Background

Youth homelessness is a major public health challenge worldwide, even in high income countries [1]. Youth experiencing homelessness are defined as, “youth between the ages of 13 to 24 who live independently of their parents or guardians, but do not have the means to acquire stable, safe or consistent residence, or the immediate prospect of it [2].” Youth pathways into homelessness are anomalous and seldom experienced as a single isolated event. Compared to the adult homeless population, youth experiencing homelessness are more likely to report leaving home due to parental conflicts, including: being “kicked out” of the home, abuse (physical, verbal, sexual and other), parental neglect due to mental health problems, or parental substance use [3,4,5,6,7,8,9,10,11]. The broader context of family dysfunction can lead to youth circumstances that further reinforce situations of homelessness, including desire for separation from unsupportive environments, financial independence, mental health challenges, substance use, and/or run-ins with the justice system [1].

Not only are youth’s pathways into homelessness different from the adult homeless population, but their experiences on the street are distinct as well. Once homeless, youth are exposed to many dangers and are at a high risk of further trauma [12]. Youth experiencing homelessness may face a number of daily stressors and have limited coping strategies and resources to deal with these stressors [13]. Youth homelessness is often invisible and includes vulnerable housing situations such as couchsurfing or staying with relatives [14]. Furthermore, youth experiencing homelessness are vulnerable to social and health inequities, which describe the fairness in the distribution of health opportunities and outcomes across populations [15]. Health inequities are differences in health status that are unfair and/or avoidable [16]. Often, the compounding effect of various stratifying characteristics can result in increased disparities between individuals.

Current research has largely focused on adult populations, with a gap in evidence on interventions for youth experiencing homelessness on a broad range of outcomes. Among the current interventions for individuals experiencing homelessness, non-abstinence contingent permanent supportive housing and case management have shown promising results in terms of improving housing stability and mental health outcomes [17]. However, youth are a distinct population and they require specifically tailored, context appropriate, equity-focused interventions and research attention [18]. From systematically searching the literature for youth interventions, this paper will introduce four main categories of interventions applied to youth experiencing homelessness: 1) individual and family therapies (ie. cognitive behavioural therapy, motivational interviewing, etc.) 2) skill building programs, 3) case management, and 4) structural interventions (such as housing support, drop-in centres, and shelters). These interventions are designed to address the complex, multifaceted pathways and contributors to youth homelessness, whether it be addressing substance use issues through motivational interviewing, mental health care through cognitive behavioural therapy, improving unstable family environments through family therapies, increasing access to resources through case management, and enhancing structural support such as income and housing support [19,20,21,22,23]. Given the complexity and interconnectedness of these outcomes, one would hope that these interventions would have an impact on not only the primary outcome, but also extend to other facets of a youth’s life. For instance, family therapies have shown promising results on both family functioning as well as substance use, by addressing the toxic family environment and thereby decreasing its contribution to unhealthy substance use patterns [24].

Current research on interventions for the population of youth experiencing homelessness lacks a comprehensive synthesis on a broad range of social and health outcomes. The objective of this review is to synthesize the existing scientific literature on interventions for homeless or vulnerably housed youth in high income countries, and assess the impacts of the interventions on housing, mental health, substance use, and family cohesion, with an equity perspective.

Methods

We established an expert working group consisting of homeless health researchers, academics, clinicians and youth with lived experience of homelessness to conduct this review. We report our results according to PRISMA-E [see Additional file 3] and published an open access protocol on the Campbell and Cochrane Equity Methods website [25, 26].

Data sources and search strategy

Without language restrictions, we systematically searched the following databases from inception until February 9, 2018: Medline, Embase, CINAHL, PsycINFO, Epistemonikos, HTA database, NHSEED, DARE, and Cochrane Central. Combinations of relevant keywords and MeSH terms were searched, including “homeless” and “homeless youth” [see Additional file 1 for search strategy]. We hand-searched included studies for primary studies and consulted experts for additional papers. We conducted a grey literature search on homeless health and public health websites.

Inclusion and exclusion criteria

We downloaded citation information into Rayyan online software [27]. All title and abstracts were screened according to our inclusion criteria (see Table 1) in duplicate by two independent reviewers, and any discrepancies were resolved. Throughout a process of several consultations, our working group, consisting of persons with lived experience and experts in the field, helped develop these inclusion criteria by identifying priority areas in which to focus this review. This study focused on youth between the ages of 13 to 24, however, the age categorizations of youth tend to differ between various definitions, with the medicolegal definition utilizing ages 16 to 21. It is important to note that the broader age range utilized in this paper may lead to risks of over-inclusion, but it was chosen as it is reflective of the currently literature on youth homelessness and includes both high school and university students who are generally still dependents living with family or relying on them for financial or moral support.

Table 1 Eligibility criteria
Full size table

Data extraction and analysis

Data extraction proceeded in duplicate using a standardized data extraction form and a third reviewer resolved discrepancies [25]. We extracted data regarding the effectiveness of interventions on a broad range of social and health outcomes. We conducted a scoping exercise to identify key outcome categories in the literature and prioritized reported outcomes with our expert working group members, which included individuals of lived experience. The outcomes rated as being of highest priority (mental health, substance use, housing, and family outcomes) are reported in the body of this paper, and the remaining outcomes (violence, sexual health, personal and social, and health and social service utilization) are reported in the appendix [see Additional file 2]. To reduce overlap between single studies and systematic reviews, we reported the results of systematic reviews and supplemented with data from randomized control trials (RCTs) that were not included in the systematic reviews. Due to heterogeneity of interventions and outcomes studied, we qualitatively synthesized the results. We created a forest plot to summarize RCTs for mental health outcomes, as sufficient data were available and it was a highly ranked outcome.

Health equity analysis

We used the PROGRESS+ framework to apply a health equity lens and enable us to identify characteristics that socially stratify youth experiencing homelessness, and various drivers of homelessness [15]. In particular, we extracted the following from studies to inform our analysis: 1) study rationale for focusing on youth-centred interventions; 2) the measures used to assess differences in outcomes for women and men; 3) the study’s gender-related findings and conclusions; and 4) the study’s incorporation of equity considerations (e.g. race/ethnicity and socioeconomic status).

Critical appraisal

We assessed the methodological quality of systematic reviews with AMSTAR II and RCTs using the Cochrane Risk of Bias Tool [28,29,30,31,32]. When assessing the overall risk of bias of RCTs, we defined the risk of bias as “not serious” when there were low risk ratings in all categories or one or two unclear risk, “serious” with one or two high risk categories, and “very serious” with more than two high risk categories.

Results

The search strategy yielded 11,934 potentially relevant citations. After we removed duplicates, we screened 7499 citations and assessed 103 full text articles. Twenty-two citations met the full inclusion criteria (See Fig. 1). Four of the included citations were systematic reviews [33,34,35,36] and the remaining 18 citations reported on 15 RCTs (see Table 2 for RCTs and Table 3 for SRs) [19, 21, 37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53].

Fig. 1
figure 1

PRISMA Flow Diagram

Full size image
Table 2 Characteristics of Included RCTs
Full size table
Table 3 Characteristics of Included Systematic Reviews
Full size table

Methodological quality of the included studies was low or very low, with serious risk of bias across most included studies (see Fig. 2 for RCTs and Table 4 for SRs). The most common domain with a high level of risk was knowledge of the allocated interventions, as blinding was often not possible or difficult with the nature of the interventions.

Fig. 2
figure 2

Methodological Quality of Included RCTs using Cochrane Risk of Bias Tool

Full size image
Table 4 Methodological Quality of Included Systematic Reviews using AMSTAR II
Full size table

The main categories of interventions applied to youth homelessness included: 1) individual and family therapy (e.g. cognitive behavioural therapy (CBT), motivational interviewing (MI), family therapy), 2) skills building (e.g. life skills, mindfulness), 3) case management and 4) structural interventions (e.g. housing support, drop-in centres, shelters). See Table 5 for the definitions of interventions. The results of RCTs have been summarized using a visual map (see Fig. 3).

Table 5 Definitions of Interventions
Full size table
Fig. 3
figure 3

Visual Summary of Results of RCTs by Outcome

Full size image

Individual and family therapy

Cognitive Behavioural therapy

CBT led to improvements in substance use and depression, and one systematic review also reported improvements in internalizing behaviours and self-efficacy [33,34,35,36]. When a CBT-based therapy (community reinforcement approach) was delivered with case management in one study, there were improvements in percentage of days being housed, psychological distress, and substance use [33]. Two systematic reviews conducted meta-analyses on CBT and CBT-based interventions and found no statistically significant difference in mental health outcomes compared to services as usual, but noted that lack of a statistically significant difference may be due to heterogeneity between studies [34,35,36].

Family therapy

Family-based therapy was delivered in an office setting, known as functional family therapy, or in the home setting, called ecologically-based family therapy. Systematic reviews reported that all three family therapy RCTs showed a reduction in substance use [34,35,36]. However, Noh (2018) conducted a subgroup meta-analysis on two family intervention studies and found no significant effect on substance use [34]. Another meta-analysis found a statistically significant improvement in family cohesion, but called it a clinically marginal effect [36]. In a three arm RCT comparing home-based family therapy with MI and a CBT-based therapy, all three groups improved over time in internalizing and externalizing behaviours, family cohesion, and substance use [47,48,49]. Furthermore, when an RCT compared functional family therapy, home-based family therapy, and services as usual, all treatments showed improvements in days living at home at three, nine and 15 months, but no group was superior to another [52].

Motivational interviewing

Brief or group MI interventions were primarily designed to address substance use and/or risky sexual behaviours. A brief intervention showed declines in non-marijuana drug use at 1-month follow up, but the reduction was no longer significant after 3 months [33,34,35]. In another RCT, both the service as usual and intervention groups showed significant improvements over time, but there were no significant and durable results in favour of the experimental group [21]. A 16-week group MI intervention found significant declines in alcohol use and increased motivation to change drug use, but no significant decreases in marijuana use [37]. A two-session individual brief MI intervention compared to an education program reported significant improvements in readiness to change alcohol use [38].

Skill building

The interventions focused on vocational and life skills, mindfulness, and strengths-based skill building. One systematic review included one study evaluating a life skills intervention and found improvements in family contact and near significant improvements in depressive symptoms [33]. Another systematic review reported similar results but noted an increase in substance use over 6 months which could not be explained [35]. A training program based on a peer influence model showed non-statistically significant decreases in drug use in the treatment group. One study evaluated a strengths-based program deployed in a shelter to identify and make use of strengths in each youth [39]. This program showed no significant differences between groups but found improvements over time in depression, substance use, and satisfaction with family relations [39]. Two RCTs evaluated a vocational and life skills program and a mindfulness skills program, though did not report promising treatment effects [40,41,42].

We attempted to conduct meta-analyses whenever possible, but due to the heterogeneity between studies, it was inappropriate to pool the results into a combined effect size. As such, we developed a forest plot for short-term mental health outcomes of a mindfulness intervention, CBT intervention, strengths-based intervention, and CBT-based intervention [39, 42, 50,51,52,53]. The figure depicts a general trend favouring the interventions but none reaching statistical significance compared to control (see Fig. 4).

Fig. 4
figure 4

Intervention vs. Usual services for Short Term (0-6 months) Mental Health Outcomes)

Full size image

Case management

Two systematic reviews reported on several case management programs, including intensive case management and multidisciplinary case management, and reported minimal additional benefit of the programs relative to their comparison interventions [33,34,35]. They noted that one program showed favourable results for substance use, but the study quality was very low due to low retention rates [33]. In a three-arm RCT, case management, a CBT-based intervention, and MI all showed significant improvements over time in housing stability, depression, and substance use, but no significant differences between groups [45]. Case management led to improvements over time in internalizing behaviours while the other groups did not [45]. Overall, there is evidence to suggest that case management may have impacts on substance use, depression, and housing stability, but different control conditions in each of the studies made it difficult to assess overall effectiveness of the intervention.

Structural support

Housing programs

A subgroup analysis of young adults in an RCT of the housing first model for adults with mental illness found that, compared to treatment as usual, housing first significantly increased the proportion of days stably housed over the 24-month trial, but had no impact on self-rated mental health [43]. One systematic review included an independent living program and reported marginal results on psychological measures, however reported some positive outcomes on housing status [33]. The same systematic review also included a study evaluating a supportive housing program, which reported lower rates of substance abuse and improvements in self-reported health, but the study quality was noted to be low. Xiang evaluated the same supportive housing program and also concluded that the lower rates of substance use may be attributed to baseline differences between control and intervention groups instead of treatment effect [35].

Drop-in and shelter services

A systematic review included three shelter services studies, two evaluating residential services and one evaluating emergency shelter and crisis services [35]. The review showed some improvements in substance use but this was not consistent over the various studies and there were no enduring effects over time. An RCT compared referrals from case management made to drop-in versus shelter services programs [44]. There were no differential treatment effects, as both groups showed decreases in depression and substance use over time [44]. However, individuals assigned to the drop-in service had greater service contacts and access to care over 6 months [44].

Gender and equity analysis

Equity variables were not consistently measured, reported, or analyzed across studies. Several studies measured equity and PROGRESS+ factors with baseline sample characteristics, but very few included them as covariates. The most examined factors were gender and ethnicity/race, with some studies mentioning place of residence and occupation. A number of RCTs included equity variables in their analysis [21, 37, 39,40,41, 43,44,45,46,47,48,49], as did three systematic reviews [34,35,36].

A number of studies indicated that females responded differently to services than males. Slesnick’s studies have showed that females initially reported higher rates of depression than males, with a greater reduction throughout the study [44,45,46]. Female adolescents showed a greater improvement in family cohesion subsequent to treatment regardless of the treatment condition [47] and appeared to derive greater benefit from shelter services than males [35].

Some variance in relation to ethnicity and employment emerged as well. While youth from ethnic minorities had greater reductions in substance use, they also relapsed more quickly than white youth [49] and had more HIV risk behaviours [44]. African Americans showed a greater reduction in percent days homeless than other ethnic groups [45]. Non-Hispanic white youth more quickly reduced their number of days drinking to intoxication [44]. Those employed or in school at baseline were more likely to remain employed at follow-up [39].

Discussion

This review identified a wide variety of interventions for youth experiencing housing instability. Regarding individual and family therapies, CBT interventions showed improvements in depression and substance use outcomes [33,34,35,36]. Family interventions led to improvements in alcohol and drug use measures and may have had an impact on family cohesion [34,35,36]. Motivational interviewing, skill-building programs and case management showed inconsistent effects on mental health and substance use when compared with services as usual and other interventions [21, 33, 35,36,37,38,39,40,41,42, 45,46,47,48,49]. Among the structural support interventions, housing first led to improved housing stability outcomes, while drop-in and shelter services led to inconsistent effects [43, 44]. The equity analysis revealed differential treatment effects based upon gender and ethnicity, with females often deriving more treatment benefit than males [44, 45, 47,48,49]. Equity analyses were limited, with very little mention of important considerations such as sexual orientation status, as LGBTQ+ youth are disproportionately represented in the homeless population [58, 59].

While in many circumstances, differences were not statistically significant between treatment groups, this does not preclude the lack of effectiveness of these interventions. It is important to note that a treatment as usual group was not the absence of an intervention, but rather involved referral to other community services and follow-up with researchers. This may lessen the differences between the intervention and control arms, and decrease the detectable effect of the intervention. Providing non-specific support for youth may be enough to improve outcomes and reduce the toxic effects of adverse childhood experiences. However, that regression to the mean may also potentially explain the changes observed over time [60]. As participants may enter the research studies during a point of crisis, they may naturally improve over time regardless of the study group, and this effect may lessen the observed differences between intervention and control groups.

Tailoring interventions to the needs of youth

The dynamics of youth homelessness are complex; pathways to housing are precarious, sociocultural backgrounds are becoming increasingly diverse and available resources are inconsistent. Research has shown that unstable family relationships underlie youth homelessness, and many youth have left homes where they experienced interpersonal violence and abuse [3,4,5, 61]. Among these difficult family issues, other personal factors arise as a result of their environmental contexts, which can interplay and lead to increased distress. These challenges include substance use, depression, and disability, and can compoundly contribute to strain [10]. The interventions identified in this review may help to address the specific needs of youth and may be tailored to their situation.

One important consideration to note is that while we have defined youth as those ages 13 to 24 for the purposes of this study, this grouping brings together minors as well as young adults of legal age. While this age categorization is reflective of the literature on the youth population, we recognize that there are differences between the experiences of younger versus older youth. Furthermore, there are medicolegal implications of the mature minor and capacity to consent. Clinicians and program implementers who work directly with this population need to consider the ethical considerations of consent for treatment participation with mature minors as well as the legal obligations provided by their governing college [62].

Strengths and limitations of the review

We conducted a high quality search, complying to PRISMA-E guidelines [26]. This review included only high quality study designs: RCTs and systematic reviews. This may, however, have limited the types of interventions that were included. Limitations include a broad range of outcomes and, thus, too few studies available for meta-analyses. There was heterogeneity in the interventions, and the available evidence was insufficient to use network meta-analysis to answer the question of the relative advantages of the different types of interventions. In our systematic review, the studies did not use placebo designs and, instead, used several different interventions/comparisons. However, there was considerable heterogeneity in the outcome measures and this prevented a pooling of the effects. The services-as-usual comparisons were often not adequately described in the primary studies, limiting the comparisons that could be made across different studies. Furthermore, our definition of youth experiencing homelessness focused on unaccompanied youth and did not include accompanied youth that enter homeless situations along with their families, as this youth population has quite distinct circumstances and needs.

Implications for future research, policy, and practice

The results suggest that tailored interventions for youth may have impacts on depression, substance use and housing. Given the diverse pathways to youth homelessness, health care policy-makers, practitioners and other stakeholders should consider the specific needs of youth during prevention and delivery of care. Furthermore, we recommend additional high quality research to be conducted in the area of family-based therapies, CBT, and housing interventions, which have shown some positive results thus far. We further recommend additional considerations for equity factors. Few studies examined equity factors, and those that did were limited largely to gender and ethnicity. There remains a large gap in data regarding the intersectionality between a variety of PROGRESS+ factors contributing to youth experiences.

There is also a large gap in research on the impact of structural interventions such as housing and case management on youth experiencing homelessness. The predominance of psychological and family interventions in this paper suggests that more work could be done to study an area in which it may be more difficult to design studies. Nonetheless, future research on these interventions are important to addressing the root causes of poverty and homelessness. Furthermore, there are emerging models of housing which have not yet been evaluated rigorously in the literature. For instance, host homes provide safe and temporary housing for up to 6 months for youth while supporting them with a case manager to identify long term solutions [63]. Rapid re-housing programs provide short-term subsidies to allow persons experiencing homelessness to acquire stable housing as quickly as possible [64, 65]. The landscape on housing models continues to evolve and future research will need to evaluate these in the context of youth experiencing homelessness.

Conclusion

This review identifies a variety of interventions targeted towards the unique needs of youth experiencing homelessness. CBT interventions may lead to improvements in depression and substance use, and family-based therapy may impact substance use and family outcomes. Housing programs may lead to improvements in housing support and stability. Other interventions such as skill building, case management, show inconsistent results on health and social outcomes.