This article challenges the accepted view that mental illness is the primary cause of homelessness. We examine critically a series of Australian studies which have attempted to estimate the prevalence of mental health problems among the homeless population and argue that the definitions of mental illness often used tend to exaggerate the numbers. We then provide our own estimate, using a study of 4,291 homeless people in Melbourne, and explore the important question of whether mental illness precedes homelessness or derives from it. Finally, we attempt to reframe the debate on mental illness and homelessness in the light of our findings and discuss the policy implications that arise from them.
Estimating the Prevalence of Mental Illness
In the international literature, it is widely recognised that people with mental health issues are over-represented in the homeless population, but there is considerable debate as to the extent of the over-representation (Snow et al. 1986; Wright 1988). There is a similar debate in the Australian literature, where some studies report that 72 to 82 per cent of homeless people are mentally ill, whereas others report that between 12 and 44 per cent have a mental illness.
Australian studies that report a low to moderate level of mental illness draw their samples in different ways and use different methods to assess whether or not people have a mental health issues. Studies that report a higher level of mental illness also draw their samples in different ways, but they have a common definition of mental illness and use a standardised diagnostic tool. First, we review six widely quoted studies to illustrate both approaches.
A large study by the Australian Institute of Health and Welfare (AIHW) (2007) examined the characteristics of 100 400 clients who used the Supported Accommodation Assistance Program (SAAP) in 2004-05. At that time, SAAP was the Australian Government's flagship program to provide accommodation and support services for homeless people. SAAP clients were classified as having a mental illness if they met one of the following criteria: they were referred from a psychiatric unit; they reported psychiatric illness as a reason for seeking assistance; they were provided with, or referred to, psychological or psychiatric services (AIHW 2007: 4). On this basis, the AIHW found that 12 per cent of SAAP clients had mental health issues.
A study by Rossiter, Mallett, Myers and Rosenthal (2003) interviewed 403 homeless young people aged 12 to 20 using SAAP services in metropolitan Melbourne. Young people were defined as homeless if they had left home for two days or longer without their parents' permission and were living in various forms of temporary accommodation, including emergency accommodation, staying with friends and relatives, and sleeping rough. Rossiter and colleagues used the Brief Symptom Inventory (BSI) to assess the teenagers' mental health. The BSI has 53 questions covering such issues as nervousness, feeling lonely, and thoughts about death and dying. Rossiter and colleagues (2003: 17) found that 26 per cent of those surveyed 'reported a level of psychological distress indicative of a psychiatric disorder'.
A study by Flatau (2007) gathered information from 173 people aged 19 or older using 31 services in Western Australia. This included 18 services providing support to homeless people and 13 services providing support to people at risk of homelessness. A survey was used to assess the prevalence of mental health issues. The survey was 'completed by members of the research team or by agency case workers' who asked clients direct questions. However, the survey was supplemented by 'case notes and entry assessments' (Flatau 2007: 13). The research found that 44 per cent had a mental health condition, and depression was the most common form of mental illness (Flatau 2007: 14).
The three studies that reported a low to moderate level of mental illness varied in their operational definition of homelessness and they used different criteria to assess whether or not people had a mental illness. Importantly, none of the studies included people who had drug and alcohol issues as mentally ill, unless there was evidence of co-morbidity.
In contrast, three other Australian studies have found that between 72 and 82 per cent of homeless people have mental health issues. These studies also draw their samples in different ways, but they have in common that they use an agreed definition of mental illness and a standardised diagnostic tool which, importantly, included a measure of substance use.
The first study was undertaken in Melbourne by Herrman and colleagues (1989). They reported findings from a sample of 382 homeless people gathered at homeless shelters in Melbourne and boarding houses in the inner-city suburb of St. Kilda. Another study by Reilly and colleagues (1994) interviewed 34 young people who were residents of a supported accommodation program in Melbourne. The most well-known study was by Hodder, Teeson and Buhrich (1998) who interviewed a sample of 210 people gathered from eight emergency hostels in Sydney. They defined someone as homeless if they had spent the preceding night in: any space not designed for shelter; with a friend or relative where they could not stay permanently; in a hotel, motel or emergency shelter; or if they reported no permanent house or flat (Hodder et al. 1998: 13).
The three studies used the same clinical definition of mental illness, which included people with mood disorders (bipolar, depression etcetera), psychotic disorders (schizophrenia, psychosis) and substance use disorders. This clinical definition also distinguished between substance abuse and substance dependence. Substance abuse refers to situations where the use of alcohol or drugs lead to a major disruption of social relationships, situations of self-harm, or other forms of extreme behaviour (ABS 2007). In contrast, substance dependence occurs when people use alcohol or drugs on a regular basis, and this 'takes on a higher priority for the person than other behaviours that once had greater value' (ABS 2007: 25). As we subsequently demonstrate, the inclusion of people with substance dependence in the definition inflates the number of people classified as mentally ill.
The studies used a structured clinical interview from the Diagnostic and Statistical Manual of Mental Disorders (DSM-III-R) to establish drug or alcohol dependence. We endeavoured to ascertain the criteria that were used to categorise someone as alcohol or drug dependent but this was not reported. However, Teeson, Hodder and Buhrich (2003: 467) subsequently revealed that they began testing for 'alcohol dependence' if people consumed 'more than 12 standard drinks in any one year'; and they began checking for drug dependence if people reported recreational drug use 'more than five times in the last 12 months'.
At what point people crossed from being 'at risk' (12 drinks in a year) to being 'dependent' is not made clear in any of the studies. However, all report very high levels of mental illness in the homeless population and in many cases this is attributed to alcohol or drug dependence. For example, Herrman and colleagues (1989: 1181) found that 72 per cent of their sample had a 'severe mental disorder' at some point during their lifetime. Of the 275 people who had experienced mental illness, 69 per cent had a substance related disorder (mainly alcohol dependence), whereas 35 per cent had a mood disorder and 30 per cent had a psychotic disorder. Thus the inclusion of people with alcohol and substance dependence significantly inflated the figures.
The study by Reilly and colleagues (1994) of 34 homeless young people in Melbourne found that 82 per cent of their sample had a mental disorder. Among mentally ill …