Roofless in a steel city: learning from the homeless mentally ill

A small, city-wide clinical outreach service for the homeless mentally ill in Sheffield, UK, attained its present configuration 6 years ago. This paper discusses the lessons learnt in the course of the service's existence. The homeless mentally ill identified by the service have disengaged from the ‘mainstream’ services and society. Most are from disturbed homes, nearly all have had prior contact with psychiatric services and as many as half have served prison terms. As service users, they must be actively sought out and engaged, which places specific demands upon a mental health team: flexibility of approach, patience and a willingness not to judge others' values. Though largely anecdotal, the inferences drawn in Sheffield may have parallels elsewhere, not least since individual lives can turn upon pivotal (‘anecdotal’) encounters and those evinced by the homeless tell us much about society, psychiatry and the values of contemporary healthcare providers. Also, most of the time, the proposed model has been successful.

If taken at face value, it seems obvious that publicly funded healthcare providers should undertake to deliver care to those who are the most impoverished. Few are more impoverished than the homeless mentally ill. However, some features of this social group and the care providers they encounter can impede satisfactory service delivery.
Sheffield is a post-industrial city in the north of England with a population in excess of half a million. Traditionally reliant upon the steel industry, and having suffered the depredations of its collapse and that of the coal industry in the 1980s, the city has recently diversified into service provision and advanced technologies. Demographically very similar to the English average, Sheffield's population is predominantly White, with a large Asian minority and many other ethnic groups represented. The city's two universities employ and educate a large proportion of the total population.

Homeless Assessment and Support Team
Each year, the Sheffield City Council categorises over 3000 people as 'homeless' (i.e. approximately 0.5% of the population). In most cases, such people reside in temporary accommodation and only a small number (less than 100) are literally 'roofless' (Sheffield City Council). The Homeless Assessment and Support Team seeks to engage with, support and treat those who are homeless and mentally ill, with a view to achieving two tangible outcomes: 1. acquisition of permanent accommodation 2. access to'mainstream' mental health services.
Growing out of a general practice project piloted in the early 1990s, the service for the single homeless individual is now jointly funded by the local authority and a National Health Service (NHS) mental health trust. It comprises three full-time equivalent keyworkers (one nurse and three social workers, two of whom are halftime), a part-time secretary, as well as input from a larger team's service manager and a consultant psychiatrist (a university-funded academic) who spends two sessions per week with the team. This 'singles' team works across Sheffield and forms the focus of this report. There is a 'families' team too, comprising health visitors and nurses, but it is not described here.

Finding the homeless
Referrals to the Homeless Assessment and Support Team come from a wide variety of sources: local authority housing officers; workers at hostels, housing associations, 'interim accommodations', bed and breakfast hotels, 'drop-in' centres; general practitioners, health visitors, keyworkers from the local substance-misuse service, probation officers and occasionally the relatives of those who are 'missing'. One member of the team has a specific remit to engage with 'rough sleepers' and actively seeks out those who are roofless (there are key locales where such people are likely to be found). There are approximately 300 referrals per annum (about 10% of the homeless population) and at any time the team will be in touch with approximately 100 individuals, of whom 30-40 are also under the care of the team's psychiatrist.

Spence Homeless mentally ill
We have audited our clinical diagnoses annually for the past 6 years and the figures have been remarkably stable: approximately half of the homeless have depression, 20% have psychoses (including schizophrenia), 30-50% have concurrent alcohol problems, a similar number have dual diagnoses (affective or psychotic disorder plus substance misuse) and 20% primarily have personality disorders. There are also usually approximately 10% with learning disabilities and a small number of other organic syndromes (Huntington's disease, dementias, Korsakoff's syndrome and Asperger syndrome). The majority are male (more than 80%), in their thirties; 20% will have been raised in 'local authority care', 10% attended 'special schools' ; most (70-90%) have used psychiatric services previously, with 40-70% having been admitted to hospital and 20-30% formally detained (under various sections of the 1983 Mental Health Act); 25-50% have history of violence; similar percentages have self-harmed and served prison sentences. Between 5 and 10% describe childhood sexual abuse but many more describe violence and unhappiness experienced as children. Most service users spend between 6 weeks and 6 months in contact with the team. The team's failed appointment rate has been 20% annually (when formally audited, over the past 3 years) and we eventually lose 25% from follow-up (often because users moved out of the city). Neither of these figures is particularly high, relative to the literature (Mitchell & Selmes, 2007).
Naturally, such a sample is likely to be highly selected. Those referred are not only homeless, they are also (in the main) referred by professionals who do not specialise in psychiatry, hence they might well comprise those most likely to attract attention, for example through disturbed behaviour, confused speech, depressed affect, demonstrable self-harm or gross neglect (Bittner, 1967). It is quite possible that those who are 'quietly ill' are systematically under-referred. However, sometimes a minor aberration serves to reveal someone who has fallen through the net: a person with Asperger syndrome who had always lived with their parents was arrested for stealing a tin of baked beans following their deaths (there was no one left to look after them); a person with dementia arrived at a railway station, carrying an empty suitcase, with only a crumpled letter from a mental health trust to identify them; a young woman with psychotic depression was found sleeping in a graveyard (she was listening to the voice of her 'dead brother', who was later traced and found to be alive); another young woman drew attention to herself in a hostel because the taps in her room were running all night -she had obsessive-compulsive disorder and was washing her hands continually; a young man with schizophrenia, who had lived almost his entire adult life on the street, known to police through his well-meaning acts of generosity (he was not an offender), only attracted referral when he developed choreiform movements, etc. Many of these occurrences are singular, yet they have one thing in common -in order for these people to be seen and helped, there has to be a service that will go out to find them; they will not be encountered in conventional out-patient clinics (Timms, 1996;Appleby, 2000;Mitchell & Selmes, 2007). Over the past 6 years we have interviewed people in hostels, bed and breakfast hotels, women's refuges, squats, drop-in centres, probation offices, police stations, general practices, public houses, cathedrals, snooker halls, graveyards, parks, alleyways, stairwells, railway sidings and wrecked cars. A homeless service needs to be adaptable.
What other lessons can we learn from the homeless?
'Man is wolf to man' { Although health and social care workers in many of Britain's cities may have grown used to meeting individuals with post-traumatic stress disorder consequent upon torture in their mother countries (Summerfield, 2002;Tribe, 2002), it may be surprising to learn that torture is also a domestic phenomenon. We have seen people held and abused for days by drug dealers and other criminals, but also by other homeless individuals. Often neither the perpetrator nor the victim volunteers a reason for abuse, although sometimes it may be sexual. This places particular demands upon a small team where the same professionals are likely to see the perpetrator for one form of assessment (the post hoc identification of forensic risk) and the victim for another (newly identified as vulnerable). It is here that we have benefited immensely from the multi-agency, public protection and planning framework, where access to police information may be most enlightening (especially where there have been previous offences that were not pursued through the courts).
Torture is relatively uncommon. More frequent are the casual acts of violence meted out to the homeless on the streets (e.g. beating up by groups of passers-by) or the punishments dispensed in certain subcultures (e.g. for infidelity, for refusing to take part in a criminal act). Indeed, the language of the streets is ripe with forensic codes and distinctions: people who emerge from prison commonly speak of having not friends but 'associates' ; offenders may draw distinctions between 'commercial' crime (robbing a warehouse) and 'domestic' crime (robbing a house), the latter often perceived as less honourable than the former.

1-Hour assessment
Given the peripatetic nature of our service users (many will have already migrated through several cities before the homeless assessment and support team meets them), it is crucial to obtain as much information as possible at the first assessment, while also recognising that some material may be too sensitive to be divulged early on. We have opted to try to include cognitive, physical and reading assessments whenever possible (Box 1), as these may inform diagnosis, prognosis and intervention: those with profound executive dysfunctions may have little prospect of modulating their behaviours unaided (Spence et al, 2004); those with learning disabilities may gain access to more supportive accommodations; some hostels may provide literacy services for those who are functionally illiterate, etc.
Additionally, a willingness to examine the homeless person points to the therapeutic aim of the encounter and the person's status -they are not perceived as untouchable. Even measuring blood pressure involves a moment of silence in which the homeless person is treated as any other medical patient. They are taken seriously.
'What's the worst thing you've ever done to someone else?' On the face of things, this is a terrible question to ask anyone, yet it emerged from a need to try to estimate forensic risk in some of our more sub-optimal interview milieu where there was always the risk of a failure of follow-up. When sparingly used and sympathetically voiced, the question requires users to cast their minds back, to think morally and to gauge what they can tell their interviewer. If they do not trust the interviewer, they will not answer the question and the interviewer rarely possesses a priori knowledge of the event described. Nevertheless, certain patterns can be identified: among homeless men in Sheffield this question frequently elicits the description of a fight; a professional criminal may begin with 'Are you writing this down?' ; for women, the worst is often a child they left behind ('It was for the best'). The question rarely elicits a response that is not suffused with some form of sadness, either at one's own failures or those of life itself. An exprofessional criminal became tearful when he admitted that the worst of all his crimes had been a domestic robbery, because when he brandished a gun he felt for an instant the terror of his female victim ('I saw it in her eyes'). This incident caused him to leave his gang.
Additionally, experience alerts one to answers that are obviously statistical outliers: over 6 years, the only trivial example offered was from a man who spoke of kicking sand in another child's face at the age of 10 -it later transpired that the man had probably killed. Hence, a trivial exemplar might well indicate obfuscation.

Bad Samaritans
There is another surprise awaiting those who attempt to treat the homeless: an apparent resistance on the part of some colleagues (Timms, 1996). We have encountered a marked therapeutic nihilism among ward staff.
When we admitted a woman with hebephrenic schizophrenia who had been found thought disordered in the street, we were approached by a nurse who said that he had 'walked past this woman every day for 3 years' so why was she being treated now? Admittedly, there is ample scope for confusion here -if a person has had psychosis for years, then when is a good time to intervene? In eight roofless individuals with psychosis admitted over the first 12 months of our service, with reported duration of untreated psychosis of 1-13 years (mean=5), it was found that despite the early discharge of two users (one because of alcohol misuse, the other through a mental health tribunal), six responded to treatment, none of whom required high-dose medication; all regained permanent accommodation (Girgis & Spence, 2003). It may be the case that not treating the homeless becomes a kind of self-fulfilling prophecy (less charitably construed as prejudice). Certainly, one of the most frustrating experiences of recent years has been attempting, over many months, to locate and engage with a man with psychosis who lived along railway lines, believing he was evading a persecuting demon (he was recurrently assaulted by gangs of youths but did not blame them because he attributed their actions to their being remotely controlled by the demon), then admitting him to hospital one morning only for the nursing staff to send him 'home on leave' the same afternoon. Some might wish to debate the semantics of home leave for people who have no home. But perhaps more salient here is the word 'leave'.
However, there is a serious problem for in-patient staff in the current NHS -the current value system places emphasis upon the duration of admission and the need to process as many admissions as quickly as possible, aiming for care in the community. It seems as if the homeless did not really fit this concept of medicine, where the model patient seems to be a person who is not very ill, who has a home to go to and a family to care for them. Perhaps the nursing staff have a pointcurrent in-patient systems are not designed for the homeless.

Absent fathers
Individuals who present to psychiatric services commonly have disturbed familial relations and experiences. However, the homeless in Sheffield exhibit one very marked feature: nearly all have no father with whom they maintained contact during childhood and adolescence. Recurrent themes are paternal abandonment, imprisonment, death, and divorce. Only 5% of those assessed by the Homeless Assessment and Support Team's psychiatrist grew up with their father. The absence of a father cannot be equated with a child's predestination towards pathology -indeed, where a father is antisocial the risk to the child might even be ameliorated by his departure (Jaffee et al, 2003). However, it seems likely that a father's absence systematically exposes the child to certain experiences: relative poverty, periods in local  Nelson & O'Connell, 1978) . Physical examination (with chaperone, as required; minimum pulse and blood pressure; muscle tone, praxis, cerebellar and eye signs) . Follow-up arrangements and contact details (written down) authority care, conflict with the stepfather, physical and sexual abuse from new 'relatives', undiluted consequences of maternal mental illness. The team have seen men who were beaten and sexually abused by their mothers' boyfriends, who saw their father solely on visiting days in prison, who were abused in residential care and rejected by the new family once they got out, sodomized by older stepsiblings.
From a psychodynamic perspective it would be unsurprising if those subjected to such experiences subsequently had difficulty in trusting male authority figures, or if they did not trust female co-workers (akin to those females who abandoned them as children). It can be particularly hard to establish rapport with these men once they are already outside society (e.g. following periods in local authority care, in prison, or in the criminal and illegal drug using groups). The homeless assessment and support team has tried to remain available but promise little, offering tangible assistance (with accommodation) while not pretending to 'understand' the experiences they have had. Most importantly, the team attempts to withhold judgement. In Winnicott's memorable terminology, we attempt to be 'good-enough' objects (Phillips, 1988).

Cycle of rejection
If one permits oneself a psychodynamic consideration of the conditions of engagement pertaining within 'homeless psychiatry', it seems clear that the service provider is attempting to undo much of what families and societies have already inflicted upon the homeless service user. Parents have often been cruel or unreliable, strangers abusive, authority figures a source of punishment and suspicion -it can be very difficult to make a fresh start. Yet, the traditional medical response to such people can often seem to recapitulate rejection (Timms, 1996): letters discharge people from follow-up when appointments are not kept, their motives are second guessed and symptoms doubted (think of pain or insomnia), their physical condition may make carers reluctant to touch them or even to remain in their presence (e.g. without the window being open). It can be very informative to ask a student to check the pulse of a malodorous person. How should the teacher respond when the student refuses to touch the person?
On one acute psychiatric ward, half of the inpatients who missed lunch each day (across 8 months) because they were still in bed at midday were homeless (constituting nearly all of those homeless on the ward at the time; Thomas & Spence, 2005). Such people withdraw from others and it may require considerable ingenuity and patience to establish a link. It was later established that returns from acute psychiatric wards in the trust usually constituted 30-40% of meals, in contrast to the older adult wards where the nurses took the food to the patients.

Lost in translation
The size of an ethnic community may constrain the probability of obtaining an accurate translation in its language.
In a city where many of the refugees seen by the Homeless Assessment and Support Team originate from the Middle East and the Horn of Africa, obtaining accurate translation poses real difficulty. There are at least three problems.
1. Lack of anonymity (in a small community, patient and translator may know each other). 2. Factionalism (in those emerging from war-torn countries, patient and translator may come from opposing sides). 3. Editing (if the patient is thought disordered, the translator may try to ameliorate that in their translation; if the subject matter is unacceptable for the translator, they may try to suppress it, particularly in cultures where suicidal ideation is considered immoral).
It is difficult to design systems of healthcare where every potential language is catered for, but the reality of such limitations should at least be acknowledged (particularly where the ethnic community is itself divided; Tribe, 2002).

Homeless pharmacology
It might be anticipated that prescribing for homeless people with a mental illness will be constrained by stark realities and risks (Timms, 1996). Clinicians are likely to avoid prescribing the potentially addictive or remunerative substances (e.g. benzodiazepines, methylphenidate or procyclidine) or those that require close monitoring (lithium, clozapine or the more recent anticonvulsants for depression, e.g. lamotrigine). In contrast, relatively safe antidepressants with a long half-life (e.g. fluoxetine) and depot preparations of antipsychotics may be favoured because intermittent non-adherence may be less disruptive.

Homeless psychodynamics
The Homeless Assessment and Support Team does not have access to the psychotherapies. Often, this is more clinically appropriate anyway (although there is some preliminary evidence that cognitive-behavioural therapy may reduce violence and offending behaviours in the homeless; Maguire, 2006). However, in some situations most of what the team encounters can be understood in psychodynamic terms. The concepts of transference, counter-transference, idealisation and splitting are recurrently made manifest through the conduct of teams and individuals dealing with this service user group. The people who are illiterate are often treated with frank disrespect, as are those with suicidal intentions and addictions. Sometimes the team member has to play the advocate, witnessing and reflecting back upon the way the person was treated by other agencies. Once a homeless person has offended someone, be they in the housing department, the out-patient clinic or the general practice reception area, it may be very difficult for them to access care (Timms, 1996). Nevertheless, we must resist the notion that only our team members understand the person.
Spence Homeless mentally ill special articles Playing for keeps Multiple exigencies are likely to be influencing every attempt at follow-up in the homeless person, for example loss of accommodation, intermittent financial imperatives, procurement and use of illegal substances, cognitive impairment, harassment and feuds. Hence, it is especially important for the staff who work with homeless people to use each contact to maximum effect. A typical pattern of engagement discernable among the old notes and records of the homeless (in particular those who have psychosis) is the repeated deferral of action. It is not unusual for a homeless person with a severe mental illness to be assessed acutely (perhaps in a police cell or else in an accident and emergency department), found to be ill but inexplicably discharged -instead of being admitted, they are offered either a so-called 'second chance' (as if having an illness was a lifestyle choice) or an out-patient appointment. When the proffered appointment is not kept, the person is 'discharged' from followup. How realistic is such a sequence of decisions? A person who believed themselves possessed was offered just this form of follow-up and then discharged; another person with acute mania was given a prescription and an appointment for 2 weeks hence -how likely was their attendance? Every opportunity to assess a homeless person should be regarded as potentially the last.

The right stuff
The nature of the work on the homeless assessment team is such that one has to be able to trust one's colleagues. Also, although it is trust policy that outreach visits should not be conducted alone, in reality this is often the case in a small team. We recommend telephone contact, agreed times of return to base and joint working with other agencies (Box 2). Disagreements may arise within the team -these usually concern the threshold of intervention (e.g. a medic may think mostly about risk, while a social worker may place greater emphasis upon a person's autonomy): when has a cognitively failing street-drinker declined sufficiently to warrant a guardianship order? How physically frail must he be for this to be feasible?

Futile referrals
A difficulty often emerges when attempting to terminate involvement with individuals who have finally obtained permanent accommodation. There is a question of how long such a person has to reside at an address before they can be admitted to mainstream mental health services. The consequences of failure can be severe. With homeless people, the routine out-patient appointment and the discharge letter when they do not attend does not seem to constitute an adequate response, presuming, as it seems to, that non-attendance is a sign that all is well. Our experience (and that of others) is that discharging the person only leads to recurrent homelessness and re-introductions to our service (Mitchell & Selmes, 2007).

Happy endings
Much of what we have rehearsed here deals with risk and failure: failure to engage, to maintain a relationship, to access housing or to treat an illness. However, even in this most highly selected and socially alienated city population, there is the prospect of redemption, even if it comprises only a place to live and a secure tenancy. For three-quarters of our service users the function of the service is fulfilled, in that a home and necessary contact with mainstream services is achieved (Girgis & Spence, 2003).
Among spontaneous expression of thanks from parents of the people we have helped, we have also received a rather more ambiguous reward: an elderly man with vascular dementia said upon discharge 'I'll never forget what you've done for me'.

Conclusions
Attempting to deliver a psychiatric service to homeless people requires a different type of practice from that of mainstream services and a lower threshold of suspicion that 'all is not well'. It requires attention to detail in obtaining as much information as possible when contact is made. It also requires an open, non-pejorative approach to people who may have had exceedingly aberrant early lives. The team must constitute 'goodenough' objects: workers who are not perfect but reliable and (hopefully) kind. The service users and their setting place specific constraints upon the pharmacology deployed. The current priorities of healthcare systems may serve to further disenfranchise the homeless, who may be hard to engage. Working with such people is not hopeless but it may be very demanding. However, in most cases it can be successful. . Keep a flexible approach (emphasis on an'outreach' model) . Maximise utility of first assessment (and each point of contact) . Maintain a non-pejorative attitude, be polite and consistent . Keep in contact (even despite non-adherence; if sending letters, make them polite and caring) . Maintain liaison with other teams (particularly housing, substance misuse services, local support agencies) . Judicious use of the multi-agency public protection planning framework . Value the team (keep in contact, have mobile telephone numbers, known appointment venues, agreed times of return to base) . Appropriate use of pharmacology (safe options, long half-life if non-adherence is an issue) . Make explicit handover arrangements with appropriate teams when housing has been secured and tenancy is stable