The evidence of the harm caused by homelessness indicates that individuals who experience chronic homelessness are at higher risk for infections (including human immunodeficiency virus [HIV]), traumatic injuries, drug overdoses, violence, death due to exposure to extreme heat or cold, and death due to chronic alcoholism. These individuals are more likely than housed persons to use the emergency department for health care needs and to be admitted to the hospital; they are also less likely to have a usual source of health care. Individuals experiencing homelessness have longer hospitalizations for the same illnesses as housed persons, often because it is simply not safe (or humane) to discharge them to the street when they are still recuperating from the condition that brought them to the hospital, even if they are no longer acutely ill (Salit et al., 1998). Additionally, individuals experiencing chronic homelessness have markedly shorter life spans. Whereas the average U.S. life expectancy is 78.8 years (Kochanek et al., 2016), individual adults who used homeless shelters in Boston, New York City, and Philadelphia were found to have an average life expectancy of between 42 and 52 years (Hwang et al., 1997).1
A myriad of reasons contribute to premature death among persons experiencing homelessness, including illnesses and injuries. According to the Centers for Disease Control and Prevention (CDC), the top five leading causes of death in the United States are heart disease, cancer, chronic lower respiratory diseases, unintentional injuries, and stroke (CDC, 2016). The leading causes of death for individuals experiencing homelessness are drug overdoses, HIV, and common chronic diseases such as heart disease and cancer, as was found in retrospective data from more than 28,000 adults experiencing homelessness in Boston (Baggett et al., 2013).2 Aside from a higher mortality rate, persons experiencing homelessness are three to six times more likely to become ill than housed persons (Zlotnick and Zerger, 2009). Diseases that are common among the homeless population include heart disease, cancer, liver disease, kidney disease, serious skin infections,
1 Because mortality rates were used, it is not specified whether the individual adults had experienced chronic or episodic homelessness.
2 See footnote 1.
Few studies have been able to assess mortality in the sub-group of homeless individuals who live on the streets. These elusive “rough sleepers,” vulnerable to the extremes of weather and violence, may have higher mortality rates than homeless persons who utilize shelters. Despite intense media scrutiny and high public visibility, little is known of the health and health care outcomes of this group of chronically homeless individuals. (p. 12)
Among older homeless adults in Oakland, the prevalence of common geriatric conditions was higher than for housed adults 20 years older (Brown et al., 2017). In addition to health problems, individuals experiencing homelessness are also more likely to be arrested and incarcerated.
Permanent supportive housing (PSH) is designed to reduce chronic homelessness. Hopper and Barrow (2003) have assessed two distinct genealogies of supported housing, including from (1) within the mental health field that advocates less-structured housing alternatives to clinically managed residential programs for persons with severe mental illness (“housing as housing”); and (2) from the movement to arrest homelessness by preserving and increasing the supply of low-income housing (“integrated housing development”). As noted earlier, many people experiencing chronic homelessness have problems that make it difficult for them to live in housing unassisted. They may be unable to find an apartment, arrange for food and utilities, or pay rent without a money manager. They may need mental health or substance abuse counseling, or simply need help getting along with their neighbors. A large body of research makes the case that in the absence of supportive services, those with chronic problems may not be able to sustain their housing even if the housing is initially provided, although this proposition has not yet been put to an experimental test (Lipton et al., 2000; Caton et al., 2007; Byrne et al., 2014; Rog et al., 2014). The “permanent” aspect of PSH addresses the problem that many people who have been homeless will not be able to enter the regular housing market in the future and are therefore not well served by time-limited transitional housing models because at the end of the transition time, they will not be able to sustain housing and will cycle back to homelessness.
In this chapter the health benefits of PSH are discussed. In reviewing the published and gray literature,3 the committee was guided by the World Health Organization (WHO) definition of health as “a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity” (WHO, 1946). Thus, evidence is presented on the degree to which PSH increases the likelihood of individuals experiencing chronic homelessness remaining housed; increases or decreases health care utilization; and affects physical health, mental
3 Gray literature is “the unpublished, non-commercial, hard-to-find information that organizations such as professional associations, research institutes, think tanks, and government departments produce” (University of Michigan Library, 2018).
health, substance abuse well-being, and incarceration. The studies below are summarized as presented in the literature. However, the committee acknowledges that many of these studies have follow-up periods that are often limited to 1 to 2 years, and in some settings may offer individuals in the control groups housing. Finally, the committee proposes a new paradigm to better understand the impact of permanent supportive housing on health, which includes the specification of “housing-sensitive conditions.” An overview of the selected studies on the effectiveness of permanent supportive housing is provided in Appendix E.
EVIDENCE OF THE IMPACT OF PERMANENT SUPPORTIVE HOUSING FOR HEALTH OUTCOMES
Permanent supportive housing can decrease the harms of homelessness only to the extent that people stay housed. Several studies have demonstrated that individuals experiencing homelessness who are also chronically ill who are randomized to PSH spend significantly fewer days homeless than those who receive usual care. For example, in the Canadian At Home/Chez Soi study, homeless adults were randomized to receive scattered-site housing with intensive case management (following the Pathways Housing First model) versus usual care in five cities (Stergiopoulos et al., 2015). In all five cities, those randomized to scattered-site housing with intensive case management spent more days stably housed than those receiving usual care, with the adjusted mean difference in days stably housed ranging from 33.0 percent to 49.5 percent over a 24-month period. Similar findings were seen in a study conducted in Chicago that randomized chronically ill homeless individuals after recovery from hospitalization to stable housing plus case management compared to usual care. Those in the intervention group had an annualized 62 more days in stable housing than those in the usual-care group (Basu et al., 2012).
While it may seem obvious that persons who receive housing would be more likely to be housed, prior to the dissemination of the results of several successful supportive housing programs, there was a common belief that individuals experiencing chronic homelessness would be unable to maintain themselves in housing because of problems stemming from mental illness and/or substance use. Those holding this belief favored a model, often referred to as “treatment first,” whereby persons experiencing homelessness suffering from mental illness and/or substance use would first receive treatment, and when they achieved sobriety they would then move to housing. However, when the Pathways Housing First (permanent supportive housing) model was compared to a treatment-first model among persons experiencing homelessness with a dual mental health and substance abuse diagnosis in New York City, the Housing First model proved superior; those randomized to PSH were housed sooner and spent more days in stable housing than those randomized to housing contingent on first achieving sobriety (Tsemberis et al., 2004; Gulcur et al., 2007).
Consistent with the results of randomized trials, observational studies of retention of persons experiencing homelessness in supportive housing showed that most programs had high annual retention rates, indicating that PSH is able to keep persons who have formerly experienced homelessness off the streets for significant periods of time. For example, Collins et al. (2013) found that in a nonrandomized controlled trial in Seattle, only 23 percent of the participants returned to homelessness over a 2-year period. Tsemberis (1999) found a retention rate of 84 percent over a 30-month period in New York City, although some participants did occasionally leave for short periods of time to receive other treatments. Aidala et al (2013) revealed that 86 percent of a Frequent Users Services Enhancement (FUSE) group was still housed after 24 months, through the FUSE initiative in New York City that provided supportive housing to roughly 200 individuals who were frequently cycling in and out of jails and homeless shelters. Finally, Wong et al. (2006) found that in a study of 943 PSH residents in Philadelphia, almost 60 percent were tracked as being “stayers.”
Except for some evidence that PSH improves health outcomes among individuals with HIV/AIDS, the committee finds that there is no substantial published evidence as yet to demonstrate that PSH improves health outcomes. However, although this was the inescapable finding based on an impartial review of the evidence available at the time of this assessment, the committee believes that housing in general improves health, and that PSH is important in increasing the ability of some individuals to become and remain housed. Remaining housed should improve the health of these individuals because housing alleviates a number of negative conditions that detract from their ability to achieve “a state of complete physical, mental and social well-being” (WHO, 1946).
Individuals who live on the street are subject to extremes of the elements (e.g., freezing temperatures, extreme heat, sun exposure, and rain); lack of places to wash, urinate, and defecate; lack of a place to lie without undue pressure on the skin; lack of refrigeration (for food or medicines) or cooking facilities; lack of privacy; lack of place for social interaction; lack of stable address for receiving services, receiving mail, or hosting family members or visitors; exposure to violence, victimization, drugs, and injection drug use; and lack of places for intravenous drug users to safely and cleanly inject with resultant increased risk for infections such as HIV, hepatitis B virus (HBV), and hepatitis C virus (HCV). Sustained housing provides a platform from which other physical, mental, and social concerns can begin to be addressed.
Health Care Utilization
Individuals experiencing chronic homelessness are often frequent users of health care services, including high-cost services. They may have longer hospital stays or be infrequent users of primary care services (Martin, 2015; Bowen, 2016). The literature provides some insights on the extent to which PSH can affect the utilization of health care services.
In a randomized study of a Housing First4 program, including case management and housing, versus usual care for individuals experiencing homelessness in Chicago, the investigators found both positive and negative results: those in the intervention group incurred 2.6 fewer hospital days, 1.2 fewer emergency room visits, 7.5 fewer days in residential substance abuse treatment, 9.8 fewer nursing home days, and 3.8 more outpatient visits each year (Basu et al., 2012) when compared to the usual-care group. However, as expected by the authors, the intervention group had higher costs in outpatient visits, housing, and case management when compared to the control group.
In the original experimental evaluation of the Pathways Housing First program in New York, 225 individuals experiencing both chronic homelessness and mental illnesses were randomized to Housing First or to usual care. Over the next 24 months, the Housing Frist group spent significantly less time in psychiatric institutions and less time homeless than the usual care group, leading to cost savings across all residential locations. Reductions in hospitalization were largest for a subsample recruited from psychiatric hospitals (who met homelessness criteria prior to admission); reductions in homelessness were largest for the subsample recruited from the street (Gulcur et al., 2003).
In a pre-test/post-test study with a control condition, a Housing-First (HF) approach program and a comparison group not provided with housing were followed over 18 months. The HF group had significantly greater reduction in emergency department visits and sobering center use than comparison group members. There was also a trend toward greater reductions in hospital admissions (Srebnik et al., 2013). However, the sample size for this study was extremely small, including 29 participants and 31 comparison group members, making it difficult to extrapolate from these findings.
In the randomized At Home/Chez Soi experiment conducted in five Canadian cities, an intervention group that received scattered-site HF5 housing and services was compared to a treatment-as-usual group (Aubry et al., 2016). In year 1, HF participants showed greater improvement in community functioning than the control group participants; by year 2, however, improvement was seen in both conditions. The authors noted that community functioning as a variable had not been examined in earlier research on this topic (Aubry et al., 2016). Interestingly, however, there were no significant differences between the two groups on a number of secondary variables, including mental health symptoms, substance use-related problems, number of arrests, and so on.
Another health care utilization study conducted in Seattle by Mackelprang and colleagues (2014) examined emergency medical services (EMS) utilization before and after entering a single-site Housing First program. The 91 program participants had severe alcohol problems. The study did not monitor health out-
4 Not a Pathways Housing First program.
comes, but examined and categorized the reasons for EMS calls through examination of administrative data, both for 2 years prior to enrollment in PSH and 2 years following enrollment in PSH. The variables of interest were trauma/injury, substance use, psychiatric difficulties, medical illness, and other. The study found a 54 percent reduction in EMS calls for those who entered supportive housing. Additionally, it concluded that each additional month in supportive housing reduced the likelihood of EMS contact by 3 percent. The authors noted that substance use problems and psychiatric difficulties were the most common use of EMS services for this sample.
Some studies have examined PSH and health service utilization and hospital admissions as indicators of health. For example, Gabrielian and colleagues (2016) analyzed the ambulatory care use of 3,631 veterans treated at the VA Greater Los Angeles facility between October 2010 and September 2011. The findings were compared between individuals who were housed and provided case management through VA Supported Housing (VASH veterans) (n = 1,904) and those who were currently experiencing homelessness (n = 1,727). Adjusting for demographics and need characteristics, VASH veterans were more likely (p < .05) than veterans experiencing homelessness to receive treatment for chronic physical illness, acute physical illness, mental illness, and substance use disorders. Among veterans treated for chronic illnesses, VASH veterans versus those veterans experiencing homelessness were more likely to have two or more visits for chronic physical illnesses, mental illnesses, and substance use disorder, indicating better follow-up medical treatment.
Rieke and colleagues (2015) examined the effect of PSH on mental and physical health emergency department use among 23 adults (12 males and 11 females) before and 1 year after housing placement in the Omaha, Nebraska, area. Findings indicated that the number of emergency department admissions decreased while the number of outpatient visits increased. Males showed greater improvement in emergency department visit decline (from 84 admissions in the year prior to placement to 40 after PSH) but had higher levels of emergency department use over females in the pre-housing period. Most emergency department visits for both males and females were related to behavioral health diagnoses. The authors concluded that supportive housing may encourage more use of health care services. Clearly, however, given the small number of participants studied, as well as the overall weakness of a pre/post study design, it is difficult to draw conclusions from this study.
A pilot study conducted in Portland, Oregon, examined the effects of single-site supportive housing on health care costs, health care utilization, and health outcomes for 98 “highly medically vulnerable” individuals experiencing homelessness (Wright et al., 2016, p. 21). This study, using retrospective survey responses and Medicaid administrative claims data, showed that placing individuals experiencing homelessness and high medical costs into supportive housing significantly reduced Medicaid expenditures for inpatient hospital and emergency department services for physical health issues, with an average annual reduction
of $8,724 in the year after moving in (Syrop, 2016). The self-reported data also showed a reduction in hospital stays and emergency department visits, indicating a shift toward using primary care services rather than acute care services. Although these results are promising, the absence of a comparison group and the use of retrospective self-reported data limit interpretations of this study.
In summary, the committee finds that providing PSH to individuals with high medical needs who are also experiencing homelessness decreases emergency department use and hospital stays. The degree to which the use of these services decreases will likely depend on the extent to which they were used prior to individuals being housed and extent to which the experience of being homeless is contributing to a worsening of conditions that results in illness (see section below on housing-sensitive conditions).
Impact on Physical Health Outcomes
Reductions in the use of ambulance transport, emergency department use, and hospitalization are important but imprecise process indicators of the impact of PSH on the health of individuals experiencing chronic homelessness. These indicators provide only indirect evidence of the physical health benefits from housing. For example, it is difficult to ascertain what portion of the reductions (e.g., emergency department visits) are related to health improvement after being housed versus the portion that is due to decreased social need to be hospitalized to escape the conditions on the street and gain safety, warmth, food, and other services. A by Rodriguez et al. (2009) found that 29 percent of homeless patients stated that hunger, safety concerns, and lack of shelter were reasons they came to the emergency department. Additionally, some portion of the decrease in hospitalizations due to the provision of housing may relate to doctors being more willing to discharge housed patients (rather than patients living on the street) from the emergency department. However, there is not yet much research to support the statement that doctors are more willing to discharge housed versus homeless patients. One qualitative study examined this issue and concluded that additional research is needed to understand practices for screening for homelessness in the emergency department and admission practices for patients who are homeless. The authors concluded that these are important areas for future research with “implications for health care costs and patient outcomes” (Doran et al., 2013).
The committee identified a number of randomized controlled trials (RCTs) and observational studies examining the impact of PSH on physical health outcomes for individuals experiencing homelessness. The majority of these studies focused on individuals living with HIV/AIDS. For example, Aidala and colleagues (2016) conducted a systematic review of studies examining the association between housing status, medical care, and health outcomes among people living with HIV/AIDS. Of the 152 studies included in their review, only two were RCTs (Buchanan et al., 2009; Wolitski et al., 2010); most studies (71 percent) did not specifically address the role of housing status in outcomes or lacked sufficient power to show statistical significance. The authors noted, however, that “worse
housing status was independently associated with worse outcomes” across virtually all studies included in the review (Aidala et al., 2016, p. e1).
The RCT by Wolitski et al. (2010) for the Housing and Health Study Team in the Centers for Disease Control and Prevention compared health outcomes and risk behaviors of 630 people experiencing homelessness and living with HIV/AIDS who were randomly assigned to immediate Housing Opportunities for People with AIDS (HOPWA) rental assistance (treatment group)6 and those assigned to traditional care (control group), which included making a plan with service providers to obtain housing. Self-reported physical and mental health data, CD4 counts,7 and HIV viral load were collected at baseline and at 6, 12, and 18 months. Results showed that although homelessness decreased in both groups, there was significantly greater improvement occurring in the treatment group. At 18 months, 51 percent of the comparison group had their own housing, thus limiting the study’s statistical power. Intent-to-treat analyses demonstrated significant reductions in medical care utilization and improvements in self-reported physical and mental health, again for both intervention and comparison groups; significant differential change benefiting the treatment group was observed for depression and perceived stress, although these results were strongest in the early phase of the study. By 18 months, there were no significant differences between the two groups.
Similarly, there were no significant differences for CD4 counts or HIV viral loads between the intervention and comparison groups. The authors concluded that HOPWA rental assistance improved housing status and demonstrated modest improvements in mental health outcomes for the intervention group.
In the second RCT, Buchanan and colleagues (2009) focused more specifically on the health benefits of PSH (a mix of single site and scattered site) on individuals experiencing homelessness and living with HIV/AIDS in Chicago. In this study, which is a subsample of the larger Chicago Housing for Health Partnership (CHHP) study described by Basu et al. (2012) in Chapter 4, 105 HIV-positive homeless hospital inpatients were randomized to usual care (discharge planning and information about short-term shelters) or permanent housing with intensive case management. The primary outcome of interest was survival with intact immunity (CD4 count greater than or equal to 200 and a viral load less than 100,000). The secondary outcomes of interest were viral load and CD4 counts. Because of to death and attrition, secondary outcomes were available for only 94 of the 105 initial participants (90 percent). At 1 year, significantly more individuals in the intervention group were alive and had intact immunity (55 percent) compared to 34 percent of the usual-care group (p = .04). In terms of viral load, 17 intervention group members (36 percent) and 9 usual-care group members (19 percent) had undetectable viral loads (p = .05). No significant differences were
6 Scattered-site housing similar to that used with Housing Choice Vouchers.
7 A laboratory test that measures the number of CD4 T lymphocytes (CD4 cells) in a sample of blood. See: https://aidsinfo.nih.gov/understanding-hiv-aids/glossary/822/cd4-count.
found for CD4 counts. The study authors concluded that supportive housing improves health outcomes for people living with HIV and experiencing homelessness, based on improved health measures related to mortality, immunity, and viral load.
Although RCTs are the gold standard for assessing cause and effect, observational studies with key health outcomes data can also be informative. For example, a study by Schwarcz and colleagues (2009) using data from the San Francisco AIDS registry showed that individuals experiencing homelessness and living with AIDS who obtained supportive housing had a lower risk of death compared to individuals experiencing homelessness. Intravenous drug use was also more common in the unhoused group, and independently predicted mortality for all individuals.
In a subsample of 676 individuals living with HIV who were also experiencing homelessness, 49 case-controlled pairs were analyzed. Of the individuals experiencing homelessness, 67 percent survived 5 years compared with 81 percent of those who received supportive housing.
Outside of the studies that focused on the HIV population, the committee found it difficult to identify high-quality data documenting the impact of housing on other aspects of physical health status. The few studies identified focused on subjective general assessments of physical health via self-reports. As described earlier, a pilot study of supportive housing in Oregon showed that the percentage of study participants who responded that they “had an unmet physical health need” dropped from 79 percent to 48 percent after moving into housing. Additionally, the proportion of respondents who marked “physical health was fair or poor” dropped from 80 percent to 54 percent after moving into supportive housing (CORE, 2013; Wright et al., 2016).
However, in the randomized At Home/Chez Soi study, there was no change in self-rated physical health (Stergiopoulos et al., 2015) over a 2-year follow-up period, nor were there significant differences between the intervention group (scattered-site housing with intensive case management) and the usual-care group. An earlier randomized study of a scatter-site supportive housing group compared to a usual-care group and a case-management-only group among veterans produced similar results of no significant change over a 3-year period for Medical Index8 scores (Rosenheck et al., 2003). There were also no significant differences between the three groups for Medical Index scores.
In summary, outside of HIV/AIDS, the research to date has not comprehensively assessed the benefits of PSH on physical health outcomes (see Table 3-1). RCTs considering a general measure of health status found no improvement in health status with a move into housing. A demonstration of the health benefits of housing would require the enrollment of patients who have health problems likely to be affected by housing status and more comprehensive assessments of health beyond self-reported scales. Self-reported assessments of health are accepted
8 The Medical Index is a component of the Addiction Severity Index. It is designed to measure physical health outcomes related to substance use.
measures in assessing quality in patient-centered outcomes. Additional details are described in Chapter 8.
|Study||Year||Randomized Controlled Trials||Key Findings|
|Aidala et al.||2016||Systematic review of studies examining the association between housing status, medical care, and health outcomes among people living with HIV/AIDS.||Of the 152 studies included in their review, most studies (71 percent) did not specifically address the role of housing status on outcomes or lacked sufficient power to show statistical significance.|
|Wolitski et al.||2010||Compared health outcomes and risk behaviors of 630 people experiencing homelessness and living with HIV/AIDS who were randomly assigned to immediate Housing Opportunities for People with AIDS (HOPWA) rental assistance (treatment group) and those assigned to traditional care (control group), which included making a plan with service providers to obtain housing.||The authors concluded that HOPWA rental assistance improved housing status and demonstrated modest improvements in mental health outcomes for the intervention group.|
|Buchanan et al.||2009||Focused on the health benefits of permanent supportive housing (a mix of single site and scattered site) to individuals experiencing homelessness and living with HIV/AIDS in Chicago.||The authors concluded that supportive housing improves health outcomes for people living with HIV and experiencing homelessness, based on improved health measures related to mortality, immunity, and viral load.|
|Schwarcz et al.||2009||Used data from the San Francisco AIDS registry.||Individuals experiencing homelessness and living with AIDS who obtained supportive housing had a lower risk of death compared to individuals experiencing homelessness.|
|Stergiopoulos et al. 2015||An intervention group that received scattered-site Housing First housing and services was compared to a treatment-as-usual group||There was no change in self-rated physical health over a 2-year follow-up period, nor were there significant differences between the intervention group (scattered-site housing with intensive case management) and the usual-care group on a number of secondary variables.|
Impact on Mental Health Outcomes and Substance Abuse
Multiple surveys of people experiencing chronic homelessness indicate that the prevalence of serious mental health conditions is much higher than in the general population (Aubry et al., 2015). Conversely, the stress of being homeless is known to worsen mental illness (Fazel et al., 2014). It is therefore not surprising that housing would be promoted as a health benefit for persons who are mentally ill (e.g., Aubry et al., 2016), although this is not necessarily supported by the available data.
However, even though supportive housing models have been found to decrease the number of days spent homeless (Rosenheck et al., 2003; Cheng et al., 2007; Stergiopoulos et al., 2015; Aubry et al., 2016) or in psychiatric hospitals (Gulcur et al., 2003) for individuals with serious mental illness and/or substance use, this has not translated to significant improvements in mental health status in most studies. For example, as part of the At Home/Chez Soi RCT, Stergiopoulos and colleagues (2015) compared individuals assigned to a Housing First model using scattered-site housing with intensive case management to a comparison group of usual care. Over a 24-month period, individuals in the Housing First program were housed between 33 percent and 50 percent more days than the usual-care group (the range of differences reflects the results from four different sites). However, there were no differences between the Housing First group and the usual-care group at 24 months on measures of severity of mental health symptoms, self-rated mental health status, community integration, or degree of recovery.
An earlier randomized study of veterans assigned to supportive housing versus two comparison groups (case management-only group and usual-care group) showed no significant differences for mental health measures between the three groups during a 3-year follow-up (Rosenheck et al., 2003). Again, although the group receiving the housing voucher spent 16 percent more days housed than the case management group, and 25 percent more housed days than the standard-care group, there were no significant differences on measures of psychiatric status or substance use.
In a previously described study in New York City, mentally ill, substance-using individuals experiencing homelessness were assigned to two conditions, a Pathways Housing First9 program and an approach where receipt of housing was contingent on treatment and sobriety (Tsemberis et al., 2004). Although individuals assigned to the Housing First program spent more days housed and fewer days in psychiatric hospitals (Gulcur et al., 2003), there were no significant differences between the two groups for substance use or psychiatric symptoms.
Several possible conclusions can be drawn from these three randomized studies investigating the association between supportive housing and mental illness. One possibility is that because mental illness is a serious, chronic disease, housing itself will not improve mental health. A second possibility is that the benefits of housing are difficult to tease out in these studies because the control group, in each case, received services that may not be available to individuals experiencing homelessness in other situations. For example, in the case of the Canadian At Home/Chez Soi study (Stergiopoulos et al., 2015), the usual-care control group also had access to existing housing and supportive services through their communities, as well as access to a more comprehensive system for sheltering individuals experiencing homelessness than exists in most parts of the United States (Stergiopolous et al., 2015).
For the study investigating the effects of housing for veterans experiencing homelessness, all of the participants across all three study conditions had access to VA services (Rosenheck et al., 2003). In the New York City study (Tsemberis et al., 2004), the comparison group received mental health and substance abuse services. In all three of these studies, the participants themselves were able to consent and agree to participate in a longitudinal study. Whether it is possible that housing decreases mental health symptoms among more severely affected individuals compared to persons continuing to live on the street without any supports is uncertain.
Mirroring the results for the findings on the association between Housing First and mental health status, all three randomized trials discussed above found no significant differences in substance use between the Housing First group and the control group. The lack of differences suggests that substance use was not enabled by living in permanent supportive housing without requirements for treatment.
9 The Pathways Housing First model provides only scattered-site housing.
In summary, the committee finds that supportive housing improves the housing status of individuals suffering from homelessness, mental illness, and substance abuse. The studies reviewed did not demonstrate improvements in psychiatric symptomatology or substance use behavior, perhaps due to homogeneity in the mental health and substance abuse services received by the intervention and control groups. More research will be needed to elucidate the role of permanent supportive housing on mental health and substance abuse outcomes.
Impact on General Well-Being
Along with physical and mental health, social well-being is a component of the WHO (1946) definition of health. The constructs of “social well-being” and, more generally, “well-being” have been defined in a number of different ways in the literature. Social science theories have described well-being in terms of a sense of belonging and identification with societal groups (Durkheim, 1997; Runquist and Reed, 2007). Well-being has also been described as “the positive aspects of a person’s life, such as positive emotions and life satisfaction” (ODPHP, 2016), and in terms of the quality of relationships, realization of individual potential, and overall life satisfaction (Diener and Seligman, 2004; Diener, 2009; CDC, 2016). Consideration of well-being as an outcome for residents in PSH is consistent with recent policy discussions in the community development and public health fields (Pastor and Morello-Frosch, 2014) and the recent approach of the U.S. Department of Housing and Urban Development (HUD) to utilize housing as a “platform for improving quality of life” (Fukuzawa and Karnas, 2015, p. 86). The merit in considering the specific construct of social well-being in particular is supported by its associations with morbidity and mortality in the general population (Yang et al., 2015).
Recognizing that there are both commonalities across descriptions of wellbeing and variability in specific research definitions and measurements, the committee initiated its review of evidence using an expansive approach. Few exclusion criteria were used except that physical and mental health conditions were not included because they are more appropriately placed within the realm of physical and mental health outcomes. Investigations of health outcomes for persons in PSH have primarily focused on indicators of physical and mental health and utilization of services. It is notable that recent systematic reviews of the literature evaluating the evidence for PSH do not include discussion of well-being indicators (Fitzpatrick-Lewis et al., 2011; Rog et al., 2014), whether due to the exclusion of such indicators in the original studies or because the studies that addressed these indicators had design limitations that precluded conclusions about effectiveness of PSH for well-being outcomes.
RCTs provide some evidence for the effect of PSH on well-being. Most of this research has focused on quality of life. A controlled trial conducted in the five Canadian cities At Home/Chez Soi study comparing participants randomized to either scattered-site Housing First, receipt of Assertive Community Treatment (ACT), or a treatment-as-usual control group found significantly improved quality
of life and community functioning at 12 months post-enrollment among the Housing First participants (Aubry et al., 2015), where quality of life was assessed with the Quality of Life Interview (QOLI) (Uttaro and Lehman, 1999) and community functioning was measured with the Multnomah Community Ability Scale (MCAS) (Barker et al., 1994; Dickerson et al., 2003). In terms of absolute gain, increases in quality of life for Housing First participants were greater for the QOLI total score and for the subscales that assessed living situation and, to a lesser extent, personal safety and leisure activities (Aubry et al., 2015). Also in terms of absolute gain, community functioning was greater among Housing First participants for social skills (i.e., social effectiveness, social network size, participation in meaningful activity) and behavior (i.e., cooperation with treatment providers, substance use, impulse control) (Aubry et al., 2015).
Social Integration and Quality of Life
In a study of Pathways Housing First in New York City, chronically homeless participants randomly assigned to either scattered-site Pathways Housing First or a control condition in which shelters and other temporary living arrangements were available completed a battery of measures representing social integration and similar constructs (Gulcur et al., 2007). Adjusted analyses indicated that residence in scattered-site Pathways Housing First locations as opposed to other places significantly predicted greater social integration at 48 months post-baseline in regression analyses. Social integration was a factor derived from measures of social support and the number of social ties from one’s neighborhood. Data from randomized controlled trials of the Pathways Housing First model therefore provide some evidence of improvements in social integration and several specific domains of quality of life.
Quasi-experimental and correlational studies have also yielded evidence of PSH impact on well-being, although results from such studies must be evaluated in light of the limited internal validity inherent in these designs. A cohort of individuals in Phoenix, Arizona, enrolled in a Housing First program (n = 47) that incorporated a peer support model, showed significantly improved quality of life from time of enrollment through 12 months (Bean et al., 2013). Quality of life was measured in that study with the WHO Quality of Life Scale (WHOQOL Group, 1998). The sample size for this study was small, however, and there was no control group.
Hwang et al. (2011) studied single adult residents, many of whom were mentally ill, who were enrolled in a supportive housing program in Canada and were followed for 18 months after entry into housing. The comparison group was potential residents who were wait-listed. Both groups were administered assessments of social functioning as part of the MOS 36-Item Short-Form Health Survey (McHorney et al., 1994) and quality of life with the EuroQol (Rabin and de Charro, 2001) and the Lehman Brief Quality of Life Interview (Uttaro and Lehman, 1999). The quality-of-life measures assessed constructs including but not
limited to social and family relations, safety issues, living situation, life satisfaction, and satisfaction with living situation. Residents demonstrated significantly increased well-being as reflected only in satisfaction with their living situation.
An observational study examined individuals experiencing chronic homelessness also with serious mental illness who received permanent supportive housing across 11 U.S. sites. Residents completed assessments on work and volunteer activities, social support, community participation, and civic activity at baseline and through 12 months in housing (Tsai et al., 2012).10 There were no significant changes over time in these dimensions after controlling for clinical symptoms. These quasi-experimental studies therefore yielded limited evidence of increased well-being for persons in housing.
Qualitative analyses of the experiences of persons living in PSH provide important insight on well-being in Canada. Research among HIV-positive individuals housed in Canada suggests risk of self-imposed social exclusion and social isolation within housing due to the ongoing stigma associated with having HIV (Chambers et al., 2014). A qualitative study involving residents of a single-site Housing First program for persons experiencing chronic homelessness with severe alcohol problems suggested the importance of meaningful and interesting social activities organized by staff for residents (Clifasefi et al., 2016). Another qualitative investigation suggests improved quality of life in Housing First programs for residents who have formerly experienced homelessness because they were able to move out of the “survival mode” that is necessary on the streets (Barr, 2004), with one respondent explaining, “You get a little bit of the criteria of what people expect from normal people: A fairly decent place to live, your own money, a job. From there, people can do just about anything they want” (Barr, 2004, p. 154).
Similarly, a qualitative study of 39 individuals who were initially part of the Tsemberis et al. (2004) randomized New York Housing Study participated in a follow-up interview study. Individuals experiencing both homelessness and mental illness were assigned to either a treatment-as-usual group (treatment-first approach) or to receive immediate housing through Pathways Housing First (Padgett, 2007). Padgett’s interviews focused on validating the concept of “ontological security,” the notion that constancy, day-to-day routines, freedom from surveillance, and the opportunity for identity construction are outcomes of the housing-first philosophy. These interview data show clear support for the components of ontological security for the individuals in the Housing First program. Participants in a pilot study of individuals moving into a single-site Housing First program in Los Angeles reported challenges in their social networks (i.e., increase in sexual relationships) after 3 months in housing (Henwood et al., 2017), with disconnecting from old relationships a major emphasis for interviewees.
Perspectives of PSH providers suggest the importance of examining social well-being as an outcome for individuals previously experiencing homelessness.
10 Article does not state if the PSH provided was single-site or scattered site housing.
According to one provider agency during the committee site visits, achieving “social integration” is considered a priority for every client in PSH, where social integration was defined broadly by this program to include well-being, getting along with neighbors in the buildings, attention to nutrition, and job skills development. These providers also perceived differences between congregate (or single-site) and scattered-site housing approaches, where congregate settings are thought to more easily facilitate getting together. Peer mentors were also seen as key to social integration attempts in scattered-site settings.
Although not a direct measure of well-being, sufficient sleep is known to be essential for well-being. Because of the difficulty of sleeping while on the street, persons experiencing homelessness have shorter sleep times than the general population, are more likely to suffer from insomnia, are more likely to take a drug to help them sleep, and are more likely to have daytime fatigue than the general population (Leger et al., 2016). Supportive housing would be expected to improve sleeping time because of the greater safety and comfort of sleep in one’s own home, although evidence on this is lacking.
In summary, despite the challenges in assessing outcomes related to wellbeing described above, the committee finds that there is evidence that housing, in general, improves the well-being of persons experiencing homelessness. This has been demonstrated using multiple different measures and study designs, including randomized controlled studies.
A NEW PARADIGM FOR CONSIDERING THE IMPACT OF PERMANENT SUPPORTIVE HOUSING
The committee deliberated on the question of why clinicians they heard from during their meetings felt certain that providing PSH to individuals experiencing chronic homelessness improved their health. Randomized and nonrandomized studies of individuals living with HIV/AIDS found benefits associated with housing. Yet randomized studies of individuals experiencing homelessness along with mental illness and/or substance abuse did not show statistically significant health benefits between the housed group and the usual-care group, except in the area of well-being. To better understand how and why providers believed PSH was a factor in improving health in this population, the committee examined the concept of medical necessity. Medical necessity is currently used by some states as a factor to determine eligibility for specific Medicaid-reimbursed services. Individuals who are experiencing chronic homelessness may qualify for services because they may suffer from specific conditions such as substance use disorders, mental illness, chronic medical conditions, or disabilities (HHS, 2014).
According to the American College of Medical Quality (2010), medical necessity is defined as “accepted health care services and supplies provided by health care entities, appropriate to the evaluation and treatment of a disease, condition, illness or injury and consistent with the applicable standard of care.” A class-action court case providing national settlement language further clarified
that medical necessity decisions must be individualized, and also noted that consideration of cost or comparative effectiveness was acceptable (Kaminiski, 2007). The associated definition agreed upon by major insurance companies and 900,000 physicians and used across the board in private market practice is as follows:
“Medically Necessary” or “Medical Necessity” shall mean health care services that a physician, exercising prudent clinical judgment, would provide to a patient for the purpose of preventing, evaluating, diagnosing or treating an illness, injury, disease or its symptoms, and that are: a) in accordance with generally accepted standards of medical practice; b) clinically appropriate, in terms of type, frequency, extent, site and duration, and considered effective for the patient’s illness, injury or disease; and c) not primarily for the convenience of the patient, physician or other health care provider, and not more costly than an alternative service or sequence of services at least as likely to produce equivalent therapeutic or diagnostic results as to the diagnosis or treatment of that patient’s illness, injury or disease. For these purposes, “generally accepted standards of medical practice” means standards that are based on credible scientific evidence published in peer-reviewed medical literature generally recognized by the relevant medical community or otherwise consistent with the standards set forth in policy issues involving clinical judgment (IOM, 2012, p. 228).
The Centers for Medicare & Medicaid Services (CMS) does not provide financial support for housing per se. However, in 2015, CMS released the Informational Bulletin on Coverage of Housing-Related Activities and Services for Individuals with Disabilities to provide guidance to states on certain housing-related activities and services (CMS, 2015; HUD, 2015b). A number of states have used this guidance to pay for housing-related services for transitioning to housing and sustaining housing for eligible individuals (CMS, 2015).
In addition, CMS has announced a future expansion of its definition of health-related benefits in its Medicare Advantage plans, which provide extra coverage, such as for vision, hearing, dental, and/or health and wellness programs, to Medicare recipients. In April 2018, CMS released a 2019 Medicare Advantage and Part D Rate Announcement and Call Letter, which announced a reinterpretation of federal statute to expand the scope of the “primarily health-related supplemental benefit” (CMS, 2018). CMS states that under this reinterpretation, the agency would “allow supplemental benefits if they are used to diagnose, prevent, or treat an illness or injury, compensate for physical impairments, act to ameliorate the functional/psychological impact of injuries or health conditions, or reduce avoidable emergency and healthcare utilization.” This expansion of benefits signals a move in the agency’s thinking toward a more integrative approach to covering services beyond those traditionally held as health-related, which could include housing.
The committee discussed a variety of scenarios where a patient’s illness would make housing a medical necessity. The scenarios included patients who
require medicines that must be refrigerated; who must receive frequent wound dressing changes; who must elevate their legs; or who are on medications known to cause extreme drowsiness, vomiting, or diarrhea (such as chemotherapy regimens). Health care providers for individuals experiencing homelessness have often used the concept of medical necessity broadly and creatively, to include, for example, the provision of housing as a medical necessity for their patients experiencing homelessness (Wilkins et al., 2014).
Beyond these acute illnesses, there may be people for whom housing prevents the onset of more serious illness. The concept of “ambulatory care-sensitive conditions” or its subset, “primary care-sensitive conditions” (Oster and Bindman, 2003; Caminal et al., 2004; Gibbons et al., 2012) refers to conditions for which primary care can exert an influence to alter the course of the condition. For example, asthma and congestive heart failure are both considered primary care-sensitive conditions by the Agency for Healthcare Research and Quality because many hospitalizations could be prevented or reduced by appropriate primary care (Fingar et al., 2015).
The committee believes that these concepts converge when considering the impact of PSH on the health of individuals experiencing chronic homelessness. HIV/AIDS, in the early years of the epidemic, is a good example of the intersection of health-related care and permanent supportive housing. HIV/AIDS was considered to worsen or at a minimum fail to improve without housing. In fact, Congress established the HOPWA program because “existing housing resources at the time were reportedly not meeting the needs of people with AIDS, who often had difficulty obtaining suitable housing because of the need for supportive services, discrimination, or other problems” (GAO, 1997, p. 1).
Similarly, HIV infection in the early years of the epidemic may have qualified as an ambulatory care-sensitive condition because the medication regimens were so complex that adherence for infected individuals experiencing homelessness was at best suboptimal. Some of the medications required refrigeration. Some had to be taken on an empty stomach, others on a full stomach. Multiple daily doses were required. Side effects such as nausea, vomiting, and diarrhea were common. Frequent clinical visits and blood draws were needed to assess efficacy. In this setting, it is not surprising that studies showed that persons who remained living on the streets had worse outcomes than those who were housed. An important and complicated next question is, which other conditions are made worse by homelessness and improved by housing?
In the committee’s judgment, homelessness worsens health through exposure to unhealthful environments and by making it difficult to care for oneself or for others to provide care. This is echoed by the Corporation for Supportive Housing (CSH, 2014a):
supportive housing provides physical safety, protection and access to basic needs. A clean, dry, safe home reduces exposure to harsh weather, communicable diseases, infections, injury, harassment and violence; it provides
a secure place to sleep and store food, clothing and medications; and it is essential to promoting personal hygiene and recuperation from illness.
A number of common health problems could be prevented or improved by stable housing. For example, specific features that are part of housing can result in better outcomes, even simple benefits such as access to a bathroom, a refrigerator/freezer, and electricity to run medical equipment. Although it is likely that housing contributes to improvements in overall health, there are not currently enough data to link stable housing to improvements in substance use or serious mental illness. Nonetheless, the committee feels that there is a moral imperative to house individuals with serious mental illness or substance use problems; as a society, there is a consensus that these individuals should not be left living on the streets.
Conversely, a lack of stable housing increases the risk of contracting tuberculosis due to potential exposure to others with a high risk for TB on the streets and in shelters. Other health issues that are exacerbated by a lack of stable housing include hypothermia and hyperthermia, skin infections, and an increased risk of assault. Chronic illness such as heart disease, cancer, and HIV/AIDS can be exacerbated because adherence to treatment is difficult without stable housing.
The concept of “housing-sensitive conditions” does not negate the fact that everyone would be healthier housed than homeless. But there is a group of chronically homeless patients for whom failure to provide housing will result in significant worsening of their conditions. Future studies are needed to understand whether there are health conditions that are more sensitive to housing.
In summary, on the basis of currently available studies, the committee found no substantial evidence that PSH contributes to improved health outcomes, notwithstanding the intuitive logic that it should do so and limited data showing that it does do so for persons with HIV/AIDS. There are significant limitations to the current research and evidentiary base on this topic.
The committee acknowledges the importance of housing in improving health in general, but it also believes that some persons experiencing homelessness have health conditions for which failure to provide housing would result in a significant worsening of their health. Said differently, notwithstanding that housing is good for health in general, the committee believes that stable housing has an especially important impact on the course and ability to care for certain specific conditions and therefore the health outcomes of persons with those conditions. The committee refers to these conditions as “housing-sensitive conditions” and recommends that high priority be given to conducting research to further explore whether there are health conditions that fall into this category and, if so, what those specific conditions are. The evidence of the impact of housing on HIV/AIDS in individuals experiencing chronic homelessness may serve as a basis for more fully examining this concept.
Research on housing-sensitive conditions could improve our understanding about how best to manage care for individuals experiencing homelessness who have conditions that fall into this category. It would also provide evidence for federal- and state-level health care policy and financing decisions related to the need for housing for those who have conditions considered housing sensitive, as well as possibly facilitate processes used for prioritizing housing for persons experiencing homelessness. Finally, this research could strengthen the connection between currently siloed health and housing systems by documenting the need for integrated data collection and decision making.
Recommendation 3-1: Research should be conducted to assess whether there are health conditions whose course and medical management are more significantly influenced than others by having safe and stable housing (i.e., housing-sensitive conditions). This research should include prospective longitudinal studies, beyond 2 years in duration, to examine health and housing data that could inform which health conditions, or combinations of conditions, should be considered especially housing-sensitive. Studies also should be undertaken to clarify linkages between the provision of both permanent housing and supportive services and specific health outcomes.
Recommendation 3-2: The Department of Health and Human Services, in collaboration with the Department of Housing and Urban Development, should call for and support a convening of subject-matter experts to assess how research and policy could be used to facilitate access to permanent supportive housing and ensure the availability of needed support services, as well as facilitate access to health care services.
Information provided by such research would be helpful in setting policies around reimbursement for housing when it is considered health-related care required to address a housing-sensitive condition.