RESEARCH ARTICLE Open Access
Service provision and barriers to care forhomeless people with mental health problemsacross 14 European capital cities
Réamonn Canavan
1*
, Margaret M Barry
1
, Aleksandra Matanov
2
, Henrique Barros
3
, Edina Gabor
4
, Tim Greacen
5
,Petra Holcnerová
6
, Ulrike Kluge
7
, Pablo Nicaise
8
, Jacek Moskalewicz
9
, José Manuel Díaz-Olalla
10
,Christa Straßmayr
11
, Aart H Schene
12
, Joaquim J F Soares
13
, Andrea Gaddini
14
and Stefan Priebe
2
Abstract
Background:
 Mental health problems are disproportionately higher amongst homeless people. Many barriers existfor homeless people with mental health problems in accessing treatment yet little research has been done onservice provision and quality of care for this group. The aim of this paper is to assess current service provision andidentify barriers to care for homeless people with mental health problems in 14 European capital cities.
Method:
 Two methods of data collection were employed; (i) In two highly deprived areas in each of the 14European capital cities, homeless-specific services providing mental health, social care or general health serviceswere assessed. Data were obtained on service characteristics, staff and programmes provided. (ii) Semi-structuredinterviews were conducted in each area with experts in mental health care provision for homeless people in orderto determine the barriers to care and ways to overcome them.
Results:
 Across the 14 capital cities, 111 homeless-specific services were assessed. Input from professionallyqualified mental health staff was reported as low, as were levels of active outreach and case finding. Out-of-hoursservice provision appears inadequate and high levels of service exclusion criteria were evident. Prejudice in theservices towards homeless people, a lack of co-ordination amongst services, and the difficulties homeless peopleface in obtaining health insurance were identified as major barriers to service provision.
Conclusions:
 While there is variability in service provision across European capital cities, the reported barriers toservice accessibility are common. Homeless-specific services are more responsive to the initial needs of homelesspeople with mental health problems, while generic services tend to be more conducive to long term care. Furtherresearch is needed to determine the effectiveness of different service delivery models, including the most effectivecoordination of homeless specific and generic services.
Background
Mental health problems are higher amongst the homelesspopulation than amongst the general population [1-7]. The more severe the level of homelessness the poorer thelevel of mental health [3,8,9]. Less than a third of homeless people with mental health problems receive treatment [3].A permanent residence still represents one of the mainrequirements for registering with the health care systemsin a number of European countries [10,11]. Homeless people are often reported as having problems registeringwith health services and often try and access mentalhealth care through accident and emergency services,where it is unlikely they will receive the appropriate careand treatment [5,7]. Entitlement to health care for homeless people doesnot always mean access [10,11]. Limited accessibility is often due to factors such as opening hours, inflexible ap-pointment procedures and location [7,11]. Homeless people may also encounter attitudinal barriers withinservices and there is often an unwillingness or difficulty 
1
Health Promotion Research Centre, National University of Ireland Galway,University Road, Galway, IrelandFull list of author information is available at the end of the article
© 2012 Canavan et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the srcinal work is properly cited.
Canavan
 et al. BMC Health Services Research
 2012,
 12
:222http://www.biomedcentral.com/1472-6963/12/222
 
in the health services to accommodate the multiple andcomplex needs presented by homeless people [10-12]. Despite the strong link between homelessness andmental health and the many barriers homeless peopleface in accessing health services, there is little informa-tion on the characteristics and quality of serviceprovision for homeless people in Europe [12]. This paperdraws on findings from the PROMO project (DG Sanco:2007
 
 2010), whose aim was to assess current serviceprovision and quality of care across 14 European capitalcities for people from the following socially marginalisedgroups who experience mental health problems: long-term unemployed, homeless, sex workers, refugees andasylum seekers, undocumented migrants and travellers.The findings specific to the homeless population are pre-sented here.
Methods
The PROMO project included the selection of twohighly deprived areas in each of the 14 participating cap-ital cities. Within these areas the study sought to: (i) ob-tain information on services providing care for homelesspeople with mental health problems; (ii) assess the over-all quality of care for homeless people with mentalhealth problems. The definition of homelessness used inthis study comprised the first two categories of the exist-ing ETHOS typology [13], i.e. roofless (people sleepingrough or in emergency accommodation) and houseless(people in temporary accommodation).
Identification of research areas
A total of 28 highly deprived geographic areas wereselected for the PROMO assessment (See Table 1). Thefocus was on highly deprived areas due to the tendency for marginalised groups to be concentrated in theseareas. The areas were identified by using the relevantlocal indices of public health and social deprivation. Thepopulation size of each research area was originallplanned to be between 80,000 and 150,000 inhabitants.However, some flexibility in population size was allowedin order to accommodate different local contexts relat-ing to administrative boundaries and service catchmentareas. Also, in some cases one or more areas were com-bined to achieve the target size.
Assessment of services
The PROMO assessment of services sought to assess allmental health, social care and general health servicesthat potentially serve marginalised groups with mentalhealth problems. While the assessment was focused onthe selected deprived areas, services located outsidethese areas but used by people from the target groupsfrom the areas were also assessed.The focus of the assessment of services in this paper ison all mental health, social care and general health ser- vices which are directed specifically at homeless people.While any health service may potentially be a resourcefor homeless people, homeless people don
t tend to ac-cess the more generic health services for reasons out-lined in the introduction e.g. accessibility, lack of healthinsurance, lack of outreach services etc. Data onprovision in the generic health services is also presented,focussing on how such provision compares to that in thehomeless specific services.Services were coded as homeless specific based onservice self-definition. The coding of each service wasdone by the researchers who carried out the interviewsin their capital city. In a small number of cases it wasnot clear which marginalised group the service wasaimed at or whether the service was group specific orgeneric. In such cases if 50% or more of the clientswere estimated to be homeless the service was desig-nated as a homeless specific service. Services were alsoclassified as either mental health, social care or generalhealth services. This distinction was once again basedon service self-definition. In cases where it was notclear whether a service was mental health specific orgeneric, if 50% of clients were estimated to have a men-tal health problem the service was classified as a mentalhealth service.
Assessment of services tool
A structured questionnaire was developed for the assess-ment of services using an iterative process involvingresearchers from all participating cities. It was translatedinto the languages of the participating countries andthree pilot interviews were carried out in each city. ThePROMO tool was designed to assess the followingaspects of service provision:
1. Provider and funding information2. Characteristics of staff 3. Service accessibility 4. Characteristics of clients5. Programmes provided to clients from target groups6. Co-ordination with other services7. Service evaluation
Services were identified according to available direc-tories of services and information from relevant localpractitioners. Information gathered during the interviewswas used to consistently update the list of services. Theassessments were carried out by PROMO researchers ineach of the 14 capital cities, either face-to-face or overthe phone, with either the manager of the service or amember of staff with the relevant knowledge.
Canavan
 et al. BMC Health Services Research
 2012,
 12
:222 Page 2 of 9http://www.biomedcentral.com/1472-6963/12/222
 
Overall quality of care for homeless people with mentalhealth problems
In order to assess the overall quality of care semi-structured interviews with
 
experts
 in mental health carefor homeless people were conducted, one in each re-search area (n=28). The experts were identified duringthe assessment of services phase of the study where ser- vice managers were asked to identify suitable intervie-wees. The criteria for inclusion were a good knowledgeof local service provision and professional experience of providing or facilitating access to mental health care forhomeless people. If such an expert could not be foundin the area, they were recruited from other areas of thecity. The experts were contacted by the researchers ineach country via telephone or email and invited to par-ticipate in the study, the purpose of which was fullexplained to them. All interviews were carried out face-to-face and audio-taped.The interviewees were employees in a wide range of services in the not-for-profit and state sector and had a variety of professional backgrounds: Social workers (8),Psychiatrists (7), Psychologists (5), Educators (2) Psychi-atric nurse (1), Medical Doctor (1), Lawyer (1), Nurse(1), Homeless service manager (1) & Therapist (1).The semi-structured interview was developed using aniterative process involving all partners and translatedinto the languages of all participating countries. Onepilot interview was conducted in each participatingcountry. The overall interview protocol consisted of (i)two case vignettes which described two patients with dif-ferent mental health problems and with different atti-tudes towards seeking care. The experts were askedabout their pathways into care, the barriers to receivingcare and ways to overcome these barriers. The same vignettes were used across all capitals to ensureconsistency; and (ii) four questions regarding thestrengths, weaknesses and co-ordination of services forhomeless people with mental health problems, and how service provision may be improved.For the purposes of this paper the research questionsanalysed were a) what are the barriers to mental healthcare for homeless people? b) what are the ways to over-come these barriers? c) what are the two most importantchanges in practice that would improve care for home-less people with mental health problems?All interviews were transcribed, ensuring the removalof any identifying information to maintain anonymity,and were translated into English. The study coordinatingcentre examined the translated transcripts and soughtany necessary clarifications from the respective centres.Ethical approval was not required in the participatingcountries for this study, as there was no health interven-tion and no personal information was collected.
Data analysis
PASW 18.0 was used to analyse the assessment of ser- vices data. Chi-Square and Mann
Whitney tests wereused to assess any significant differences in serviceprovision between the homeless specific services and thegeneric services assessed as part of the wider project. Allstatistical tests were two-tailed and the significance levelwas set at (p
<
0.05).The semi-structured interview transcripts were ana-lyzed using thematic analysis [14]. The data from the ini-tial 12 transcripts were coded independently bresearchers from two capital cities and a coding framewas produced. This coding frame was then used to codethe remaining 16 transcripts. This was carried out by 
Table 1 Target areas identified for assessment in 14 EUcapital cities
COUNTRY/CAPITAL AREA 1 AREA 2(census year) (population) (population)Austria/ Vienna (2008)
District 1694,735District 2082,369
Belgium/ Brussels (2007)
Schaerbeek+St Josse113,493+ 23,785Molenbeek 81,632
Czech Republic/ Prague (2006)
Praha 3+Praha 769,939+39,425Praha 8100,255
France/Paris(2006)
Secteur Flandrepsychiatric sector102,387La Courneuve+Aubervilliersin Seine Saint Denis37,347+73,506
Germany/ Berlin (2006)
Wedding (thesub area of 
Schillerpark 
removed)123,191Kreuzberg147,798
Hungary/ Budapest (2001)
District VIII.81,787District VII. and IX.64,137+ 62,995
Italy/Rome(2007)
District 7117,479District 15146,090
Ireland/ Dublin (2006)
Dublin NorthCentral126,572Dublin West134,020
Netherlands/ Amsterdam (2006)
Bos en Lommer+De Baarsjes+Geuzenveld-Slotermeer30,045+33,767+ 41,314AmsterdamZuid Oost78,922
Poland/ Warsaw (2006)
Praga Polnoc73,207Wola142,025
Portugal/ Lisbon (2001)
A group of smaller areas85,177Marvila+SantaMaria dos Oliváis82,753
Spain/ Madrid (2006)
Villaverde146,859Centro149,797
Sweden/ Stockholm (2010*)
Rinkeby-Kysta+Spånga-Tensta+Skarpnäk 45,500+36,000+40,000Södermalm118,000
UK/London (2001)
 Hackney202,824 Tower Hamlets196,106
*Figures provided were updated during the project using 2010 data.
Canavan
 et al. BMC Health Services Research
 2012,
 12
:222 Page 3 of 9http://www.biomedcentral.com/1472-6963/12/222
 
one researcher at the project co-ordinating centre. Theinitial codes were then merged into categories and fur-ther refined into conceptual themes. [15,16]. The fre- quency counts of themes and the correspondingcategories were recorded. The emerging themes wereregularly revised at the co-ordinating centre and dis-cussed within the wider international project group.
Results
Assessment of services
In total 111 homeless-specific services were assessed. Ascan be seen in Figure 1, there was a high level of vari-ability in the number of services assessed across the par-ticipating capital cities.Nineteen of the 111 services assessed were describedas homeless-specific mental health services. These ser- vices were found in seven of the 14 participating capitalcities: Berlin (5), Paris (4), Dublin (3), London (3), Stock-holm (2), Amsterdam (1) and Rome (1). Eighty-four ser- vices were described as homeless-specific social careservices and were found in all participating capital citiesbar one. These services included 30 accommodationrelated services, 17 day centres, 13 social support ser- vices providing for example advice and support, and 5outreach services. The remaining 19 services providedmore than one service type e.g., accommodation plusoutreach. Eight services were described as homeless-specific general health services, all of which were pri-mary care services. These services were found in fourparticipating capital cities: Berlin (4), London (2), Dublin(1) and Warsaw (1).A summary of the main findings relating to serviceprovision in the research areas of the 14 participatingcapital cities is shown in Table 2.Twelve of the assessed services were provided specific-ally for women and 10 specifically for men. Overall 64%of services reported some form of exclusion criteria
,
with aggressive behaviour being the most prominent.There was a high level of variability across countries interms of active outreach provision. The research areas inParis and London reported the highest levels of outreachprovision (71% and 62% respectively) with the lowestreported in Warsaw and Vienna (13% and 0% respect-ively). High levels of variability were also reported in theprovision of internal supervision for staff and whetheraggressive behaviour on the part of homeless people wasa reason for exclusion from service.The median number of whole time equivalent(WTE) staff reported across all services was 7.25(interquartile range 12.00). The percentage of servicesproviding professionally qualified health care staff canbe seen in Figure 2. The majority of services (70%)reported that they do not employ any professionally qualified mental health staff (a psychiatrist, apsychologist/psychotherapist or a counsellor to a com-bined WTE of at least 0.5). 29% reported that they donot employ social care staff (an occupational therapistor a social worker to a combined WTE of at least0.5). In terms of peer support 9 (8%) of servicesreported that former clients are involved in direct de-livery and contact with clients in a paid role and 16(14%) in an unpaid role.92% of services assessed reported providing some typeof social care programme (social welfare support, hous-ing support, legal advice and support or job coaching/finding). 21% of services reported that they provide sometype of addiction treatment programme (detoxificationtreatments, drug addiction treatments or alcohol addic-tion treatments).Of the 350 generic services assessed as part of the widerproject, 148 (42.3%) reported that they document whetherthe client is homeless and 28 (8%) reported that they pro- vide a specific programme for homeless people. In com-parison to the generic services assessed, homeless-specificservices (n=111) were significantly more likely to be pro- vided by 
 
not for profit private organisations
 (71.2% vs47.7%, p
<
0.01**); to engage in case finding (27.0% vs16.9%, p
<
0.05*); to provide some type of social careprogramme (92.0% vs 76.1%**); and were less likely to re-port having a waiting list (27.9% vs 47.0%**).On the other hand, the generic services assessed weresignificantly more likely to report providing addictionprogrammes (42.1% vs 20.9%**); individual psychother-apy (48.0% vs 16.2%**); and to have a higher number of paid staff (median 9.5 [interquartile range 21.5] vs 7.3[12.0]*). They were also significantly more likely to re-port having doctors (26.7% vs 16.5%*); psychiatrists(34.9% vs 11.0%**); psychologists/psychotherapists(50.6% vs 22.0%**); and occupational therapists (22.9% vs6.4% **) as part of staff. On the other hand the homelessspecific services were significantly more likely to reporthaving social workers (68.8% vs 57.2%*) as part of staff than the generic services assessed.
05101520253035
Figure 1
 Number of services assessed in each capital city(median=7, SD=8.29).
Canavan
 et al. BMC Health Services Research
 2012,
 12
:222 Page 4 of 9http://www.biomedcentral.com/1472-6963/12/222
 
Services which were defined as homeless-specific mentalhealth services (n=19) were significantly more likely toreport providing active outreach (63.2% vs 33.6%*) andsupport around housing (84.2% vs 58.9%*) than the gen-eric mental health services assessed as part of the widerproject (n=221). However, they were significantly lesslikely to report having psychiatrists (21.1% vs 47.5%*), psy-chologists/psychotherapists (36.8% vs 60.2%*) and occupa-tional therapists as part of their staff (5.3% vs 28.9%*), andto provide psychotherapy (5.3% vs 63.5%**).
Barriers to care and ways to overcome them
The most common themes and corresponding categoriesarising in the interviews are presented here. The numberof experts who highlighted the issue is indicated in eachcase (n=28).
Barriers
A common theme identified was the difficult and chaoticlife circumstances of homeless people (23 experts), in-cluding alcohol and substance abuse issues (13) and dif-ficulties in maintaining medication compliance (10). Theunwillingness amongst the homeless population to en-gage with the services was also seen as a barrier (17),often due to a lack of trust in health professionals (12).Barriers relating to health insurance were frequently reported (15), mainly relating to not having insurance ornot being registered with a General Practitioner (GP)(14). Admission and discharge procedures in the healthservices were also highlighted (11), with the main barrierhere being a lack of clear responsibility within the ser- vices in relation to the treatment of homeless peopleand complex rules in relation to catchment areas (8).Lack of collaboration between mental health, socialwelfare and homeless services (14) was also highlightedfrequently by the experts, as was a lack of mental healthoutreach provision (10)
.
 Prejudice/negative responses by health professionals towards homeless people were regu-larly highlighted (15). Barriers linked to the provision of 
Table 2 Characteristics of assessed services across all research areas in participating capital cities (n=111)
Variable N% Variable N (%)Accessibility Programmes provided
Accepting self-referrals 89 (80.2%) Active outreach 42 (37.8%)Open outside office hours Mon-Fri 58 (52.3%) Case finding 30 (27%)Open and time at weekends 67 (60.4%) Counselling 70 (63.1%)Services requiring out of 
 
pocket
 fee for payment 40 (36%) Individual psychotherapy 18 (16.2%)Waiting lists for any aspect of the service 31 (27.9%) Detoxification treatment 12 (10.8%)Drug addiction treatment 15 (13.5%)
Exclusion Criteria
 Alcohol addiction treatment 17 (15.3%)Addiction 25 (22.5%) Support around social welfare 85 (76.6%)Aggressive behaviour 49 (44.1%) Housing advice and support 95 (85.5%)Criminal history 7 (6.3%) Legal advice and support 59 (53.2%)Command of language of host country 14 (12.6%) Job coaching and finding 67 (60.4%)Lack of motivation 28 (25.2%) Mental health advocacy 44 (39.6%)
Co-ordination Staff Supervision
Routine meetings at least once a month concerningthe care of homeless people60 (54.1%) Internal supervision 58 (52.3%)External Supervision 50 (45.5%)
Evaluation: Systemic recording of Services provided by
Socio-demographic characteristics of clients 97 (87.4%) State 32 (26.1%)Attendance and care provided 85 (76.6%) Not for profit/private 79 (71.2%)Clients satisfaction and experience 45 (40.5%)
00.511.522.533.5CounsellorsDoctorsNurses OTPsychiatristsPsychologistsSocial Workers
Figure 2
 Average number of reported professional staff perservice (whole time equivalent) (n=111).
Canavan
 et al. BMC Health Services Research
 2012,
 12
:222 Page 5 of 9http://www.biomedcentral.com/1472-6963/12/222
of 9