Is the Housing First Model Effective? Different Evidence for Different Outcomes

For more than two decades since the development of the Housing First model, there have been debates about the model’s effectiveness in serving individuals experiencing homelessness. Although the Housing First model has various fidelity standards, its hallmark feature is the provision of immediate access to permanent, subsidized, independent housing with no prerequisites such as mandating treatment participation or requiring sobriety. This feature is theorized to provide an effective pathway for homeless individuals to achieve positive outcomes. Various time-trend analyses have been conducted showing that increases and decreases in homelessness have coincided with increases and decreases in housing vouchers, housing units, or implementation of the Housing First model during the same period. However, the old adage remains true that “correlation does not equal causation,” and these analyses are subject to threats to internal validity, such as history effects or other confounding factors occurring concurrently. Instead, one should look to the gold standard of research designs—the randomized controlled trial. I provide a brief synthesis of the evidence (or lack thereof) from randomized controlled trials for the Housing First model to further discussions and inform policymaking.

STRONG EVIDENCE

Of the four total major randomized controlled trials of the Housing First model,¹ three have been conducted in the United States, including the original trial of the Pathways to Housing program of Housing First in New York. Two of the randomized trials in the United States found that Housing First led to a quicker exit from homelessness and greater housing stability over time compared with treatment as usual.^2,3

In addition to these trials in the United States, a $110 million five-city randomized controlled trial was conducted in Canada called At Home/Chez Soi. Similar to studies conducted in the United States, this trial found that Housing First participants spent 73% of their time in stable housing compared with 32% of those who received treatment as usual.⁴

MODERATE EVIDENCE

A meta-analysis of randomized controlled trials of Housing First concluded that Housing First may result in reduced use of emergency department services, fewer hospitalizations, and less time hospitalized compared with treatment as usual, although variability between studies was considerable.¹ These trials are supported by a handful of observational studies that have reported similar results.⁵

The Department of Veterans Affairs (VA) serves as the largest provider of homeless services in the United States, serving more than 100 000 homeless and at-risk veterans annually for the past five years. To date, no large randomized controlled trial of Housing First has been done in VA settings. It is unknown whether research on Housing First in non-VA settings is generalizable to VA settings because the VA is unique in having an integrated, comprehensive health care system for homeless veterans unlike traditional brokered care systems for other homeless adults. One demonstration project in 2010 in the VA that used a nonequivalent groups design (https://bit.ly/2Y74ryp) found that Housing First led to reduced time to housing placement (from 223 to 35 days) and higher housing retention rates than treatment as usual (98% vs 86%).

WEAK EVIDENCE

The first randomized trial of Housing First conducted in the United States found that Housing First did not lead to greater improvements in substance use or psychiatric symptoms compared with treatment as usual.² Other trials have had similar findings on mental health, substance abuse, and physical health outcomes consistent with a National Academies of Sciences report (https://bit.ly/2Y8iaVJ) that concluded the following of permanent supportive housing (which is a broader term that includes Housing First, and the report included the Housing First studies mentioned here): “There is no substantial published evidence as yet to demonstrate that PSH [permanent supportive housing] improves health outcomes or reduces healthcare costs.” The one exception is a randomized trial of Housing First that found improved health outcomes for patients with HIV/AIDS,³ so this may be an important subgroup that experiences health benefits from Housing First. A systematic review of randomized and nonrandomized studies of Housing First also concluded that little evidence indicates that Housing First improves criminal justice outcomes.⁶

LACK OF EVIDENCE

All existing randomized controlled trials have compared Housing First with treatment as usual, which has been vaguely defined and has not used a structured approach. A meta-analysis of 44 studies involving unique community housing models, including Housing First and “non–model housing,” found that all housing models were associated with greater housing stability than no housing model, but no one model emerged as better than the others.⁷ Related to this is an important concern that some programs reportedly offering Housing First have experienced “program drift” and have deviated from model fidelity for Housing First, which is a common occurrence across many defined service models and treatments in the field.

A few observational studies have reported that Housing First is more effective for those with no major substance use disorders or particular substance use disorders over others (i.e., stimulants vs depressants), but more specific research in this area is needed. Very few studies, including observational studies, have examined heterogeneity of treatment effects to identify important subgroup differences in Housing First outcomes. If one is to assume that a one-size-fits-all approach will not work, the question of who benefits most from Housing First is important and yet has not been answered.

CONCLUSIONS

Studies have found that Housing First results in greater improvements in housing outcomes for homeless adults in North America. Housing First may lead to greater reductions in inpatient and emergency health care services but may have limited effects on clinical and social outcomes. Although supportive services are typically provided as part of the Housing First model, services are voluntary and can vary greatly between clients. Homeless adults who need Housing First also may need crucial health care and social services to help them live meaningful, sustainable, and productive lives. The debate about Housing First needs to be furthered through research to identify who benefits most from Housing First, what services are needed in addition to Housing First, and which housing models can serve as effective alternatives to the Housing First model when appropriate or necessary.