (Treatment* n1 Control*) OR AB (Treatment* n1 Control*) OR TI (Control n1 (Group OR Condition*)) OR AB (Control n1 (Group OR Condition*)) OR TI (Comparison n1 (Group* OR Condition*)) OR AB (Comparison n1 (Group* OR Condition*)) OR TI Int* Time Series OR AB Int* Time Series OR TI Comparative Effectiv* OR AB Comparative Effectiv* OR TI Experiment* OR AB Experiment* OR TI Difference in Difference* OR AB Difference in Difference* OR TI Instrumental Variable* OR AB Instrumental Variable* OR TI (Propensity n1 (Score OR Match*)) OR AB (Propensity n1 (Score OR Match*)) OR TI (Match* n2 (Control OR Comparison)) OR AB (Match* n2 (Control OR Comparison)) OR TI (Control* n1 Treat*) OR AB (Control* n1 Treat*) OR TI Wait* list OR AB Wait* list OR TI Quasi exp* OR AB Quasi exp*

6. 4 OR 5

7. 3 AND 6

 8 June 2023 17

3.1.2 Search strategy and results for grey sources

Sources and search terms used for grey sources are detailed in Table S4 below.

3.2 Data extraction

Pairs of reviewers performed data extraction (DT and SV), with one reviewer checking the other’s results.

We developed a data extraction template using Google’s Sheets app to facilitate collaboration between team members. Each tab of the sheet contained specific information.

All reported measures corresponding to outcomes of interest at all available time points were extracted into a shared spreadsheet which is available in Table S9.

The template in Table S5 was used to extract information about the study.

Table S5: Data extraction template: study-level data

The template in Table S6 was used to extract information about reported outcomes that would allow us to transform each of the reported results into a common effect size.

Table S6: Data extraction template: Outcome-level data

3.3 Effect size transformation

In most cases, formulas from Borenstein, Hedges, Higgins, & Rothstein (2009) were used to convert reported results into a common effect size (Hedges’ g). This was done using the R Project for Statistical Computing (R Core Team, 2025), and specifically the esc package (Lüdecke, 2019).

In cases where studies lacked sufficient data for effect size computation or conversion, the study’s authors were approached to obtain the missing information. If this information was unavailable or insufficient to calculate an effect size, reasonable assumptions were made.

The results in the study by Kozloff, Adair, et al. (2016) are one such case where assumptions were made. Results in this study were presented as treatment effects (i.e., mean differences, ratios or rate ratios and ratios of odds ratios and their confidence intervals). The review team was unable to obtain the necessary information from the authors to transform these using the formulas in the esc package. Accordingly we approximated these results using information reported in the included studies. The approach used for each type of result, and the assumptions they rely on are detailed below.

3.3.1 Estimating Cohen’s d or Hedges’ g from a Ratio of Rate Ratios

The esc_ratio_rate_ratio function estimates Cohen’s d or Hedges’ g effect sizes from a reported ratio of rate ratios, its standard error or confidence interval, sample sizes, and the population baseline rate. The method relies on several assumptions and approximations to convert the ratio of rate ratios to a standardised mean difference.

3.3.1.1 Assumptions

  1. The ratio of rate ratios is a measure of the relative difference in event rates between the treatment and comparison groups, with the comparison group as the reference (Lash, VanderWeele, Haneause, & Rothman, 2021).
  2. The standard error of the ratio of rate ratios is either provided or can be calculated from the confidence interval assuming a normal distribution (Altman & Bland, 2011).
  3. The population baseline rate represents the proportion of the population experiencing the event at baseline and is known or can be estimated (Lash et al., 2021).
  4. The event counts in the treatment and comparison groups follow a Poisson distribution (Fleiss, Levin, & Paik, 2003).
  5. The sample sizes of the treatment and comparison groups are known.

3.3.1.2 Step 1: Obtain the standard error of the ratio of rate ratios

The function first obtains the standard error of the ratio of rate ratios either directly from the provided ratio_rate_ratio_se argument or by calculating it from the confidence interval using the following formula (Altman & Bland, 2011):

\[ SE_{RRR} = \frac{RRR_{upper} - RRR_{lower}}{2 \times 1.96} \]

se_ratio_rate_ratio <- (ratio_rate_ratio_ci_upper - ratio_rate_ratio_ci_lower) / (2 * 1.96)

where \(SE_{RRR}\) is the standard error of the ratio of rate ratios, \(RRR_{upper}\) and \(RRR_{lower}\) are the upper and lower bounds of the ratio of rate ratios confidence interval, respectively.

3.3.1.3 Step 2: Estimate the event counts

The event count for the comparison group is estimated using the population baseline rate and the comparison group sample size:

\[ E_C = n_C \times BR \]

where \(E_C\) is the event count for the comparison group, \(n_C\) is the comparison group sample size, and \(BR\) is the population baseline rate.

comparison_event_count <- comparison_n * population_baseline_rate

The event count for the treatment group is then estimated using the ratio of rate ratios and the comparison event count:

\[ E_T = E_C \times RRR \]

where \(E_T\) is the event count for the treatment group, and \(RRR\) is the ratio of rate ratios.

treatment_event_count <- comparison_event_count * ratio_rate_ratio

3.3.1.4 Step 3: Calculate the pooled event rate and standard deviation

The pooled event count is calculated as the average of the treatment and comparison event counts:

\[ E_P = \frac{E_T + E_C}{2} \]

where \(E_P\) is the pooled event count.

pooled_event_count <- (treatment_event_count + comparison_event_count) / 2

The pooled event rate is then estimated by dividing the pooled event count by the average of the treatment and comparison sample sizes:

\[ R_P = \frac{E_P}{\frac{n_T + n_C}{2}} \]

where \(R_P\) is the pooled event rate, and \(n_T\) and \(n_C\) are the treatment and comparison group sample sizes, respectively.

pooled_event_rate <- pooled_event_count / ((treatment_n + comparison_n) / 2)

The pooled standard deviation is estimated using the pooled event rate, assuming a Poisson distribution (Fleiss et al., 2003):

\[ SD_P = \sqrt{R_P \times (1 - R_P)} \]

where \(SD_P\) is the pooled standard deviation.

pooled_sd <- sqrt(pooled_event_rate * (1 - pooled_event_rate))

3.3.1.5 Step 4: Calculate Cohen’s d

Cohen’s d is calculated using the estimated event rates for the treatment and comparison groups and the pooled standard deviation (Cohen, 1988):

\[ d = \frac{\frac{E_T}{n_T} - \frac{E_C}{n_C}}{SD_P} \]

where \(d\) is Cohen’s d.

cohens_d <- (treatment_event_count / treatment_n - comparison_event_count / comparison_n) / pooled_sd

3.3.1.6 Step 5: Calculate the standard error and confidence interval for Cohen’s d

The standard error for Cohen’s d is estimated using the event rates and sample sizes:

\[ SE_d = \sqrt{\frac{E_T}{n_T^2} + \frac{E_C}{n_C^2}} \times \frac{1}{SD_P} \]

where \(SE_d\) is the standard error for Cohen’s d, \(E_T\) and \(E_C\) are the event counts in the treatment and comparison groups, respectively, \(n_T\) and \(n_C\) are the sample sizes of the treatment and comparison groups, respectively, and \(SD_P\) is the pooled standard deviation.

standard_error_d <- sqrt((treatment_event_count / treatment_n^2) + (comparison_event_count / comparison_n^2)) / pooled_sd

The confidence interval for Cohen’s d is then calculated using the standard error (Nakagawa & Cuthill, 2007):

\[ CI_d = [d - 1.96 \times SE_d, d + 1.96 \times SE_d] \]

where \(CI_d\) is the confidence interval for Cohen’s d.

cohens_d_ci_lower <- cohens_d - (1.96 * standard_error_d)
cohens_d_ci_upper <- cohens_d + (1.96 * standard_error_d)

3.3.1.7 Step 6: Calculate Hedges’ g (optional)

If the esc_type argument is set to ‘g’, the function also calculates Hedges’ g, which applies Hedges’ correction to an estimate of the standardised mean difference (Hedges, 1981):

\[ g = d \times \left(1 - \frac{3}{4 \times (n_T + n_C) - 9}\right) \]

where \(g\) is Hedges’ g.

hedges_g <- cohens_d * (1 - (3 / (4 * (treatment_n + comparison_n) - 9)))

The confidence interval for Hedges’ g is calculated using the same standard error as Cohen’s d.

3.3.2 Estimating Cohen’s d or Hedges’ g from a Mean Difference

The esc_mean_difference function estimates Cohen’s d or Hedges’ g effect sizes from a reported mean difference, its standard error or confidence interval, and the sample sizes of the treatment and comparison groups. The function calculates the pooled standard deviation, effect size, standard error, confidence interval, and p-value.

3.3.2.1 Assumptions

  1. The mean difference between the treatment and comparison groups is provided.
  2. The standard error of the mean difference is either provided or can be calculated from the confidence interval assuming a normal distribution (Altman & Bland, 2011).
  3. The sample sizes of the treatment and comparison groups are known.
  4. The variance of the outcome is assumed to be equal in both groups.

3.3.2.2 Step 1: Obtain the standard error of the mean difference

The function first obtains the standard error of the mean difference either directly from the provided mean_diff_se argument or by calculating it from the confidence interval using the following formula (Altman & Bland, 2011):

\[ SE_{MD} = \frac{MD_{upper} - MD_{lower}}{2 \times 1.96} \]

standard_error <- (mean_diff_ci_upper - mean_diff_ci_lower) / (2 * 1.96)

where \(SE_{MD}\) is the standard error of the mean difference, \(MD_{upper}\) and \(MD_{lower}\) are the upper and lower bounds of the mean difference confidence interval, respectively.

3.3.2.3 Step 2: Estimate the pooled standard deviation

The pooled standard deviation is estimated using the standard error and the sample sizes of the treatment and comparison groups:

\[ SD_{pooled} = \frac{SE_{MD}}{\sqrt{\frac{1}{n_T} + \frac{1}{n_C}}} \]

where \(SD_{pooled}\) is the pooled standard deviation, \(SE_{MD}\) is the standard error of the mean difference, and \(n_T\) and \(n_C\) are the sample sizes of the treatment and comparison groups, respectively.

standard_deviation_pooled <- standard_error / sqrt((1 / treatment_n) + (1 / comparison_n))

3.3.2.4 Step 3: Calculate Cohen’s d

Cohen’s d is calculated by dividing the mean difference by the pooled standard deviation (Cohen, 1988):

\[ d = \frac{MD}{SD_{pooled}} \]

where \(d\) is Cohen’s d, \(MD\) is the mean difference, and \(SD_{pooled}\) is the pooled standard deviation.

cohens_d <- mean_difference / standard_deviation_pooled

3.3.2.5 Step 4: Calculate the standard error and confidence interval for Cohen’s d

The standard error for Cohen’s d is calculated by dividing the standard error of the mean difference by the pooled standard deviation:

\[ SE_d = \frac{SE_{MD}}{SD_{pooled}} \]

where \(SE_d\) is the standard error for Cohen’s d.

standard_error_d <- standard_error / standard_deviation_pooled

The confidence interval for Cohen’s d is then calculated using the standard error (Nakagawa & Cuthill, 2007):

\[ CI_d = [d - 1.96 \times SE_d, d + 1.96 \times SE_d] \]

where \(CI_d\) is the confidence interval for Cohen’s d.

mean_diff_ci_lower_d <- cohens_d - (1.96 * standard_error_d)
mean_diff_ci_upper_d <- cohens_d + (1.96 * standard_error_d)

3.3.2.6 Step 5: Calculate the p-value

The t-statistic is calculated by dividing the mean difference by its standard error:

\[ t = \frac{MD}{SE_{MD}} \]

where \(t\) is the t-statistic.

t_stat <- mean_difference / standard_error

The degrees of freedom are calculated as the sum of the sample sizes minus 2:

\[ df = n_T + n_C - 2 \]

where \(df\) is the degrees of freedom.

df <- treatment_n + comparison_n - 2

The p-value is then calculated using the t-distribution with the calculated degrees of freedom:

p_value <- 2 * (1 - pt(abs(t_stat), df))

3.3.2.7 Step 6: Calculate Hedges’ g (optional)

If the esc_type argument is set to ‘g’, the function also calculates Hedges’ g, which is an unbiased estimate of the standardised mean difference (Hedges, 1981):

\[ g = d \times \left(1 - \frac{3}{4 \times (n_T + n_C) - 9}\right) \]

where \(g\) is Hedges’ g.

hedges_g <- cohens_d * (1 - (3 / (4 * (treatment_n + comparison_n) - 9)))

The confidence interval for Hedges’ g is calculated using the same standard error as Cohen’s d.

3.3.3 Estimating Cohen’s d or Hedges’ g from a Ratio of Odds Ratios

The esc_ratio_odds_ratio function estimates Cohen’s d or Hedges’ g effect sizes from a reported ratio of odds ratios, its standard error or confidence interval, sample sizes, and the population baseline rate. The method relies on several assumptions and approximations to convert the ratio of odds ratios to a standardised mean difference.

3.3.3.1 Assumptions

  1. The ratio of odds ratios is a measure of the relative difference in odds between the treatment and comparison groups, with the comparison group as the reference (Szumilas, 2010).
  2. The standard error of the ratio of odds ratios is either provided or can be calculated from the confidence interval assuming a normal distribution of the log ratio of odds ratios (Altman & Bland, 2011).
  3. The population baseline rate represents the proportion of the population experiencing the event at baseline and is known or can be estimated (Lash et al., 2021).
  4. The event counts in the treatment and comparison groups follow a binomial distribution (Fleiss et al., 2003).
  5. The sample sizes of the treatment and comparison groups are known.

3.3.3.2 Step 1: Obtain the standard error of the log ratio of odds ratios

The function first obtains the standard error of the log ratio of odds ratios either directly from the provided ratio_odds_ratio_se argument or by calculating it from the confidence interval using the following formula (Altman & Bland, 2011):

\[ SE_{log(ROR)} = \frac{log(ROR_{upper}) - log(ROR_{lower})}{2 \times 1.96} \]

se_log_ratio_odds_ratio <- (log_ratio_odds_ratio_ci_upper - log_ratio_odds_ratio_ci_lower) / (2 * 1.96)

where \(SE_{log(ROR)}\) is the standard error of the log ratio of odds ratios, \(ROR_{upper}\) and \(ROR_{lower}\) are the upper and lower bounds of the ratio of odds ratios confidence interval, respectively.

3.3.3.3 Step 2: Convert the ratio of odds ratios to the log scale

The ratio of odds ratios is converted to the log scale:

\[ log(ROR) = log(ROR) \]

log_ratio_odds_ratio <- log(ratio_odds_ratio)

where \(log(ROR)\) is the log ratio of odds ratios.

3.3.3.4 Step 3: Estimate the event probabilities and odds

The event probability in the comparison group is estimated using the population baseline rate:

\[ P_C = BR \]

where \(P_C\) is the event probability in the comparison group, and \(BR\) is the population baseline rate.

comparison_event_prob <- population_baseline_rate

The odds in the comparison group are estimated using the event probability:

\[ O_C = \frac{P_C}{1 - P_C} \]

where \(O_C\) is the odds in the comparison group.

comparison_odds <- comparison_event_prob / (1 - comparison_event_prob)

The odds in the treatment group are estimated using the ratio of odds ratios and the comparison odds:

\[ O_T = O_C \times ROR \]

where \(O_T\) is the odds in the treatment group, and \(ROR\) is the ratio of odds ratios.

treatment_odds <- comparison_odds * ratio_odds_ratio

The event probability in the treatment group is estimated using the treatment odds:

\[ P_T = \frac{O_T}{1 + O_T} \]

where \(P_T\) is the event probability in the treatment group.

treatment_event_prob <- treatment_odds / (1 + treatment_odds)

3.3.3.5 Step 4: Convert event probabilities to approximate event counts

The event probabilities are converted to approximate event counts using the sample sizes:

\[ E_T = n_T \times P_T \\ E_C = n_C \times P_C \]

where \(E_T\) and \(E_C\) are the event counts in the treatment and comparison groups, respectively, and \(n_T\) and \(n_C\) are the sample sizes of the treatment and comparison groups, respectively.

treatment_event_count <- treatment_n * treatment_event_prob
comparison_event_count <- comparison_n * comparison_event_prob

3.3.3.6 Step 5: Calculate the pooled event rate and standard deviation

The pooled event count is calculated as the average of the treatment and comparison event counts:

\[ E_P = \frac{E_T + E_C}{2} \]

where \(E_P\) is the pooled event count.

pooled_event_count <- (treatment_event_count + comparison_event_count) / 2

The pooled event rate is then estimated by dividing the pooled event count by the average of the treatment and comparison sample sizes:

\[ R_P = \frac{E_P}{\frac{n_T + n_C}{2}} \]

where \(R_P\) is the pooled event rate.

pooled_event_rate <- pooled_event_count / ((treatment_n + comparison_n) / 2)

The pooled standard deviation is estimated using the pooled event rate, assuming a binomial distribution (Fleiss et al., 2003):

\[ SD_P = \sqrt{R_P \times (1 - R_P)} \]

where \(SD_P\) is the pooled standard deviation.

pooled_sd <- sqrt(pooled_event_rate * (1 - pooled_event_rate))

3.3.3.7 Step 6: Calculate Cohen’s d

Cohen’s d is calculated using the estimated event probabilities for the treatment and comparison groups and the pooled standard deviation (Cohen, 1988):

\[ d = \frac{P_T - P_C}{SD_P} \]

where \(d\) is Cohen’s d.

cohens_d <- (treatment_event_prob - comparison_event_prob) / pooled_sd

3.3.3.8 Step 7: Calculate the standard error and confidence interval for Cohen’s d

The standard error for Cohen’s d is estimated using the event probabilities and sample sizes (Hasselblad & Hedges, 1995):

\[ SE_d = \sqrt{\frac{P_T \times (1 - P_T)}{n_T} + \frac{P_C \times (1 - P_C)}{n_C}} \times \frac{1}{SD_P} \]

where \(SE_d\) is the standard error for Cohen’s d, \(P_T\) and \(P_C\) are the event probabilities in the treatment and comparison groups, respectively, \(n_T\) and \(n_C\) are the sample sizes of the treatment and comparison groups, respectively, and \(SD_P\) is the pooled standard deviation.

standard_error_d <- sqrt((treatment_event_prob * (1 - treatment_event_prob)) / treatment_n + (comparison_event_prob * (1 - comparison_event_prob)) / comparison_n) / pooled_sd

The confidence interval for Cohen’s d is then calculated using the standard error (Nakagawa & Cuthill, 2007):

\[ CI_d = [d - 1.96 \times SE_d, d + 1.96 \times SE_d] \]

where \(CI_d\) is the confidence interval for Cohen’s d.

cohens_d_ci_lower <- cohens_d - (1.96 * standard_error_d)
cohens_d_ci_upper <- cohens_d + (1.96 * standard_error_d)

3.3.3.9 Step 8: Calculate the p-value

The Z-statistic is calculated using the log ratio of odds ratios and its standard error:

\[ Z = \frac{log(ROR)}{SE_{log(ROR)}} \]

where \(Z\) is the Z-statistic.

z_stat <- log_ratio_odds_ratio / se_log_ratio_odds_ratio

The p-value is then calculated using the standard normal distribution:

p_value <- 2 * (1 - pnorm(abs(z_stat)))

3.3.3.10 Step 9: Calculate Hedges’ g (optional)

If the esc_type argument is set to ‘g’, the function also calculates Hedges’ g, which is an unbiased estimate of the standardised mean difference (Hedges, 1981):

\[ g = d \times \left(1 - \frac{3}{4 \times (n_T + n_C) - 9}\right) \]

where \(g\) is Hedges’ g.

hedges_g <- cohens_d * (1 - (3 / (4 * (treatment_n + comparison_n) - 9)))

The confidence interval for Hedges’ g is calculated using the same standard error as Cohen’s d.

3.3.4 Analytic Code Implementation

The code chunk below details the R function that implements the methods described above.

Code 
# Purpose: Most common transformations can be handled using those in the esc() R package. 
# However, the results in the study by Kozloff (2016) are presented as treatment effects (i.e., mean differences, ratios or rate ratios and ratios of odds ratios and their confidence intervals).
# These three functions transform these results into a common measure of the standardised mean difference: Cohen's d or Hedge's g

## Transform ratio of odds ratios

# Declare function to estimate either Cohen's d or Hedges' g, along with their 95% confidence interval and derived p-value from reported ratio of odds ratios and baseline rates
esc_ratio_odds_ratio <- function(
  ratio_odds_ratio,
  ratio_odds_ratio_se = NULL,
  ratio_odds_ratio_ci_lower = NULL,
  ratio_odds_ratio_ci_upper = NULL,
  treatment_n,
  comparison_n,
  population_baseline_rate,
  esc_type = 'd') {
  
  # Use reported standard error if provided, otherwise calculate standard error from confidence intervals
  if (!is.null(ratio_odds_ratio_se)) {
    se_log_ratio_odds_ratio <- ratio_odds_ratio_se / ratio_odds_ratio
  } else if (!is.null(ratio_odds_ratio_ci_lower) & !is.null(ratio_odds_ratio_ci_upper)) {
    log_ratio_odds_ratio_ci_lower <- log(ratio_odds_ratio_ci_lower)
    log_ratio_odds_ratio_ci_upper <- log(ratio_odds_ratio_ci_upper)
    se_log_ratio_odds_ratio <- (log_ratio_odds_ratio_ci_upper - log_ratio_odds_ratio_ci_lower) / (2 * 1.96)
  } else {
    stop("Either ratio_odds_ratio_se or both ratio_odds_ratio_ci_lower and ratio_odds_ratio_ci_upper must be provided.")
  }
  
  # Convert ratio of odds ratios to log scale
  log_ratio_odds_ratio <- log(ratio_odds_ratio)
  
  # Estimate the event probability in the comparison group using the population baseline rate
  comparison_event_prob <- population_baseline_rate
  
  # Estimate the odds in the comparison group
  comparison_odds <- comparison_event_prob / (1 - comparison_event_prob)
  
  # Estimate the odds in the treatment group using the ratio of odds ratios and the comparison odds
  treatment_odds <- comparison_odds * ratio_odds_ratio
  
  # Convert odds to event probabilities
  treatment_event_prob <- treatment_odds / (1 + treatment_odds)
  
  # Convert event probabilities to approximate event counts
  treatment_event_count <- treatment_n * treatment_event_prob
  comparison_event_count <- comparison_n * comparison_event_prob
  
  # Calculate the pooled event count
  pooled_event_count <- (treatment_event_count + comparison_event_count) / 2
  
  # Estimate the pooled event rate
  pooled_event_rate <- pooled_event_count / ((treatment_n + comparison_n) / 2)
  
  # Estimate the pooled standard deviation using the pooled event rate
  pooled_sd <- sqrt(pooled_event_rate * (1 - pooled_event_rate))
  
  # Calculate Cohen's d
  cohens_d <- (treatment_event_prob - comparison_event_prob) / pooled_sd
  
  # Calculate the standard error for Cohen's d using the event probabilities and sample sizes
  standard_error_d <- sqrt((treatment_event_prob * (1 - treatment_event_prob)) / treatment_n + (comparison_event_prob * (1 - comparison_event_prob)) / comparison_n) / pooled_sd
  
  # Calculate the confidence interval for Cohen's d
  cohens_d_ci_lower <- cohens_d - (1.96 * standard_error_d)
  cohens_d_ci_upper <- cohens_d + (1.96 * standard_error_d)
  
  # Calculate the Z-statistic
  z_stat <- log_ratio_odds_ratio / se_log_ratio_odds_ratio
  
  # Calculate the p-value using the standard normal distribution
  p_value <- 2 * (1 - pnorm(abs(z_stat)))
  
  # Check if Hedges' g is requested
  if (esc_type == 'g') {
    # Calculate Hedges' g
    hedges_g <- cohens_d * (1 - (3 / (4 * (treatment_n + comparison_n) - 9)))
    return(list(effect_size = hedges_g, conf_interval = c(cohens_d_ci_lower, cohens_d_ci_upper), p_value = p_value))
  } else {
    return(list(effect_size = cohens_d, conf_interval = c(cohens_d_ci_lower, cohens_d_ci_upper), p_value = p_value))
  }
}

## Transform mean differences

# Declare function to estimate either Cohen's d or Hedges' g, along with their 95% confidence interval and derived p-value from reported mean differences
esc_mean_difference <- function(
  mean_difference, 
  mean_diff_se = NULL, 
  mean_diff_ci_lower = NULL, 
  mean_diff_ci_upper = NULL, 
  treatment_n, 
  comparison_n, 
  esc_type) {
  # Use reported standard error if provided, otherwise calculate standard error from confidence intervals
  if (!is.null(mean_diff_se)) {
    standard_error <- mean_diff_se
  } else if (!is.null(mean_diff_ci_lower) & !is.null(mean_diff_ci_upper)) {
    standard_error <- (mean_diff_ci_upper - mean_diff_ci_lower) / (2 * 1.96)
  } else {
    stop("Either mean_diff_se or both mean_diff_ci_lower and mean_diff_ci_upper must be provided.")
  }
  
  # Estimate the pooled standard deviation using the standard error
  standard_deviation_pooled <- standard_error / sqrt((1 / treatment_n) + (1 / comparison_n))
  
  # Calculate Cohen's d
  cohens_d <- mean_difference / standard_deviation_pooled
  
  # Calculate the standard error for Cohen's d
  standard_error_d <- standard_error / standard_deviation_pooled
  
  # Calculate the confidence interval for Cohen's d (95% confidence level, Z = 1.96)
  mean_diff_ci_lower_d <- cohens_d - (1.96 * standard_error_d)
  mean_diff_ci_upper_d <- cohens_d + (1.96 * standard_error_d)
  
  # Calculate the t-statistic
  t_stat <- mean_difference / standard_error
  
  # Calculate degrees of freedom
  df <- treatment_n + comparison_n - 2
  
  # Calculate the p-value using the t-distribution
  p_value <- 2 * (1 - pt(abs(t_stat), df))
  
  # Check if Hedges' g is requested
  if (esc_type == 'g') {
    # Calculate Hedges' g
    hedges_g <- cohens_d * (1 - (3 / (4 * (treatment_n + comparison_n) - 9)))
    return(list(effect_size = hedges_g, conf_interval = c(mean_diff_ci_lower_d, mean_diff_ci_upper_d), p_value = p_value))
  } else {
    return(list(effect_size = cohens_d, conf_interval = c(mean_diff_ci_lower_d, mean_diff_ci_upper_d), p_value = p_value))
  }
}

## Transform ratio of rate ratios

# Declare function to estimate either Cohen's d or Hedges' g, along with their 95% confidence interval and derived p-value from reported differences in rate ratios and baseline rates
esc_ratio_rate_ratio <- function(
  ratio_rate_ratio,
  ratio_rate_ratio_se = NULL,
  ratio_rate_ratio_ci_lower = NULL,
  ratio_rate_ratio_ci_upper = NULL,
  treatment_n,
  comparison_n,
  population_baseline_rate,
  esc_type = 'd') {
  
  # Use reported standard error if provided, otherwise calculate standard error from confidence intervals
  if (!is.null(ratio_rate_ratio_se)) {
    se_ratio_rate_ratio <- ratio_rate_ratio_se
  } else if (!is.null(ratio_rate_ratio_ci_lower) & !is.null(ratio_rate_ratio_ci_upper)) {
    se_ratio_rate_ratio <- (ratio_rate_ratio_ci_upper - ratio_rate_ratio_ci_lower) / (2 * 1.96)
  } else {
    stop("Either ratio_rate_ratio_se or both ratio_rate_ratio_ci_lower and ratio_rate_ratio_ci_upper must be provided.")
  }
  
  # Estimate the event count for the comparison group using the population baseline rate
  comparison_event_count <- comparison_n * population_baseline_rate
  
  # Estimate the event count for the treatment group using the rate ratio and the comparison event count
  treatment_event_count <- comparison_event_count * ratio_rate_ratio
  
  # Calculate the pooled event count
  pooled_event_count <- (treatment_event_count + comparison_event_count) / 2
  
  # Estimate the pooled event rate
  pooled_event_rate <- pooled_event_count / ((treatment_n + comparison_n) / 2)
  
  # Estimate the pooled standard deviation using the pooled event rate
  pooled_sd <- sqrt(pooled_event_rate * (1 - pooled_event_rate))
  
  # Calculate Cohen's d
  cohens_d <- (treatment_event_count / treatment_n - comparison_event_count / comparison_n) / pooled_sd
  
  # Calculate the standard error for Cohen's d using the event rates and sample sizes
  standard_error_d <- sqrt((treatment_event_count / treatment_n^2) + (comparison_event_count / comparison_n^2)) / pooled_sd
  
  # Calculate the confidence interval for Cohen's d
  cohens_d_ci_lower <- cohens_d - (1.96 * standard_error_d)
  cohens_d_ci_upper <- cohens_d + (1.96 * standard_error_d)
  
  # Calculate the Z-statistic
  z_stat <- (ratio_rate_ratio - 1) / se_ratio_rate_ratio
  
  # Calculate the p-value using the standard normal distribution
  p_value <- 2 * (1 - pnorm(abs(z_stat)))
  
  # Check if Hedges' g is requested
  if (esc_type == 'g') {
    # Calculate Hedges' g
    hedges_g <- cohens_d * (1 - (3 / (4 * (treatment_n + comparison_n) - 9)))
    return(list(effect_size = hedges_g, conf_interval = c(hedges_g - (1.96 * standard_error_d), hedges_g + (1.96 * standard_error_d)), p_value = p_value))
  } else {
    return(list(effect_size = cohens_d, conf_interval = c(cohens_d_ci_lower, cohens_d_ci_upper), p_value = p_value))
  }
}
 
# testing sandpit 
diff <- 0.67
diff_lower <- 0.22
diff_upper <- 2.07
treatment_n <- 87
comparison_n <- 69
population_baseline_rate <- 0.2

esc_mean_difference(
  mean_difference = diff, 
  mean_diff_ci_lower = diff_lower, 
  mean_diff_ci_upper = diff_upper, 
  treatment_n = treatment_n, 
  comparison_n = comparison_n, 
  esc_type = "g")

esc_ratio_rate_ratio(
  ratio_rate_ratio = diff,
  ratio_rate_ratio_ci_lower = diff_lower,
  ratio_rate_ratio_ci_upper = diff_upper,
  population_baseline_rate = population_baseline_rate,
  treatment_n = treatment_n,
  comparison_n = comparison_n,
  esc_type = "g")

esc_ratio_odds_ratio(
  ratio_odds_ratio = diff,
  ratio_odds_ratio_ci_lower = diff_lower,
  ratio_odds_ratio_ci_upper = diff_upper,
  population_baseline_rate = population_baseline_rate,
  treatment_n = treatment_n,
  comparison_n = comparison_n,
  esc_type = "g")

The code below details the complete process for effect-size transformation.

Code 
# load custom es transformation functions
source("./review/code/es_transformation/custom_es_transformation_functions.R")

# read raw data
data_extraction_sheet_location <- "./review/inputs/yh_review_data_extraction_sheet.xlsx"
data_extraction_sheet <- read_excel(
  path = data_extraction_sheet_location,
  sheet = "outcome_data",
  col_names = TRUE
)

# clean data
clean_data_extraction_sheet <- data_extraction_sheet |>
  # remove instructions
  filter(!row_number() == 1) |>
  # remove superfluous columns
  select(
    -primary_extractor,
    -secondary_extractor,
    -meta,
    -required_information,
    -comment,
    -additional_extraction_items
  ) |>
  # specify variable types
  mutate(
    across(
      c(
        outcome_timing,
        grp1n,
        grp2n,
        prop1event,
        prop2event,
        grp1m,
        grp1sd,
        grp1se,
        grp2m,
        grp2sd,
        grp2se,
        pre1mean,
        pre1sd,
        post1mean,
        post1sd,
        pre2mean,
        pre2sd,
        post2mean,
        post2sd,
        gain1mean,
        gain1se,
        gain2mean,
        gain2se,
        or,
        f,
        se,
        totaln,
        b,
        beta,
        sdy,
        chisq,
        t,
        t_pvalue,
        diff,
        diff_lower,
        diff_upper,
        diff_se,
        population_baseline_rate
        ),
      as.numeric
    )
  ) |>
  # exclude baseline measurements
  filter(
    !outcome_timing == 0
  )

# group data by effect size type
mean_gain <- clean_data_extraction_sheet |>
  filter(
    esc_type == "Mean Gain"
  ) |>
  mutate(
    id = row_number()
  )

mean_sd <- clean_data_extraction_sheet |>
  filter(
    esc_type == "Mean SD"
  ) |>
  mutate(
    id = row_number()
  )

binary_proportions <- clean_data_extraction_sheet |>
  filter(
    esc_type == "Binary proportions"
  ) |>
  mutate(
    id = row_number()
  )

mean_difference <- clean_data_extraction_sheet |>
  filter(
    esc_type == "Mean Difference"
  ) |>
  mutate(
    id = row_number()
  )

ratio_rate_ratio <- clean_data_extraction_sheet |>
  filter(
    esc_type == "Ratio of Rate Ratios"
  ) |>
  mutate(
    id = row_number()
  )

ratio_odds_ratio <- clean_data_extraction_sheet |>
  filter(
    esc_type == "Ratio of Odds Ratios"
  ) |>
  mutate(
    id = row_number()
  )

# transform results reported as mean gains to hedges' g using esc package
mean_gain_hedges_g <- mean_gain |>
  # run es function
  effect_sizes(
    study = study_id,
    fun = "esc_mean_gain",
    grp1n = grp1n,
    grp2n = grp2n,
    pre1mean = pre1mean,
    pre1sd = pre1sd,
    post1mean = post1mean,
    post1sd = post1sd,
    pre2mean = pre2mean,
    pre2sd = pre2sd,
    post2mean = post2mean,
    post2sd = post2sd,
    es.type = "g") |>
  # rename var
  rename(
    study_id = study,
    es_ci_low = ci.lo,
    es_ci_high = ci.hi,
    sample_size = sample.size,
  ) |>
  # add row number as id
  mutate(
    id = row_number()
  ) |>
  # merge back with raw data
  left_join(
    mean_gain,
    by = c(
      "id",
      "study_id")
  ) |>
  # select vars for reporting
  select(
    study_id,
    outcome_domain,
    outcome_construct,
    outcome_measure,
    outcome_timing,
    estimand,
    es,
    es_ci_low,
    es_ci_high,
    sample_size,
    favourable_direction
  ) 

# transform results reported as mean and standard deviation gains to hedges' g using esc package
mean_sd_hedges_g <- mean_sd |>
  # run es function
  effect_sizes(
    study = study_id,
    fun = "esc_mean_sd",
    grp1m = grp1m,
    grp1sd = grp1sd,
    grp1n = grp1n,
    grp2m = grp2m,
    grp2sd = grp2sd,
    grp2n = grp2n,
    es.type = "g") |>
  # rename var
  rename(
    study_id = study,
    es_ci_low = ci.lo,
    es_ci_high = ci.hi,
    sample_size = sample.size,
  ) |>
  # add row number as id
  mutate(
    id = row_number()
  ) |>
  # merge back with raw data
  left_join(
    mean_sd,
    by = c(
      "id",
      "study_id")
  ) |>
  # select vars for reporting
  select(
    study_id,
    outcome_domain,
    outcome_construct,
    outcome_measure,
    outcome_timing,
    estimand,
    es,
    es_ci_low,
    es_ci_high,
    sample_size,
    favourable_direction
  ) 

# transform results reported as binary proportions to hedges' g using esc package
binary_proportions_hedges_g <- binary_proportions |>
  # run es function
  effect_sizes(
    study = study_id,
    fun = "esc_bin_prop",
    prop1event = prop1event,
    grp1n = grp1n,
    prop2event = prop2event,
    grp2n = grp2n,
    es.type = "g") |>
  # rename var
  rename(
    study_id = study,
    es_ci_low = ci.lo,
    es_ci_high = ci.hi,
    sample_size = sample.size,
  ) |>
  # add row number as id
  mutate(
    id = row_number()
  ) |>
  # merge back with raw data
  left_join(
    binary_proportions,
    by = c(
      "id",
      "study_id")
  ) |>
  # select vars for reporting
  select(
    study_id,
    outcome_domain,
    outcome_construct,
    outcome_measure,
    outcome_timing,
    estimand,
    es,
    es_ci_low,
    es_ci_high,
    sample_size,
    favourable_direction
  ) 

# transform results reported as mean differences from Kozloff 2016 et al that reports CI's using custom function
mean_difference_hedges_g_kozloff_2016 <- mean_difference |>
  # filter results from Gilmer 2016
  filter(
    study_id == "kozloff_2016"
  ) |>
  # run custom function
  rowwise() |>
  mutate(result = list(
    esc_mean_difference(
      mean_difference = diff,
      mean_diff_ci_lower = diff_lower,
      mean_diff_ci_upper = diff_upper,
      treatment_n = grp1n,
      comparison_n = grp2n,
      esc_type = "g"
    )
  )) 

# transform results reported as mean differences from Gilmer 2016 that reports SE's using custom function
mean_difference_hedges_g_gilmer_2016 <- mean_difference |>
  # filter results from Gilmer 2016
  filter(
    study_id == "gilmer_2016"
  ) |>
  # run custom function
  rowwise() |>
  mutate(result = list(
    esc_mean_difference(
      mean_difference = diff,
      mean_diff_se = diff_se,
      treatment_n = grp1n,
      comparison_n = grp2n,
      esc_type = "g"
    )
  )) 

# merge results back together and clean up
mean_difference_hedges_g <- bind_rows(
  mean_difference_hedges_g_kozloff_2016,
  mean_difference_hedges_g_gilmer_2016
  ) |>
  ungroup() |>
  unnest_wider(
    result
  ) |>
  # Convert the character column to a list of numeric vectors
  mutate(
    conf_interval = as.character(conf_interval),
    conf_interval = str_extract_all(conf_interval, "-?[0-9.]+") |> 
      map(~ as.numeric(.))
  ) |>
  # split the list-column into two new columns
  unnest_wider(
    conf_interval, 
    names_sep = "_") |>
  rename(
    es = effect_size,
    es_ci_low = conf_interval_1,
    es_ci_high = conf_interval_2
  ) |>
  # add sample size var
  mutate(
    sample_size = grp1n + grp2n,
    p_value = round(p_value, 4)
  ) |>
  # select vars for reporting
  select(
    study_id,
    outcome_domain,
    outcome_construct,
    outcome_measure,
    outcome_timing,
    estimand,
    es,
    es_ci_low,
    es_ci_high,
    p_value,
    sample_size,
    favourable_direction
  ) 

# transform results reported as ratios of rate ratios using custom function
ratio_rate_ratio_hedges_g <- ratio_rate_ratio |>
  # run custom function
  rowwise() |>
  mutate(result = list(
    esc_ratio_rate_ratio(
      ratio_rate_ratio = diff, 
      ratio_rate_ratio_ci_lower = diff_lower, 
      ratio_rate_ratio_ci_upper = diff_upper, 
      population_baseline_rate = population_baseline_rate,
      treatment_n = grp1n, 
      comparison_n = grp2n, 
      esc_type = "g"
    )
  )) |>
  # clean up results
  ungroup() |>
  unnest_wider(
    result
  ) |>
  # Convert the character column to a list of numeric vectors
  mutate(
    conf_interval = as.character(conf_interval),
    conf_interval = str_extract_all(conf_interval, "-?[0-9.]+") |> 
      map(~ as.numeric(.))
  ) |>
  # split the list-column into two new columns
  unnest_wider(
    conf_interval, 
    names_sep = "_") |>
  rename(
    es = effect_size,
    es_ci_low = conf_interval_1,
    es_ci_high = conf_interval_2
  ) |>
  # add sample size var
  mutate(
    sample_size = grp1n + grp2n,
    p_value = round(p_value, 4)
  ) |>
  # select vars for reporting
  select(
    study_id,
    outcome_domain,
    outcome_construct,
    outcome_measure,
    outcome_timing,
    estimand,
    es,
    es_ci_low,
    es_ci_high,
    p_value,
    sample_size,
    favourable_direction
  ) 

# transform results reported as ratios of odds ratios using custom function
ratio_odds_ratio_hedges_g <- ratio_odds_ratio |>
  # run custom function
  rowwise() |>
  mutate(result = list(
    esc_ratio_odds_ratio(
      ratio_odds_ratio = diff, 
      ratio_odds_ratio_ci_lower = diff_lower, 
      ratio_odds_ratio_ci_upper = diff_upper, 
      population_baseline_rate = population_baseline_rate,
      treatment_n = grp1n, 
      comparison_n = grp2n, 
      esc_type = "g"
    )
  )) |>
  # clean up results
  ungroup() |>
  unnest_wider(
    result
  ) |>
  # Convert the character column to a list of numeric vectors
  mutate(
    conf_interval = as.character(conf_interval),
    conf_interval = str_extract_all(conf_interval, "-?[0-9.]+") |> 
      map(~ as.numeric(.))
  ) |>
  # split the list-column into two new columns
  unnest_wider(
    conf_interval, 
    names_sep = "_") |>
  rename(
    es = effect_size,
    es_ci_low = conf_interval_1,
    es_ci_high = conf_interval_2
  ) |>
  # add sample size var
  mutate(
    sample_size = grp1n + grp2n,
    p_value = round(p_value, 4)
  ) |>
  # select vars for reporting
  select(
    study_id,
    outcome_domain,
    outcome_construct,
    outcome_measure,
    outcome_timing,
    estimand,
    es,
    es_ci_low,
    es_ci_high,
    p_value,
    sample_size,
    favourable_direction
  ) 

# clean up data for export 
effect_sizes <- dplyr::bind_rows(
  mean_gain_hedges_g,
  mean_sd_hedges_g,
  binary_proportions_hedges_g,
  mean_difference_hedges_g,
  ratio_rate_ratio_hedges_g,
  ratio_odds_ratio_hedges_g
  ) |>
  # round quant outcomes for reporting and combine in single column
  mutate(
    es = round(es, 2),
    es_ci_low = round(es_ci_low, 2),
    es_ci_high = round(es_ci_high, 2)
  )

# add flags for discussion
effect_sizes_info <- effect_sizes |>
  mutate(
    # create variable that determines the precision of the estimate based on whether or not it crosses zero
    estimate_precision = case_when(
      es_ci_low < 0 & es_ci_high > 0 ~ "imprecise",
      TRUE ~ "precise"
    ),
    # create variable that determines the size of the effect based on Cohen's benchmarks
    es_magnitude = case_when(
      abs(es) < 0.2 ~ "very small",
      abs(es) >= 0.2 & abs(es) < 0.5 ~ "small",
      abs(es) >= 0.5 & abs(es) < 0.8 ~ "medium",
      abs(es) > 0.8 ~ "large"
    ),
    # create variable that shows if direction of effect favours the intervention or comparison
    direction_of_effect = case_when(
      es < 0 & favourable_direction == "Negative" ~ "Favours intervention",
      es > 0 & favourable_direction == "Negative" ~ "Favours comparison",
      es < 0 & favourable_direction == "Positive" ~ "Favours comparison",
      es > 0 & favourable_direction == "Positive" ~ "Favours intervention"
    )
  ) |>
  # drop redundant vars
  select(
    -p_value
  )
    
# export effect size data
saveRDS(
  effect_sizes_info,
  "./review/output/tables/effect_size_results.RDS"
  )

write_csv(
  effect_sizes_info,
  "./review/output/tables/effect_size_results.csv"
  )

4 Supplementary Results

4.1 Studies excluded at full text review

Studies excluded at full-text review stage are detailed in Table S7, with a reason and rationale for their exclusion.

Table S7: Studies excluded at full-text review
Reference Reason for exclusion Rationale
Aparicio et al. (2019) Wrong Study Design Intervention does not include a housing component
Aubry et al. (2019) Wrong Population Mean age treatment group ~29; Mean age comparison group ~22
Baer, Garrett, Beadnell, Wells, & Peterson (2007) Wrong Intervention Intervention does not include a housing component
Baggett et al. (2018) Wrong Intervention Intervention does not include a housing component
Balasuriya & Dixon (2021) Wrong Study Design Not a study
Bender et al. (2018) Wrong Intervention Intervention does not include a housing component
Brown, Rowe, Cunningham, & Ponce (2018) Wrong Population Study population mean age ~48.8
Burt (2012) Wrong Population Only ~2% of treatment group and ~5% of comparison group aged 18-24 at baseline
Chum et al. (2020) Wrong Population Only 13.6% of treatment group and 11.3% of comparison group aged <25 at baseline
Ciaranello et al. (2006) Wrong Population Mean age treatment group ~41.6; Mean age comparison group ~41.3
D’Amico, Houck, Tucker, Ewing, & Pedersen (2017) Wrong Intervention Intervention does not include a housing component
DiGuiseppi et al. (2021) Wrong Intervention Intervention does not include a housing component
Dizon-Ross (2020) Wrong Population Intervention provided to families, outcomes measured for children
Doré-Gauthier, Miron, Jutras-Aswad, Ouellet-Plamondon, & Abdel-Baki (2020) Wrong Study Design Counterfactual relied on historical comparison group, did not meet study design inclusion criteria
Drake, Yovetich, Bebout, Harris, & Mchugo (1997) Wrong Population Mean age treatment group ~36.2; Mean age comparison group ~34.4
Dunt et al. (2022) Wrong Population Only ~33% of treatment group and ~31% of comparison group aged 18-30 at baseline
Ferguson, Xie, & Glynn (2012) Wrong Intervention Intervention does not explicitly include a housing component
Ferguson (2018) Wrong Intervention Intervention does not include a housing component
Glendening et al. (2020) Wrong Population Intervention designed for families in child welfare system
Grace & Gill (2014) Wrong Intervention Intervention does not include a housing component
Grace & Gill (2016) Wrong Intervention Intervention does not include a housing component
Guo, Slesnick, & Feng (2016) Wrong Population Secondary report of Slesnick & Erdem (2013)
Guo & Slesnick (2017) Wrong Intervention Intervention does not include a housing component
Hawk & Davis (2012) Wrong Study Design Study does not include a counterfactual, also adult population with mean age ~46.
Hyun, Chung, & Lee (2005) Wrong Intervention Intervention does not include a housing component
Kennedy et al. (2018) Wrong Intervention Intervention does not include a housing component
Kidd et al. (2020) Wrong Intervention Intervention does not include a housing component
Kidd (2021) Wrong Intervention Intervention does not include a housing component
Krabbenborg, Boersma, & Wolf (2013) Wrong Intervention Intervention does not include a housing component
Krabbenborg, Boersma, Beijersbergen, Goscha, & Wolf (2015) Wrong Intervention Intervention does not include a housing component
Krabbenborg et al. (2017) Wrong Intervention Intervention does not include a housing component
C. T. Lee et al. (2018) Wrong Population Only 12.9% of treatment group and 14.6% of comparison group aged 18-29 at baseline
Linnemayr et al. (2021) Wrong Intervention Intervention does not include a housing component
Lund et al. (2024) Wrong Study Design Implementation study, secondary report of Kidd et al. (2020)
MacDonald (2020) Study protocol Ongoing study that would likely meet our inclusion criteria
Mackelprang, Collins, & Clifasefi (2014) Wrong Population Study population mean age ~48.4
McCay et al. (2011) Wrong Intervention Intervention does not include a housing component
McCay et al. (2015) Wrong Intervention Intervention does not include a housing component
Medalia, Saperstein, Huang, Lee, & Ronan (2017) Wrong Intervention Intervention does not include a housing component
Metraux, Marcus, & Culhane (2003) Wrong Population Only 25.7% of treatment group and 30.4% of comparison group aged 18-29 at baseline
Nolan (2006) Wrong Study Design Study does not include a counterfactual
O’Campo et al. (2017) Wrong Population Only ~10% of treatment group and ~9% of comparison group aged 18-24 at baseline
Osilla, Kennedy, Hunter, & Maksabedian (2016) Wrong Study Design Study does not include a counterfactual, Intervention does not include a housing component
Padgett, Gulcur, & Tsemberis (2006) Wrong Population Only ~19.1% of study population was aged 18-30 at baseline
Padgett, Stanhope, Henwood, & Stefancic (2011) Wrong Population Mean age treatment group ~44; Mean age comparison group ~40
Parpouchi, Moniruzzaman, McCandless, Patterson, & Somers (2016) Wrong Population Only ~7% of study population was aged 18-24 at baseline
Parpouchi, Moniruzzaman, & Somers (2021) Wrong Study Design Study does not include a counterfactual, or test an intervention.
Patterson, Currie, Rezansoff, & Somers (2015) Wrong Population Adult population in Housing First Trial
Pochetti (2018) Wrong Study Design Study does not include a counterfactual, or test an intervention.
Poremski, Rabouin, & Latimer (2017) Wrong Population Mean age treatment group ~45.2; Mean age comparison group ~47.1
Samuels, Fowler, Ault-Brutus, Tang, & Marcal (2015) Wrong Population Study population mean age ~32.5
Schick, Wiginton, Crouch, Haider, & Isbell (2019) Wrong Population Only 5.3% of study population was aged 18-29 at baseline
Simard, Chouinard-Thivierge, & Tanguay (2023) Wrong Study Design Study does not include a counterfactual, or test an intervention.
Slesnick & Erdem (2012) Wrong Study Design Study does not include a counterfactual
Slesnick & Erdem (2013) Wrong Population Study population mean age ~26.3
Slesnick et al. (2021) Study protocol Ongoing study that would likely meet our inclusion criteria
Smith, North, & Fox (1996) Wrong Population Study population mean age ~29.8
Sullivan et al. (2023) Wrong Population Study population mean age ~34.5
Thompson & Hasin (2017) Wrong Intervention Intervention does not included housing component
Tsemberis, Gulcur, & Nakae (2004) Wrong Population Only ~19% of study population was aged 18-30 at baseline
Tucker, D’Amico, Ewing, & Pedersen (2016) Wrong Intervention Intervention does not include a housing component
Tucker, D’Amico, Ewing, Miles, & Pedersen (2017) Wrong Intervention Intervention does not include a housing component
Tucker, D’Amico, Pedersen, Rodriguez, & Garvey (2020) Wrong Intervention Intervention does not include a housing component
Tucker, Kennedy, Osilla, & Golinelli (2021) Wrong Intervention Intervention does not include a housing component
Tucker et al. (2022) Wrong Intervention Intervention does not include a housing component
Upshur (1986) Wrong Study Design Study design does not include a counterfactual
Wolitski et al. (2010) Wrong Population Only 11.1% of treatment group and 9.6% of comparison group aged 18-29 at baseline

4.2 Housing First components of included interventions

Details of Housing First components in included studies are detailed in Table S8.

Table S8: Details of Housing First components in included studies
Study
Components Kozloff, Adair, et al. (2016) Thulien et al. (2019) Slesnick et al. (2023) Raithel, Yates, Dworsky, Schretzman, & Welshimer (2015) Gilmer (2016)
Housing-related components
Secure access to housing Sought to place youth in secure housing Provision of rental subsidies is deemed as securing access to housing Sought to place youth in housing of their choice and provided utility and rental assistance Provision of housing for up to 2 years. During the program they assisted individuals to secure more stable housing Sought to place youth in permanent housing
Support to maintain tenancy Sought to support the maintenance of tenancy through development of life skills Rent subsidies to help maintain tenancy Youth were provided services to assist with maintaining tenancy as well as utility and rental assistance Sought to support the maintenance of tenancy through development of life skills, money management skills, etc —
Clients are re-housed Sought to provide alternative housing for youth who were unable to maintain tenancy Rent subsidies are portable and therefore was not limited to a specific location Utility and rental assistance are portable and therefore they were not tied to a specific location Sought to help youth secure stable housing —
Conditionality of access to housing Not mandated to achieve or maintain sobriety or accept psychiatric treatment There were no conditions to receiving rent subsidies There were no prerequisites for receiving housing support — There were no conditions to access housing
Relationship between housing and other supports Housing and support services were not contingent on one another - mobile, off-site support services Rent subsidies were not contingent on mentorship and vice versa Housing placement and rental assistance were not contingent on support sessions and vice versa — —
Varied housing options available Scattered-site housing was offered at all sites, one site offered transitional/congregate housing option also — Rent subsidies are portable and not tied to a specific location — —
Clients are matched to housing based on their preferences Youth were able to choose from available scattered-site housing Youth had control over where they wanted to live Youth were able to choose from available apartment options — —
Support services components
Support to re-engage in services — — — — —
Young people are supported to reunify with their families Assistance in helping youth establish family connections as desired by the individual — — — Sought to assist youth with individualised goals, of which can include reunification with family
Uses a recovery-focused approach Sought to improve social and family connections, development of life-skills, etc Sought to support recovery through enhancing and encouraging choice and control Sought to support recovery via the building self-efficacy, problem-solving skills, etc Sought to assist youth to develop an individualised “action plan” outlining educational/vocational, money management, housing, health/wellness, personal goals, life skills development, community building, etc Focus of case management was rehabilitation and recovery
Support services encourage social and community integration Sought to provide services that work toward recovery and reintegration into the community including educational, employment and other life skills training Mentors sought to work towards recovery and reintegration into the community Sought to provide strengths-based advocacy support and sought to connect youth with other needed support in the community Sought to assist youth to develop life skills, build community networks, educational/vocational support services, etc Sought to provide services that work toward recovery and reintegration into the community including educational, employment and other life skills training
Services are provided at suitable intensity Service delivery varied in terms of intensity depending on need. Youth had weekly contact with mental health worker — — — —
Type of case management engagement ACT and ICM was used — — — ICM was used. A multidisciplinary treatment team model was also used but did not specify ACT approach
Components applicable to both housing and support services
Involvement of youth in consumer choice and self-determination Sought to promote individual choice and control over decisions, housing and support services Youth had high degree of control over the housing and support they received Sought to encourage and facilitate youth choice in the housing they were placed — Sought to encourage and facilitate choice in the supports and services they received
Access to housing or services is ongoing, for as long as required No limited duration, services ongoing — — Access to transitional housing for up to 2 years, during this time staff assisted individuals to secure more stable housing. Sought to provide aftercare —
Case managers have sufficient time to devote to clients — — — — —

4.3 Instruments used in included studies

Instruments used to measure outcomes in included studies are detailed in Table S9.

Table S9: Instruments used in included studies
Construct Measure Publicly available Evidence of validation identified Evidence of validation among population with complex needs identified Evidence of validation for homeless youth population identified Free to use Included studies using instrument Reference
Stable Housing
Housing Stability Residential Time-Line Follow-Back Inventory No Yes Yes No Unclear Kozloff, Adair, et al. (2016) Tsemberis, McHugo, Williams, Hanrahan, & Stefancic (2007)
Housing Quality Perceived Housing Quality Scale Yes Yes No No Unclear Thulien et al. (2019) Caffaro, Galati, & Roccato (2016)
Health
Mental Health GAIN Short Screener (GAIN-SS) – Substance Use Yes Yes Yes No Yes Kozloff, Adair, et al. (2016) Dennis, Chan, & Funk (2006)
Quality of Life EuroQoL 5 Dimensions Yes Yes No No Yes Kozloff, Adair, et al. (2016) Lamers, Bouwmans, van Straten, Donker, & Hakkaart (2006)
Quality of Life Lehman Quality of Life Interview 20 (QOLI-20) Yes Yes Yes No Unclear Kozloff, Adair, et al. (2016) Lehman (1996)
Recovery Recovery Assessment Scale Yes Yes Yes No Yes Kozloff, Adair, et al. (2016) Corrigan, Giffort, Rashid, Leary, & Okeke (1999)
Self-rated Physical Health Short Form 12 – Physical Component Yes Yes No No No Kozloff, Adair, et al. (2016) Ware, Kosinski, & Keller (1996)
Self-rated Mental Health Short Form 12 – Mental Component Yes Yes No No No Kozloff, Adair, et al. (2016) Ware et al. (1996)
Mental Health Symptoms Colorado Symptom Index Yes Yes Yes No Yes Kozloff, Adair, et al. (2016); Thulien et al. (2019) Conrad et al. (2001)
Health and Criminal Justice System Involvement Health, Social and Justice Service Use Inventory No No No No Unclear Kozloff, Adair, et al. (2016) Goering et al. (2011)
Substance Use Form-90 Yes Yes No No Yes Slesnick et al. (2023) Miller (1996)
Social and emotional wellbeing
Global Self-Worth Rosenberg Self-Esteem Scale Yes Yes No No Unclear Thulien et al. (2019) Rosenberg (1965)
Belongingness Social Connectedness Scale Yes Yes No No Unclear Thulien et al. (2019) R. M. Lee & Robbins (1995)
Motivation and expectations about future Beck Hopelessness Scale Yes Yes Yes No No Thulien et al. (2019) Beck, Weissman, Lester, & Trexler (1974)
Degree to which self-concept is defined by homelessness Modified Engulfment Scale Yes Yes Yes No Unclear Thulien et al. (2019) McCay & Seeman (1998)
Self-efficacy Mastery Scale Yes Yes No No Yes Slesnick et al. (2023) Pearlin & Schooler (1978)
Employment and earnings
Employment Vocational Time-Line Follow-Back Inventory No No No No Unclear Kozloff, Adair, et al. (2016) Goering et al. (2011)
Legal and justice
Victimisation Canadian General Social Survey Yes No No No Yes Kozloff, Adair, et al. (2016) Statistics Canada (2019)
Living skills and independence
Community Integration Community Integration Scale Unclear No No No Unclear Kozloff, Adair, et al. (2016); Thulien et al. (2019) Stergiopoulos et al. (2014)
Community Functioning Multnomah Community Ability Scale Yes Yes Yes No No Kozloff, Adair, et al. (2016) Barker, Barron, McFarland, & Bigelow (1994)

4.4 Risk of bias in included studies

In a similar fashion to the data extraction templates, we used a shared Google Sheet to assess the risk of bias in each reported result.

For randomised controlled trials, we used the Cochrane Risk of Bias 2 tool to assess the risk of bias in all reported results Sterne et al. (2019). The template and results are available in Table S10.

Table S10: Assessing Risk of Bias in randomised controlled trials

For non-randomised studies of interventions, we used the ROBINS-I tool Sterne et al. (2016). The template and results are available in Table S11.

Table S11: Assessing Risk of Bias in non-randomised studies of interventions

4.5 Barriers and Facilitators

Barriers and facilitators grouped by EPIS domain are detailed in Table S12.

Table S12: Barriers and facilitators by EPIS domain
Housing components
Support services components
Housing and support service components
EPIS domain Inner context Outer context Inner context Outer context Inner context Outer context
Exploration
Barriers — Legal status and informed consent issues complicate implementing Housing First with youth under 18 (Kozloff, Stergiopoulos, et al., 2016). — — — —
Facilitators — — — — — —
Preparation
Barriers — — — —

Designing training that was relevant across sites proved challenging due to the heterogenous skills and capabilities of participants and the varying local, social and systemic factors unique to each location (Nelson et al., 2012).

Poor communication about the limited scope for local adaptation affected staff buy-in to the intervention (Nelson et al., 2013).

Challenges were faced in hiring culturally competent staff to accommodate the needs of Aboriginal participants in the program (Nelson et al., 2012; Nelson et al., 2014) .

Tight timelines impacted ability to engage staff with lived experience in the development of the intervention, and led to issues such as mistrust and poor communication between sites and head office (Nelson et al., 2013).

Discomfort arose for some stakeholders regarding the project’s relationship to existing local services, particularly the head-to-head comparisons between the externally developed interventions, Housing First and local approaches (Nelson et al., 2013).

Some sites faced challenges in completing proposals due to a lack of a local housing and homelessness services infrastructure, leadership, service capacity, or human resources (Nelson et al., 2013).

Sites who were otherwise well equipped felt they had insufficient time to consult with the existing service providers during the proposal phase (Nelson et al., 2013).
Facilitators — — — The strength of existing services in the community was a key factor in the project’s success (Nelson et al., 2012; Nelson et al., 2014).

Effective leaders and strong staff teams with the right combination of technical and interpersonal skills, including decision-making abilities, clear direction, fostering a shared learning, and extensive experience working with various consumer populations, were crucial for success (Macnaughton et al., 2015; Nelson et al., 2012; Nelson et al., 2014).

Timelines and expectation promoted teamwork and productivity the during planning stages (Nelson et al., 2013).

Site-specific programs addressing the needs of CALD and Indigenous communities facilitated engagement and success (Nelson et al., 2012; Nelson et al., 2014).

Flexibility for sites to develop unique interventions or third arms was valuable in adjusting to individual needs (Nelson et al., 2013).

At the national level, Site Coordinators with a strong understanding of the project-wide logic model and the ability to move their site towards implementation of that model were critical to success (Nelson et al., 2013, 2012).

A site operations team, local advisory committee and various work groups facilitated good organisational structure and governance (Nelson et al., 2012).

Stakeholders deemed the program a valuable opportunity with potential for transformative change, and the initial excitement and enthusiasm were beneficial for moving the project forward (Nelson et al., 2013).

The community valued collaboration and collective effort which aligned with the project (Nelson et al., 2012).

Recruiting consumers in a manner that allowed them to maintain existing service relationships with established community service providers was important (Nelson et al., 2012) .

Collaboration between project programs and the major shelters facilitated recruitment of participants (Nelson et al., 2012).
Implementation
Barriers

The lack of established program protocols prior to implementation made it difficult to manage the intake of new consumers while simultaneously finding housing for participants, supporting existing consumers and helping them maintain housing (Nelson et al., 2012; Nelson et al., 2014).

Bottlenecks in the referral process from screening to referral slowed down access to housing (Macnaughton, Goering, & Nelson, 2012).

ACT and ICM teams needed closer relationships with housing team to better identify and engage yet-to-be housed individuals (Macnaughton et al., 2012).

The lack of a clear governance model sometimes led to delays in obtaining housing for consumers and a lack of cohesion and contact among clinical housing teams (Nelson et al., 2014).

There was a lack of after-hours staff support to assist with crisis situations and issues with landlords (Nelson et al., 2012; Nelson et al., 2014).

Housing availability was limited due to a lack of affordability and the reluctance of some housing agencies to risk established relationships with landlords by housing higher-risk Housing First clients (Macnaughton et al., 2012, 2015; Nelson et al., 2012; Nelson et al., 2014)

In small communities, privacy was a concern when certain information about the project or consumers was shared. For example, landlords would sometimes share information about difficult consumers within their network, making it more challenging to secure housing for those individuals (Nelson et al., 2012).

Some landlords exhibited stigma and racism, particularly towards Aboriginal people with mental health issues or substance use disorders, which led to them avoiding housing these individuals (Macnaughton et al., 2012; Nelson et al., 2012; Nelson et al., 2014)

Program staff had to work hard to sustain/repair relationships with landlords, as they were unable to fulfill all the guarantees that initially attracted them to the program, such as consistent visits with the participants or prompt repairs to damaged property (Nelson et al., 2012; Nelson et al., 2014).

There was insufficient planning and discussion on how to assist people not doing well in the program due to the diversity of consumer needs and functioning (Nelson et al., 2014).

Varying support needs meant that vocational specialists’ work was often dominated crisis management for a significant minority of unstable clients, rather than focusing on proactive, long-term goal planning with participants ready to move further in their recovery (Macnaughton et al., 2015).

ACT and ICM teams struggled to reconcile a recovery approach with heavy workload demands (Nelson et al., 2012).

The high consumers to staff ratio resulted in reduced one-to-one attention for participants, the impact of travel time between staff offices and participants’ homes was a particular burden for service teams (Nelson et al., 2012; Nelson et al., 2014).

Busy team members found it difficult to find time to participate in training and improve their practice approaches without further increasing workload and stress (Nelson et al., 2012).

Service providers’ irregular work hours and lack of opportunities for debriefing concerns were challenging (Nelson et al., 2012).

Lack of public transport was a barrier for consumers to get to appointments to receive health and social services, visit food bank, maintain relationships with family and friends (Nelson et al., 2012; Nelson et al., 2014).

Better collaboration and communication with local health care providers in the community were needed to address challenged in partnerships with community agencies (Nelson et al., 2012).

Important service programs, such as addiction treatment, vocational and educational support, were lacking in the community (Nelson et al., 2014).

COVID pandemic restrictions impacted the ability to have face to face meetings with mentors (Thulien et al., 2022, 2019).

The project design made transferring participants between service teams and housing types very difficult, even when a different model would better suit clients’ needs (Nelson et al., 2012).

There was a lack of training and difficulty meeting the cultural and linguistic requirements of diverse populations (Nelson et al., 2012; Nelson et al., 2014).

Limited opportunities for informal communication between teams, housing and support services being at different sites, and tensions between support teams due to competition for housing and perceived comparisons of project outcomes hindered collaboration and cohesion (Nelson et al., 2012; Nelson et al., 2014).

High staff turnover was experienced, possibly driven by weak leadership and poor team dynamics (Macnaughton et al., 2015).

Peer specialists were not seen as equal team members and were not meaningfully integrated, with a general lack of communication about their role. One site noted the importance of supporting peers to adopt a full-service role through proper training and workplace support, especially given their susceptibility to re-traumatisation during work (Macnaughton et al., 2015; Nelson et al., 2012) 		—
Facilitators —

Rent subsidies, housing team procurement strategies and community relationships were crucial in both expanding potential housing supply and achieving quicker access to housing (Macnaughton et al., 2012).

Collaborating with landlords, landlord associations and property management companies was beneficial for implementation. These partnerships: a) made landlords more likely to consult with service team members to resolve issues rather than notifying police or initiating eviction proceedings, and b) led to successful, high quality housing procurement (Macnaughton et al., 2015; Nelson et al., 2012; Nelson et al., 2014).

Providing consistent support to landlords, including prompt responses to problematic tenancy issues, paying for any damage, and guaranteeing rent payment, were strong incentives for landlord participation (Macnaughton et al., 2015). 		Partnerships with consumers enabled peer-driven initiatives like social gatherings, peer support programs for individuals with substance use issues, participant-produced newsletters, and participant-led focus groups on eviction prevention. Consumers could help fellow consumers and offer staff insider information about community resources (Nelson et al., 2012; Nelson et al., 2014). Across sites, participants highlighted the importance of having a diverse group of partners in providing a comprehensive continuum of care to participants (Nelson et al., 2012).

Implementation support from project’s national implementation team was beneficial (Macnaughton et al., 2015).

Staff members with lived experience were able to build trust, relate and empathise with participants (Nelson et al., 2012).

Organisational structure and governance were critical in defining roles, responsibilities, facilitating collaboration, partnership building, effective communication and conflict resolution (Nelson et al., 2012; Nelson et al., 2014).

A good organisational culture facilitated team dynamics – where the values were consistent with Housing First principles and common attitude to do whatever it takes to help participants achieve their goals, teams worked better together and had higher fidelity (Macnaughton et al., 2015).

Team diversity, fostering cross-team learning and sharing, breaking down hierarchies, along with team cohesion through structured meetings, formal training, all-team events and shared office space, nurtured teamwork and supported service delivery that leveraged expertise across areas (Nelson et al., 2012; Nelson et al., 2014).

Staff stability allowed service providers to integrate new approaches, practices and tools effectively (Nelson et al., 2012).

Incorporating additional expertise into the team, such as a psychiatrist to support clients with complex needs, or a home economist to help participants develop tenancy skills, was beneficial (Macnaughton et al., 2015).

Partnerships with government agencies and departments enhanced the project’s ability to secure consumer access to housing units, mental health and homelessness services, and government income supports (Nelson et al., 2012; Nelson et al., 2014).

Participants stated that local partners, including business owners, police, hospital staff, community mental health teams, shelters, meal providers, churches, the United Way, and mobile crisis teams, were crucial in contributing expertise and experience. This increased the project’s ability to integrate into the community’s network of supports and services (Nelson et al., 2012; Nelson et al., 2014).

The small size of the community facilitated easier contact and collaboration with existing community service providers, enabling efficient information transfer among professionals, the public and potential participants (Nelson et al., 2012).

Training and technical assistance provided by the Mental Health Commission of Canada were responsive, fair, generous and supportive. Positive and productive communication, funding, practical guidance and training opportunities were beneficial (Nelson et al., 2012; Nelson et al., 2014).
Sustainment
Barriers Short project funding horizon reduced landlords’ willingness to sign year-long leases as the project neared its end (Macnaughton et al., 2015) — — —

Program costs were identified as a barrier to program sustainability (Nelson et al., 2012)

Staff expressed significant concerns about the lack of information regarding project sustainability. They felt a critical need for knowledge dissemination about the steps being taken to ensure that consumers were not abandoned once the research project ended (Nelson et al., 2013, 2012). 		—
Facilitators — — — Forming and maintaining good community partnerships with various stakeholders and existing agencies with various social service systems, (e.g., justice system personnel, employment/income assistance agencies, and wider mental health and substance use systems). These collaborations led host agencies to take “ownership” over the project and this was deemed as potentially important for sustaining the project long-term (Macnaughton et al., 2015) Staff demonstrated a high-level of commitment to the project and its values, despite the challenging nature of the work (Macnaughton et al., 2015) —

References

Altman, D. G., & Bland, J. M. (2011). How to obtain the confidence interval from a P value. BMJ, 343, d2090. https://doi.org/10.1136/bmj.d2090
Aparicio, E. M., Kachingwe, O. N., Phillips, D. R., Fleishman, J., Novick, J., Okimoto, T., … Anderson, K. (2019). Holistic, trauma-informed adolescent pregnancy prevention and sexual health promotion for female youth experiencing homelessness: Initial outcomes of Wahine Talk. Children and Youth Services Review, 107, 104509. https://doi.org/10.1016/j.childyouth.2019.104509
Aubry, T., Bourque, J., Goering, P., Crouse, S., Veldhuizen, S., LeBlanc, S., … Bradshaw, C. (2019). A randomized controlled trial of the effectiveness of Housing First in a small Canadian city. BMC Public Health, 19(1), 1154. https://doi.org/10.1186/s12889-019-7492-8
Baer, J. S., Garrett, S. B., Beadnell, B., Wells, E. A., & Peterson, P. L. (2007). Brief motivational intervention with homeless adolescents: Evaluating effects on substance use and service utilization. Psychology of Addictive Behaviors, 21(4), 582–586. https://doi.org/10.1037/0893-164X.21.4.582
Baggett, T. P., Chang, Y., Yaqubi, A., McGlave, C., Higgins, S. T., & Rigotti, N. A. (2018). Financial incentives for smoking abstinence in homeless smokers: A pilot randomized controlled trial. Nicotine & Tobacco Research, 20(12), 1442–1450. https://doi.org/10.1093/ntr/ntx178
Balasuriya, L., & Dixon, L. B. (2021). Homelessness and mental health: Part 2. The impact of housing interventions. Psychiatric Services, 72(5), 618–619.
Barker, S., Barron, N., McFarland, B. H., & Bigelow, D. A. (1994). A community ability scale for chronically mentally Ill consumers: Part I. Reliability and validity. Community Mental Health Journal, 30(4), 363–383. https://doi.org/10.1007/BF02207489
Beck, A. T., Weissman, A., Lester, D., & Trexler, L. (1974). The measurement of pessimism: The Hopelessness Scale. Journal of Consulting and Clinical Psychology, 42(6), 861–865. https://doi.org/10.1037/h0037562
Bender, K. A., DePrince, A., Begun, S., Hathaway, J., Haffejee, B., & Schau, N. (2018). Enhancing risk detection among homeless youth: A randomized clinical trial of a promising pilot intervention. Journal of Interpersonal Violence, 33(19), 2945–2967. https://doi.org/10.1177/0886260516633208
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Introduction to meta-analysis. Chichester: John Wiley & Sons. https://doi.org/10.1002/9780470743386
Brown, M., Rowe, M., Cunningham, A., & Ponce, A. N. (2018). Evaluation of a comprehensive SAMHSA service program for individuals experiencing chronic homelessness. The Journal of Behavioral Health Services & Research, 45(4), 605–613. https://doi.org/10.1007/s11414-018-9589-8
Burt, M. R. (2012). Impact of housing and work supports on outcomes for chronically homeless adults with mental illness: LA’s HOPE. Psychiatric Services, 63(3), 209–215. https://doi.org/10.1176/appi.ps.201100100
Caffaro, F., Galati, D., & Roccato, M. (2016). Development and validation of the Perception of Housing Quality Scale (PHQS). TPM - Testing, Psychometrics, Methodology in Applied Psychology, (1), 37–51. https://doi.org/10.4473/TPM23.1.3
Campbell, M., McKenzie, J. E., Sowden, A., Katikireddi, S. V., Brennan, S. E., Ellis, S., … Thomson, H. (2020). Synthesis without meta-analysis (SWiM) in systematic reviews: Reporting guideline. BMJ, 368, l6890. https://doi.org/10.1136/bmj.l6890
Chum, A., Wang, R., Nisenbaum, R., O’Campo, P., Stergiopoulos, V., & Hwang, S. (2020). Effect of a housing intervention on selected cardiovascular risk factors among homeless adults with mental illness: 24-month follow-up of a randomized controlled trial. Journal of the American Heart Association, 9(19), e016896. https://doi.org/10.1161/JAHA.119.016896
Ciaranello, A. L., Molitor, F., Leamon, M., Kuenneth, C., Tancredi, D., Diamant, A. L., & Kravitz, R. (2006). Providing health care services to the formerly homeless: A quasi-experimental evaluation. Journal of Health Care for the Poor and Underserved, 17(2), 441–461. https://doi.org/10.1353/hpu.2006.0056
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). New York, NY: Routledge Academic.
Conrad, K. J., Yagelka, J. R., Matters, M. D., Rich, A. R., Williams, V., & Buchanan, M. (2001). Reliability and validity of a modified Colorado Symptom Index in a national homeless sample. Mental Health Services Research, 3(3), 141–153. https://doi.org/10.1023/A:1011571531303
Corrigan, P. W., Giffort, D., Rashid, F., Leary, M., & Okeke, I. (1999). Recovery as a psychological construct. Community Mental Health Journal, 35(3), 231–239.
D’Amico, E. J., Houck, J. M., Tucker, J. S., Ewing, B. A., & Pedersen, E. R. (2017). Group motivational interviewing for homeless young adults: Associations of change talk with substance use and sexual risk behavior. Psychology of Addictive Behaviors, 31(6), 688–698. https://doi.org/10.1037/adb0000288
Dennis, M. L., Chan, Y.-F., & Funk, R. R. (2006). Development and validation of the GAIN Short Screener (GSS) for internalizing, externalizing and substance use disorders and crime/violence problems among adolescents and adults. American Journal on Addictions, 15(s1), 80–91. https://doi.org/10.1080/10550490601006055
DiGuiseppi, G. T., Tucker, J. S., Prindle, J. J., Henwood, B. F., Huey, S. J., Rice, E. R., & Davis, J. P. (2021). Comparing the effectiveness of three substance use interventions for youth with and without homelessness experiences prior to treatment. Journal of Consulting and Clinical Psychology, 89(12), 995–1006. https://doi.org/10.1037/ccp0000704
Dizon-Ross, E. (2020). The impacts of homelessness prevention and rapid rehousing on homeless and highly mobile students. Palo Alto CA: Stanford Center for Education Policy Analysis.
Doré-Gauthier, V., Miron, J.-P., Jutras-Aswad, D., Ouellet-Plamondon, C., & Abdel-Baki, A. (2020). Specialized assertive community treatment intervention for homeless youth with first episode psychosis and substance use disorder: A 2-year follow-up study. Early Intervention in Psychiatry, 14(2), 203–210. https://doi.org/10.1111/eip.12846
Drake, R. E., Yovetich, N. A., Bebout, R. R., Harris, M., & Mchugo, G. J. (1997). Integrated treatment for dually diagnosed homeless adults. The Journal of Nervous & Mental Disease, 185(5), 298–305. https://doi.org/10.1097/00005053-199705000-00003
Dunt, D. R., Day, S. E., Collister, L., Fogerty, B., Frankish, R., Castle, D. J., … Redston, S. (2022). Evaluation of a Housing First programme for people from the public mental health sector with severe and persistent mental illnesses and precarious housing: Housing, health and service use outcomes. Australian & New Zealand Journal of Psychiatry, 56(3), 281–291. https://doi.org/10.1177/00048674211011702
Ferguson, K. M. (2018). Nonvocational outcomes from a randomized controlled trial of two employment interventions for homeless youth. Research on Social Work Practice, 28(5), 603–618. https://doi.org/10.1177/1049731517709076
Ferguson, K. M., Xie, B., & Glynn, S. (2012). Adapting the individual placement and support model with homeless young adults. Child & Youth Care Forum, 41(3), 277–294. https://doi.org/10.1007/s10566-011-9163-5
Fleiss, J. L., Levin, B., & Paik, M. C. (2003). Statistical methods for rates and proportions (Second Edition). Hoboken NJ: John Wiley & Sons.
Gilmer, T. P. (2016). Permanent supportive housing for transition-age youths: Service costs and fidelity to the Housing First model. Psychiatric Services, 67(6), 615–621. https://doi.org/10.1176/appi.ps.201500200
Glendening, Z. S., Shinn, M., Brown, S. R., Cleveland, K. C., Cunningham, M. K., & Pergamit, M. R. (2020). Supportive housing for precariously housed families in the child welfare system: Who benefits most? Children and Youth Services Review, 116, 105206. https://doi.org/10.1016/j.childyouth.2020.105206
Goering, P. N., Streiner, D. L., Adair, C., Aubry, T., Barker, J., Distasio, J., … Zabkiewicz, D. M. (2011). The At Home/Chez Soi trial protocol: A pragmatic, multi-site, randomised controlled trial of a Housing First intervention for homeless individuals with mental illness in five Canadian cities. BMJ Open, 1(e000323). https://doi.org/10.1136/bmjopen-2011-000323
Grace, M., & Gill, P. R. (2014). Improving outcomes for unemployed and homeless young people: Findings of the YP4 clinical controlled trial of joined up case management. Australian Social Work, 67(3), 419–437. https://doi.org/10.1080/0312407X.2014.911926
Grace, M., & Gill, P. R. (2016). Client-centred case management: How much makes a difference to outcomes for homeless jobseekers? Australian Social Work, 69(1), 11–26. https://doi.org/10.1080/0312407X.2015.1016445
Guo, X., & Slesnick, N. (2017). Reductions in hard drug use among homeless youth receiving a strength-based outreach intervention: Comparing the long-term effects of shelter linkage versus drop-in center linkage. Substance Use & Misuse, 52(7), 905–915. https://doi.org/10.1080/10826084.2016.1267219
Guo, X., Slesnick, N., & Feng, X. (2016). Housing and support services with homeless mothers: Benefits to the mother and her children. Community Mental Health Journal, 52(1), 73–83. https://doi.org/10.1007/s10597-015-9830-3
Hasselblad, V., & Hedges, L. V. (1995). Meta-analysis of screening and diagnostic tests. Psychological Bulletin, 117(1), 167–178. https://doi.org/10.1037/0033-2909.117.1.167
Hawk, M., & Davis, D. (2012). The effects of a harm reduction housing program on the viral loads of homeless individuals living with HIV/AIDS. AIDS Care, 24(5), 577–582. https://doi.org/10.1080/09540121.2011.630352
Hedges, L. V. (1981). Distribution theory for Glass’s estimator of effect size and related estimators. Journal of Educational Statistics, 6(2), 107–128. https://doi.org/10.3102/10769986006002107
Hyun, M.-S., Chung, H.-I. C., & Lee, Y.-J. (2005). The effect of cognitive–behavioral group therapy on the self-esteem, depression, and self-efficacy of runaway adolescents in a shelter in South Korea. Applied Nursing Research, 18(3), 160–166. https://doi.org/10.1016/j.apnr.2004.07.006
Kennedy, D. P., Osilla, K. C., Hunter, S. B., Golinelli, D., Maksabedian Hernandez, E., & Tucker, J. S. (2018). A pilot test of a motivational interviewing social network intervention to reduce substance use among Housing First residents. Journal of Substance Abuse Treatment, 86, 36–44. https://doi.org/10.1016/j.jsat.2017.12.005
Kidd, S. A. (2021). Examining the effectiveness of a critical time intervention to stabilize trajectories out of homelessness for youth. https://clinicaltrials.gov/study/NCT04755361.
Kidd, S. A., Vitopoulos, N., Frederick, T., Leon, S., Wang, W., Mushquash, C., & McKenzie, K. (2020). Trialing the feasibility of a critical time intervention for youth transitioning out of homelessness. American Journal of Orthopsychiatry, 90(5), 535–545. https://doi.org/10.1037/ort0000454
Kozloff, N., Adair, C. E., Palma Lazgare, L. I., Poremski, D., Cheung, A. H., Sandu, R., & Stergiopoulos, V. (2016). "Housing First" for homeless youth with mental illness. Pediatrics, 138(4), e20161514. https://doi.org/10.1542/peds.2016-1514
Kozloff, N., Stergiopoulos, V., Adair, C. E., Cheung, A. H., Misir, V., Townley, G., … Goering, P. (2016). The unique needs of homeless youths with mental illness: Baseline findings from a Housing First trial. Psychiatric Services, 67(10), 1083–1090. https://doi.org/10.1176/appi.ps.201500461
Krabbenborg, M. A. M., Boersma, S. N., Beijersbergen, M. D., Goscha, R. J., & Wolf, J. R. L. M. (2015). Fidelity of a strengths-based intervention used by Dutch shelters for homeless young adults. Psychiatric Services, 66(5), 470–476. https://doi.org/10.1176/appi.ps.201300425
Krabbenborg, M. A. M., Boersma, S. N., Van Der Veld, W. M., Van Hulst, B., Vollebergh, W. A. M., & Wolf, J. R. L. M. (2017). A cluster randomized controlled trial testing the effectiveness of Houvast: A strengths-based intervention for homeless young adults. Research on Social Work Practice, 27(6), 639–652. https://doi.org/10.1177/1049731515622263
Krabbenborg, M. A. M., Boersma, S. N., & Wolf, J. R. L. M. (2013). A strengths based method for homeless youth: Effectiveness and fidelity of Houvast. BMC Public Health, 13(1), 1–10. https://doi.org/10.1186/1471-2458-13-359
Lamers, L. M., Bouwmans, C. a. M., van Straten, A., Donker, M. C. H., & Hakkaart, L. (2006). Comparison of EQ-5D and SF-6D utilities in mental health patients. Health Economics, 15(11), 1229–1236. https://doi.org/10.1002/hec.1125
Lash, T. L., VanderWeele, T. J., Haneause, S., & Rothman, K. (2021). Modern Epidemiology (4th ed.). Philadelphia PA: Wolters Kluwer Health.
Lee, C. T., Winquist, A., Wiewel, E. W., Braunstein, S., Jordan, H. T., Gould, L. H., … Lim, S. (2018). Long-term supportive housing is associated with decreased risk for new HIV diagnoses among a large cohort of homeless persons in New York city. AIDS and Behavior, 22(9), 3083–3090. https://doi.org/10.1007/s10461-018-2138-x
Lee, R. M., & Robbins, S. B. (1995). Measuring belongingness: The Social Connectedness and the Social Assurance scales. Journal of Counseling Psychology, 42(2), 232–241.
Lehman, A. F. (1996). Measures of quality of life among persons with severe and persistent mental disorders. Social Psychiatry and Psychiatric Epidemiology, 31(2), 78–88. https://doi.org/10.1007/BF00801903
Linnemayr, S., Zutshi, R., Shadel, W., Pedersen, E., DeYoreo, M., & Tucker, J. (2021). Text messaging intervention for young smokers experiencing homelessness: Lessons learned from a randomized controlled trial. JMIR mHealth and uHealth, 9(4), e23989. https://doi.org/10.2196/23989
Lüdecke, D. (2019). Esc: Effect size computation for meta analysis.
Lund, J. I., Toombs, E., Mushquash, C. J., Pitura, V., Toneguzzi, K., Bobinski, T., … Kidd, S. A. (2024). Cultural adaptation considerations of a comprehensive housing outreach program for Indigenous youth exiting homelessness. Transcultural Psychiatry, 61(3), 457–472. https://doi.org/10.1177/13634615221135438
MacDonald, C. (2020). Research demonstration projects in youth homelessness. https://doi.org/10.1186/ISRCTN10505930
Mackelprang, J. L., Collins, S. E., & Clifasefi, S. L. (2014). Housing First is associated with reduced use of emergency medical services. Prehospital Emergency Care, 18(4), 476–482. https://doi.org/10.3109/10903127.2014.916020
Macnaughton, E. L., Goering, P. N., & Nelson, G. B. (2012). Exploring the value of mixed methods within the At Home/Chez soi Housing First project: A strategy to evaluate the implementation of a complex population health intervention for people with mental illness who have been homeless. Canadian Journal of Public Health, 103(1), S57–S62. https://doi.org/10.1007/BF03404461
Macnaughton, E. L., Stefancic, A., Nelson, G., Caplan, R., Townley, G., Aubry, T., … Goering, P. (2015). Implementing Housing First across sites and over time: Later fidelity and implementation evaluation of a pan-Canadian multi-site Housing First program for homeless people with mental illness. American Journal of Community Psychology, 55(3), 279–291. https://doi.org/10.1007/s10464-015-9709-z
McCay, E. A., Carter, C., Aiello, A., Quesnel, S., Langley, J., Hwang, S., … Karabanow, J. (2015). Dialectical Behavior Therapy as a catalyst for change in street-involved youth: A mixed methods study. Children and Youth Services Review, 58, 187–199. https://doi.org/10.1016/j.childyouth.2015.09.021
McCay, E. A., Quesnel, S., Langley, J., Beanlands, H., Cooper, L., Blidner, R., … Bach, K. (2011). A relationship-based intervention to improve social connectedness in street-involved youth: A pilot study. Journal of Child and Adolescent Psychiatric Nursing, 24(4), 208–215. https://doi.org/10.1111/j.1744-6171.2011.00301.x
McCay, E. A., & Seeman, M. V. (1998). A scale to measure the impact of a schizophrenic lllness on an individual’s self-concept. Archives of Psychiatric Nursing, 12(1), 41–49. https://doi.org/10.1016/S0883-9417(98)80007-1
Medalia, A., Saperstein, A. M., Huang, Y., Lee, S., & Ronan, E. J. (2017). Cognitive skills training for homeless transition-age youth: Feasibility and pilot efficacy of a community based randomized controlled trial. Journal of Nervous & Mental Disease, 205(11), 859–866. https://doi.org/10.1097/NMD.0000000000000741
Metraux, S., Marcus, S. C., & Culhane, D. P. (2003). The New York-New York Housing Initiative and use of public shelters by persons with severe mental illness. Psychiatric Services, 54(1), 67–71. https://doi.org/10.1176/appi.ps.54.1.67
Miller, W. R. (1996). Form 90: A structured assessment interview for drinking and related behaviors — Test manual. American Psychological Association. https://doi.org/10.1037/e563242012-001
Nakagawa, S., & Cuthill, I. C. (2007). Effect size, confidence interval and statistical significance: A practical guide for biologists. Biological Reviews, 82(4), 591–605. https://doi.org/10.1111/j.1469-185X.2007.00027.x
Nelson, G., Macnaughton, E., Goering, P., Dudley, M., O’Campo, P., Patterson, M., … Vallée, C. (2013). Planning a multi-site, complex intervention for homeless people with mental illness: The relationships between the national team and local sites in Canada’s At Home/Chez soi project. American Journal of Community Psychology, 51(3-4), 347–358. https://doi.org/10.1007/s10464-012-9554-2
Nelson, G., Rae, J., Townley, G., Goering, P., Piat, M., Egalité, N., … Tsemberis, S. (2012). Implementation and fidelity evaluation of the Mental Health Commission of Canada’s At Home/Chez Soi project: Cross site report. Ottawa: Mental Health Commission of Canada.
Nelson, G., Stefancic, A., Rae, J., Townley, G., Tsemberis, S., Macnaughton, E., … Goering, P. (2014). Early implementation evaluation of a multi-site housing first intervention for homeless people with mental illness: A mixed methods approach. Evaluation and Program Planning, 43, 16–26. https://doi.org/10.1016/j.evalprogplan.2013.10.004
Nolan, T. C. (2006). Outcomes for a transitional living program serving LGBTQ youth in New York city. Child Welfare, 85(2), 385–406. Retrieved from https://www.jstor.org/stable/45398770
O’Campo, P., Hwang, S. W., Gozdzik, A., Schuler, A., Kaufman-Shriqui, V., Poremski, D., … Addorisio, S. (2017). Food security among individuals experiencing homelessness and mental illness in the At Home/Chez Soi Trial. Public Health Nutrition, 20(11), 2023–2033. https://doi.org/10.1017/S1368980017000489
Osilla, K. C., Kennedy, D. P., Hunter, S. B., & Maksabedian, E. (2016). Feasibility of a computer-assisted social network motivational interviewing intervention for substance use and HIV risk behaviors for Housing First residents. Addiction Science & Clinical Practice, 11(1), 14. https://doi.org/10.1186/s13722-016-0061-x
Padgett, D. K., Gulcur, L., & Tsemberis, S. (2006). Housing First services for people who are homeless with co-occurring serious mental illness and substance abuse. Research on Social Work Practice, 16(1), 74–83. https://doi.org/10.1177/1049731505282593
Padgett, D. K., Stanhope, V., Henwood, B. F., & Stefancic, A. (2011). Substance use outcomes among homeless clients with serious mental illness: Comparing Housing First with Treatment First programs. Community Mental Health Journal, 47(2), 227–232. https://doi.org/10.1007/s10597-009-9283-7
Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., … Moher, D. (2021). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. Systematic Reviews, 10(1), 89. https://doi.org/10.1186/s13643-021-01626-4
Parpouchi, M., Moniruzzaman, A., McCandless, L., Patterson, M., & Somers, J. M. (2016). Housing First and unprotected sex: A structural intervention. Journal of Health Care for the Poor and Underserved, 27(3), 1278–1302. https://doi.org/10.1353/hpu.2016.0113
Parpouchi, M., Moniruzzaman, A., & Somers, J. M. (2021). The association between experiencing homelessness in childhood or youth and adult housing stability in Housing First. BMC Psychiatry, 21(1), 1–14. https://doi.org/10.1186/s12888-021-03142-0
Patterson, M. L., Currie, L., Rezansoff, S., & Somers, J. M. (2015). Exiting homelessness: Perceived changes, barriers, and facilitators among formerly homeless adults with mental disorders. Psychiatric Rehabilitation Journal, 38(1), 81–87. https://doi.org/10.1037/prj0000101
Pearlin, L. I., & Schooler, C. (1978). The structure of coping. Journal of Health and Social Behavior, 19(1), 2. https://doi.org/10.2307/2136319
Pochetti, I. (2018). Perspective des droits, genre et travail social auprès des enfants des rues au Mexique: Interprétations locales des normes globales. L’Homme & la Société, 206(1), 213–240. https://doi.org/10.3917/lhs.206.0213
Poremski, D., Rabouin, D., & Latimer, E. (2017). A randomised controlled trial of evidence based supported employment for people who have recently been homeless and have a mental illness. Administration and Policy in Mental Health and Mental Health Services Research, 44(2), 217–224. https://doi.org/10.1007/s10488-015-0713-2
R Core Team. (2025). R: A language and environment for statistical computing. Vienna: R Foundation for Statistical Computing.
Raithel, J., Yates, M., Dworsky, A., Schretzman, M., & Welshimer, W. (2015). Partnering to leverage multiple data sources: Preliminary findings from a supportive housing impact study. Child Welfare, 94(1), 73–85.
Rosenberg, M. (1965). Society and the adolescent self-image. Princeton NJ: Princeton University Press. https://doi.org/10.1515/9781400876136
Samuels, J., Fowler, P. J., Ault-Brutus, A., Tang, D.-I., & Marcal, K. (2015). Time-limited case management for homeless mothers with mental health problems: Effects on maternal mental health. Journal of the Society for Social Work and Research, 6(4), 515–539. https://doi.org/10.1086/684122
Schick, V., Wiginton, L., Crouch, C., Haider, A., & Isbell, F. (2019). Integrated service delivery and health-related quality of life of individuals in permanent supportive housing who were formerly chronically homeless. American Journal of Public Health, 109(2), 313–319. https://doi.org/10.2105/AJPH.2018.304817
Simard, M.-C., Chouinard-Thivierge, S., & Tanguay, P. (2023). La réadaptation au coeur de nos préoccupations: Portrait et analyse des besoins d’adolescents hébergés en centre de réadaptation et en foyer de groupe. Criminologie, 56(1), 215–244. https://doi.org/10.7202/1099012ar
Slesnick, N., Chavez, L., Bunger, A., Famelia, R., Ford, J., Feng, X., … Kelleher, K. (2021). Housing, opportunities, motivation and engagement (HOME) for homeless youth at-risk for opioid use disorder: Study protocol for a randomized controlled trial. Addiction Science & Clinical Practice, 16(1), 30. https://doi.org/10.1186/s13722-021-00237-7
Slesnick, N., & Erdem, G. (2012). Intervention for homeless, substance abusing mothers: Findings from a non-randomized pilot. Behavioral Medicine, 38(2), 36–48. https://doi.org/10.1080/08964289.2012.657724
Slesnick, N., & Erdem, G. (2013). Efficacy of ecologically-based treatment with substance-abusing homeless mothers: Substance use and housing outcomes. Journal of Substance Abuse Treatment, 45(5), 416–425. https://doi.org/10.1016/j.jsat.2013.05.008
Slesnick, N., Zhang, J., Feng, X., Mallory, A., Martin, J., Famelia, R., … Kelleher, K. (2023). Housing and supportive services for substance use and self-efficacy among young mothers experiencing homelessness: A randomized controlled trial. Journal of Substance Abuse Treatment, 144, 108917. https://doi.org/10.1016/j.jsat.2022.108917
Smith, E. M., North, C. S., & Fox, L. W. (1996). Eighteen-month follow-up data on a treatment program for homeless substance abusing mothers. Journal of Addictive Diseases, 14(4), 57–72. https://doi.org/10.1300/J069v14n04_04
Statistics Canada. (2019). General Social Survey: Canadians’ safety. https://www23.statcan.gc.ca/imdb/p3Instr.pl?Function=assembleInstr&lang=en&Item_Id=1236284.
Stergiopoulos, V., Gozdzik, A., O’Campo, P., Holtby, A. R., Jeyaratnam, J., & Tsemberis, S. (2014). Housing First: Exploring participants’ early support needs. BMC Health Services Research, 14(1), 167. https://doi.org/10.1186/1472-6963-14-167
Sterne, J. A. C., Hernán, M. A., Reeves, B. C., Savović, J., Berkman, N. D., Viswanathan, M., … Higgins, J. P. T. (2016). ROBINS-I: A tool for assessing risk of bias in non-randomised studies of interventions. BMJ, 355(i4919). https://doi.org/10.1136/bmj.i4919
Sterne, J. A. C., Savović, J., Page, M. J., Elbers, R. G., Blencowe, N. S., Boutron, I., … Higgins, J. P. T. (2019). RoB 2: A revised tool for assessing risk of bias in randomised trials. BMJ, 366(l4898). https://doi.org/10.1136/bmj.l4898
Sullivan, C. M., Guerrero, M., Simmons, C., López-Zerón, G., Ayeni, O. O., Farero, A., … Sprecher, M. (2023). Impact of the domestic violence housing first model on survivors’ safety and housing stability: 12-month findings. Journal of Interpersonal Violence, 38(5-6), 4790–4813. https://doi.org/10.1177/08862605221119520
Szumilas, M. (2010). Explaining odds ratios. Journal of the Canadian Academy of Child and Adolescent Psychiatry, 19(3), 227–229.
Thompson, R. G., & Hasin, D. S. (2017). 630 Effects of a smartphone application plus BMI in reducing substance and sexual risk among homeless young adults. Alcoholism: Clinical and Experimental Research, 41(S1), 175A. https://doi.org/10.1111/acer.13391
Thulien, N. S., Amiri, A., Hwang, S. W., Kozloff, N., Wang, A., Akdikmen, A., … Nisenbaum, R. (2022). Effect of portable rent subsidies and mentorship on socioeconomic inclusion for young people exiting homelessness: A community-based pilot randomized clinical trial. JAMA Network Open, 5(10), e2238670. https://doi.org/10.1001/jamanetworkopen.2022.38670
Thulien, N. S., Kozloff, N., McCay, E. A., Nisenbaum, R., Wang, A., & Hwang, S. W. (2019). Evaluating the effects of a rent subsidy and mentoring intervention for youth transitioning out of homelessness: Protocol for a mixed methods, community-based pilot randomized controlled trial. JMIR Research Protocols, 8(12), e15557. https://doi.org/10.2196/15557
Tsemberis, S., Gulcur, L., & Nakae, M. (2004). Housing First, consumer choice, and harm reduction for homeless individuals with a dual diagnosis. American Journal of Public Health, 94(4), 651–656. https://doi.org/10.2105/AJPH.94.4.651
Tsemberis, S., McHugo, G., Williams, V., Hanrahan, P., & Stefancic, A. (2007). Measuring homelessness and residential stability: The residential time-line follow-back inventory. Journal of Community Psychology, 35(1), 29–42. https://doi.org/10.1002/jcop.20132
Tucker, J. S., D’Amico, E. J., Ewing, B. A., Miles, J. N. V., & Pedersen, E. R. (2017). A group-based Motivational Interviewing brief intervention to reduce substance use and sexual risk behavior among homeless young adults. Journal of Substance Abuse Treatment, 76, 20–27. https://doi.org/10.1016/j.jsat.2017.02.008
Tucker, J. S., D’Amico, E. J., Ewing, B. A., & Pedersen, E. R. (2016). 251 A brief group-MI intervention reduces alcohol use among homeless young adults. Alcoholism: Clinical and Experimental Research, 40(S1), 311A. https://doi.org/10.1111/acer.13085
Tucker, J. S., D’Amico, E. J., Pedersen, E. R., Rodriguez, A., & Garvey, R. (2020). Study protocol for a group-based motivational interviewing brief intervention to reduce substance use and sexual risk behavior among young adults experiencing homelessness. Addiction Science & Clinical Practice, 15(1), 26. https://doi.org/10.1186/s13722-020-00201-x
Tucker, J. S., D’Amico, E. J., Rodriguez, A., Pedersen, E. R., Garvey, R., & Klein, D. J. (2022). Reducing alcohol use and sexual risk behavior among emerging adults experiencing homelessness: Results from a cluster crossover randomized controlled. Alcoholism: Clinical and Experimental Research, 46(S1), 14–69. https://doi.org/10.1111/acer.14831
Tucker, J. S., Kennedy, D. P., Osilla, K. C., & Golinelli, D. (2021). Motivational network intervention to reduce substance use and increase supportive connections among formerly homeless emerging adults transitioning to housing: Study protocol for a pilot randomized controlled trial. Addiction Science & Clinical Practice, 16(1), 18. https://doi.org/10.1186/s13722-021-00227-9
Upshur, C. C. (1986). The Bridge, Inc. Residential independent living project evaluation. Second year follow-up report. Washington DC: Office of Human Development Services, Department of Health and Human Services.
Ware, J. E., Kosinski, M., & Keller, S. D. (1996). A 12-item short-form health survey: Construction of scales and preliminary tests of reliability and validity. Medical Care, 34(3), 220.
Wolitski, R. J., Kidder, D. P., Pals, S. L., Royal, S., Aidala, A., Stall, R., … Courtenay-Quirk, C. (2010). Randomized trial of the effects of housing assistance on the health and risk behaviors of homeless and unstably housed people living with HIV. AIDS and Behavior, 14(3), 493–503. https://doi.org/10.1007/s10461-009-9643-x

Citation

BibTeX citation:
@online{taylor2026,
  author = {Taylor, David and Vecchio, Stephanie and Baidawi, Susan and
    Shlonsky, Aron},
  title = {Housing and {Support} {Interventions} for {Homeless} {Youth}
    in {High-Income} {Countries:} A {Systematic} {Review} —
    Supplementary Material},
  date = {2026-01-15},
  url = {https://davetayl-r.github.io/youth-homelessness-review},
  doi = {10.5281/zenodo.14503767},
  langid = {en}
}
For attribution, please cite this work as:
Taylor, D., Vecchio, S., Baidawi, S., & Shlonsky, A. (2026, January 15). Housing and Support Interventions for Homeless Youth in High-Income Countries: a Systematic Review — supplementary material. https://doi.org/10.5281/zenodo.14503767