Free from Homelessness : Is “ Housing First ” the Solution ?-A Comparison with the “ Staircase Model ” . A Feasibility Study

Introduction: Homelessness is as large political question globally as it is in Sweden with a growing number of homeless. “Housing First” is a municipality based program aiming to offer stable and enduring housing to homeless people immediately. This model was imported to Sweden and was tried in Stockholm and Helsingborg, a provincial town in the southwest of Sweden. This prevailing study is to test the feasibility of making longitudinal studies on homeless individuals. Method: A longitudinal study using repeated measures of selfreports on questionnaires. Result: This study report results from baseline and a 24 month follow-up of the program compared to a treatment-as-usual control group. It was shown that although the former homeless in the Housing First group perceived an increased housing normality and empowerment there were no significant differences between groups across time. Conclusion: Although individuals living in a Housing First apartment tended to have a more normal way to live they seem to not improve psychologically. However some problems that is connected to longitudinal research on the homeless was detected. Since Housing First is a program consisting of housing and support the effect is dependent on both. The null result in this study was discussed as a consequence of foremost insufficient support in housing.

In itself, unstable housing co-varies with many different psychosocial problems and health risks, whereas stable housing has a positive effect on health [11] and for instance, substance use has the opposite effect, accordingly, loss of housing [12][13][14].
Homeless people suffer from a high prevalence of physical and mental problems and a convincing amount of research indicates which risk factors are associated with homelessness [15][16][17]. Studies shows that the prevalence of mental illness and addictions is higher among homeless people than the general population and that 25-70% have co-occurring mental health and substance use problems [18]. Furthermore, homeless people also constitute a highly underprivileged group with substantial service use and costs to public and an important question is to what degree welfare policies and practices actually support or impede social inclusion [19]. The present article investigates the possible effects of an innovative way to manage homelessness in Sweden and the underlying question is what an effective way to eliminate homelessness is?
Studies have been made of different housing programs for the homeless [19][20][21], which broadly can be divided into two types of models: linear and non-linear. This also coincides with the division between housing as a means to an end -a more rehabilitating abstinence approach -with housing as the goal -which would be a more recovery oriented harm reduction approach.

Introduction
The importance of suitable housing has been discussed before [1][2][3] and is often described as a "fundamental human right" [4][5][6] or even in terms of "a right to housing" [7]. The United Nations agree with this view and advocate that a home which is safe, habitable and affordable is a fundamental human right. However, as the UN also notes, this right is out of reach for many individuals [8]. The UN's general declaration of human rights (article 25) states that everybody has a right to a home, yet the number of homeless people in the world is estimated to be between 100 million and one billion people [9], depending on which definition is used [10]. without the strong focus on care and treatment there is no researchbased evidence for that it really achieves its purpose [35,36] and the critique of this approach points to the problem that those who do not improve in this system are stuck, while those who misbehave are either relegated to a lower step or pushed down to the bottom floor, often a shelter or homelessness [37]. The staircase model is "treatment as usual" in the prevailing study.
In contrast, an innovative approach to solving the problem of homelessness has evolved viewing housing as a fundamental human right as a starting point. Housing First is a counter-reaction to the linear or continuum programs, and reversed the housing continuum with a focus on quick access to housing, builds on the notion that everyone has the right to a secure place to live. The idea originated in the U.S. and evolved in the early nineties within psychiatric health care as an alternative way of handling the effects of the deinstitutionalization process. Apart from everyone's right to housing, the idea of user participation, harm reduction and an intense wrap around service, usually in the form of Assertive Community Treatment model, is central [38,39]. The Housing First model is the intervention in the prevailing study.
Evaluations of Housing First programs have been done in an international context. During the development of the program, most of the evaluations were made by the inventor of the program, and all showed good results, suggesting that the program reached its stated aim of getting people out of homelessness and into stable housing, even though other problems remained unchanged [24,40,41]. In a follow-up study over three years, Fichter & Quadflieg (2006) found that 86% of homeless people maintained housing stability but those who drank heavily were worse off than lighter drinking people in the program. [42]. However, this division into groups was made ad hoc and no control group was used. More recently, independent evaluations and more complex studies of Housing First programs have been introduced, which discuss the possibility of implementing the model outside the U.S. [21,31,37,39,[43][44][45].
Critiques has pointed out the lack of Housing First program effects on substance use [14,21] and specifically on illicit drugs [18]. The Housing First model has become an innovative solution to the problem of homelessness in the U.S. and has been implemented in more than 300 cities in connection with the introduction of a 10 year plan for managing the problem. In Canada the government conducted the largest and most comprehensive experiment of Housing First across a variety of settings, and allocated $110 million to implement an RCT. In the At Home/Chez Soi experiment, which took place over four years and across five cities, the authors argue that Housing First appears to have eliminated homelessness for dually diagnosed, single adult, and chronically homeless individuals [14,46]. The model has recently been implemented in the Nordic countries with more or less success [47,48] and in Sweden in a few municipalities [49].
As mentioned above an important part of the Housing First programs is the option of a highly qualified treatment team. In a review of the research literature [34], it was apparent that regardless of the type of housing program used, some sort of personal support needs to be added, most often in the form of a case manager or housing support. From a psycho-social perspective, one can assume that coordinated treatment programs including stable housing combined with social support is important for a person's well-being, sense of security and motivation to abstain from (or reduce) alcohol or drug use.

The Swedish Situation
In a recent survey carried out by the National Board of Health and Welfare (NBHW), the number of people in a homeless situation was approximately 34 000 [50]; a considerable increase compared to the earlier estimate made in 2005 of 17 800 homeless individuals during a given week [51].
The National Board of Health and Welfare allocated considerable funding to local projects between the years 2002-2009, all of which aim to reduce homelessness. Despite this, the number of homeless people in Sweden has doubled during the past 6 years (2005-2011). However, the operationalisation of the definition was different when these estimates were made in 2005 and 2011, which means that it is uncertain whether these municipal projects have been effective in reducing homelessness. The lack of a control group in many projects means that we do not know what the situation would have been like without these interventions. It was concluded that few projects had used the knowledge gained through recent research on homelessness and none of the projects was about the "Housing First" program [52].
Nevertheless, over the last years, the "Housing First" model has gained increased attention in Scandinavia [19,32,49,53] and although widely disseminated, implementation of Housing First has been slow in Sweden and only 4 percent of Sweden's 290 municipalities has so far implemented the model [54]. Ending homelessness requires a focused, organized response from all those involved, especially on a national level from the policymakers and this is not happening in Sweden now.
The purpose of this study is to test the feasibility of evaluate the difference between an intervention consisting of the program "Housing first in Stockholm City"/Helsingborg, and the usual way of handling homeless people in Stockholm (the staircase housing model). Only the quantitative survey is reported in this prevailing study. The outcomes variables are the housing situation, the experience of having control over one's situation (locus of control), the sense of coherence (KASAM) and also the use of alcohol and drugs. Both groups of homeless people are being assessed during a 24 months follow-up period after one group acquired housing and social support. Since research has shown that satisfactory housing is a prerequisite of good health, we anticipate that the Housing First program will lead to a better housing situation and, at the same time, to improved mental well-being and reduced use of drugs.

Method Participants
The selection of homeless people to be included in the Housing First condition (H F) was made by the social services in four districts of the city and by the unit for homeless people serving the center of Stockholm. The criteria were that the participants should have been without stable housing for an extended period, and that they had difficulties obtaining housing within the usual homeless program, e.g. the housing staircase [55]. Therefore, the participants come from emergency housing or low threshold housing permitting substance use. However, participants also need to express a desire for an apartment of their own, and be willing to have a support/ case management contact at least once a week and to cooperate with the social services. Because of the gender distribution among the homeless in Stockholm, one fifth of the participants were women. At the time of starting the evaluation a decision to include Housing First tenants from Helsingborg was made. This was decided to increase the power of the study. However, the differences between the cities are large both in geographical location (west coast vs east coast) and in population (135000 vs 912000).
According to political intentions first 15 and one year later another 15 apartments in Stockholm and 25 in Helsingborg were decided to be allocated to the Housing First program. Apartments were distributed to the project in time when they were left by the previous tenant. Due to lack of apartments and problems to match apartments to homeless individuals, the date when tenants moved into their apartments differed and not all of apartments were matched with a homeless person. The matching problems mainly consisted of disabilities that made the persons to require an elevator but in the allocated houses elevators were rare. Another matching problem was when an allocated apartment was situated in an area where the tenant had many homeless friends. In that situation it is easy to let the friends sleep in the apartments that often lead to a distortion to alternatives were from least normal to most normal: lived roof less, lived in emergency housing, lived with acquaintances, lived at a assigned hotel, lived in low threshold housing, lived in training apartment, lived with second hand contract and lived with an own contract.

Locus of control
Rotter's "locus of control" scale [56] was used to assess the general sense of being in control over one's life circumstances. It was used as an indication of 'having a sense of control' of the life. This was used as a proxy to empowerment. A low ability to control one's circumstances has been associated with low confidence in public institutions [57] and high consumption of alcohol and drugs [58]. The scale consists of 10 statements, and the respondent was asked to estimate how well each statement described their behaviors. The assessment was made on a 4-item Likert scale ranging from "do not agree at all" to "agree completely". The scale showed a high internal consistency in the selection of homeless people at baseline (Cronbach's alpha = 0.76) and at 24 month follow-up (Cronbach's alpha = 0.851), which means that participants answered the statements in a consistent manner.

Sense of coherence
In order to estimate the participants' sense of being in a situation that is intelligible, manageable and meaningful, the short (13 statements) version of Antonovsky's Sense of Coherence Scale (KASAM-13) was used. Respondents were asked to assess how well each statement describes their own situation. The assessment was made on a seven steps semantic differential scale, with scores ranging between 1 and 7. The sum of the scale was accordingly between 13 and 91 points. KASAM-13 has been reported to have good construct validity and reliability [59]. A high sense of coherence has been associated with good health and positive mental well-being [60].
The participants assessed their sense of coherence on the scale with a high degree of reliability on both occasions (Cronbach's alpha = 0.74 at the baseline and 0.78 in at the 24 month follow-up).

Alcohol Use Disorder Identification Test
The Alcohol Use Disorders Identification Test (AUDIT) consists of 10 questions about alcohol and is a questionnaire recommended by the World Health Organization. It is also useful in non-clinical settings. The Swedish version has proved to be valid and reliable, and norms for the Swedish population have been published [61] The test has proved to be two-dimensional in a selection of normal alcohol consumers, but one-dimensional among high consumers. The recommended cut-off for risk consumption is 8 points for men, and in Sweden 6 is used for women. AUDIT has been used in surveys of the Swedish population between 1997 and 2009 [62]. The questionnaire format is a 5-graded Likert scale (0-4) with a maximum of 40 points.
The participants in this study answered the 10 AUDIT questions consistently at both baseline (Cronbach's alpha = 0.92) and 24 month follow-up (Cronbach's alpha = 0.91).

Drug Use Disorder Identification Test
The Drug Use Disorder Identification Test (DUDIT) was developed as a parallel test to the AUDIT to measure drug habits; the neighbors. Another problem was that homeless people with a psychiatric problem and/or drug abuse don't show-up on scheduled appointments. This made the homeless persons to be included in the control group on availability basis. This implies that the study was underpowered from the beginning since a calculation showed that at least 40 persons in each group was needed to reach 80% power.
At the end of the project, 24 months after the start, 24 homeless in Stockholm and 14 in Helsingborg had moved into their apartments. Their distribution of gender and age is shown in table 1. These tenants were used as the intervention group (HF) and as a treatment as usual control group (TAU) homeless persons matched on length of homelessness, difficulties of to get housing in treatment as usual, age and gender was used. The TAU was planned to originally consist of 75 eligible homeless people (males and females) who were in contact with the social services in Stockholm at the time. However, some of them were not at available at every assessment.
The intervention group was asked to give their answers to a survey at scheduled meetings with the provider of daily life support [2] from the city mission (Stadsmissionen) and the control group was surveyed at scheduled appointments with the social services. At baseline 13 tenants out of 15, 9 men and 4 women, from the intervention group from Stockholm responded (87%), in addition to 5 male tenants from Helsingborg, and 26 male and 9 females from the control group (response rate = 47%). In table 2 below only subjects that participated in both the baseline measurement and 24 month follow-up is analyzed. This subgroup consisted of 9 men and 1 woman among the intervention group and 15 men and three women among the controls. At the follow-up a response rate is difficult to estimate due to off registration from social service and not only non-response.

Design
A quasi-experimental longitudinal design with a control group was used, which means that only data from clients who participated in both the baseline and the 24 month follow-up were analyzed and reported here. For political and administrative reasons, some individuals moved into their apartments before the baseline assessment. Therefore, the data for the housing situation in table 2 has been taken from a register of these peoples housing before they acquired their apartments.

Data analyses
Data were analyzed using SPSS v. 20.0. The analyses of the differences between Housing First and TAU across time were made by using repeated measures ANOVA, both main effects and interactions time by group was evaluated. Due to the small sample no further dichotomization of the sample was possible. Identical items were used both at baseline and 24 month follow-up making it possible to assess changes over time.

Methods of Measurement The housing situation
The 'normality' of the housing situation was assessed with a single question: "Where have you spent the night?" The answers were given on a scale developed specifically for this study and builds on the staircase model that is used in Stockholm [55]. The response both illegal drugs and medicines that are taken without a doctor's prescription. DUDIT has proved to be valid and reliable for use both in the Swedish population [63] and among clients sentenced to undergo forensic assessment [64]. The average score in the Swedish population for women is 2 and 6 for men and these scores are used as cut-offs, and 25 has been proposed as the cut-off point for problematic consumption. Among patients in forensic psychiatry settings, the mean value was 16.9.
The participants in this study all responded consistently to the 11 statements in DUDIT, both at baseline (Cronbach's alpha = 0.93) and 24 month follow-up (Cronbach's alpha = 0.94).

Results
As shown in table 2 there were no statistically significant interaction effects between group and time. This means that there were no differences between groups across 24 months in any of the measures. However, there was a strong tendency that the intervention group improved their housing normality more compared to the control group. There were significant main effects of time in housing normality (p = 0.002) and empowerment (p = 0.041) which means that both groups in average improved their housing normality and empowerment. Cross-sectionally, compared to the housing first group the controls had a statistically significant higher AUDIT score at both baseline and 24 month follow-up (p < 0.04). This indicates that the remaining subjects were not effectively matched into groups. The initial difference in AUDIT score remains after 2 years (Table 2).
Due to the small number of participants in both groups the analyses were made regardless of gender. The improved average housing normality and empowerment across two years indicates that something common in the treatment of both groups had the effect.

Discussion
Housing interventions in Sweden are usually in accordance with the so-called "housing staircase" model, in which the individual must prove that he or she can handle successively more permanent and normal housing. The experiences from New York and "Pathways for Housing" have shown that through giving the homeless a place to live and housing support in order to be able to stay, "Housing First", the likelihood is higher that the individual will still live in the same place after a few years. The present study was an attempt to evaluate the feasibility to study the consequences of giving the homeless people a permanent housing linked to individually adapted support in order to stay and compare it with those who only get the usual support from the social services. The expectations were that the formerly homeless people (the tenants) would have a more normal housing situation 24 months after the baseline. Since a place to live can give a person a certain basic security in life, and since it is a basic condition for mental well-being and motivation, it was also expected that the tenants would have better mental health and would reduce their addiction compared with the homeless people in the housing staircase.
The result shows that, regardless of whether they had got housing through Housing First or not, in average all homeless people improved their housing situation on average between the baseline and the two year follow-up. However, the tenants (formerly homeless) tended to have a better housing situation than those who had not got permanent housing. On average, life-control and empowerment increased significantly for both groups, but there was no difference between the groups over time. This result indicates that at least some people in the control group can manage to advance in the housing staircase program and may also experience their situation as more controllable, which means that the housing staircase cannot be disregarded altogether as a form of housing alternative for homeless people.
These results may be due to a low statistical power, since too few individuals had responded at both baseline and follow-up. Regarding the alcohol habits according to AUDIT, all participants in average showed a weak and not significant decrease during the follow-up period. This may indicate that responding to items asking about alcohol use awakens thoughts of the extent of the consumption and a decision to reduce it. Mc Cambridge & Day (2007) showed a reducing effect on the self-reported consumption in a 3 month follow-up as an effect of filling in AUDIT at the baseline [65]. However, the alcohol habits showed no difference between the groups over the 24 month period. According to the DUDIT estimation, the drug habits did not change during the follow-up period and there was no difference between groups across time. This indicates that the relationship between housing and drug use is complicated and not a drugpreventive factor per se, which also Kirst et al. (2015) discuss. This may be due to the fact that the right to housing is not linked to any demand that the individual should be free from drugs in order to keep his or her "Housing First" apartment. This wakens expectations of support to handle drug problems actually may work as a motivating factor, which has been demonstrated in several other studies [66]. The central aspect is that the program should be extensive and actually include care, which is often lacking as the guiding principle in the Swedish versions of the housing staircase.
The theory that one needs a stable housing situation before one can come to grips with additional problems -apart from being homeless, that is -is probably correct, but there is a risk involved in a harm reduction intervention like "Housing First", and that is that individuals are abandoned in their housing and may continue with their addiction. From this follows that the quality of the housing support (Case Management) is crucial to the result of the intervention.
There is unanimous support in research for the view that if homeless people shall be able to handle the transition to living in stable housing situations in the long run, they need both housing and support [32,34]. This involves getting the help they need in the shape of a personal resource in order to live as normal a life as possible.
One cannot disregard the problems that most homeless people have, regardless of whether these problems have led to homelessness or sprung from it. The set of problems may be complex to such an extent that housing only solves a small part of it, which is why the improvements as regards mental well-being and addiction that one might expect after a person got a place to live did not become visible in this relatively short follow-up study.
Social support is given in the project as individually adapted support with access 24 hours a day. It is different from the support Table 2: Means (M) and standard deviations (SD) for scores on the scales for the groups, 10 tenants (HF) and 18 controls (TAU) , at baseline and two year followup and p-values for the group-time interaction . that the control group gets through their housing solutions and contacts with social secretaries. An assumption that the difference in social support should have any major effects on other factors apart from continued housing is contradicted by the fact that there are no group differences in the tenants' favour at follow-up apart from the normality of the housing. Since we have no pure factorized design and the "Housing First" program also includes support interventions, it is not possible to evaluate whether it is the housing or the support or the combination of both which has had an effect on the higher housing normality.
Most homeless people who have acquired a permanent place to live in the "Housing First" project still live there after two years and experience a better housing situation than those who have not acquired housing. From the viewpoint of the project's purpose, which is to give homeless people with addiction and psychiatric problems a home of their own, the project is successful. However, the mental well-being of those who have acquired housing seems not to have improved, which may be interpreted as saying that the change in the problems is only marginal.
The addiction problem seems not to improve as an effect of having acquired housing. Instead, the general reduction of drug and alcohol use seems to be the effect of surveillance and demands and making the individuals aware of their addiction. A larger reduction of drug habits by the control group than the intervention group is probably due to the fact that demands to reduce their drug use are being made on those who are in the housing staircase in order to improve their situation. Those who have acquired apartments of their own have no such demand to fulfill in order to keep those apartments.

Strengths and Limitations
This study is, as we know, one of the first assessing the consequences of implementation of Housing First for empowerment, sense of coherence, and use of alcohol and drugs. However, the validity of the study was threatened by the difficulties to get the same persons to respond at each measure. In the longitudinal design participants decreased due to low motivation to respond on repeated questionnaires. The consequences of the trial for psychological factors and substance use were evaluated across a 24-month period. This may be too short period to change behavior learned during many years as homeless. Therefore, the evaluation period was afterwards prolonged with 2 years but the proportion of non-response was high. Only a few persons were possible to follow across the whole evaluation period and that is why the period after 24 months is not shown. The decision to include tenants in the Housing First program in Helsingborg was made initially to increase the power although differences between cities are large both in population and localization. Some of the problems listed above could be avoided if the data collection was made by researchers and not, as in this study, by staff at the social service or city mission.

Conclusions
In this study no significant differences across time between Housing First (HF) and the Staircase model (TAU) was shown. Since HF is a program consisting of housing and support the explanation could be low power or insufficient support.