Abstract
The mental health of homeless men, and to a lesser extent homeless women, has been assessed in many Western cities [1–11]. The results of these studies have consistently demonstrated high rates of mental disorders among people who are homeless.
One of the weaknesses of research into homelessness has been the inconsistent application of rigorous methodology. The application of consistent and rigorous methodologies may allow for comparisons across studies and countries. Such comparisons may be used to assess the impact of existing health and welfare systems on homelessness. For example, the fact that Australia has a universal health care system may increase the health service utilization of homeless people compared to other countries. For these reasons, the use of standardized methodologies has been emphasized in contemporary literature on homelessness [12], [13] and is used in the present study.
Lack of rigorous methodology has also restricted comparisons of the prevalence of mental disorders among the homeless and other populations within the same country [12], [13]. Such comparisons, with control where possible for confounding variables such as age and sex, allows the enumeration of the extent of the problem and may highlight areas of particular need.
In 1997, the first National Study of Mental Health and Well-Being (NSMHWB) was conducted among the Australian general population [14]. This study assessed the prevalence of mental disorders in a representative sample of 10 641 adult Australians. They were interviewed using the Composite International Diagnostic Interview (CIDI) [15]. For our study we employed the same version of the CIDI and it is the first study to use this diagnostic instrument among homeless persons in this country. Both studies were undertaken in the same year. This provided a unique opportunity to have comparable data from the Australian general population and a sample of homeless people in Sydney.
Research concerning the mental health of homeless men and women in Sydney has been restricted to rates of schizophrenia. Studies to date have shown that schizophrenia is common within these two populations. Two studies were conducted at an inner city refuge for homeless men in 1985 and 1990 [2], [3]. They found the rates of schizophrenia among the men to be 14–16% and 21–26%, respectively. In a separate study in 1993 the prevalence of schizophrenia among homeless women was found to be 26–30% [5]. In another study in 1993, the prevalence of cognitive impairment among homeless men in inner Sydney was found to be 25% [6]. While disorders such as alcohol and other drug dependence are reportedly common among homeless populations [8], [9], they have not been measured within the homeless population in inner Sydney. Nor has the prevalence of mood and anxiety disorders, although these conditions occur commonly in the general population [14], [16], [17]. It is reasonable to expect that these disorders would be more prevalent among homeless people due to their deprived living circumstances, social isolation, and traumatic life experiences [18].
The first aim of this study was to assess the 12-month prevalence of mental disorders among homeless men and women in inner Sydney using a valid and reliable structured diagnostic interview. The second aim was to compare this data with: (i) data collected on homeless populations in other western cities; and (ii) data on a representative sample of the Australian general population, using the same instrument and DSM-IV criteria.
Method
Sydney is the largest city in Australia and has a population of approximately 3.8 million. Inner Sydney has a population of 72 000 [19] within approximately 16 square kilometres (10 square miles) surrounding the central business district. It is characterized by extreme wealth and poverty. This inner Sydney area has the highest concentration of homeless people and services for homeless people in Sydney.
Sample
The size and composition of the homeless population in Sydney has not been formally assessed, as is the case in most large cities. Homeless people in Sydney fall into three main groups: (i) those who seek accommodation in refuges for homeless people; (ii) those who use these refuges for food and support but not accommodation (refuges for homeless people provide the majority of free meal services in Sydney); and (iii) those who are homeless but make no use of refuge services. In Australia, the number of homeless persons who do not use a refuge for accommodation or support is very small [4]. Our sample was drawn from groups (i) and (ii), that is, all persons eating a meal at refuges for homeless people in inner Sydney. The group did not necessarily reside in inner Sydney.
The seven major refuges for adult (age 18 and older) homeless men and women in inner Sydney participated in the study. In total they provide 407 beds, 350 for men and 57 for women. Each refuge provided dining facilities for 50–500 people; therefore, sampling for men was conducted proportionally to the number of persons eating at the different refuges. Women were over-sampled compared to men as less was known about homeless women in Sydney. The minimum sample size was set at 50 for the women so as to allow meaningful analysis. For this reason, the rates for men and women are described separately in this paper.
The chairs in the dining rooms of the refuges were individually numbered. A random sample of chair numbers was generated for each dining room at breakfast, lunch, and dinner. The occupant of the selected chair was approached as they left the dining room and asked to participate in the study. Sampling took place over 7 months. All subjects were approached and interviewed by the one interviewer (TH) who had worked in the refuges and knew many of the men and women. The previous contact improved the participation rate for the study and reduced the possibility that the same person was interviewed twice.
Our definition of a homeless person, based on that of Gelberg and Linn [20], was a person who had ‘spent the previous night in: (i) emergency shelter; (ii) outdoors; (iii) any space not designed for shelter; (iv) hotel, motel, or home of a relative or friend and was uncertain whether he or she could continue to live there for at least the next 60 days; (v) stated that he/she did not have a permanent house or apartment to which he or she could go; or (vi) had stable accommodation yet required regular support from services for homeless persons’.
People were not included in the study if they were less than 18 years or spoke little English or were too intoxicated to interview.
Measures
The study measures included five modules from the National Survey of Mental Health and Wellbeing Composite International Diagnostic Interview 2.0 (CIDI 2.0) 12 month Version [15], namely: anxiety disorders; mood disorders; alcohol use disorders; other drug use disorders; and Mini Mental State Exam. The interview was designed for use by lay interviewers in a variety of cultures and settings, and has demonstrated validity and reliability for anxiety disorders, mood disorders and substance use disorders [21]. It employs the diagnostic criteria of DSM-IV [22]. Due to the location of interviews in refuges for the homeless the paper and pencil version of the interview was administered. In addition to the CIDI, questions on service utilization from the Australian National Health Survey [23] were also asked.
The CIDI is considered unreliable in diagnosing schizophrenia [15]. We therefore used the following method to assess schizophrenia: a fully structured screen for psychotic disorders [24], including interviewer observations, and a clinician rating by a psychiatrist (NB). The psychosis screener comprises seven items relating to symptoms that may have occurred in the past 12 months. They are delusions of control; thought interference and passivity; delusions of reference or persecution; grandiose delusions; and (lifetime) diagnosis of schizophrenia. It has been shown to be a sensitive and specific screener for schizophrenia [24]. This process of screening was accompanied by observations based on the interviewer observations section of the CIDI. If a person responded ‘Yes’ to at least one item on the screener or interviewer observations then they were asked if they would allow the psychiatrist to interview them. The psychiatrist using DSM-IV criteria, rated the person as having ‘definite schizophrenia, possible schizophrenia or unlikely schizophrenia’ in their lifetime. Definite schizophrenia was rated when an individual met all the criteria for DSM-IV schizophrenia, possible was rated when there were some symptoms (e.g. negative symptoms) but not sufficient for diagnosis. Unlikely was rated when there was no current evidence of schizophrenia.
Clinical diagnosis was based on an assessment of symptoms of the DSM-IV criteria with the diagnostic requirement of a 6-month history based on previous admission for schizophrenia or mental illness. The assessment took between 20 and 40 min.
The Brief Psychiatric Rating Scale [25] was completed to assess severity of symptoms. This method replaced the schizophrenia section of the CIDI which has been shown to have poor validity and reliability [21].
Procedure
Subjects selected for the study were approached by the interviewer who outlined the study and gave them an Information Sheet which explained in detail its aims, procedure, and confidentiality agreement. All interviews were completed by TH, a lay interviewer who was trained in the delivery of the CIDI. The average duration of the interview was 75 min. The study was approved by the appropriate ethics committees.
Data analysis
The CIDI was scored using algorithms provided by the Australian Bureau of Statistics and was scored consistently with scoring from the Australian National Survey of Mental Health and Wellbeing. The prevalence estimate for any given mental disorder is based on the number of people who were assessed for that disorder. The prevalence estimates for comorbid disorders are based on the number of people who were assessed for all disorders, that is, who completed the entire interview.
Results
Two hundred and eighty-nine people were approached to participate in the study of whom 79 refused or were excluded. Two hundred and ten people (160 men and 50 women) agreed to take part in the study, a participation rate of 73%.
The mean age of the men was 44 years (SD ± 16) and of the women was 39 years (SD ± 14); 59% of men and 56% of women who were interviewed were Australian-born. Most of the subjects had left school before age 17 years (65% men and 62% women). Few people were married or in a defacto relationship (never married 67% men and 46% women), 41% of men and 54% of women had one or more children. With respect to living conditions, the sample spent the previous night accommodated in many places: 75% in a hostel; 8% in their own place; 6% in a boarding or rooming house, hotel, or motel; 5%; in a car/park/ street/‘squat’; 1% in a psychiatric hospital and 4% stayed elsewhere such as with friends.
Prevalence of mental disorders
Table 1 gives the 12-month prevalence rates of four mental disorders among homeless people in inner Sydney and in the Australian general population using DSM-IV criteria. For reasons discussed above, we did not use the CIDI to diagnose schizophrenia. Of the 210 people interviewed, 124 (84 male and 40 female) screened positive for psychosis using the Psychosis Screener and interviewer observation. Of the 124, 81 (52 male and 29 females) were assessed by the psychiatrist. The remaining 43 refused interview or could not be located.
Men
Seventy-three percent of men met criteria for at least one mental disorder in the past 12 months or lifetime for schizophrenia. Forty percent of men met criteria for at least two mental disorders. The most common disorders among men were substance use disorders, primarily alcohol, cannabis, or opiate use disorders.
Women
Eighty-one percent of women met criteria for at least one mental disorder in the past 12 months. Fifty percent of women met criteria for at least two mental disorders. The most common disorders among women were affective disorders followed by substance use disorders other than alcohol. Thirty-six percent of women met criteria for one or more anxiety disorders.
Comorbidity
There was considerable comorbidity between mental disorders. One in five men (20%) and 29% of women met criteria for two mental disorders, predominantly alcohol and affective disorders in men and mood and anxiety disorders in women. One in five men (20%) and 21% of women met criteria for three or more mental disorders.
Comparison with the Australian general population
Table 1 also presents data from the National Survey of Mental Health and Well-Being [14]. The prevalence of any mental disorder is four times higher among homeless men and women in inner Sydney than within the Australian general population. Twelve-month prevalence estimates for alcohol use disorders, other drug use disorders, affective and anxiety disorders, are significantly higher among the homeless population in inner Sydney than among the Australian general population. The prevalence of opiate dependence–abuse and sedative dependence–abuse are far greater among homeless people in inner Sydney than among the domiciled population in this country.
Prevalence of mental disorders among homeless people in inner Sydney and the Australian general population
Comparison with other international studies
Tables 2,3 compare the results of four prevalence studies among the homeless to the present study, for both men and women. While each study used validated measures, one of the major limitations for direct comparison is the fact that the timeframes for assessment differ as do the measures themselves. This comparison can therefore only be indicative. There has been one study from Madrid using the CIDI to assess mental disorders among the homeless [13]. The Madrid study therefore provides the strongest international comparison with the present study. The Madrid study found substance use disorders and affective disorders to be consistently lower than in the current Sydney study.
Comparison of prevalence rates across studies of homeless men
Comparison of prevalence rates of mental disorders across studies of homeless women
Discussion
The rates of mental disorders and of comorbid mental disorders are high among homeless people in inner Sydney. These rates are consistent with studies among homeless populations in the US [1], [9], Australia [2–6], and Europe [7], [12], [13].
This is the first study to assess the prevalence of mental disorders among homeless people in parallel with a national study in the general population, using the same instruments. It allows us to conclude with confidence that the 12-month prevalence rates of DSM-IV mental disorders are much higher among homeless people than in the general population in Australia. This is true for all of the disorders assessed. The high prevalence of schizophrenia among homeless persons in Sydney is of considerable concern. High prevalence rates of schizophrenia within this population have been reported by these authors over the past 10 years. Despite the provision of early intervention strategies, reduction in long stay hospitalization and provision of community care, a minority of people with schizophrenia are still homeless in the inner city of Sydney. Schizophrenia is clearly a very debilitating disorder. In the year 2000, the first national estimates on psychosis reinforced the debilitating consequences of schizophrenia [24]. The study estimated that 0.5% of the Australian population had schizophrenia and that they were disadvantaged across employment, income, broader social outcomes, and general health outcomes.
A strength of this study is that it enables data on homeless people in Sydney to be compared with data from other studies. We have compared our data with data on the homeless populations in a number of countries. The strongest comparisons can be made with Madrid [13]. The most frequent mental disorders among homeless men in Sydney and Madrid were alcohol use disorders. Homeless women in Madrid were most likely to meet criteria for an affective disorder whereas homeless women in Sydney were most likely to meet criteria for a drug use disorder. The prevalence of substance use disorders (other than alcohol) was much higher among homeless men and women in Sydney, than in Madrid.
The consistency in the methodology, age, and sex profiles of the studies conducted in Sydney and Madrid would indicate that the differences in prevalence rates are not due solely to differences in methodology. The reasons for the differences are likely to differ across the different disorders. The large differences in the prevalence rates for schizophrenia are likely to reflect the fact that the Madrid study relied on self-report, whereas the Sydney study incorporated clinical assessment. The rates of depressive disorders vary widely across the samples, with higher rates in Sydney compared to the Madrid study. The wide variation is not obviously explained but the higher Sydney rates are consistent with the high rates of suicide (highly correlated with depression), found in Australia compared to other countries [26]. While alcohol use disorders are highly prevalent in both the Madrid and Sydney studies, the most striking differences occur in drug use disorders. We hypothesise this to be the result of recent trends in drug availability in Australia, including an increase in heroin availability as outlined below. This emphasizes the potential causal role of drug dependence in homelessness and highlights the importance of cross-national comparisons.
The most common drugs of dependence were cannabis and opiates. Nearly 20% of homeless men in Sydney were dependent on opiates (mostly heroin) compared to 5% of homeless men in Madrid. This difference is striking given the similar rates of heroin use among the adult population in Australia (1.3%) and Spain (1.1%) [14], [27]. There was a major initiation of heroin use in Australia from 1982 to 1985. Since late 1997, there have been indications of rising heroin use among young adults in Australia [28]. The recent increases in use of heroin in Australia may be reflected in the high rates of use and dependence observed among homeless people, in this study.
The low drug use disorder rates among homeless people in Madrid were also highlighted by Vazquez et al. [13]. They suggest that one reason for the low prevalence of these disorders was that ‘younger, heavy drug users are not disproportionately incorporated into the homeless population, since most of them live with their families’. Yet the age profile of the samples drawn for both studies is strikingly similar. There is little to suggest that family support for young drug users differs in Sydney and Madrid. Factors contributing to the different rates of drug use disorders in these two cities therefore need to be elucidated.
There remain a number of limitations to the present study. The study does not take into account varying patterns of attendance at the refuges; that is, some people will use the refuges for meals once a day, some twice, and some may use it three times a day. The concern is that heavy users may be different and more likely to be mentally ill than infrequent users of the food services. Although such weighting would provide some level of greater precision in the prevalence estimates, asking the homeless individuals such information would be unreliable and directly observing would have been prohibitively expensive, and not feasible in the setting of a homeless refuge. However, the prevalence rates of mental disorders are so much higher than the Australian general population and comparable to international studies on the homeless that such weighting, while it would account for the possibility that frequent users may be more likely to have mental disorders, it is unlikely to substantively change any conclusions of the paper.
Intervention strategies among homeless persons in Australia have tended to focus on psychotic disorders [29] and this is appropriate. Now more intensive efforts are needed to address the high rates of other mental disorders and comorbid mental disorders within this population. We have shown that these high rates, especially of comorbid mental disorders and substance use disorders, are not just restricted to Australia. The comorbidity and high rates of substance use and other mental disorders constitutes a significant challenge for appropriate services, treatment strategies and public health research both in Australia and internationally. Vazquez et al. [13] argue that the complexity of the problem of homelessness benefits from not only within-country but also between-country studies. Such comparisons were made in the current paper, however, further international public health research is required if we are to understand the major international social issue of homelessness.
Footnotes
Acknowledgements
Sydney City Mission, The Society of St Vincent de Paul, Salvation Army, Wesley Mission, Haymarket Foundation, and the Centre for Mental Health.