Treatment First is a stairway process. The clients are assisted through a structured process to determine goals and plans and to achieve step-by-step the housing readiness at the end of this process. The stairway model which meet the pre-requisitions that the clients are sobriety or treatment compliance is applied frequently in the treatment in Europe [5] .

The program often misses the needs of people with mental illness, drug or alcohol problems and other complex needs because treatment and housing depend on abstinence and on the ability of the client to keep the rules of the Treatment First program [6] . There is a little range of consumer choice with regard to goals and plans but no exception in cases of alcohol or drug relapses [7] . This residential treatment demands an abstinence-based support.

The Treatment First condition is based on two assumptions:

・ to teach the daily living skills to homeless adults it is necessary that they have a transitional or permanent housing and

・ housing readiness is connected with sobriety and compliance with the (psychiatric) treatment.

If the clients agreed with the Treatment First conditions and adheres to the agreements on sobriety and on compliance, they complete the stages step-by-step and move on into a less restrictive residence. In cases of relapse or due non-compliance the client returns to a previous stage of restricted residence. The way to an independent life can be a long-lasting process for the service user [8] . Studies from Padgett [9] and Collins [10] have shown that an abstinence-based treatment condition is not more effective than Housing First programs.

In the early 1990s this consumer driven approach was designed to provide a long-term support for people with complex needs [6] . Therefore the caregivers from the Pathway Inc. rented apartments from community landlords. They started to support people with severe mental illness and chronical homelessness.

The service users often had a poor physical health condition, suffered from drug abuse and learning disabilities. Furthermore, this people had a poor social condition such as unemployment, debt, homelessness and they experienced a lack of social support and community integration. These clients were supported to move directly in one of these apartments.

The Pathways program was named Housing First program. It distinguishes from other conventional supporting programs which demand a sobriety and compliance with the psychiatric treatment rules. The Housing First program is based on four theoretical principles [6] :

1) Immediate provision of housing and consumer-driven service.

2) Separation of housing and clinical service.

3) Providing support and treatment with a recovery orientation.

4) Facilitation of community integration.

Greenwood et al. [8] emphasized that mastery and perceived choice is another important factor for the client well-being and mental health stability.

According to the Housing First philosophy housing is a basic right. This right should apply to each person and it is not a reward at the end of a treatment condition. The Treatment First approach demands that clients have to achieve appointed goals before the service provider decides that the clients are “housing ready” [11] .

However, the Housing First program separates housing from treatment support and contributes a secure leasing agreement. The consumer-driven support offers harm reduction and at least one on-site contact with a caregiver per month.

The meaning of the abbreviation C.O.M.E. is Choice to cooperation, Observation, Mutual-support and Evaluation. From the two competing approaches Treatment First and Housing First we derived the C.O.M.E. program with the aim to enable the clients to live a more independent life. This non-abstinence approach promotes an individual need-oriented care by teaching social skills and social participation and by providing support in health-topics and everyday-matters. Housing and support are separated [11] . Furthermore, this program includes an assessment-sheet which reflects the topics of a self-determined housing. The assessment-sheet describes issues which often occur in home treatment situations.

One of the biggest challenges in the daily care of supportive housing is to keep the contact to the clients [8] . To handle this challenge it requires a stable cooperation-relationship between the clients and the caregivers. Our assumption of a sustainable relationship is built on three factors:

1) safety,

2) control and,

3) confidence.

Already at the beginning it is agreed that the service offer will be upheld even if the client relapse or reject it at times. Generally there will be one weekly face-to-face contact in the flat. The caregivers at least ensure the housing retention and react in time in cases of emergency. According to the needs of the clients and to the current situation there will be more or less contacts appointed. This procedure promotes control and autonomy of the clients. Moreover, the caregivers clarify that all issues and other topics are treated confidential.

The C.O.M.E. program approach involves a minimum willingness to cooperation between the service workforce (e.g. psychiatrists, psychologists, nurses), relatives, guardians and the clients of the program. Self and external observing during the process is also an important part of this program [12] . The mutual support as an unprofessional support can be useful in daily situations when the caregivers are not on site (e.g. support by neighbours). In cases of crisis it is necessary to meet the individual needs and therefore the caregiver has to respond flexibly and react in time (e.g. if the symptoms get worse or if conflicts occur) [13] . This requires that the caregivers have to evaluate and to reappraisal current situations and possibly to adjust the intervention to prevent a sudden interruption of the home-treatment support.

About 150 people are supported in the context of the community psychiatry home treatment service in Vorarlberg. 52 of them took part in our long-term study at the baseline in the year 2016. All of the probands, were able to provide informed consent and had a history of mental illness and were currently housed. Participants were drawn from the service providers to the long-term study and have to be at least 18 years old at the baseline. To reduce the loss at the beginning and during the data collection the caregiver of the service providers conduct the Interview and the ongoing measurements together with the probands. This procedure matches to the agreement and contents of the C.O.M.E. program―caregiver training. A total of 36 probands could be included into the further analysis. The age range was 20 to 68 years at the beginning of the study. The participants were both women (N = 20) and men (N = 16). In the case of a psychiatric and/or a drug diagnosis they get the social welfare benefit from the Government of Vorarlberg which covers the costs of the home-treatment-service. The participant requires a lease contract, own a home or have a use-permit from a service provider.

The federal state Vorarlberg counts 400,000 residents and is divided into four districts. In sum, four service providers ensure the community treatment service for people with mental health problems. One of the four service provider operates with use-permits. Among the rest three providers the service user needs a lease contract or an own home.

All of the participating caregivers had different education and practical experiences. In order to reduce the confounding variables of this heterogeneous group it was necessary to teach them the theoretical basics and the practical application of the assessment sheet and the WHODAS [15] . Therefore we contributed workshops for caregivers. One of the course content was to emphasize that a regular weekly scoring is as necessary as a related notice report for the data collection and its evaluation. Further we offered a troubleshooting and a coding-list. Another important point was that the caregivers accept that the participants decide the scope of service and that the caregivers have to work in a non-abstinence treatment condition (drugs, alcohol), comparable with the housing first principle [6] .

A sustainable supporting-cooperation is based on the three attitudes: safety, control and confidence. The client-led planning of interventions and goals take into consideration that the caregivers have to know the different stages of the participant’s medical-history [13] . Therefore, we designed a psychosocial-interview according to Carlat’s guidelines for anamnestic surveys of psychologists, social workers, nurses and psychiatric issues [16] . Demographic data include descriptive information of age, gender, education, employment and marital status. The history of housing and of illness, alcohol abuse and the drug history were assessed during a single interview. The demographic variables were also used for sample descriptions and for inferential statistical models. The descriptive data were processed using SPSS (Version 23).

In addition we used an assessment-sheet which was scored weekly for every participant. Each assessment-sheet contains 13 items which are related to the general steps and goals of the C.O.M.E. program and complies with four dimensions (as shown in Table 1):

・ household management,

・ personal hygiene,

・ (social) participation,

・ psychosocial stability.

The data were collected in the context of the home treatment service. There was no randomised controlled trial. The target groups pursued the same goals. Only the boundary-conditions were different (no lease contract, abstinent or not). Additionally, we use the caregivers report to interpret the weekly measurements.

Figure 1 shows the example of the assessment-sheet bringing more light

how to use and how this tool is constructed. The items are divided into serval categories to make the changings of the probands during the home treatment contacts visible. The assessment-sheet is designed as a paper-pencil application.

Table 2 shows the factor analysis of the 13 item assessment-sheet confirmed the assignment of the items to 3 dimensions. The Kaiser-Meyer-Olkin (KMO) measure of the sample which suggested a minimum value of KMO = 0.600 to use the data for any further analysis was met. Values closer to KMO = 1.000 are better. The data from the current study has a KMO value which can be used for further analysis (KMO = 0.814). The Bartlett-Test of Sphericity was significant (p ≤ 0.001). The Bartlett-Test measures indicate that the correlation matrix is an identity matrix. All diagonal elements are 1 and all off diagonal elements are 0. The two tests are a minimum standard which should be passed before a principal components analysis should be conducted. The extraction retained 3 factors which in sum explain 54.65% of the variance.

In addition, a Cronbach’s alpha test was conducted. The Cronbach’s alpha (α) is a measure of internal consistency, that is, how closely related a set of items are as a group. The measure is considered as the scale reliability. A value of α = 0.70 and higher is acceptable. The alpha coefficient for the 13 items is α = 0.812. A crucial part of the assessment sheet is to score the types of contacts between the participants and the caregivers because it is the basis for cooperation especially in crisis. The quality of contact can change through cooperation. The evaluation

Extraction Method: Principal Axis Factoring; Rotation Method: Varimax with Kaiser Normalization^a.

of the assessment-sheet item “Type of contacts” makes changes in contact-types visible. This item is splitted into four different scaling types (Table 3).

The WHODAS is a generic cross cultures assessment instrument for health and disability. It corresponds to the requirements of International Classification of Functioning, Disability and Health (ICF). The scale includes mental, neurological and addictive disorders and comprises six domains of functioning (Table 4).

The WHODAS is a well-studied schedule which captures the health difficulties over the past 30 days [15] . The plan includes 30 items. It takes about 20 minutes to complete it and the plan is easy to handle. The caregivers have to use the WHODAS at the beginning and at the end of the home treatment service.

Table 5 shows the characteristics of the probands. There is no correlation between the characteristics and the items of the assessment sheet.

The sample of 36 probands have an age average of 46.76 years (SD = 11.72). This long-term study includes 43.2% men and 56.8% women. The education level shows that most of the probands (52.8%) have only a secondary school without further education. 36% of the probands completed a trainee program. Only 11.2% of the probands graduated a high school or a college. The marital status shows that 22 probands (61%) are singles, two are married (5.6%), eleven are divorced (30.6%) and one is actually in a relationship.

The result of the income analysis is that 32 probands (88.8%) gets a pension or rehabilitation welfare, three of the probands (8.4%) get a social-welfare benefit and only one participant is an employer. Over 97% of the participants don’t have any regular work or activity. In seven cases a guardian was appointed (19.4%).

The analysis of the diagnoses shows that 22 probands suffer from psychotic disorder (61.1%), seven persons suffer from depressive disorder (19.4), five probands suffers from personality disorder (13.9%) and two probands have a substance use disorder (5.6%). According to the medical history report 52.8% of the probands were admitted in a psychiatric ward within the last year, 33.3% during a period of one to five years.

The further analysis was computed with the Spearman rank-order correlations (r_s) of the 13 item of the assessment-sheet (Table 6). The result suggests that the correlation within all 13 items is high significant. (p ≤ 0.001). The evaluation of the item psychopathological state shows a positive trend indicative of a stabilization of the associated items (Table 6). Except for the correlation between the items “Type of Contact/Grocery” (r_s = 0.270*; p ≤ 0.05) and “Laundry/Medication” (r_s = 0.069*; p ≤ 0.05) there were a high significance on a 95% level (p ≤ 0.05). There was a nonsignificant correlation (r_s = 0.68; p = n.s) between the items: “Dealing with Drugs/Home cleanness”; “Dealing with Drugs/Activities (social)”; “Activities (social)/Trash” and “Activities (social)/Medication”. This is due to the circumstance

**high significant (p ≤ 0.001) correlation; *significant (p ≤ 0.05).

that the items above are also not correlated within the rotated factor matrix (Table 2). The positive course of the other items reflects the increase in advisory contacts. Using the Cohen’s guidelines, the effect size ranged from small (r_s = 0.10) to medium (r_s = 0.30) [17] .

In some cases the loss of the flat or home is related to a decreased cleanness of the home. So the landlords or service providers could quit the lease contract. We assume that there is a relation between the items “Type of Contact”, “Home-cleanness”, “Psychopathological state” and “Dealing with drugs”. The analysis of the assessment-sheet enable to show results which are related if e.g. drug or alcohol consume increase or contacts between the caregivers and clients decrease. The following analysis was conducted with only the categories in which the probands neglect the contacts (“Type of Contacts―retreat”), show symptom worsening (“Psychopathological state―instable”), increasing drug or alcohol consume (“Dealing with drugs―instable”) and less scoring within the item “Home-cleanness”. The results suggest that in case of less or no contacts the cleanness of the home decrease. The spearman correlation between the items “Home-cleanness” and “Type of Contacts―retreat” is high significant (r_s = −0.382**; p ≤ 0.001). The spearman correlation between the items “Home-cleanness” and “Psychopathological state―instable” is high significant (r_s = −0.740**; p ≤ 0.001). The spearman correlation of the item “Home-cleanness” and “Dealing with drugs―instable” show also a high significant result (r_s = −0.357**; p ≤ 0.001).

Figure 2 shows the frequency analysis of the item “Type of Contacts”. It reveals a number of 1583 contacts between the caregivers and the participants and displays that in 39% the “Caregiver stays in contact” with the probands. In addition, 36% of coding this item belongs to the category “Proband seeks contact”. In 21% of the cases there was “No need” and in 3% of coding this item the probands retreat the contacts to the caregiver. Only 1% was an invalid coding.

Figure 3 shows the coded categories of the item “Types of Psychopathological State” in which the caregiver rates the state of the proband. The analysis counts 1350 codings. The categories are distributed as followed. The category “stable” was coded in 42.90%, the “moderat” category was coded by the caregiver in 46.10% and in 9.60% of the cases the caregiver coded the “instable” category. In 1.40% there was an invalid coding. The spearman correlation between these two items as shown in Figure 2 and Figure 3 is high significant (r_s = 0.264, p ≤ 0.001).

In addition to the spearman rank-order correlation we use an ordinal regression to predict the variables which influences the stability of the psychopathological state which was used for the dependent variable. The independent variables are

the categories of the assessment-sheet as shown in Figure 1. The ordinal regression is a member of the family of regression analyses. It describes data and explains the relationship between one dependent variable (psychopathological state) and two or more independent variables (Type of contacts, activity, sleep habits, daily structure, cleanness of the home and body washing). This is a predictive analysis. The 2-Log-Liklihood value was significant (p < 0.001) for the items of the assessment sheet. So we can assume that there is a significant improvement of the psychopathological state. The Nagelkerke pseudo was R² = 0.39 (percentage of variance declared by ordinal regression). Further analysis show that the independent variables (items listed * in Table 7) have a significant (p < 0.05) effect to predict a stable psychopathological state.

Table 7 shows the independent variables which predict a stability of the psycho-pathological state on a 95% significance level.

The category “caregiver stays in/takes contact” of the item “Type of contacts” influences the psychopathological state in a positive direction. This category records the contact (face to face, telephone) which is initiated by the caregiver to

prevent a symptom worsening, social withdrawal or other similar situations. The category “proband seeks contact” describe the initiative contact by the proband. The extended availability of the workforce during the day made it possible for the participants to react in time in cases of symptom worsening to call support from the caregiver.

Most of the other independent variables have a motivational or guidance content which results only from the contact with the caregivers. This points the importance of contacts between the probands and the clients, too.