Disability and Service use among Homeless People Living with Psychotic Disorders

Abstract

Background: The prevalence of psychosis and needs for care among homeless people were studied in inner Melbourne.

Method: This was a two-stage nested study within the Australian National Survey of People Living with Psychotic Illness. A screen for psychosis was administered to a representative sample of men and women living in marginal housing in a mental health service catchment area. A selected subsample of 82 screen-positive respondents was interviewed using the Diagnostic Interview for Psychosis (DIP), a semistructured, standardized interview with three modules: (i) demography, functioning and quality of life; (ii) diagnosis; and (iii) service use.

Results: An unexpectedly high prevalence of people living with psychotic disorders (estimated lifetime prevalence 42%, 95% CI = 37 − 47%) may reflect a concentration of vulnerable people in the shrinking marginal housing supply in the inner city areas. Disability in everyday, occupational and social functioning is greater for this subgroup than for other people living with psychosis in Australia. Most people were single and unemployed, and many reported social isolation and feeling unsafe. Substance use disorders were common. Most people were using health services, including specialist mental health services, but few were receiving rehabilitation, vocational or housing support.

Conclusions: Despite high levels of contact with a well-organized, sectorized mental health service in an affluent country, this pocket of several hundred people had high levels of persisting disability and needs. The literature and local experience suggest that changing this situation is likely to require co-ordinated policy and practice between the health, welfare and housing sectors.

Clinicians, policy-makers and politicians remain uncertain and divided about the best ways to respond to the problems of a subgroup of homeless people living with mental disorders in Europe [1–3], the US [4–8] and Australia [9–11].

In Australia, the prevalence of psychosis in general and schizophrenia in particular, among homeless people has remained high over the last two decades. In 1985, Doutney et al. [12] reported the point-prevalence of schizophrenia among men living in a homeless refuge in Sydney as 15%. In 1987, a Melbourne study found a lifetime prevalence of approximately 20% for psychotic disorders among people living in marginal accommodation [13]. More recently, in Sydney, the 12-month prevalence of schizophrenia in hostel users was 29% [14] and 23% among those using homeless refuges [11].

Even though mentally ill homeless people are diverse in many ways, including diagnosis, treatment histories, disabilities and service needs, and share many characteristics with the majority of homeless people (who are not mentally ill), they have service requirements that may be overlooked unless they are recognized as a special population. They by-and-large require mental health services in addition to social support, housing and economic security, to alter their life circumstances [15]. Homeless mentally ill people with psychotic disorders in developed countries are likely to have had earlier contact with psychiatric services, but past studies in Australia and elsewhere have typically found limited current contact and low rates of psychotropic medication use [16–19].

On the other hand, accessible mental health services for homeless people including outreach services can encourage helpful service contact [19–21], and homeless mentally ill people in Australian settings for instance, will use accessible primary health care [16]. Demonstration programs in the US concluded that homeless adults with severe mental illnesses, often thought to be beyond the reach of existing outreach services, are willing to accept psychiatric treatment and can remain in community-based housing with appropriate support. Housing stability, appropriate psychiatric treatment and increased income, can lead to an improved quality of life [4–8].

The rate of disorder among homeless people is useful information for advocacy and for monitoring policy and practice change in a community. However, it does not in itself give a good indication of the service needs, and contact with psychiatric services does not necessarily imply that health needs are being met. To answer these questions requires the assessment of disabilities, including self-reports [17–19],[22]. This study aimed to estimate the prevalence of psychotic disorders among homeless people in inner Melbourne, a decade after an earlier study [13], as well as assess the disability and service use of these people.

Method

The study was nested within the Australian National Survey of People Living with Psychotic Disorders [23]. The National Survey (a component of the National Survey of Mental Health and Wellbeing) was designed to identify profiles of symptoms, other impairments and disabilities; collect information on services received and needed; and explore quality of life in a broadly representative sample of people with psychotic illness. The National Survey was conducted in four urban or predominantly urban service ‘catchment areas’: in the metropolitan areas of Brisbane, Melbourne and Perth, and in the Australian Capital Territory. The Survey had two phases: a census and screen of people who were predominantly service users, followed by interviews with a sample of those screened. The nested study reported here was based in the Melbourne catchment area, an inner city and suburban site. The Melbourne site conducted as an add-on study, a systematic survey of the residents of marginal accommodation in the catchment area.

Design and definitions

The study was designed to screen a representative sample of people who were using marginal accommodation places in the catchment area. The definition of homelessness and the process of sampling were equivalent to those used in the earlier study [13]. We defined homelessness according to an Australian definition, widely used in this country, which includes people in insecure accommodation with shared bathroom and kitchen and living space, as well as roofless people [24]. Roofless people, a small minority in this community, were not included in the study.

The study design had three components. First, the sampling frame of accommodation in the catchment area was established. Second, a random sample of people was selected from that frame and these people were then approached for screening for psychosis. A subsample of those screened as positive for psychosis were then interviewed. The identification of the sample took place over a six-week period, including the census period for the National Survey, in winter 1997.

As in the core National Study [23], the criteria for inclusion were set so as to identify and interview people aged 18–64 with psychotic disorders that could be classified according to International Classification of Diseases, 10th revision (ICD-10) as affective or nonaffective psychotic disorder (schizophrenia, bipolar disorder and related disorders). Brief psychotic episodes secondary to psychoactive substance intoxication or a medical condition were excluded. Other exclusion criteria were: temporary visitor status in Australia; significant cognitive deficit (dementia or moderate to profound intellectual disability); severe communication impairment (e.g. post-stroke aphasia); and insufficient English language to allow a valid interview. The same instruments as in the National Study were used: a screening protocol for psychosis and the Diagnostic Interview for Psychosis.

Ethical aspects

We contacted key non-government agencies, including a large housing support co-operative, and religious organizations providing long-term and short-term accommodation. A focus group of residents expressed a strong view that screening be conducted by direct contact between interviewers and residents. Confidentiality, privacy and advance information were critical considerations for residents and organizations. The interviewers met staff from St Vincent's Mental Health Service (SVMHS) and other outreach agencies for advice. The St Vincent's Hospital, Melbourne institutional ethics committee granted approval for the study.

Catchment area

The area comprised two local government areas in Melbourne, the inner City of Yarra, population 65 000, and the contiguous more affluent suburban City of Boroondara, population 145 000 [25]. The Index of Relative Social Disadvantage (IRSD), which for Australia is 1000 and Victoria 1082, is 984 for Yarra and 1134 for Boroondara [26], where lower scores indicate greater social disadvantage. Specialist mental health services for adults aged 16–64 living or transient in the catchment area are provided by SVMHS. In the Yarra sector this includes a team of two psychiatric nurses who work specifically to engage homeless people and assist with providing care, as well as liasing with the relevant accommodation and welfare services. SVMHS is an integrated community-based service with continuity of care between the local inpatient and community teams. Emphasis is placed on communication and referral between local general practitioners (primary health care), hospital emergency departments, drug and alcohol services, and disability support, housing and welfare services, despite the lack of formal organizational links between these sectors.

Instruments

The screening protocol was developed for the national study [23] from psychosis screening items of the Composite International Diagnostic Interview (CIDI [27]), and the Psychosis Screening Questionnaire [28]. It includes five questions about specific psychotic symptoms (current or at any time in the past), a sixth question asking whether a doctor had ever told the person he/she may have schizophrenia, schizoaffective disorder or bipolar disorder (manic depression), and an item to record the rater's judgement about whether psychotic symptoms were present. As with the National Survey, scoring two or more items as positive was considered a positive screen. With sensitivity and specificity at 0.67 and 0.84, respectively, the positive predictive value of the screen was calculated as 0.70 and the negative predictive value as 0.80.

The instrument used to interview the selected subsample of screenpositive respondents was the Diagnostic Interview for Psychosis (DIP), a semistructured, standardized interview with three modules: (i) demography, functioning and quality of life; (ii) diagnostic module; and (iii) service utilization. The diagnostic module of the DIP allows the rating of the computerized algorithmic 90-item Operational Criteria for Psychosis (OPCRIT) checklist using the World Health Organization (WHO) Schedules for Clinical Assessment in Neuropsychiatry (SCAN) interview questions, probes and definitions (OPCRITSCAN). Two sets of computer-generated diagnoses based on the ICD-10 were produced, one on lifetime occurrence and the other on current occurrence (the latter using symptoms present in the prior month). The WHO Disability Assessment Schedule (WHO-DAS) [29]) was the main component of the disability schedule [30], [31]. The DIP was designed to be suitable for use by trained mental health professionals. In Melbourne the interviewers were psychologists, social workers and psychiatric nurses trained according to procedures established for the national study [23]. These training procedures were designed because of the complexity of using a semistructured diagnostic instrument by clinicians of any professional background. The level of diagnostic agreement achieved among interviewers was acceptably high (generalized κ of 0.73 for ICD-10 diagnoses).

Sample selection

Cluster sampling was used to identify a representative sample of residents from marginal accommodation in the service catchment area. Marginal accommodation was defined as residential accommodation affordable on a pension, consistent with the definition outlined earlier. This included rooming houses (which provide room only), boarding houses (which provide accommodation and some services such as cleaning, laundry and meals), crisis accommodation (shelters and other short-term accommodation run by religious and non-government agencies), and supported residential services for homeless people (privately owned for profit and offering meals, basic housekeeping and some assistance with medication, but no specialized help). It excluded a small number of hostels specifically funded and designated to accommodate and support clients of psychiatric services. The marginal accommodation has varying levels of support from housing workers.

All facilities in the catchment area meeting the definition were listed, using information from state government housing authorities, local councils, non-government community and religious agencies, clinicians and other researchers [32]. Information was collected about the number of rooms, amenities, ‘turnover’ of residents, and staffing and support in each place. The list was shortened substantially when we discovered that some places were converted into apartments or other use.

Houses were then stratified by local government area (Yarra or Boroondara) and size (bed numbers <20, 20–49, ≥50). From this sampling frame, a random list of houses was produced for each of the six strata. Using a stratified random cluster sample design gave an equal chance of inclusion to each person living in a particular stratum (defined by city and size of house), while giving different weights to the strata to allow inclusion of a minimum number of people from the different settings. Women were over-sampled because they were a small minority in the population, and all women from the four womenonly dwellings were included.

A target of 500 residents was used as a basis for obtaining 360 screens (assuming the prevalence of psychotic disorders as 30%). This target of 500 was divided among the strata, and the number of houses required to provide the estimated number of respondents drawn from the randomly ordered list of all houses in that cluster. If the proprietor of a selected house did not agree to the study proceeding, the next rooming house on the list was contacted.

Interviewers knocked on all doors in each selected house during the study period, aiming to approach all residents. Three separate attempts were made to contact the occupants of each room. On contact, the interviewer explained the purpose of the visit, described the study and administered the screening questionnaire to those who gave verbal consent.

All those who met the study criteria and screened positive were invited to take part in an interview. When individuals agreed to participate, they were offered a small reimbursement for their time. Interviews were conducted as soon as possible after the screen and according to the resident's preference, either at the house or a neighbourhood venue. The interviews were mostly completed in one session of 1–2 hours' duration.

The final listing was 80 houses or other facilities (‘houses’) with 1355 beds. Of these, the managers or owners of 13 did not respond to inquiries and 17 refused to participate. These non-participating houses were of varying sizes and standards and were not systematically different from the group that participated. From the remaining 50 houses, 37 houses with a total of 757 beds were randomly selected and participated in the study, 28 from 34 houses in Yarra, and 9 from 16 houses in Boroondara. Twenty-seven selected houses (with 322 beds) had fewer than 20 beds, 8 houses (with 272 beds) had 20–49 beds, and two houses (with 163 beds) had more than 50 beds.

From the total sample of 757 (Table 1), 160 people (21%) did not meet the criteria for screening. Out of the maximum 597 people that met study criteria, 348 were screened (58%); the remaining 42% comprised 117 people (20%) who refused to participate, and 132 (22%) who were not contacted after three attempts. Of the 348 people screened, 164 (47%) were positive.

Table 1.

Sampling results and overview of ‘case-finding’

Results

Details of the numbers of people screened and interviewed are displayed in Table 1. Ninety-four per cent of those who were screened positive and then interviewed were confirmed as having a lifetime diagnosis of psychosis according to the ICD-10 criteria described above.

Prevalence of people living with psychosis in marginal accommodation

Standard formulae for cluster sampling [33] were used to obtain estimates and standard errors of prevalence within each stratum separately, using the information presented in Table 1 and its notes. These estimates and standard errors were then combined using the known (population) weights for the numbers of houses within each stratum, using standard formulae for stratified sampling [33].

The estimated prevalence for lifetime psychotic disorders among those living in marginal accommodation is 42% (95% CI = 37–47, standard error [SE] = 0.024). Using current (within the prior month) occurrence of symptoms as ascertained by the OPCRIT-SCAN, the estimated prevalence of current psychotic disorders is 34% (95% CI = 29–40, SE = 0.028). This means (given the estimate noted in Table 1 for the number of eligible beds in the cities' marginal accommodation) that in these two local government areas in Melbourne, where the number of accommodation places of this nature is small and declining, there would likely be at least 300–400 people in these settings living with current psychotic disorders. An additional small number would be living rough or in other settings.

Characteristics of the interviewed sample with a current or lifetime psychotic disorder present (respondents)

Socio-demographic characteristics and diagnosis

Most of the respondents with a lifetime psychotic disorder were single unemployed men (Table 2), as in the main survey [23]. Women respondents were also likely to have never married (53%). The vast majority (97% of men and all women) were receiving government benefits or pensions. This reflects the very low proportion of people employed at the time of interview, and contrasts with the much higher proportion of people employed prior to illness. The pattern of findings is similar for the subgroup of 48 men and 12 women who met the criteria for a current psychotic disorder.

Table 2.

Percentage rates of socio-demographic and diagnostic characteristics of interview respondents

About 40% of the respondents had a lifetime diagnosis of schizophrenia. A higher proportion of women received this diagnosis and the diagnosis of psychotic depression (major depression with psychotic features) than men, while more men than women received a diagnosis of ‘other psychosis’ (30%). Those with a current diagnosis were more likely than the others to be diagnosed with schizophrenia.

Substance use

A high proportion of the respondents, men and women, with lifetime and current psychotic disorders, reported ‘ever using illicit drugs’ and ‘daily use during the past year’ (Table 3). The rates for tobacco smoking are very high (90% for men and 80% for women, compared with 73% and 56%, respectively, in the main survey, and 27% and 20%, respectively, in the general population of Australia [23]); and for a lifetime diagnosis of alcohol abuse or dependence (43% of men and 20% of the women met these criteria compared with 39% and 17%, respectively, in the main survey, and a 12-month prevalence of 9% in men and 3% in women [6.5% overall] in the general population [34]).

Table 3.

Percentage rates of substance use among interview respondents

Relationships with others, home duties, self-care and outside interests

Most respondents did not see themselves as part of a household (Table 4), even though most were living in a rooming house with other tenants and shared kitchen and sometimes dining facilities. A high proportion, higher in women, reported social withdrawal and difficulties socializing with others. Nearly half reported having no intimate relationship (a special person, either a friend or caregiver, with whom they could share their thoughts and feelings). Almost one in four said that no friends were available and, contrary to stereotype, over half the men and women said they needed and wanted more friends. Women (53%) and men (34%) reported lack of outside interests, and 36% of men and 40% of women reported difficulties with self-care.

Table 4.

Percentage rates of function (relationships, home duties, self care, other interests) and quality of life among interview respondents

Quality of life

Almost one in four respondents reported feeling unsafe in the living environment and a similar proportion reported being a victim of violence in the preceding year (Table 4). Between 20% and 30% of men reported being arrested or charged by police, and on the other hand, being unable to access police or legal assistance as required during the year. Most men and women said they were dissatisfied with their life as a whole and half the men were dissatisfied with their independence.

Service use

Almost all respondents had used some health service in the preceding year (Table 5). Contact rates with specialist mental health services in the previous year were high, as were those with general practitioners (all over 75%). Just one individual used specialist inpatient services only. Most had contact with specialist mental health services in the community and half of these had a case manager. The rates differed only slightly whether or not they satisfied the criteria for present-state psychosis.

Table 5.

Percentage rates of services used and received by interview respondents

In contrast to the high prevalence of substance use, fewer than 1-in-10 respondents reported using specialist drug and alcohol services (which are organized separately from mental health services). Approximately 1-in-5 received some other services specifically for drug and alcohol problems, including visits to a general hospital emergency department for a drug and alcohol problem or after a drug overdose, or attending a counsellor or general practitioner. Over half the respondents had attended a general hospital emergency department; the annual rate for the general population is 3% [35]. One-in-three respondents reported deliberate self-harm or overdose in the preceding year (Table 4), and 1-in-5 said they took a drug overdose and did not seek help.

While most people were prescribed medicines including psychotropic medicines, relatively few received counselling, help with work or time use, or other support and rehabilitation services. Fewer than 1-in-5 attended a defined rehabilitation or day program, to which they would have had access through the specialist mental health services and other agencies.

Discussion

The findings suggest an unexpectedly high prevalence (42% ± 5%) of people living with psychotic disorders in marginal accommodation in Melbourne. This compares with findings 10 years earlier when 22% of people living in marginal settings in Melbourne had a lifetime diagnosis of psychosis [13]. Housing for people on low incomes became much scarcer over the intervening 10 years. For instance, in one part of the study area, there were 53 houses with 786 beds in 1987, but only 20 houses with 419 beds in 1997 (field work for this study in the City of Boroondara). There is a possible link between these observations given the local welfare provisions. As marginal housing becomes scarcer, it may be occupied to a greater extent by those who are particularly disabled or disadvantaged, of whom those with psychosis are an important subgroup.

The limitations of research work in this diverse and mobile population are acknowledged and need to be considered in interpreting these findings. The limitations are similar to those in the earlier studies in a nearby area of inner Melbourne, where sampling and case-finding methods were also similar. Whether the selection process (whereby residents were included when we made contact and they agreed to participate) will mean we have an over- or under-estimate of prevalence is unknown. While the houses where managers agreed to participate were not systematically different in material ways from those where they refused, we have very little information about the characteristics of those people who refused to participate or could not be found. However, the findings are useful at an indicative level. The precise estimate of prevalence is of limited interest, and the methods allow us to have faith in the broad conclusion that the prevalence of psychotic disorders in this subgroup has risen rather than fallen over the decade. In these studies we took the relatively unusual step of studying as representative a group as possible of people living in the catchment area, rather than confining the study to one setting where the selection can be more carefully monitored.

The broad definition of homelessness used in this study, to include people in marginal accommodation, is consistent with that used in the earlier studies in Melbourne and elsewhere [13] and is a widely used Australian definition. A definition of homelessness will usefully acknowledge the lack of appropriate housing and the social marginalization of the individual [1]. Nonetheless, the definition of homelessness varies between studies, and sometimes includes only those sleeping in designated shelters or public spaces [e.g. 36]. People may be transiently or episodically or chronically homeless [37], and entry to or exit from the homeless state is usually part of a process rather than a single jump.

Disability in everyday, occupational and social functioning is even higher for this subgroup than for other people living with psychosis in Australia [23]. The rate of unemployment and the degree of difficulty with social relationships (both the perceived lack of success in social interactions and the level of avoidance) are particularly striking. Respondents also felt unsafe in their day-to-day environment. This perceived lack of safety and other adverse features of the environment no doubt contribute to the high levels of dissatisfaction described by respondents.

The extensive use of addictive substances and the high rates of lifetime abuse and dependence are comparable with those reported previously in Melbourne and Sydney [10], [13], although several times higher than those found among comparable international populations of people living with psychotic disorders [10], [13]. The consistency of this finding, and the low rate of use of drug and alcohol services, has important implications for treatment and support needs. The overlap in individual homeless people between substance-related disorders and mood and psychotic disorders is probably an important factor in the genesis of homelessness [36].

Homeless women with psychotic disorders are a minority, but overall the profile of disability is at least as severe as that of men. This is consistent with other studies of homeless mentally ill women in Australia [14] and elsewhere [38]. Homeless mentally ill women have been characterized as more deviant in communities where women are generally less prone to isolation than men. Women in this study have particular characteristics pointing to a different profile of needs. Co-occurrence of psychotic disorder and substance use is common, although less than for homeless men. There is a different profile of psychotic diagnoses, with more women diagnosed with psychotic depression. Furthermore, more homeless women than men are partnered or previously so, indicating a possible role of marital breakdown in their progression to homelessness. Finally, the women in this study have greater difficulties than the men in their social functioning.

Whether or not the people in these settings have a current diagnosis as well as a lifetime diagnosis of psychosis, apparently makes little difference to the profile of disabilities. This finding is consistent with the common clinical wisdom that assessing needs for care should extend beyond the current symptoms to take account of the history of disorder and the extent and type of disability. Diagnosis and disability occur on separate axes [31]. Many of the people with a lifetime diagnosis and few current symptoms, as well as those with persisting symptoms, require continuing psychosocial support and medication in a safe, secure and stimulating environment to maintain and recover function.

The study extends the scope of the Low Prevalence Disorders survey by moving outside the sampling frame of health services. The literature, earlier work and service experience, all suggest that the subgroup of homeless or marginally accommodated people is likely to have higher rates of psychotic disorders than the general population, and to be in contact with primary health care but not with specialist mental health services. While the first two of these observations were confirmed, this study found high rates of current contact with the local mental health service, despite the pattern of disability.

The findings raise the question: is this the best we can do? The study area has an organized, sectorized, specialist mental health service. There were high rates of contact with the community mental health services among those interviewed and also high rates of contact with primary health and hospital emergency services. However, we do not know the focus of these latter contacts, whether they were addressing mental disorders and associated disabilities and indeed, whether these services were in contact with each other. The unmet need for drug and alcohol services is noted above. Whereas we could assume that an intensive rehabilitation approach would be needed for most of these very disabled people, relatively few reported that they were receiving personal support or rehabilitation or vocational services. The people interviewed had access to specialist mental health services, but had at the same time high levels of disability and unmet needs.

These findings support the growing realization that even well-organized clinical services are failing to meet the needs of those with the most complex problems. Some of the ‘essential ingredients’ of good psychiatric and health services may remain unclear [39]. But also links with housing and welfare services are characteristically weak, to the detriment of the most vulnerable groups. Evidence from the literature suggests that disabilities in this subgroup can be addressed with careful assessment and treatment planning. To illustrate, the social disability persisting among mentally ill hostel residents in the UK, despite good contacts with specialized mental health and primary health services [22], was amenable to adequate treatment [19]. Community-based services require accurate planning and monitoring, and this in turn depends on a detailed assessment of individual needs for care, including self-reports and the views of staff [22]. Welfare and housing support can also contribute to an improved quality of life, as demonstrated in the US [4–8].

Co-ordinated services are critical for people with complex needs. Substance abuse treatment must be an integral part of comprehensive psychiatric services for people with severe mental illnesses. A range of housing options is required, including independent living with the availability of support services, including rehabilitation and opportunities for social contact and activity as appropriate. Early intervention is also important. The pathways to homelessness involve the interaction of societal factors and personal factors, including illness. Homelessness for any individual may follow an accumulation of problems, including mental illness and substance abuse, which limit function and choice [36]. While wider factors are involved, attention to risk factors at the individual level may reverse, or prevent, a drift towards homelessness [36].

It is clear from the literature that well-organized services with adequate resources can improve function and quality of life for homeless people with psychotic disorders, including those with high and complex needs as found in this study. The challenge is to achieve these interventions in real service systems with resource constraints and organizational boundaries. It is likely that this will be achieved if policy and practice can encourage co-ordinated intervention across health, welfare and housing sector boundaries. Follow-up studies of programs and interventions are now required in local settings [40], [41].