Abstract
There is conflicting evidence as to whether compulsory treatment of psychiatric patients in the community increases compliance with treatment or reduces health service use [1]. Some matched or randomised controlled designs demonstrated a significant reduction in health service use compared to controls [2], [3], while others did not [4], [5]. In a further group of studies, including one from New South Wales, it was not possible to determine whether any effects were due to the compulsory nature of the treatment or greater intensity of management [1],[6–9].
Data from one randomised controlled trial suggested that treatment orders have to be maintained for at least 180 days to be effective [1]. However, this was based on post hoc analyses of the non-random subsample whose orders were extended beyond the first 6 months. Such analyses are subject to bias and confounding that randomised trials are designed to minimize [10]. For instance, analysis of subjects who have not been randomly assigned to compulsory community treatment of less than, and more than, 180 days may reflect a bias where treatment was selectively extended when it seemed to be helping the patient [11].
Many studies only consider patients discharged from inpatient units rather than living in the community. We investigate the characteristics of all patients placed on community treatment orders across a whole jurisdiction (Western Australia [WA]), to determine why certain patients are placed on compulsory treatment.
Community Treatment Orders in Western Australia
The current Mental Health Act of Western Australia was implemented in November 1997. As in other parts of Australia and New Zealand, it includes the provision of involuntary treatment in the community through the introduction of a community treatment order. The order is made out by a qualified mental health specialist, who must specify which medical practitioner is to supervise the patient's treatment or care, where the patient is to receive care, the frequency at which the medical practitioner is to report to the specialist, and the duration of the order, which must not exceed 3 months in the first instance. The Mental Health Review Board must conduct a review between four and six months after instigation of the order. A lawyer, a psychiatrist, and a lay-person make up the review board and hear evidence from both treating doctors and patients. The prevalence of CTOs in WA is similar to that of New South Wales but considerably higher than the use of similar legislation in Canada [9], [12].
Mental health services in Western Australia
The data used in this study were collected two years after a major reorganization of mental health services in WA with a shift from institutional to community and mainstreamed care in line with national and state policy [13–15]. This included the closure of one major psychiatric hospital and the reduction in the size of another. By the end of 1996, 80% of care occurred in community settings, with inpatient care being reserved for acute emergencies [16], [17]. The introduction of CTOs was the final step in the move to greater community care.
Method
Data sources and case selection
Western Australia is fortunate in having a comprehensive health services database that links a mental health register with inpatient and outpatient activity in psychiatric units and general hospitals throughout the state. Western Australia is also unique in that the bulk of its population of 1.7 million is isolated by desert from the rest of Australia, so limiting cross-boundary flow. We combined this database with one administered by the Mental Health Review Board that records all involuntary treatment under the Mental Health Act and the offenders database held by the Police Department, which records all offences and convictions. Use of these three databases enabled us to assess the psychiatric and forensic history of all patients who had been placed on a Community Treatment Order (CTO) in the state.
We compared two groups of subjects. The first group (n = 265) consisted of all patients placed on a CTO between 1997 and 1998 in Western Australia. The date of first placement on an order from either an authorized mental health hospital or the community was referred to as the ‘index date’. Patients placed on orders between 13 November 1997 (the date of implementation of the Mental Health Act 1996) and 31 December 1998 were selected.
The second was a consecutive control group (CCG). This consisted of the next patient who was consecutively discharged from an authorized mental health hospital on the same day as a CTO case was discharged, or placed on CTO in the community (n = 265). The index date for the CCG was the same date as their respective CTO case.
Data preparation
Some of the data elements selected for our study had categorical values (e.g. gender, marital status) and others were continuous (e.g. illness duration, or length of hospitalization). After testing various model configurations, we needed to reduce the number of value categories for some of the variables such as marital status. In most cases, we transformed these continuous variables to three levels, that is, those values smaller than 25th percentile being the first level (usually taken as a reference category), those values between 25th and 75th percentiles being the second, and those values greater than 75th percentile being the third. If the continuous variable demonstrated high skewness, the median of the variable was taken as a dividing point to divide all values of the variable into two levels.
Patient characteristics included age, gender and aboriginality. We divided marital status into those who had ever been married or in a de facto relationship and those who had not. We also considered country of birth (Australian vs. other countries) and location of service (metropolitan vs. rural or remote).
Case complexity variables included rates of admission, inpatient bed days, outpatient contacts, residential bed days and whether a subject ever received aftercare treatment (suspended or conditional discharge) in the year prior to the index date. We also examined diagnostic complexity using the same methodology as previous epidemiological studies [18]. The final diagnosis in each episode of care was coded using ICD9 according to the following hierarchy:
ICD9 290, 293–296: dementia, organic psychotic conditions, schizophrenia, and affective psychosis. ICD9 291–292, 297–305, 312, 315: alcohol and drug psychoses, paranoid states, other organic psychoses, neurotic disorders, personality disorders, sexual deviations, alcohol and drug dependence, childhood disorders. ICD9 306–312, 317–319: adjustment reaction, reaction to stress, depressive disorders, conduct disorders, special syndromes, and mental retardation. Non-chapter 5 ‘mental disorders’ diagnoses.
We also considered the number of three-digit primary or secondary ICD-9 psychiatric diagnoses in the year prior to the index date and whether subjects had ever been diagnosed as having a personality disorder.
Forensic history included whether a subject had an offence recorded, was imprisoned and the number of offences recorded for the year before the index date. Lifetime ‘ever’ history of imprisonment was also included. In addition, we examined the type of offence (antisocial, property or person) and the most serious one that had occurred in the previous year.
Logistic regression analysis
We used logistic regression analysis to examine whether demographic, case complexity or forensic history contributed to an increased risk of CTO placement. Scatter plots were used to screen data for multicollinearity among the predictors. No multicollinearity was identified, which means estimates and their respective standard errors are reliable for each analysis.
Results
Descriptive statistics
Patients placed on a CTO were more likely to be younger, male, have a high diagnostic complexity, have more aftercare placements prior to index date, be diagnosed having schizophrenia and have committed an offence on a person. They also had greater health service use in the year prior to the index date even though they had shorter overall psychiatric history (Tables 1 and 2).
Comparison of characteristics of risk factors between two groups (categorical variables)
Comparison of characteristics of risk factors between two groups (continuous variables)†
Logistic regression analysis
Variables used in the descriptive analysis were placed within a multivariate model to investigate their contribution to CTO placement. The effects of the presence or absence of each variable under investigation on outcome were examined using adjusted odds ratios (OR). These were derived using forward stepwise logistic regression to determine the relative importance of clinical and socio-demographic variables on outcome. Table 3 reports coefficients for significant predictors within the equation. A positive coefficient means that compared with its reference category, patients with a specific level of a risk factor are more likely to be placed on CTOs. The opposite effect is true for negative coefficients.
Logistic regression results for case and consecutive control groups†
Factors that increased the chance of being placed on a CTO included a history of aftercare placement, a history of schizophrenia, offences on a person, a greater number of inpatient admissions, longer inpatient stays and a greater number of outpatient visits. Having never been married and longer mental disorder history decreased the chance of being placed on a CTO (Table 3). Patient's age, gender, and diagnostic complexity have not been found to be significant.
Discussion
We are aware of only one study of determinants of CTO placement that covers a whole jurisdiction [19]. However, Schied-Cook [19] studied court-ordered community treatment and not situations where the decision is made by a clinician [20]. Other studies have considered a particular hospital [5] or region [1], [3]. The one previous study of CTOs in Australia was restricted to Hornsby in the northern suburbs of Sydney, New South Wales [9].
Only one previous study has directly compared the characteristics of patients on compulsory community treatment and controls in clinical settings [3]. However, this was not strictly speaking a study of community treatment but of extended leave from hospital under the Mental Health Act in England and Wales. This use of the Mental Health Act was subsequently judged illegal and the practice stopped. The paper therefore reported differences between patients who had been placed on extended leave and those selected by psychiatrists as not requiring compulsory community treatment, matched on age, sex and diagnosis. It was therefore not possible to investigate for the effect of these variables in the study.
We used an epidemiological sampling frame covering all patients placed on such an order so that the sample was representative of patients placed on a CTO, and not subject to referral or selection bias. We were able to link mental health and forensic databases so we had an accurate record of all offences committed in a whole jurisdiction. We were therefore not reliant on self-report or information from health records that could be subject to reporting bias [3]. We also used regression modelling to allow for multivariate examination of a wide range of predictors of CTO placement. However, as this was an epidemiological study, we could only compare information on demographic characteristics, diagnosis, health service use and forensic history. We did not have any information on other clinical features or social disability.
We used data collected from 1997 to 1998 reflecting the time it takes to ensure data-sets are complete and obtain permission to link databases from different government agencies. The data were collected after a time of considerable change in mental health services in WA that has not occurred subsequently. Importantly, a review of the Mental Health Act in 2003 did not lead to any substantial change in the legislation. However, we cannot exclude the possibility that clinicians' use of the Act may have changed in the last 5 years.
As in previous studies using descriptive analyses, patients placed on CTOs tended to be less than 40 years old [1], [5], male [1], [9] and have a high diagnostic complexity, including mental disorders such as schizophrenia [1], [2], [5]. Previous health service use and a positive forensic history, particularly against persons, also predicted compulsory community treatment [1], [3], [9], [17]. This patient profile applies equally to studies that report positive findings on the effect of CTOs on health service use [2], [3], and those with more equivocal results [1], [6], [7], [9]. Differences in the reported characteristics of subjects do not seem to explain the variable efficacy of CTOs, although it is possible that the results might be affected by other variables for which we had no data, such as social disability.
On multivariate analyses, gender, age and diagnostic complexity did not emerge as predictors of CTO placement. This finding was unexpected as male gender has been shown to independently influence compulsory admission to hospital [21]. Sensky et al. [3] in their study of extended leave also found few differences between subjects and controls other than in the areas of dangerousness and compliance with treatment. However, these authors specifically matched their intervention and control groups on age, sex and diagnosis.
Being in a present or past relationship also increased the likelihood of CTO placement. This result supports previous research in the US where people living with their families are more likely to have their conditional releases revoked than those living alone [22], but is in contrast to the findings of Vaughan et al. [9] in New South Wales. This may indicate that placement and revocation of CTOs are more likely to occur where the patient is consistently interacting with others, which may be a source of stress or provocation.
Our study supports previous findings that recent dangerousness, particularly violence against others, increases the probability of compulsory treatment in either hospital or community settings [3], [20]. We do not know whether previous dangerousness is truly predictive of future dangerousness when related to mental illness [23]. Without these linkages, it is difficult to say whether compulsory community treatment really protects patients from self-harm or harm to others.
We need further research to establish the relative contribution of patient characteristics, legislation and service setting to understand the use and outcome of compulsory community treatment. This appears to be the only way to make sense of the variations in the reported efficacy of compulsory community treatment.
For instance, compulsory community treatment can cover a wide range of interventions including community treatment orders, involuntary outpatient treatment, involuntary outpatient commitment, extended leave and supervised discharge. Extended leave provisions or supervised discharge apply at the time of discharge from compulsory inpatient treatment. They are used in Canada [24] and Great Britain [3], and give mental health professionals the right to return patients to hospital against their wishes if they do not comply with treatment. Community treatment orders are used in Canada [24] and Australia [9] and give mental health professionals the right to place an individual on an order, whether they are in hospital or not. This is in contrast to extended leave or supervised discharge, which only apply to patients who are being discharged from inpatient care [24]. Community treatment orders are designed to divert people from possibly having to be admitted as inpatients, and the individual may not have to meet the same criteria for compulsory admission to hospital [24]. In Australia, it can include the power to force medication in the community [4]. Involuntary outpatient treatment or commitment is the preferred term in the US and covers courtordered community treatment [25]. In this case, a judge, not a healthcare professional, decides on the appropriateness of the order.
We still do not know for whom and why these interventions might work, and if particular models are associated with better outcomes. Given these remaining questions, evaluation of the use and effect of compulsory community treatment should be included whenever this legislation is introduced.
Footnotes
Acknowledgements
Research funded by the Medical Research Foundation, Fremantle Hospital and Health Service, Western Australia.