Housing the Chronically Homeless: Title and subTitle BreakHigh Hopes, Complex Realities

During the last decade, federal policy has concentrated on ending “chronic homelessness,” a designation that applies to approximately 18% of the 671 888 persons identified as homeless on a single night in 2007.¹ Few interventions have attracted more interest than an approach to permanent housing known as Housing First.² Two articles in JAMA, one published in the April 1, 2009, issue³ and one in this issue,⁴ advance understanding of this housing approach and raise questions about where and to whom it should be applied.

Housing First offers permanent housing to homeless individuals or families, with few requirements for participation or success in rehabilitative activities such as medical, addictive, or psychiatric treatment.⁵ In contrast, traditional approaches require homeless persons to demonstrate such success before providing long-term housing support or helping clients obtain their own housing on the private market. The traditional approaches have worked for many persons but have failed repeatedly for others, especially those who have lived for years on the streets or in shelters. Communities incur steep costs when severely debilitated homeless persons cycle repeatedly through treatment programs, hospitals, and jails, often in quick succession. A 2002 study⁶ reported that mentally ill homeless persons in New York accrued $40 451 per person in yearly expenses across the judicial and health sectors, but service costs declined appreciably with housing support, although this pattern has not been seen uniformly in other data.⁷

Housing First is grounded on the premise that individuals who have stable housing will have reduced need for other public resources, saving taxpayers money. The Seattle Housing First study³ powerfully demonstrated the savings that accrue when individuals with especially costly patterns of health and judicial service utilization are housed. In that study, Seattle's Downtown Emergency Services Center recruited chronically homeless persons with severe alcohol dependence and histories of heavy service utilization across the health and judicial sectors (sometimes termed “chronic public inebriates”). Recruited individuals averaged 16 prior addiction treatment episodes and reported consuming more than 15 alcoholic drinks per day. Such persons match the stereotype of a small but painfully public subgroup of chronically homeless individuals. Despite controversy, the Downtown Emergency Services Center offered permanent housing in a rehabilitated building, explicitly permitting residents to drink alcohol in their rooms. Clients were offered an array of services, including addiction treatment. A preliminary report of the initial 75 clients showed that two-thirds remained housed after 1 year.⁸ In their published report, the Seattle team compared health and judicial costs for 91 housed clients with costs for 35 wait-listed controls. Adjusting for differences between these groups, the authors reported a net savings of $3822 per housed person per month, after accounting for housing program costs. Additionally, drinking declined by a median of 5.1 drinks per day at 12 months among those housed. This is the second study to show declining alcohol intake when homeless persons with refractory alcohol dependence are offered housing and permission to drink.⁹

A caution to communities seeking to replicate the Seattle experience: capital costs ($11.2 million¹⁰ ) are not included in their analysis. This is not necessarily a flaw in the research, because capital expenditures are typically drawn from separate funding pools. However, the capacity to raise such capital will determine which communities are able to consider Housing First at all. Additionally, the Seattle study³ was not a prospective randomized controlled trial, and individuals offered housing differed from wait-listed controls in ways that might have contributed to the favorable results. Nevertheless, careful analyses helped to minimize this bias.

In light of the unavoidable uncertainties of making observational comparisons without random assignment, the report in this issue of JAMA from a prospective randomized controlled trial of Housing First in Chicago is especially welcome.⁴ To our knowledge, this is only the second published randomized controlled trial of Housing First and the first to focus on chronically homeless persons with medical illness in medical settings. The previous trial focused on severely mentally ill individuals.⁵

In the Chicago study, Sadowski and colleagues⁴ recruited 407 homeless adults with chronic medical illness and randomly assigned half to a medical respite care program,¹¹ followed by expedited placement to one of several community-based housing programs. Participants in the control group had access to usual care, including a hospital social worker. Among 176 participants in the housing intervention group alive at 18 months, 66% were housed at 18 months, compared with 10% among in the usual care group. As in Seattle, housing success was accompanied by reductions in the use of health services. After 18 months, significantly fewer participants in the intervention group had 2 or more hospitalizations (48% vs 59%) or 3 or more emergency department visits (33% vs 50%), compared with the usual care group. Contrary to expectations, however, the housed group did not attain superior health status.

The results reported by Sadowski et al⁴ confirm the decrease in health service utilization achievable with Housing First. Although not calculated by the authors, the potential cost savings are likely to be less impressive than those obtained in Seattle because individuals enrolled in the Chicago study had not been as dependent on public resources prior to enrollment. Before the housing intervention, Seattle participants averaged nearly a dozen annual contacts with a single major medical center (not counting numerous other health and judicial contacts). Those in Chicago's Housing First intervention group had a median of only 1 emergency department visit and 1 hospitalization at the 2 primary study sites over the course of a year (the median was 1 emergency department visit and zero admissions for the usual care group).

Even with generous assumptions favoring the Housing First intervention, back-of-the-envelope estimates suggest that overall cost savings in Chicago would not approach the magnitude seen in Seattle. In the Chicago study, compared with usual care participants, those in the intervention group had 2.7 fewer hospital days per person per year and 1.2 fewer emergency department visits per person per year. Using national averages of $688 for emergency department visits and $3320 for a hospital night including physician fees (based on 2006 data from the Medical Expenditure Panel Survey),¹² the health cost offsets could increase to as high as $9790 per person per year. Estimated yearly costs of offering Housing First typically range from $12 000 to $16 000 ($13 440 in the Seattle study). From a financial perspective, offering Housing First to persons similar to those comprising the Chicago sample seems less likely to produce substantial overall cost savings. The probable reduction in savings illustrates how financial returns are likely to decrease when Housing First is offered to less severely debilitated individuals.

These studies compellingly demonstrate how the provision of secure housing to the most vulnerable members of society—the sickest of the chronically homeless—can be a win-win situation for all parties concerned. When Housing First is offered to a wider subset of the homeless population, the strictly economic benefits will likely diminish or disappear. A central question is thus raised: how far should such programs be expanded?

The economic argument for Housing First programs is most powerful when communities enroll individuals who are frequent users of health and judicial services, particularly those with debilitating medical conditions as well as severe mental illness, severe alcoholism, or both. When housing other subgroups of the homeless population, cost savings are uncertain, and important questions remain unanswered. Most Housing First data derive from cities that spend generously on social services (eg, New York, Seattle, San Francisco). Will similar savings be realized in cities without such historic spending patterns or with fewer resources? How will decisions be made about which frequent users to house? The ethical value of relieving misery and the civic impact of reducing homelessness justify additional societal expense, but sometimes the ethical argument is less persuasive to policy makers than a case for cost savings.

Uncertainty still remains about the applicability of Housing First programs for homeless persons with active and severe addictions.¹³ Severely alcohol-dependent persons may be accommodated, but what if the substance is an illegal drug? What if the user is the parent of a homeless family, the most rapidly increasing segment of the homeless population? Evidence-based treatments for addiction in homeless persons exist and also promote housing success,¹⁴ so the question arises: can the woefully underfunded addiction treatment system in the United States be refurbished or integrated with housing in ways that would help?

The findings of the Seattle and Chicago studies do not apply to the 82% of homeless individuals who are not chronically homeless. The continuing plight of these individuals should recall the systemic factors that contributed to the precipitous increase in homelessness that began nearly 30 years ago: the vanishing supply of affordable rental housing, changing family arrangements, declining workforce participation by men, increasing dependence on illicit drugs, diminished cash supports for the poor and disabled, and deinstitutionalization of the mentally ill.¹⁵ To these social stresses can now be added a major recession and the return of more than 1.5 million combat veterans from Iraq and Afghanistan, many of whom will need clinical as well as social services.

The studies by Larimer et al³ and Sadowski et al⁴ add to the increasing evidence that at least some large US cities cannot afford not to house some who live on their streets. These studies demonstrate that for the most frequent users of costly public services, service use substantially abates when individuals have stable housing. The challenge now is to determine which subgroups of the homeless population could benefit most from Housing First, a valuable new approach—if not a panacea—in the quest to end homelessness.