Originally published in Journal of Mental Health (1992) 1, 265-275

 TOM CRAIG & PHILIP W. TIMMS

 United Medical & Dental Schools, St Thomas's Hospital, Lambeth Palace Road, London SE1 7EH

In the last decade there has been growing concern about the numbers of mentally ill homeless people on the streets in Britain. It is widely believed that this is a direct consequence of the closure of hospital asylums. In this report we examine the existing evidence for such a link drawing on material collected in recent British studies. Several tentative conclusions can be drawn from the available information: (1) rates of severe mental illness amongst the long-term homeless are considerably in excess of what might be expected given general population rates of these disorders; (2) schizophrenia accounts for the majority of these illnesses; (3) levels of disability (both social and symptomatic) are broadly similar to those found in long-stay populations; (4) the majority have never experienced long periods of hospitalisation and are not those people who have been discharged as part of the planned closure of a large mental hospital; (5) recent studies demonstrating an apparent failure to deliver effective ongoing care to this population precisely echo studies conducted four decades ago. It is likely that the present crisis of visibility is the consequence of long-standing failures to provide assertive community care coupled with the less widely publicised reduction in direct access hostels which since the mid-1950s have served as unacknowledged asylums for large numbers of mentally ill people.

Over the last decade there has been mounting concern in the national press, on television and in professional journals regarding the numbers of psychologically disturbed homeless people on the streets of London and other major cities. It is widely believed that this is a direct result of the run down and closure of psychiatric hospital accommodation, particularly that which formerly dealt with the long-stay population. There have consequently been calls to halt this process before further damage is done (Weller, 1989). In this paper, we will review evidence from British studies for such an association, drawing on existing literature and previously unpublished survey and clinical service data.

1954 was the high-water mark of psychiatric bed-occupancy in this country, with 148,100 occupied beds (Tooth & Brooke, 1961). Since then the number of long-stay patients in psychiatric hospitals in England and Wales has declined dramatically. For the most part, this has not been a 'top down' process but has followed a gradual shift in the way professionals have chosen to deliver services in the light of advances in methods of rehabilitation and the introduction of effective medication. In 1962, this gradual process was given official policy support in the publication of Hospital Plan for England (Department of Health and Social Security, 1962). Enoch Powell, then Minister for Health, suggested in his famous "water towers" speech that halving the number of mental hospital beds would be a desirable and achievable target (Powell, 1962) and by the mid-1970s other policy documents, notably 'Better Services for the Mentally Ill', (D.H.S.S., 1975) were suggesting targets of 47,900 hospital beds with an emphasis on final closure of those hospitals which were not well placed to provide services for their local district and which were already near the end of their useful life.

The total number of psychiatric in-patient places remaining in 1986 was 97,064 (Department of Health, 1990) and although there are no official statistics for the last 4 years, a recent reply to a Parliamentary question (Dorrell, 1991) suggested that the total number of mental illness beds for 1989/90 was 59,290, of which 37,350 were occupied by patients staying longer than 1 year or in secure facilities. Thirty mental hospitals closed between 1980 and 1989 and a further 38, amounting to 12,500 beds, are scheduled for closure before 1995 (National Schizophrenia Fellowship 1989). So, the major part of the shrinkage of the mental hospital system had already occurred by the mid-1980s and it seems likely that the target for in-patient places set in 1975 will shortly be met.

Some commentators (eg Weller, 1989) take the line that each of the 67,000 or so bed-spaces that have disappeared represents a long-stay patient discharged to care in the community, needing the same degree of intensive input that they might have had in the institution. This seems unlikely as many of those "discharges" would have been due to the deaths of elderly long-stay patients and some reduction in the numbers of acute beds. Furthermore, long-stay patients discharged from the mental hospitals as part of a planned closure programme have typically been accommodated in supported housing schemes, with residential care staff and the additional services of specialised continuing care or rehabilitation community teams. A very small number of these patients subsequently become homeless. In the longitudinal study of the closure and reprovision of Friern Barnet Hospital, only 2% of the 278 patients were thought to have been lost to follow-up because of becoming homeless (TAPS, 1990; Leff 1991) and in a contemporaneous study of the closure of Cane Hill Hospital, none of the 103 patients followed-up one year after discharge to community-based facilities had become homeless (Pickard et al., 1991). Finally, preliminary data from our follow-up study of the fate of long-stay patients discharged from Tooting Bec Hospital indicates that only 1 of 150 patients discharged over the last 4 years adopted anything approaching a vagrant life-style. This man continued to occupy a room in supported accommodation but disappeared for a few days at a time during which he would sport a placard declaring himself 'homeless and hungry' so managing to supplement his meagre DSS income quite successfully.

A major difficulty facing any reviewer is the lack of any general consensus as to what constitutes homelessness. The legal definition as set down in the 1985 Housing Act includes people who have no right to occupy any accommodation and those at risk of losing their home within 28 days. In practice, the term has been applied to individuals in a broad spectrum of unsatisfactory housing conditions ranging from cardboard boxes and park benches through night-shelters and direct access hostels, to bed and breakfast accommodation or even sleeping on friends' floors, (Watson & Austerberry, 1986). It is also widely accepted that each step on the accommodation ladder encompasses several different sub-populations with varying characteristics and needs. For instance, the needs of homeless families placed in bed and breakfast accommodation are very different from those of the single men and women who also use these institutions (Wall, 1991) and while plausible distinctions are sometimes made between 'rough sleepers' and other groups it has to be acknowledged that the people who sleep rough on the street are neither a static population nor a homogenous one. People move from the streets to hostels or other temporary accommodation and back again. In short, no single classification system has proven satisfactory though it continues to be common practice to distinguish rough sleepers from people in hostels and people in board and lodgings such as private hostels, lodging houses and bed and breakfast hotels. There is also wide consensus that the core notion of homelessness includes not only the simple fact of houselessness but also a lack of resources and community ties resulting in both isolation and disaffiliation with society (Milburn & Watts, 1986).

Accurate counts of the numbers of people sleeping rough are particularly difficult to obtain because of the wide variety of sites involved and because many move between sites and between the streets and a variety of temporary accommodation. A street survey of 17 London boroughs in 1989 found 751 people sleeping out on the streets or in railway stations (Canter et al., 1989) and the 1991 Census enumerated 2,703 people sleeping rough in England and Wales, of whom 1, 275 were in London (OPCS, 1991). However, these figures are based on surveys of sites known to be popular with rough sleepers and it is likely that significant numbers of people sleeping in abandoned buildings or less accessible sites were missed by both surveys. On the basis of these and other surveys, one of the national charitable organisations for the homeless in Britain has suggested that as many as 6,000 people may sleep rough at some time during the course of a year (Brace et al., 1991).

The official figures on hostel occupancy are somewhat more reliable than those of rough sleepers, though even these are subject to errors of classification and of missing data. In answer to a parliamentary question in February 1991, the DoE reported a total of 22,383 hostel bed-spaces for single people in London and a further 37,759 in the rest of England (Hansard, 1991) while a recent postal study carried out on behalf of two charitable organisations suggests rather fewer bed-spaces outside of London, reflecting difficulties in ascertaining a complete census of voluntary and private sector hostel provision (Randall, 1992). These hostels run at very high rates of occupancy - seldom less than 90% - and frequently have to turn people away. Finally, data provided by the DSS and reported by Randall (1992) indicates that in 1990 there were some 76,000 single claimants in board and lodgings in Britain, of whom some 11,694 were resident in London. Nationally, the numbers of households placed in temporary accommodation by local authorities virtually doubled from 23,000 in 1986 to 40,000 in 1989 (HMSO: Social Trends, 1991).

Although the estimates of absolute numbers of homeless people vary according to whether they are taken solely from the official statistics or from the generalisations of surveys carried out by the charitable organisations which provide front-line services, there is broad agreement that the number of homeless people has grown during the past decade and that this is mainly accounted for by younger age groups. Unlike the residents of the more traditional hostels and the clientele of the reception centres and soup kitchens as reported in the mental health surveys of the 1960s and 1970s, these younger populations include significant numbers of women and people from ethnic minorities. Some 1700 people aged between 16 and 25 arrive in London each year with nowhere to stay, no gainful occupation and no close family to turn to for support and advice. One in four are women, three-quarters originate outside of London and the Home Counties and one in ten are from ethnic minority groups. In 1988 there were approximately 5,700 people aged 16-19 staying in temporary accommodation, the majority in board and lodgings (1800) or squatting (2000) and the remainder sleeping rough or in hostels and short-life properties (Randall, 1988; 1989).

The research into mental illness amongst homeless people in the UK has concentrated on the problems of residents of hostels (Crossley & Denmark, 1969; Lodge Patch, 1970;1971; Timms & Fry, 1989; Marshall, 1989; Marshall & Reed, 1992), reception centres (Tidmarsh & Wood, 1972), lodging houses (Laidlaw, 1956; Priest, 1976) and soup kitchens or night shelters (Edwards et al., 1966; 1968; Weller et al., 1987). Reflecting the setting of the surveys, the majority of the subjects are male, in their late 40s, with high rates of unemployment and few family or friendship ties. All the studies have found high rates of chronic psychiatric morbidity of a severity similar to that seen in hospitalised patients. Schizophrenia is the most common single diagnostic group followed closely by alcohol problems.

The majority of these studies were snap-shot surveys with little information regarding past history or prior contact with mental health services. Crossley & Denmark (1969) reported that two thirds of the 55 men reported a past history of admission to a mental hospital; Tidmarsh & Wood (1972) in their survey of the Camberwell Reception Centre found that 29% of 359 men had been admitted to a mental hospital at some time in their life (9% within the previous 12 months); and Lodge Patch (1970) noted that 14 of 18 schizophrenic men had previously been hospitalised though only 3 continued to receive any psychiatric care. None of these studies reports figures on length or numbers of previous hospitalisations and, although it seems likely that their psychiatric disorder antedated their homelessness, it is impossible to know whether the latter was related to the lack of long-stay hospital alternatives to community placement and subsequent hostel residency.

The majority of studies reviewed so far have dealt exclusively with men. Homelessness is less common in women and it appears that homeless women are on the whole younger and have rather more stable social ties (Drake et al., 1982). In one of the very few studies of female hostel populations, Marshall & Reed (1992) interviewed 70 residents of two direct access hostels which cater specifically for homeless women. The duration of stay in the hostel varied from one week to 50 years with a median stay of 2 years. Twenty-seven of the women had moved to their present accommodation from other hostels and 8 had come directly from psychiatric hospitals. Forty-five women met DSM-III criteria for schizophrenia and a further 18 suffered from other classifiable psychiatric disorder. Forty-five of the women reported at least one previous psychiatric admission (median 2; range from 1 to 30), with 5 subjects having been in hospital during the previous 12 months. At least 15 had been detained formally under mental health legislation in the past. As with the studies referred to earlier, it was not possible to determine how much the changes in local long-stay psychiatric provision contributed to their problems, though the authors report that hostel staff found increasing difficulty in arranging psychiatric admissions and women who were admitted to hospital were often precipitously discharged back to the hostel with inadequate after-care arrangements.

In parallel with these surveys of homeless populations, studies of mental hospital admissions since the mid-1950s have reported that as many as 10% are of people with 'no fixed abode' (Whiteley, 1955; Berry & Orwin, 1966). This was recently confirmed by an audit of admissions to a central London Teaching Hospital (Neville & Masters, 1990). Studies such as this based on the analysis of aggregated patient admission statistics (PAS) have also suggested that those registered as homeless make higher demands on hospital in-patient services than the domiciled population (Glover, 1989) and this observation is borne out by more detailed studies of such admissions (Herzberg, 1987). Unfortunately, the registration of a patient as having no fixed abode is largely confined to those who are literally roofless and so ignores many others living in any other parts of the spectrum of unsatisfactory accommodation. Figures generated from patient administration systems and from more detailed surveys are, therefore, likely to under-estimate the size of the problem and to focus on a restricted sub-group of the whole. With this caveat in mind, it appears that common characteristics of this population include prolonged unemployment, lack of close social ties and a criminal record. As many as half of these patients suffer from schizophrenia, often with lengthy histories of illness but with poor compliance with treatment, higher rates of discharge against medical advice and consequently a history of repeated brief hospitalisations and discharge to unsatisfactory living circumstances (Whiteley, 1955; Betty & Orwin, 1966; Herzberg, 1987).

While the early literature contains only very sketchy details concerning the pattern of psychiatric service use amongst homeless people the question can be partially addressed from previously unpublished data collected by three recent studies carried out by Timms and his colleagues.

In the first of these, Timms and Fry (1989) surveyed a representative sample of men newly arriving at a Salvation Army hostel and compared these with a sample of men who had been resident in the hostel for longer than a year. In all, 124 men were interviewed. In keeping with previous studies of hostel populations, over half were suffering from mental disorder. Schizophrenia was the most common single diagnosis, present in a quarter of the new arrivals and 37% of the hostel residents. Although all the schizophrenic patients suffered from chronic disorders in the sense that their illness had begun more than 5 years previously, more than half of the patients had spent less than 2 years in hospital in their life-time and the mean length of stay per admission (aggregated time in hospital/number of admissions) was 11.33 months. Only 7 men had been continuously hospitalised for greater than 1 year in their life-time (5 with hospitalisations of 5 or more years). One of these men had been hospitalised for 12 years in Germany and another was currently an in-patient of a local hospital where he had resided for the past 6 years. He had been reported missing shortly before he turned up at the hostel.

In the second study, data on prior psychiatric treatment were collected in the context of an experimental case management service for homeless mentally ill people. In the course of a year, The Psychiatric Team for Homeless People (Brent-Smith & Dean, 1990) received 94 referrals of the most disabled or disturbed clients resident in hostels or using a variety of day services in Lewisham & North Southwark. All the patients suffered from chronic illnesses. Two-thirds had a primary research diagnosis of schizophrenia and four-fifths were severely socially disabled. Despite the severity of their illnesses, 15 had never been admitted to hospital and of the remainder the most common pattern of hospitalisation reported was one of multiple brief admissions - with no subsequent improvement in accommodation. During the entire course of their illness, these patients had an average of 6.4 admissions to hospital (median 3; range 0 - 57) with a mean length of stay of 6.7 months. Only 2 patients reported previous hospitalisation of a year or longer. Both had spent more than 20 years in hospital.

Finally, current data from the first year of the London Homeless Mentally Ill Initiative confirm the low proportion of ex-long stay patients in the hostel or street homeless population. In the first year of this study, three clinical teams providing services to street homeless, hostel residents and users of day centres and soup-kitchens have assessed a total of 800 homeless people. Of these 544 have been found to be suffering from severe mental illness (over 60% with a diagnosis of schizophrenia). In the "ill" group, only 3 have been found whose longest continuous hospitalisation was greater than 5 years. The great majority of cases with histories of previous hospitalisation report multiple brief admissions. In keeping with earlier survey work and the findings of other homeless services, there is a depressing consistency in the inadequacy of follow-up and after care provided by main stream services.

Several tentative conclusions can be drawn from the available information. First, the rates of severe mental illness amongst the long-term homeless are considerably in excess of what might be expected given general population rates of these disorders. Second, schizophrenia accounts for the majority of these illness. Third, the levels of disability (both social and symptomatic) are broadly similar to those seen in patients who have been hospitalised in long-term settings. Fourth, and perhaps most significantly, recent studies echo those of four decades ago - institutions catering for the long-term homeless population attract sizeable numbers of people with chronic severe mental illness, who once in these settings are largely ignored by mainstream psychiatric services.

It also appears that the mentally ill residents of these hostels are not those who have been discharged as part of the planned closure of the large mental hospital. Most have not experienced long periods of continuous hospitalisation but appear instead to have found a niche within a very similar institutional environment. Like the old hospital asylum, a typical hostel provides a low-key, non-interfering environment in which it is possible to stay for many months or even years with only a minimum requirement of social interaction with other residents or staff. Bizarre behaviour is usually ignored and residents are not subjected to medical or social work demands to make changes in their lives, take medication or participate in rehabilitation programmes (Timms & Fry, 1989). The mentally ill residents of such hostels have been 'managed' with multiple hospitalisations over a number of years interspersed with periods where psychiatric care has been virtually non-existent. This failure to provide continuing active treatment has sometimes been attributed to the homeless person's reluctance to accept help (Whiteley, 1955) and always to the lack of sufficiently assertive community treatment from a system which continues to be largely hospital orientated. The need for effective outreach services is particularly important for homeless people suffering from schizophrenia who are much less likely to maintain contact with psychiatric services than are people suffering from other mental health problems (Priest, 1976). In short, the fundamental difficulties in providing continuing care to this population are not new and should not lead to simplistic calls for halts to the 'hospital closure programme'.

But while it seems clear that the hospital closure programme per se has not contributed large numbers of homeless mentally ill people, there can be little doubt that deinstitutionalisation in a wider sense plays an important part in creating and maintaining the problem (Bachrach, 1992; Lamb, 1984). The deficiencies in the implementation of community care policy for people with severe mental illness have been well documented and widely discussed (eg Marks, 1992; Thornicroft & Bebbington, 1989; Sayce et al., 1991; Wing, 1990). Some of these difficulties are particularly pertinent for homeless populations.

In some districts there is a shortage of suitable environments in which active treatment and rehabilitation can be carried on over periods of a year or more. With the move towards District General Hospital (DGH) sites there was a corresponding shift in emphasis towards providing intensive but brief treatment for acute psychosis. Surveys around London suggest that these units operate at a very high level of occupancy, sometimes exceeding 100% (Holander et al., 1990) and in such a climate it is not that unusual for patients to be prematurely discharged in order to admit another whose problems appear more pressing or dangerous (eg. Patrick et al., 1989). While some patients do remain in hospital as efforts are made to sort out a suitable community placement, health care staff and their managers often resent what they perceive as a 'bed blockage' due to the lack of a social care provision which is not in their immediate control.

There is a need for assertive outreach to engage groups of patients who are reluctant to accept services and patients who are unlikely to 'recover' with brief interventions. This reluctance has many sources and includes factors as diverse as the expansion in community services for patients with acute neurotic illnesses and transient stress reactions (Sayce et al, 1991); the ethical dilemmas of 'forcing' treatment on people who appear reluctant to accept it; and the sheer complexity of arranging and delivering effective care when the organisations providing such care are no longer contained within a single asylum but are scattered around a community, often operating in isolation and complete ignorance of each other (Clifford & Craig, 1989).

One of the core principles behind the creation of the DGH and district community services is the notion that professional teams should be responsible for providing care within defined geographical boundaries. The original purpose was to ensure that no needy individual could be overlooked. But while this policy ensures that assessment and initial treatment services work in the interests of the local population, it also serves to disrupt continuity of care for people who move between catchment areas and it is particularly problematic for the homeless person who moves between the streets and a variety of temporary accommodation. Disputes can arise between teams as to who 'owns' the case and even when there is good will, inadequate hand-over of care can result in patients simply being lost from care altogether.

In addition to these organisational barriers, the prejudices of some health professionals may make services unacceptable to homeless mentally ill people (Stearn & Stilwell, 1989); and the emphasis on treatment and active involvement is often at odds with the homeless individual's inability to look beyond the next meal or bed (Bachrach, 1987).

The last 15 years has seen a dramatic reduction in the number of beds in DHSS reception centres/resettlement units, Salvation Army Hostels, Rowton Houses and nightshelters. Nationally, the DHSS has had a policy of closing resettlement units for some time and the numbers of such units fell steadily from 215 in 1948 to only 21 in 1970 (Hewettson, 1975). In 1985 it was decided that all remaining resettlement units should close and be replaced by smaller and more appropriate accommodation provided by local authorities and the voluntary sector. There was to be no overall reduction in funding, but potential service-providers were slow in coming forward. Five years later only 8 such schemes had been approved and none had sufficient bed-spaces to replace the units that were already scheduled for closure (HMSO, 1991).

In London in 1981 there were 9,751 bed-spaces in direct access hostels, 6000 of which were in large, traditional hostels for the homeless. By 1985 the numbers of direct-access bed-spaces had declined to 4885 (SHIL, 1986) and by 1990 to around 2000 (Harrison et al., 1992). A report by the London Borough Association described them as "at once a resource and a problem" (LBA, 1981) - a resource because of the shelter provided, a problem because of the often appalling physical conditions and catastrophically inadequate staffing. The report recommended the opening of 600 beds yearly to replace the old hostels, but noted that such schemes were facing financial and planning difficulties. Belated enforcement of fire regulations led to the contraction of some hostels and the precipitate closure of others, most recently the 280-bedded Salvation Army hostel in Blackfriars Road in 1991. The closure of the Camberwell Reception Centre alone resulted in the loss of 900 bed-spaces of which only 62 were reprovided in new direct access accommodation, the rest being scheduled for reprovision in specialist referral-only units. Furthermore, the replacement schemes may well be seriously under-funded. Despite the well documented similarity between patients in long-stay wards and mentally ill residents of these hostels, the financial allowances available for closures of these beds are vastly inferior, sometimes 5 times less than those provided for the closure of long stay hospitals (SHIL, 1988).

In all, even including the new hostels and cold-weather shelters constructed under the current government initiative there has been a 75% reduction in direct-access hostel spaces over the last decade. Surprisingly, there are no published studies of the fate of the mentally ill ex-residents of the units which have closed. In South London, some were certainly placed in specialised accommodation with support from the local authority and some were transferred to other hostels. As with hospital closures, replacement schemes have tended to resettle the more able clients, leaving behind an increasingly disabled and disadvantaged group, some of whom are now to be found amongst the rough sleepers on our streets. 'Thamesreach', a voluntary organisation with a long history of direct work with the street homeless, reports an increase in the numbers of ex-hostel residents amongst its clients and this observation is echoed by similar teams around London (Hall - personal communication). Not only are these people poorly served by the mental health services but it seems increasingly likely that they have also been bypassed by the best intentions of the hostel reprovision process as well.

In conclusion, during the last two decades the process of mental hospital closure has been paralleled by a similar attrition of direct access hostels which since the mid-1950s have served as unacknowledged asylums for large numbers of mentally ill people. It is hardly surprising that with the disappearance of these facilities, the homeless mentally ill have been increasingly visible on our streets. This recent crisis of visibility has unmasked a fundamental failure to provide effective psychiatric services for a substantial minority of people with long-term mental illness. The psychiatry of homelessness should not be seen as a micro-speciality in its own right, but as an integral part of the spectrum of community care. The specialist health care teams that have been set up as a consequence of this crisis are a first step towards developing co-ordinated community care for people whose disorders simply will not go away.