The purpose of this rapid review was to identify new research examining interventions to increase access to health and healthcare for people who are homeless or at risk of homelessness published since the 2005 systematic review by Hwang et al. , with an additional focus on the effect of these interventions on housing status. A total of 1546 new and unique articles were identified, however, less than 30% were relevant and the majority of the relevant articles were methodologically weak. As a result < 1% met inclusion criteria for this review. None of the studies were rated to be of strong methodological quality while 10 were of moderate quality. These 10 studies represent new data since the 2005 systematic review of the literature .
Concurrent issues of substance abuse, mental illness, and infectious disease make designing interventions to improve the health and housing status of homeless individuals challenging. New data included in this review indicates that provision of housing is associated with decreased substance use, relapses from periods of substance abstinence, health services utilization, and increased housing tenure [87, 88]. In addition, abstinence-contingent housing appears to provide greater impact on sustained abstinence than non-abstinence-contingent housing [87, 88]. In the review by Hwang et al. , the evidence supporting the effectiveness of case management on substance use was equivocal [104, 105], however, interventions that included post-detoxification stabilization , abstinence-contingent work therapy , or an intensive residential treatment program [108, 109] all showed significantly greater reductions in substance use than the usual care groups. These interventions all have a component of abstinence-contingency and thus are consistent with the abstinence-contingent housing interventions discussed in the current review.
The recent data included in this review [89–92] suggests that for homeless people living with mental illness, provision of housing during discharge planning from hospital is associated with maintaining stable housing. In light of the small sample size for the pilot RCT for homeless people living with mental illness , it should be noted that using the validated quality assessment tool, sample size specifically is not a quality assessment criterion but is part of the first criterion that assesses 'selection bias'. As such, this did impact the global quality rating. It should be noted, however, that despite the small sample size the pilot study was able to detect highly statistically significant differences, which is remarkable given the very limited statistical power.
These results from this review are in contrast with Hwang et al.  who reported that housing interventions did not improve health-related outcomes for homeless people with mental illness. Nevertheless, Hwang et al. found that case management with additional services, such as outreach supports or drop-in centre services, improved health outcomes [110–112]. The integration of intensive case management that includes the provision of housing may be a more effective intervention for improving health-related outcomes in homeless people with mental illness as it targets multiple factors that can affect health and healthcare utilization.
This review identified two new studies that both found structured education modules to be effective at reducing risk behaviour in homeless youth with HIV [102, 103]. These data are consistent with findings reported by Hwang et al.  that attending sessions of an educational program aimed at reducing sexual risk behaviours for HIV was associated with reduced sexual risk behaviour for HIV in homeless runaway youth when compared to usual care [113, 114]. As with other homeless sub-populations, case management appears to be an effective intervention with benefits that include improving mental health outcomes, lowering levels of aggression, aiding in social adjustment, and increasing satisfaction with quality of life [113, 114]. Unfortunately, research on this sub-population of homeless remains limited and no data exists to indicate what effect, if any, these interventions have on the housing status of homeless youth. It is also of note that, despite a wide literature search, no new methodologically strong or moderate studies were found that examined interventions for homeless women, families or children. In the previous review , only two studies examined interventions for homeless women and these found no impact following educational interventions on HIV risk behaviours or mental health outcomes [115, 116].
The current review identified 4 new studies that examined interventions for homeless people living with HIV [97–99, 101]. There does, however, remain a paucity of literature examining interventions on homeless people with HIV or other infectious diseases (e.g., tuberculosis and hepatitis-B). The 2005 review by Hwang et al.  identified 2 relevant studies that both reported no effect of educational interventions in reducing HIV risk behaviours in homeless women [115, 116]. In contrast, one study from the current review reported that individual counselling was associated with reduced substance abuse and the number of risk behaviours . Case management, when used consistently, appears to be a very effective intervention for homeless people with HIV. Case management has been found to improve mental health, use of health services, and improved overall health [99, 101]. Moreover, the addition of housing services is associated with improvement in housing status, with programs targeted for individuals with HIV/AIDS being even more effective .
Implications for Research
Of the studies identified as relevant for this review, study methodology was rated as moderate for only 10 of them, with the remaining rated to be of weak quality. Issues that resulted in studies being rated as methodologically weak were generally related to either study design or statistical analyses. Researchers should be aware of these challenges so they can address or mitigate these limitations in future investigations.
With respect to study design, it was difficult to ascertain the extent of selection bias, which would threaten the external validity of the results of many of the studies. Selection bias may have arisen as a result of recruiting participants from a single program, shelter, or city. Moreover, some studies had such stringent inclusion criteria that those criteria themselves introduced the high potential for selection bias. Blinding of outcome assessors was also rarely addressed which introduces another potential source of bias.
In terms of statistical analysis procedures, most studies lacked statistical power, as they did not have adequate sample sizes. As such, it was difficult to attribute a lack of between-group differences to the intervention not being effective and not simply a type II statistical error. Given the recruitment and follow-up challenges associated with individuals in this population, researchers can attempt to maximize statistical power by increasing actual effect size, decreasing sample variability, and increasing precision of outcome measurements. Many studies also failed to include basic statistical data such as effect size and lacked specific detail regarding outcome measures, thereby limiting the outcome analysis. Finally, the use of intention-to-treat analysis was rarely specified even in appropriate situations.
Taking the unique methodological challenges associated with studying those who are homeless or at risk of homelessness into consideration, findings from studies rated as methodologically weak were briefly presented in this review where appropriate. Regardless of quality, these data contribute to the overall body of knowledge of how to best increase access to health, healthcare, and housing for those who are homeless or at risk of homelessness. Moreover, the lack of new studies of moderate or strong quality informs future research directions by identifying knowledge gaps. Additional research is warranted examining homeless subgroups of women, families, and children who have thus far been understudied. Furthermore, as homelessness is associated with a wide range of chronic disease such as HIV/AIDS, tuberculosis, schizophrenia, diabetes and hepatitis C [117–119], specific interventions targeting these conditions are needed.
As a rapid review, this literature synthesis has a number of limitations. Some of these limitations are a product of the short timelines determined by the contracting agency to conduct the review. For example, grey literature searching was limited in its scope, conference proceedings and trial registers were excluded, and a limited number of relevant websites were selected for searching. In instances where data were unclear and/or incomplete, time constraints prohibited contacting authors to clarify data and citation tracking for subsequently published studies was not feasible. As a result of these limitations, it is possible that some potentially relevant studies were missed in the search. A further limitation of this review is that it synthesizes only methodologically moderate articles, as no methodologically strong studies were found and weak studies were not discussed in detail.