The estimation of homeless people's mental health problems became a significant issue in the psychiatric literature of the 1980s and 1990s, ^{Reference Weller and Weller1-Reference Weller, Weller, Coker and Mahomed4} and remains important today. ^{Reference Timms and Fry5,Reference Marshall6} There are concerns that, with the deepening economic crisis, the issue of homeless people with mental health problems will feature strongly again in psychiatric literature. Many surveys have been conducted to estimate these needs, but the majority have used diagnosis as the sole indicator of these needs. Diagnosis is, however, limited for the following reasons:

1. • it does not denote a plan of clinical action nor does it define the person as needing medical care; ^{Reference Desai and Rosenheck7}

2. • case definition in psychiatric community surveys tends to overestimate the need for psychiatric services; ^{Reference Henderson, Bainbridge, Keaton, Kenton, Guz and Kanis8}

3. • social disablement is a better indicator of need for psychiatric services. ^{Reference Kendell9,Reference Bebbington10}

Because of these problems, Bebbington suggested the use of ‘needs assessment’ procedures rather than diagnostic procedures to assess needs for psychiatric treatment in community surveys. ^{Reference Henderson, Bainbridge, Keaton, Kenton, Guz and Kanis8} Few studies have specifically assessed the psychiatric needs of homeless people. ^{Reference Timms and Fry5,Reference Wykes, Creer and Sturt11}

Psychiatric surveys of homeless people have had many methodological limitations in terms of the lack of epidemiologically representative samples of the different types of homeless people as well as the problems with using case definition highlighted above. ^{Reference Wing12} These studies have suggested - but did not measure - the presence of unmet need for mental health services among homeless people. McQuistion et al called for a study of the epidemiological trends in the mental healthcare needs of homeless people. ^{Reference Abdul-Hamid13} This, in the authors' opinion, is the only effective and credible way that will enable communities to assess and deal with the problem posed by homeless people with psychiatric disorders. A report by Professor Siân Griffiths suggested that 30-50% of people who sleep rough have some form of mental health need, and suggested that ‘tackling homelessness needs more than just a housing response: it needs determined and coordinated action over a period of time by a wide range of partners’. ^{Reference Abdul-Hamid, Wykes and Stansfeld14}

Many needs assessment tools have been developed since the 1990s; we used one that made a judgement of need based on assessment of patients' social behaviour, the burden on relatives and patients' attitude to services, as well as the staff perception of the degree of need. ^{Reference Kendell9,Reference McQuistion, Finnerty, Hirschowitz and Susser15} This judgement was made for 13 interventions that constituted professionally defined needs equivalents. We used the Wykes version (which was the basis of later editions) because it was more suitable for community surveys of people not in contact with psychiatric services. There are also comparison data available on the needs of the people in contact with different community psychiatric services. ^{Reference Kendell9,Reference McQuistion, Finnerty, Hirschowitz and Susser15} We tried to fill gaps in the previous literature by addressing the following aims:

1. • assessing the feasibility of using an established and simple needs assessment method on a sample of homeless people;

2. • describing the needs of a representative sample from hostels for homeless men in an inner-London health district;

3. • describing the needs of those in the sample who presented with florid psychotic problems.

Method

Homeless individuals (n = 110) living in four hostels for homeless men in the Bloomsbury health district were selected randomly from the room-lists of these hostels. From this randomly selected sample, 101 residents agreed to participate in the study (92% participation rate).

The Hostel Interview Schedule (HIS) was completed by all participants. It included demographic data, health and contacts with health services data, staff burden, the Social Behaviour Schedule (SBS) ^{Reference Kendell9} and the Need Schedule. The Needs Schedule is a semi-structured questionnaire adapted from Wykes et al. ^{Reference Kendell9,Reference McQuistion, Finnerty, Hirschowitz and Susser15} It covers 13 needs that describe community mental healthcare and support. The data for the Needs Schedule were collated both from the residents under study and from the staff members who knew them best. The judgement of the needs each resident was based on assessment of the patient's social behaviour, the burden on staff and the resident's attitude to services, as well as the staff perception of the degree of need.

These qualitative and quantitative data were then assessed by the research team for each resident and the need for each item of community mental healthcare coded into five categories: no need, met need, unmet need, overmet need and unclear need. Items were recorded as ‘no need’ if there was no deficit in a resident's functioning or health problem that required this intervention. ‘Met need’ was recorded as any need in which an intervention had been provided that effectively alleviated or treated the resident's problem. ‘Unmet need’ was recorded when there was a problem that required an intervention that was not currently being provided, or had not been provided effectively. ‘Overmet need’ was recorded when an intervention was provided when no problem was identified that might require that intervention. The ‘unclear need’ category was recorded when there was a problem that required a specific intervention but something hindered the provision of this intervention. This included the resident's refusal or if the intervention was secondary to another intervention - for example, the need for day care or counselling when the resident had not received any psychiatric assessment that would have revealed more clearly the need for either of these items.

The reliability of the Needs Schedule was assessed by an independent evaluation of the needs of 10 randomly selected participants from the study sample. For individual items the percentage agreement ranged from 50 to 100%. The calculated kappa was 0.4-0.8, which constitutes fair to good agreement.

Results

Sociodemographic characteristics of the sample

The men in the sample were aged 29-84 years (mean 59, s.d. = 13); 77% were over 50 years old and 37% were over 65 years old. The majority were single, and unemployed or retired. Half of the hostel residents were born in the UK and a third of them were from Ireland (Table 1). The mean length of stay in the study hostels was 8 years but half of the residents had been there for more than 5 years.

Table 1 Demographic characteristics of the sample (n = 101)

	n (%)
Marital status
Single	74 (74)
Widowed	5 (5)
Divorced	21 (21)
Employment
Regular	5 (5)
Casual	6 (6)
Unemployed	48 (48)
Old age pensioner	35 (35)
Other	7 (7)
Ethnic origin
UK	54 (54)
Ireland	36 (36)
West Indies	1 (2)
Asia	2 (2)
Europe	2 (2)
Other	5 (5)

Needs

The highest recorded need was for housing and residential care (Table 2). In 60% of residents this need was met, as both residents and staff agreed that the hostel was the appropriate placement. Fourteen per cent of the residents needed residential care with higher levels of supervision. In 14% of the residents this need was coded as overmet, as residents needed more independent accommodation. Social assessment was the next most prevalent need, in 30% of the sample; in 29% this need was unmet. One per cent of the sample had a met need in a hostel run by social workers. Twenty-four per cent of the sample needed day care; in nine per cent this need was met by hostel day care or liaison with a day centre. Four per cent of the sample needed training in domestic skills and eight per cent needed help with self-care; both these needs were unmet.

Table 2 Needs for mental health services of the men in the survey hostels (n = 101)

	No need n (%)	Met need n (%)	Unmet need n (%)	Overmet need n (%)	Unclear need n (%)
Security	99 (98)	0 (0)	2 (2)	0 (0)	0 (0)
Psychiatric assessment	66 (65)	5 (5)	26 (26)	0 (0)	3 (3)
Training in domestic skills	94 (93)	0 (0)	4 (4)	0 (0)	3 (3)
Self-care	93 (92)	0 (0)	8 (8)	0 (0)	0 (0)
Behavioural therapy	99 (98)	0 (0)	0 (0)	0 (0)	2 (2)
Social activity	88 (87)	1 (1)	8 (8)	0 (0)	4 (4)
Counselling	80 (79)	2 (2)	11 (11)	14 (13.9)	8 (8)
Residential care	0 (0)	61 (60)	14 (14)	0 (0)	7 (7)
Day care	65 (64)	9 (9)	15 (15)	0 (0)	12 (12)
Domiciliary psychiatric service	87 (86)	0 (0)	1 (1)	0 (0)	13 (13)
Staff support	73 (72)	1 (1)	27 (27)	0 (0)	0 (0)
Alcohol and drug services	69 (68)	1 (1)	20 (20)	0 (0)	11 (11)
Other needs	80 (79)	2 (2)	19 (19)	0 (0)	0 (0)
Social assessment	70 (69)	1 (1)	29 (29)	0 (0)	1 (1)

Thirty-one per cent of the sample needed psychiatric assessment and in twenty-seven per cent the need was unmet. Thirteen per cent of the sample needed counselling; in eleven per cent the needs were unmet. Most of the needs for domiciliary psychiatric assessment and for behavioural therapy were unclear. This was because psychiatric assessment is needed before the need for these interventions can be addressed. Two residents had an unmet need for supervision or protection (security), one because of fire-setting behaviour and the other because of repeated expression of suicidal thoughts.

Of the sample, 20% had an unmet need for alcohol or drug services (19% and 1% respectively). In addition, 11% of the sample had unclear need because they refused to consider such services despite staff being clear about the individuals' need for these services. Staff needed support and advice from mental health services to enable them to cope with the problems of 27% of the residents.

Needs of hostel residents with psychotic problems

Because strict diagnostic measures were avoided in this study, participants with psychosis were identified using the social behaviour items that are associated with psychotic illnesses. These items are incoherence of speech, odd or inappropriate conversation, laughing and talking to self, and acting out bizarre ideas. Any participant who had one or more of these problems was selected. Among the 13 participants identified as having such problems, the highest total needs were for psychiatric assessment (in 85%) and for residential care (in 85%) (Table 3). Nearly half (46%) of the residents needed day care and 23% needed social activities. The more prevalent unmet needs of these residents were for psychiatric assessment (69%) (Table 3), social assessment (46%), day care (23%) and social activity (15%).

Table 3 Needs for mental health services of the hostel residents with psychotic problems (n = 13)

	Total needs n (%)	Unmet needs n (%)
Security	1 (8)	1 (8)
Psychiatric assessment	11 (85)	9 (69)
Training in domestic skills	1 (8)	1 (8)
Self-care	2 (15)	2 (15)
Behavioural therapy	0 (0)	0 (0)
Social activity	3 (23)	2 (15)
Counselling	1 (8)	1 (8)
Residential care	11 (85)	4 (31)
Day care	6 (46)	3 (23)
Social assessment	6 (46)	6 (46)

These results show that none of the needs of the residents with psychotic problems for social assessment were met. In only 15% out of 85% of the sample were the needs for psychiatric assessment met. In contrast, the majority of the needs for residential care were met (54% out of 85%) and half of the needs for day care (23% out of 46%). Both the residential and day-care needs were met completely or partly by the hostels' activities.

Discussion

Needs assessment is an importance step in planning and providing services for the homeless. Although an old, simple needs assessment procedure was used in this study, it proved to be feasible, practical and reliable for homeless people with no current contact with psychiatric services. The simplicity of the procedure and its use of direct and important contributing measures to needs assessment made it very suitable for our community surveys. The use of residents' perceptions of needs added an important dimension that might increase the utilisation of services. Although the study was completed in the 1990s, the authors felt that this study is still relevant and useful to publish because of the renewed discussion on homelessness highlighted above.

The majority of the needs of the sample for mental health services were unmet. Highest unmet needs were recorded for social assessment, psychiatric assessment and assessment of drinking problems. This finding reflects the multidimensional nature of homeless people's mental health problems and supports the demand for a multidisciplinary service that works with the homeless and incorporates mental, social and alcohol misuse services.

Linhorst suggested that hostels with single-room occupancy allow residents personal freedom and privacy, and offer a sense of community that can meet a variety of needs. ^{Reference Griffith16} In our study, the item that showed relatively high met needs was residential care in which the greater proportion of needs was met by hostels (which are considered primarily as temporary accommodation). Also, hostels helped to meet many of the needs for day care by providing day-care facilities. The continuous decline in hostel bed spaces without alternative provision will leave more people roofless and more needs unmet. The denominator population of this study is older than studies that were done in other areas. The age distribution of the sample might explain the ethnic composition of the sample (cohort effect). The catchment area, which traditionally attracts elderly Irish homeless people, could also explain these differences.

The residents with psychotic problems showed high levels of needs and of unmet needs. The majority of needs were for psychiatric assessment and were unmet, whereas many of the residential and day-care needs were met by the hostels. Previous research in the Bloomsbury area suggested that many homeless people are reluctant to use traditional psychiatric services and prefer the services of a visiting community psychiatric nurse. ^{Reference Wykes, Sturt and Creer17} This had also been shown by the high utilisation of the community psychiatric nursing service in Bloomsbury by homeless people. ^{Reference Linhorst18} The levels and pattern of sample needs together with their ‘wants’ indicate that these needs could be met within a community-based multidisciplinary service. This service uses care management in linking homeless people to services and maintaining their contact with these services following placement in housing. ^{Reference Satchell19}

Homeless men in this sample showed high levels of unmet needs for psychiatric, social and alcohol services which should be incorporated in such services. Nevertheless, these needs were significantly different from those of people in old long-stay institutions. ^{Reference Kendell9} This clearly suggests the need for specialist multidisciplinary community mental health teams that work specifically to meet the mental health needs of homeless people. As a result of the findings of this and other studies on homeless people in inner London, there are now community-based multidisciplinary services that work with homeless people in this area. Examples of such services are the Focus Homeless Outreach Team (London N1) and the Westminster Joint Homelessness Team (London W1). ^{Reference Abdul-Hamid and McCarthy20} This study provides an historical baseline on which to judge the success of these services.