Reappraising community treatment orders — can there be consensus?

Community treatment orders have become standard practice without serious consideration of the underlying research base

Community treatment orders (CTOs) require someone with a mental illness to follow a treatment plan while living in the community.1 Initially, debate focused on the ethical justification for CTOs, but subsequently shifted to their effectiveness. These considerations are particularly relevant to Australia as a few states, such as Victoria, have among the highest rates of CTO use in the world.2 Further, CTOs may also be incompatible with Australia’s obligations to the United Nations Convention on the Rights of Persons with Disabilities.3 Here, we present a consensus from two authors who have previously expressed very different views on the use of CTOs.1

We note, at the outset, a lack of clarity about the purpose, or purposes, of CTOs. Are they to reduce revolving-door admissions, provide a less restrictive alternative to involuntary admission, prevent violence by people with severe mental illness, or increase stability and promote recovery? These different aims involve a range of different outcomes: hospital use, perceived coercion, violent acts and quality of life. Uncertainty about their purpose is compounded by the range of different interventions. Interventions include clinician-initiated CTOs or supervised discharge and court-ordered outpatient committal. Supervised discharge coexists with CTOs in several countries including Canada and the United Kingdom.4

CTOs can be either “least restrictive” or “preventative” in function. “Least restrictive” orders are an alternative to involuntary hospitalisation in a person who would otherwise meet the criteria for inpatient commitment. By contrast, “preventative” orders aim to avoid relapse and hospitalisation where someone is at risk but does not currently meet the criteria for compulsory admission. Australian CTOs are clinician-initiated and contain elements of both. They are used either on discharge from hospital, or in the community, and last up to 12 months in Victoria, but less elsewhere. In contrast to many other countries where CTOs can only be used if the person has been previously hospitalised, CTOs are often used as an alternative to admission.

Australian patients on CTOs are typically younger males, with schizophrenia or serious affective disorders, frequent admissions, poor adherence and a forensic history. Indigenous status and/or rurality are not associated with CTO placement, possibly reflecting the availability of services to administer the order in remote areas.5

Research findings of health service use

Before-and-after studies suggest that compulsory community treatment led to increased follow-up with clinical services, a reduction in inpatient admissions and reduced lengths of stay in hospital. Subsequent controlled studies, using matching or multivariate analysis, confirmed increased follow-up with mental health services, as well as improved forensic outcomes, but not reduced hospitalisation6,7

Only randomised controlled trials (RCTs) can fully control for potential confounders. There have been three pragmatic studies on the effectiveness of CTOs: two in the late 1990s in North Carolina and New York,8,9and one more recently in the UK.10 None of these demonstrated statistically significant reductions in hospitalisation or improvements in health and forensic outcomes for patients on CTOs.

The two US studies have been the subject of considerable debate. Both excluded patients with a history of violence from randomisation and had high attrition rates resulting in selection bias and reduced study power. In one study, there was also concern about adherence to the research protocol and a smaller than expected sample size.9 A subsequent meta-analysis of the two US studies did not change the negative findings, but only addressed study power, not the other issues.4,7

A non-randomised post-hoc analysis of the North Carolina study reported that CTO placement of more than 180 days was associated with positive outcomes, including decreased hospitalisation and violence.8 However, such an analysis potentially introduced a type of selection bias that randomisation is designed to avoid: individuals may have been maintained on a CTO precisely because they were doing well.

The most recent RCT was the Oxford Community Treatment Order Evaluation Trial (OCTET) in the UK.10 This was a study of clinician-initiated treatment, as opposed to court-ordered interventions, like those examined in the two US RCTs, and therefore is more applicable to other countries with similar provisions. The study randomised patients discharged from hospital to an experimental group (CTOs) or a control group (extended leave under Section 17 of the UK Mental Health Act 1983) and compared their outcomes at 12 months.10 Unlike the US studies, patients with a history of violence were included.

There were no differences in readmissions or length of stay between patients on CTOs and patients in the control group. However, subjects were only included if they were equally suitable for a relatively short period of Section 17 leave or a CTO. Patients who might have especially benefited from a CTO may therefore have been excluded. Further, around 20% of those eligible lacked the capacity to consent or refused to participate. In addition, physicians could make clinical decisions irrespective of the initial randomisation, resulting in over a fifth of those in either arm swapping treatments. A sensitivity analysis to remove these protocol violations may, in turn, have left the study underpowered given the authors’ own calculations.10It will also have not removed the possibility that Section 17 patients who were swapped to a CTO might have been more severely ill than those remaining on Section 17 leave as per the protocol. Finally, the UK trial did not compare CTOs with voluntary treatment but with another type of compulsory community treatment. Although the length of initial compulsory outpatient treatment differed widely between the two groups (medians of 183 days v 8 days), Section 17 patients averaged 4 months on some form of compulsory treatment — a long time for control patients supposedly under voluntary care. This time in compulsory care was mostly due to protocol violations such as the Section 17 patients being changed to CTOs. The lack of difference in outcomes might therefore be explained by the two types of intervention being similar.

In conclusion, the evidence suggesting that CTOs reduce hospitalisation comes primarily from before-and-after studies and the post-hoc analysis of the North Carolina RCT. Most studies with randomised or matched controls have failed to show an effect. However, methodological issues with all the study designs, including the three RCTs, mean that the results have to be interpreted with caution.

Other outcomes

Outcomes in health service use may not reflect outcomes in other important areas such as quality of life or social functioning. The North Carolina RCT reported that subjects on a CTO were significantly less likely to be victimised, but also more likely to report perceived coercion.4,8 The effect on stigma was not reported. In terms of other outcomes, there is no RCT evidence for differences in treatment adherence, social functioning, homelessness, mental state, quality of life and arrests.4,8,9 However, in the case of the forensic outcomes, it is important to note that two studies excluded patients with a history of violence.

Other important or rare outcomes, such as mortality, cannot easily be assessed by an RCT. However, epidemiological studies have suggested that CTO cases have reduced mortality rates compared with control patients after adjusting for confounders, possibly by improving physical care through increased contact with community psychiatric services.1 1

Implications

We suspect that CTOs may initially increase the likelihood of admission because of increased monitoring in the community. Their effect on other outcomes is less clear. In view of the continued uncertainty about the effectiveness of CTOs, what should clinicians and legislators do? As CTOs are coercive, clinicians have a responsibility, irrespective of their effectiveness, to use CTOs judiciously and only if less coercive approaches have failed. Of concern is the marked variation in the use of CTOs in different jurisdictions.2 While it is unclear if this indicates overuse or underuse, clinicians should avoid the temptation to routinely place patients on CTOs. In addition, one potential conclusion from the UK OCTET study is that briefer conditional leave may suffice for some patients and should therefore be considered when both CTOs and conditional leave are available.

Policymakers must recognise that CTOs are a vehicle for delivering services and are not an alternative to providing care. Governments should also avoid naming the relevant CTO legislation after victims of violence by people with mental illness; examples include Laura’s, Kendra’s and Brian’s Law.1Naming CTO legislation after victims of violence is stigmatising and perpetuates an erroneous belief that people with mental illness are dangerous. Also, while there is non-randomised evidence that CTOs may reduce violence by individuals with untreated illness, this has not been confirmed in subsequent RCTs.8,9 Further, CTOs cannot prevent all potential violence as many individuals only come under professional care after they have offended, and others are inaccurately assessed as at low risk of violence.

Future directions

There is sufficient uncertainty about the effectiveness of CTOs to warrant further RCTs. Trials should have adequate sample sizes with minimal exclusion criteria and compare patients on CTOs with control patients receiving entirely voluntary care, as opposed to compulsory treatment of any form. Multiple outcomes should be assessed — not simply hospitalisation — and the optimum length of treatment should also be evaluated. Given the difficulties of achieving a balance between intervention and control arms, as well as the complexities of obtaining informed voluntary consent, we also need complementary well conducted, large-scale, quasi-experimental and naturalistic studies with rigorous multivariable statistical controls. In addition, while individuals on CTOs feel coerced, there is less understanding about the relative importance of this coercion vis-a-vis other negative outcomes.8

Conclusions

CTOs have become standard practice in Australia without questioning of the wide variations in their use or their research base. When evidence of clinical effectiveness is unclear, health policy is more likely to be shaped by political and social factors. Given several Australian Mental Health Acts are under review, the issue is too important to be uninformed by reliable data.

Let children cry

To the Editor: We believe that Jureidini1 challenges one of the fundamental ethical principles of medicine; namely, a doctor’s duty to relieve suffering whenever possible. While we willingly accept this principle when dealing with physical pain, why should it be called into question when dealing with mental suffering? For young people and families directly affected by suicidal behaviour, social isolation, exclusion from education and employment, and worsening mental and physical health, a staged approach to intervention is entirely appropriate.2

We now have extensive evidence from epidemiological and clinical studies that mental disorders affect more than 50% of young people during their transition to adulthood.3,4 When clinically significant, these disorders result directly in premature death, widespread disability, failure to reach potential, and huge economic costs.2

In Australia, headspace services2 and other online psychological and social support options (eg, http://www.eheadspace.org.au; http://www.youthbeyondblue.com; http://au.reachout.com) ensure that distressed young people can access measured and appropriately targeted clinical assessment and psychosocial care. headspace services prioritise strengths-based approaches for young people and provide psychological and behavioural interventions that focus on building coping skills, resilience and a healthy lifestyle. Indeed, the majority of headspace clients receive these interventions rather than psychotropic medication.5 Newly available data highlight the functional gains associated with these approaches.6,7

Let children cry

In reply: Doctors have a duty to relieve suffering, but with qualifications: the target is overall, not immediate, suffering; and the primary injunction is to do no harm.

McGorry and colleagues would presumably agree that interfering with healthy mourning does more harm than good, even if it lessens immediate suffering. Where we disagree is that I have faith in families’ own resources to deal with a broader range of distress, while McGorry and colleagues claim there are benefits from pre-emptively attracting distressed individuals into the mental health system. Evidence needs to be provided to support the idea that medical intervention does more good than harm for less than severe impairment. Too often, selective or exaggerated evidence is offered.1

While it is true that the majority of headspace clients at one site received interventions other than psychotropic medication,2 nevertheless, 20% (168/858) of those who fell short of threshold diagnosis were medicated. De-identified headspace data should be made accessible to allow independent research groups to analyse headspace‘s impact on disability and functioning.

Where are general practitioners when disaster strikes?

GPs, inevitably involved in disasters, should be appropriately engaged in preparedness, response and recovery systems

In the past two decades it is estimated that Australians have experienced 1.5 million disaster exposures to natural disasters alone.1 General practitioners are a widely dispersed, inevitably involved medical resource who have the capacity to deal with both emergency need and long-term disaster-related health concerns. Despite the high likelihood of spontaneous involvement, formal systems of disaster response do not systematically include GPs.

An Australian Government review of the national health sector response to pandemic (H1N1) 2009 influenza suggested: “General practice had a larger role than had been considered in planning”.2 It commented that “structures . . . in place to liaise with, support and provide information to GPs were not well developed”; personal protective equipment provision to GPs was “a significant issue”; and planned administration of vaccinations through mass vaccination clinics was instead administered through GP surgeries.2

GPs are well positioned to help

As of the financial year 2013–14, Australia had 32 401 GPs,3 distributed through rural and urban communities. GPs are onsite with local knowledge when disaster affects their communities. External assistance may be delayed, and the local doctor may be integral in initial community response and feel compelled to act, yet have a poorly defined role.

GPs can identify vulnerable community members, and are situated in local medical infrastructure with medical resources. When other agencies withdraw in the months after disaster, GPs remain, providing continuity of care, which is likely to be important at this time of high distress and medical need (Box 1). Primary health care during extreme events can support preparedness, response and recovery, with the potential to improve health outcomes.4 The challenge lies in linking GPs with the existing medical assistance response.

Australian GPs’ experience of responding to disasters

Australian GPs have a strong sense of responsibility and moral obligation to their patients. They have spontaneously demonstrated willingness and capacity to respond in recent disasters, including the 2011 Australian floods, the 2009 pandemic influenza, and recent bushfires. In interviews with 60 Tasmanian GPs, 100% of GPs surveyed intended to contribute to patient care in the event of a pandemic, with expression of a strong sense that to do otherwise was unethical, although this was dependent on provision of appropriate personal protective equipment.5

What is lacking is consistent support for GPs, their families and their practices. Local GPs may be personally affected and immersed in the disaster, or experience repetitive exposure to their patients’ trauma. Changes in patient presentations, workload, income and working conditions create additional stress, particularly if compounded by personal loss or injury.6 GPs involved in ad hoc spontaneous response may experience uncertainty of their role or efficacy, reluctance to stand down, or may prefer no involvement. GPs interviewed after the 2011 Christchurch earthquake noted experiencing “emotional exhaustion” and physical fatigue; some were aware of the need for personal care at the time, and others only in retrospect.6

Principles of disaster management

The principles of disaster management follow the internationally accepted all-hazards, all-agencies approach through the phases of prevention, preparedness, response and recovery (PPRR).7 Despite the variation in GP roles due to practice locations and context, the GP role in disaster management is most evident across the time frames of PPRR. As shown in Box 1, GPs provide continuity of care across these periods, but with the least consistency in the response phase.

Preparedness

Our discussions with key GP groups and leaders in the field suggest that despite a rapid increase in the number of practices engaging in disaster planning over the past year, most GPs are currently underprepared for disasters (Box 2). Lack of preparedness increases vulnerability. To redress this global problem, the World Medical Association recommends disaster medicine training for medical students and postgraduates. This could include education on existing disaster response systems, mass casualty triage skills, psychological first aid and the epidemiology of disaster morbidity in the first instance.

Response

In the response phase, it is important that GPs are aware of the overarching plan following the incident management system that coordinates multiple disciplines (including fire, police, ambulance and health) to respond to all types of emergencies, from natural disasters to terrorism. With this in mind, roles for GPs have previously included accepting patients from a neighbouring affected practice, assisting at other practices or with surges in hospital emergency department presentations and at GP after-hours services, or keeping patients out of hospitals through “hospital in the home” services. It may involve providing prescriptions and medical treatment in an evacuation centre, being included in medical teams such as St John Ambulance or identifying more vulnerable patients for evacuation assistance. Most importantly, GPs should maintain usual practice activities where possible. These response models are aligned with the range of GP skills and have clear operational requirements.

Recovery

GP involvement is imperative in the recovery phase, ensuring continuity of physical and psychosocial health care during the ensuing months to years. While most patients recover with minimal assistance, it is crucial that individuals in need of increased support are recognised, particularly those with pre-existing chronic disease. Some presentations may be related to particular hazards, eg, smoke inhalation after bushfire, but many others are risks regardless of the hazard. These include increased substance use, anxiety, depression, acute or post-traumatic stress disorder, chronic disease deterioration, and the emergence of new conditions, including hypertension, ischaemic heart disease and respiratory conditions.8 Children are particularly vulnerable, and changes in behaviour or school performance may indicate residual problems.

Support from general practice organisations (GPOs)

During the 2009 Victorian bushfires, Divisions of General Practice provided strong support to enable general practices affected by the fires to continue to offer health care, by providing human and material resources, skills training, advocacy and media liaison. During the 2013 New South Wales bushfires, there was strong GP linkage by the Nepean-Blue Mountains Medicare Local to existing systems through the Nepean Blue Mountains Local Health District and the state health emergency operations centre, as well as to GPOs at a state level. Lessons learnt need to be incorporated into systems planning.

The need for unified disaster planning is increasingly recognised at both individual GP and GPO levels. The General Practice Roundtable, with input from all the major GPOs, has diverse GP representation, providing an opportunity for broad input into disaster planning across PPRR. Important recent initiatives by GPOs include position statements for GPs,9 and ongoing development of disaster resources, promotion of general practice disaster planning, and the recent formation of a national Disaster Management Special Interest Group within the Royal Australian College of General Practitioners.

Where to from here?

Disasters are devastating events and by nature are unpredictable. While recognising and acknowledging the critical role of the formal emergency response agencies in the existing system of specialised health response and management, the strength of general practice lies in the provision of comprehensive continuity of care, and this lends itself to greatest involvement in the preparedness and recovery phases. There is a need for a clear definition of roles in the response stage. GPs as local medical providers in disaster-affected communities need to be systematically integrated into the existing stages of PPRR with clear responsibilities, lines of communication, and support from GPOs, avoiding duplication of other responders’ tasks. Valuing and using the expertise and resources that GPs can bring to disasters may improve long-term patient and community health outcomes.

1 Current defined roles for general practitioners in disasters

2 Potential roles for general practitioners and GP-related groups in disasters

Prevention and preparedness — before the disaster

national position on the role of GPs in disasters across PPRR;
clearly defined roles that integrate with other responding agencies;
GPO representation on national, state and local disaster management committees;
unified disaster planning across GPOs through the GPRT;
information for other agencies on GPs’ skills and roles through the GPRT and GPOs;
education and training in core aspects of disaster medicine for GPs and medical students;
involvement of local GPs in local disaster planning and exercises through ML or PHN;
general practice business continuity and disaster response practice planning;
assisting patient preparedness to reduce vulnerability;
GP personal and family preparedness; and
vaccination, infection control measures and surveillance in infectious events.

Response — during the disaster

representation in EOCs for communication and coordination with other responders (including ambulance, mental health, public health, etc);
unified disaster response from GPOs, including information, resources and phone support;
coordination through GP networks for workforce support for affected practices;
clearly defined integrated roles in existing systems for GPs involved in response, such as:
- maintaining usual practice activities where possible to help surge capacity
- expanding practice capacity to treat extra patients if needed
- expanded use of practice infrastructure, medical resources and trained staff as appropriate
- supporting existing medical teams such as St John Ambulance
- assisting at the scene, evacuation centre or local clinic as appropriate;
assistance in identification of potentially vulnerable and at-risk individuals and families;
ongoing communication with and referral between other local primary care health providers;
patient education on hazard-related health matters, eg, asbestos, infectious outbreaks, etc;
preventive vaccination — tetanus (clean-up injuries); and
surveillance for future outbreaks and emerging community disease threats.

Recovery — after the disaster

inclusion in the review process to improve future PPRR;2
representation on recovery committees to improve interagency referral and communication;
ongoing support from GPOs for affected GPs and staff through regular contact and resources;
GPOs and ML or PHN support for those practices that are more affected;
management of deterioration of pre-existing physical and mental health conditions;
surveillance for new physical and psychological conditions to improve patient outcomes;
surveillance for emerging community disease threats; and
linkage and communication with community groups and allied health on recovery activities.

EOC = emergency operations centre. GPO = general practice organisations. GPRT = General Practice Roundtable. ML = Medicare Locals. PHN = Primary Health Networks. PPRR = prevention, preparedness, response and recovery.

A prospective cohort study of trends in self-poisoning, Newcastle, Australia, 1987–2012: plus ça change, plus c’est la même chose

Intentional poisoning is a major public health problem and generally occurs in the context of deliberate self-harm and drug misuse. There are 60 International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10) codes for drug-related deaths. In 2009, these codes together accounted for 6.4% of male and 5.5% of female total years of potential life lost in Australia.1 Most deaths are in young people, and drug-related deaths account for a large proportion of lost years of life — causing about 25% of completed suicides.1 The estimated Australian rate of people hospitalised for self-poisoning of 119 per 100 000 population per year2 substantially underestimates total numbers as many patients are not admitted or do not present.3 Most poisonings are in young adults and are impulsive or unplanned. Morbidity and mortality from poisoning has proved surprisingly responsive to targeted public health interventions to reduce the availability of means to poison oneself accidently or deliberately.4,5 Identification of drugs causing disproportionate numbers of poisonings, morbidity or deaths is thus a key aspect of an effective toxicovigilance system.

We aimed to examine inhospital morbidity and mortality associated with poisoning in the greater Newcastle region over 26 years and to broadly determine what factors have (and have not) changed over this time, during which there have been substantial changes in medication use. In particular, the use of psychotropic drugs has changed and there have been large increases in antidepressant prescribing in Australia over the past three decades.6,7 Favourable and unfavourable effects on population suicide rates have been postulated for antidepressants.8,9 Thus we assessed the effect of the increase in antidepressant prescribing on total and antidepressant self-poisoning, and examined changes in prescribing of and self-poisoning with several drugs that have previously been identified as having higher relative toxicity in overdose: short-acting barbiturates, dextropropoxyphene, chloral hydrate, dothiepin, thioridazine, pheniramine, temazepam, amisulpride, alprazolam, venlafaxine and citalopram.10–15

Methods

This is a cohort study of patients presenting consecutively after self-poisoning to the Hunter Area Toxicology Service (HATS) between January 1987 and December 2012. Since 1987, HATS has provided a comprehensive 24 hours/day toxicology treatment service for a population of about 500 000. From 1992, ambulances diverted all poisoning presentations in the lower Hunter to this service. HATS currently has direct clinical responsibility for all poisoned adult patients in all hospitals in the greater Newcastle region and provides a tertiary referral service to Maitland and the Hunter Valley. HATS routinely records data on patients who present to hospital (even if the poisoning is uncomplicated).16 Previous studies on poisoning in Newcastle17,18 have shown that no patients were treated exclusively in private hospitals or by their family doctor, indicating that most presentations to medical care facilities are recorded. This cohort does not comprehensively cover unintentional childhood (age < 14 years) poisonings. The local human research ethics committee has previously granted an exemption regarding use of the database and patient information for research.

A structured data collection form is used by HATS to prospectively capture information on patient demographics (age, sex, postcode), drugs ingested (including doses), co-ingested substances, regular medications and management and complications of poisoning.19 At discharge, further information is collected (eg, hospital length of stay [LOS], psychiatric and substance misuse diagnoses). Data are routinely entered into a fully relational Microsoft Access database separate to the hospital’s main medical record system. Data on all patients aged ≥ 14 years who presented following self-poisoning were analysed.

Analyses of population-referenced data (ie, rates) were restricted to postcodes that predominantly cover Newcastle, Lake Macquarie and Port Stephens. Changes in total self-poisoning rates in the four statistical subdivisions in this area were examined between 1991 and 2011. Changes in rates of self-poisoning using the main antidepressant drug classes (tricyclic antidepressants [TCAs], selective serotonin reuptake inhibitors [SSRIs], serotonin–noradrenaline reuptake inhibitors [SNRIs], monoamine oxidase inhibitors [MAOIs] and other) were also examined. Data on rates of antidepressant drug use in these drug classes (standardised by the defined daily dose [DDD]) in Australia from 1991 to 2011 were taken from Australian government publications. We have previously shown that these data agreed within two significant digits with Newcastle-specific data for a range of medications.11

Results

Over the study period, there were 17 266 admissions of patients who had self-poisoned and 11 049 individual patients; the median number of admissions per patient was one (range, 1–115). The number of admissions increased over the first 8 years, but since 1995 has been quite stable (HATS became well established in 1994) (Appendix 1).

Of the total admissions, 15 327 (88.8%) were attempts at self-harm and the remainder were a mixture of unintentional, iatrogenic and recreational self-poisonings. (Data are generally presented for admissions, and may thus include the same patient with different poisonings.) The median age of admitted patients was 32 years (range, 14–97 years) and the female : male ratio was 1.6 : 1 (10 514 female, 6711 male and 39 transgender patient admissions). The median LOS was 16 hours (interquartile range, 9.3–25.7 hours; total time spent in hospital, 15 688 hours). Of the total admissions, 2101 involved admission of the patient to an intensive care unit (ICU) (12.2%) and 1281 involved ventilation of the patient (7.4%). There were 78 inpatient deaths (0.45% of admissions).

We investigated the changes in morbidity and mortality over the study period by dividing admissions into those in the first 6 years, reported previously,20 and four subsequent 5-year periods (Box 1). Over this period, the rate of admission to ICU dropped from 19.2% (376/1955) to 6.9% (280/4060) and rate of mechanical ventilation from 13.7% (268/1955) to 4.8% (193/4060). The fatality rate dropped from 0.77% (15/1955) to 0.17% (7/4060). The median LOS decreased from 20.5 hours in the first 6 years to about 16 hours in all subsequent 5-year periods.

In the 17 266 admissions, a previous history of psychiatric illness (9692, 56.1%), previous admission for a psychiatric episode (6426, 37.2%), previous suicide attempt (9665, 56.0%) and history of alcohol or drug misuse (8466, 49.0%) were commonly recorded. Few admitted patients were in full-time paid work (2421, 14.0%); some were unemployed (3622, 21.0%), pensioners or retired (4163, 24.1%), students (1058, 6.1%), doing home duties (920, 5.3%) or other (703, 4.1%); data were missing for the remainder (4379, 25.4%). Only 23.8% (4104) were married; others were single (9567, 55.4%), separated (1325, 7.7%), divorced (1302, 7.5%), widowed (409, 2.4%), in de facto relationships (79, 0.5%), other (13, 0.1%); and data were missing for the remainder (467, 2.7%). These demographic, social and psychiatric factors remained stable, except for a slight rise in the proportion of patients reporting previous self-harm (Appendix 2).

Including co-ingested alcohol, 34 342 substances were involved (mean, 1.99 per patient; range, 1–18 per patient). The major groups of agents involved in self-poisonings and ICU admissions are shown in Appendix 3 (drugs usually available on prescription) and Appendix 4 (non-prescription drugs and other substances). The most commonly ingested substances were benzodiazepines (5470, 15.9%), alcohol (5461, 15.9%), paracetamol (4619, 13.5%), antidepressants (4477, 13.0%), antipsychotics (3180, 9.3%), anticonvulsants (1514, 4.4%), opioids (1232, 3.6%), non-steroidal anti-inflammatory drugs (1104, 3.2%) and antihistamines (743, 2.2%). Prescription items accounted for 18 950 agents (55.2%), of which 14 445 (76.2%) were known to have been prescribed for the patient.

There were major changes over time in the patterns of drugs ingested (Box 2, Box 3), especially for psychotropic drugs and sedatives. Psychotropic drugs consistently accounted for about 50% of all drugs ingested but newer antidepressants and atypical antipsychotics have largely replaced the older drugs (TCAs and conventional antipsychotics). Several of the drug classes for which frequency of ingestion declined (eg, barbiturates, theophylline, TCAs) have disproportionately high toxicity (Appendix 3). In some drug classes, there were larger declines in individual drugs identified as having greater toxicity in the mid 1990s10–13 (Appendix 5). This is only partly explained by falling prescriptions for these agents (Appendix 6).

There was a more than sixfold increase in antidepressant DDDs per 1000 people per day between 1991 and 2010 (from 12 to 77). However, the increase in the proportion of poisonings due to antidepressants was very modest (about 1.34-fold). There were no corresponding changes in the population rates of self-poisoning (Box 4, Appendix 6), which fluctuated around the long-term mean in each district. There was thus a large decrease in the rate of self-poisoning per DDD prescribed per 1000 population per day for antidepressants (Appendix 7). In Box 4, the increase in total antidepressant prescriptions is illustrated by the total shaded area and changes in individual classes are illustrated by the coloured shading.

Discussion

The rates of self-poisonings in Newcastle were stable over the past two decades, and the features of the population presenting with self-poisoning were constant. This suggests a long-term ongoing and reasonably predictable need for clinical toxicology treatment and ancillary psychiatric and drug and alcohol support services. Despite large increases in prescriptions for drugs used to treat psychiatric illness (and a range of other major mental health interventions), there appears to have been no positive result in terms of reducing episodes of self-harm.

Interestingly, there was a more than sixfold increase in the use of antidepressants, and while the agents taken in overdose changed substantially, there were only small changes in rates of antidepressant overdoses. Interpreting this surprising finding is not straightforward. It probably indicates that antidepressants are increasingly being prescribed for patients who have minimal risk of self-harm. Reassuringly, there is no evidence in our population to support concerns about pro-suicidal effects of new antidepressant prescriptions. The lack of any change in overall self-harm rates also suggests that increased antidepressant use for depression is not an effective public health strategy to reduce rates of self-harm. The only strategy to prevent fatal poisoning with consistent supporting evidence is restricting the availability of high-lethality methods.4,5

Identification of high toxicity in overdose is a problem that can only be studied after approval for marketing is granted. Postmarketing surveillance by pharmaceutical companies of toxicity in overdose is not a requirement for drug registration in Australia or any other country. There is little incentive for voluntary surveillance. Although most companies record case reports of overdoses of their drugs, this does not facilitate comparisons. Reporting biases mean that such cases may be atypical of the usual clinical picture.

The many new psychiatric medications coming onto the market should mandate coordinated collection of timely information on self-poisoning and suicide. Ideally, this should be done at three levels. First, a national coronial register of drug-related deaths is essential to enable an analysis of relative mortality (as done in the United Kingdom21). Second, data on poisonings reported to poison centres are essential, particularly for childhood poisonings that rarely require admission and for assessing the effect of primary prevention measures. Poison centre data have limitations due to referral bias, lack of uniformity of assessment and lack of clinical information.22 Third, for these reasons, systematic use of clinical databases to record hospital admissions for cases of poisoning is needed to measure relative clinical toxicity.

The HATS clinical database has identified disproportionate effects in overdoses with many drugs. Translation of HATS data into clinical risk assessment and guidelines has occurred, but with lengthy delays. For example, the risk of QT prolongation and torsades de pointes with thioridazine, citalopram and escitalopram10,14,23 were detected in the database 5–10 years after these drugs became available. However, to reduce the time to identify toxicological problems to 1–2 years, collaborating centres in Australia and overseas will be needed to accelerate collection of data on self-poisoning.

Most deaths due to poisoning occur outside hospital. Any significant decrease in mortality from self-poisoning will result from primary or secondary prevention. Efforts at decreasing morbidity and mortality from self-poisoning should continue to target drugs that are frequently taken or are lethal in overdose.

The falls in prescriptions and poisonings with several drugs with greater relative toxicity occurred several years after the problems relating to overdoses with these drugs were identified in the HATS database around 1994–1995 (Appendix 6, Appendix 8). For example, large reductions were due to withdrawal of drug subsidies (those for dextropropoxyphene and thioridazine were withdrawn in 2000) and removal of a formulation that was misused (temazepam gel-filled capsules were withdrawn in 2001). Publication of toxicity data alone had limited (if any) effect in terms of reducing prescriptions of the more toxic drugs. No drugs were banned in Australia on the basis of HATS data, although manufacturers did voluntarily withdraw some of the highlighted drugs (barbiturates and chloral hydrate were withdrawn in 1994–1995, thioridazine was withdrawn in 2009). Further, there has been no drop yet in prescriptions of or poisoning with drugs identified as having greater toxicity in the mid 2000s14,15 (data not shown).

Our data show large drops in rates of poisoning with some of the more lethal drugs, such as TCAs and barbiturates (Box 2, Box 4). The introduction of less toxic antidepressants and sedatives dramatically changed prescribing, which in turn changed the types of drugs taken in self-poisoning. This trend has presumably also been reflected in changes in drug-related deaths. However, the nature of coding in official death statistics means that there are no published Australian data to support this contention. For example, poisoning by “antiepileptic, sedative-hypnotic, antiparkinsonism and psychotropic drugs” are all lumped together under one code in Australia.1 Most fatal poisonings are classified as due to unspecified or multiple agents. Some improvement in coding of drug-related deaths should not be difficult. Much finer detail is provided in other ICD-10 codes. For example, the Australian Bureau of Statistics records deaths from crocodile and rat bites (three and zero, respectively, between 1999 and 2008) separately from those caused by other animal bites.1 The development of the National Coronial Information System, launched in 2000, may make fatal poisoning comparisons possible in future, but only if data are accurately and consistently coded. Assessing the impact that clinical toxicology has on direct patient management and on public health is hindered by the lack of reliable epidemiological and clinical data.

Our data have inherent limitations. There is likely to be selection bias against less severe poisonings (the types of cases where patients might not present for medical attention) and rapidly lethal poisonings (cases in which patients die outside hospital). In surveys of self-harm and anecdotal reports from patients, it has been estimated that a significant proportion of people who have self-poisoned (5%–15%) don’t present for medical care.3 Also, although there is a prospective data collection form, retrospective review of medical records is often required to complement prospectively collected data. Data on the ingested drugs were based on patient history, including corroborating history obtained from ambulance officers and accompanying people and from information on drug containers. Drug concentrations were not measured in most patients, although previous research on specific drugs has found the patient history of drugs ingested to generally be confirmed by an appropriate assay.24

The key strengths of our study are its long duration and the consistent core data fields. There are few similar attempts to gather data longitudinally on self-poisoning over prolonged periods. Many have retrospective identification of cases from hospital coding and thus rely entirely on the completeness of medical records.25 Others have not been collected continuously26 or have focused on psychiatric factors and treatment (rather than drugs ingested and toxicity)27 or were conducted in developing countries where agents ingested differ substantially from those used in Australia.28

However, the uniqueness of the HATS database also highlights a weakness of our study. As there are no comparable current datasets, it is difficult to determine the extent to which the Hunter experience represents that of the developed world. Expansion of the database could be facilitated by database systems integrated with electronic medical records.

We identified interesting and important patterns relating to drug prescriptions, epidemiology of overdose patients and importance of relative toxicity. A massive increase in antidepressant prescriptions has had little impact on rates of self-harm or antidepressant poisoning. Changes in antidepressant classes (generally from more to less toxic) have had significant effects on morbidity and mortality from antidepressant poisoning, and therefore from all poisonings. We were also able to generate information regarding relative toxicity and patient management. However, for many rare poisons, gaining sufficient numbers of patients to generate reliable information about management and prognosis will take decades from one centre. We believe that, in Australia and overseas, there is a need for a coordinated approach to address the toxicity of drugs in overdose. The public health benefits would greatly outweigh the modest costs of enhancing postmarketing surveillance through more widespread systematic collection of poisoning and overdose data.

1 Morbidity and mortality due to self-poisoning, 1987–2012, measured by need for intensive care and ventilation, length of stay and fatality*

* Data were divided into five periods with roughly equal patient numbers.

2 Use of sedatives and psychotropic drug classes in self-poisoning, 1987–2012

3 Use of alcohol, analgesics and selected other drug classes in self-poisoning, 1987–2012

4 Rates of antidepressant prescribing (1990–2011, shaded areas) and total rates of self-poisoning (1992–2011, solid lines and symbols)*

* Total numbers of admissions for self-poisoning before 1993 are slightly lower as the Hunter Area Toxicology Service was just being established. Dotted lines indicate 20-year mean for each district.

Firearms, mental illness, dementia and the clinician

To the Editor: In their recent article in the Journal, Wand and colleagues suggest that the medical profession should play a more active role in the regulation of firearm licences held by older Australians.1 However, the authors underestimate the rate of firearm ownership in Australia by a factor of 1000 when they state that 3.9 per 100 000 people held a firearm license in 2001. In reality, about three-quarters of a million Australians held a firearm licence in 2001.2

While the reported vignettes seem compelling enough, the authors’ recommendations need some scrutiny. Almost 15% of the population are aged over 65 years, yet these older people commit about 3% of the roughly 250 homicides per year. 3,4 Further, only about 15% of Australian homicides involve a gun.3 Hence, the potential number of lives saved by the measures they suggest can only be tiny.

In contrast, the downside of their recommendations might be significant. First, obligations on doctors to play a more active role in firearm ownership might deter some patients from seeking medical care. Second, even if people were not deterred from seeking health care, more active involvement by doctors in firearm regulation would come at the opportunity cost of ordinary medical care — care that could be focused on common and lethal medical conditions.

Firearm control in Australia has been singularly successful. While it may be the case that firearm regulations should be tightened, this is not really the responsibility of the medical profession, nor is it fair to focus on older Australians.

Firearms, mental illness, dementia and the clinican

In reply: There was a transcription error in our article.1 The rate of licensed firearm ownership in Australia in 2001 was indeed 3.9 per 100 people,2 although this is likely to be an underestimate, as unregistered, unlicensed and illegal firearms are not captured by official statistics.

Although the overall rate of homicide by firearm owners is low, we argue that the stakes are high. Other potential adverse outcomes of a person who lacks the capacity to safely handle firearms continuing to have a firearm include accidental injury and suicide.

We acknowledged the ethical implications of doctors having a role in assessing suitability for firearm licences.1 However, there is already an expectation that doctors should notify police when concerns about risk to the community or individuals arise from a patient’s access to firearms.3 Risk assessment alone is inadequate, but doctors better meet their obligations when risk assessment is combined with capacity assessment.

Older adults are more likely to have complex cognitive and physical comorbid conditions that affect their ability to safely use a firearm. Screening is important, and doctors will use their clinical judgement to identify patients who may need a closer examination of their capacity in relation to firearm access.

Firearms, mental illness, dementia and the clinician

Correction

Incorrect rate of licensed firearm owners: In “Firearms, mental illness, dementia and the clinician” in the 15 December 2014 issue of the Journal (https://www.mja.com.au/journal/2014/201/11/firearms-mental-illness-dementia-and-clinician), the rate of licensed firearm owners was incorrectly reported as 3.9/100 000 rather than the correct rate of 3.96/100 population.

The journey from moral inferiority to post-traumatic stress disorder

What has been learned over the past century about the psychological injuries of war?

One of the striking differences between recent conflicts and the conflagrations of the past century is that the number of physical casualties has been drastically reduced, a change that has placed greater focus on the psychological costs of war. To place the enormity of earlier losses into context: around 14% of those who served in World War I died (over 60 000 Australians), and a further 40% (more than 156 000 men and women) were wounded, gassed or taken prisoner — at a time when the nation’s population was only 4.9 million.1

Those who survived the Great War were regarded as fortunate, even if they returned home with crippling injuries. There was often little empathy for the psychological wounds of the veterans, construed by many as reflecting moral inferiority, compensation-seeking or “poor seed”. In particular, there was considerable debate within the medical profession as to whether the traumatic neurosis of war — shell shock — was organic or psychogenic in origin.2

Recognition of post-traumatic stress disorder

Veterans of the Vietnam War confronted the medical establishment in the United States and Australia about the lack of understanding for their mental suffering shown by veterans’ affairs officials. They demanded that quality scientific research be undertaken to better characterise the nature and cause of their psychological injuries, as well as to facilitate the development of effective treatments. Independent research, including the seminal US National Vietnam Veterans’ Readjustment Study (NVVRS), showed that post-traumatic stress disorder (PTSD) was not explained by a desire for compensation, but was caused by the cumulative burden of exposure to the horrors and privations of combat.3–5 The NVVRS found that the lifetime prevalence of PTSD in Vietnam War veterans was 18.7%.3

It is ironic that “traumatic neurosis” (an earlier name for PTSD) had been well characterised as early as 1890,6 but social and political factors had subsequently prevented its being systematically investigated. Over the past three decades, much has been learned by a generation of high-quality research that has used epidemiological tools to assess the aftermath of both combat and peacekeeping missions.5 The mental health sequelae of major deployments since 1990 have been systematically studied, and the mental health of the entire Australian Defence Force (ADF) was documented in 2010.7 The findings of these investigations highlight that exposure to human suffering, the risk of death and active involvement in combat are not only the major risk factors for PTSD, but also for major depressive disorder and other anxiety disorders.4,5,7 Furthermore, the similarities of traumatic exposures during modern peacekeeping operations are often greater than the differences experienced during deployment in a declared combat zone; the rates of psychiatric disorder in veterans following the two types of deployment are therefore quite similar.7

Suicidal ideation and suicide are of particular concern in veterans with PTSD, and the number of active duty suicides in the US armed forces over the past decade has caused increasing concern.8

While investigators routinely find substantial levels of psychiatric disorder in former members of the armed forces, the rates in currently serving personnel often do not differ substantially from those for the general community; nor are the rates of psychiatric disorder in deployed and non-deployed military personnel strikingly different, despite traumatic combat experiences. It should be remembered, however, that those who enlist in the defence forces are generally fitter than the population from which they are recruited, and this healthy worker effect is amplified by selecting the more resilient for deployment.7–9

Prevention and support

Selection processes — no matter how rigorous — cannot entirely avert the psychological injuries caused by repeated exposure to traumatic events. Lifetime exposure to traumatic stress (including pre-enlistment events) needs to be taken into account when calculating the risk of PTSD. Deployment rotation cycles that limit the periods of exposure are equally critical for managing risk.

A recent review by the US Institute of Medicine of military programs that aim to reduce the risk of PTSD highlighted the fact that there is little systematic evidence for the effectiveness of the most widely used approaches, such as psychoeducation, emotional decompression, psychological debriefing and resilience training, so that more research is needed.10 Postdeployment screening was the only approach endorsed by the review, and used by the ADF since 1998. Early recognition of PTSD not only enables rapid diagnosis and intervention, but also assists when dealing with other barriers to effective care, such as stigmatisation of the patient.11 Adequate training of medical and mental health staff in understanding the various manifestations of traumatic reactions (not just PTSD) is critical. Medical providers are often unaware of the military service of discharged veterans, so it is also essential that this information is routinely requested. High-risk groups, such as the physically ill and injured, require regular screening and follow-up, as delayed-onset PTSD is now recognised to be much more prevalent than was previously thought, particularly in veterans.11

More research into treatment is needed

The compelling need for better treatment outcomes has led to substantial investment in research programs by departments of defence and veterans’ affairs in the US, the Netherlands, the United Kingdom and Australia.12 Recent developments include the identification of epigenetic markers for PTSD, and recognition of the role of inflammation in its aetiology.13 Mild traumatic brain injury as a risk factor is also being extensively studied in military populations, highlighting the importance of the integrity of neural networks.14

Some effective treatments for PTSD have been comprehensively investigated, including cognitive behaviour therapy, eye movement desensitisation and reprocessing therapy, and antidepressant medication therapies, and it has been found that the effect size of these treatments is often smaller in veterans than in patients with PTSD following single-incident traumas.11 The prolonged hypervigilance and repeated exposure to traumatic experiences that characterise military deployments cause more significant neurobiological disturbances (eg, reduced anterior cingulate cortex volume), and conditioned traumatic triggers are less amenable to extinction.15 This profound neurobiological dysregulation needs to be targeted by treatment, with potential roles for both pharmacological and psychological interventions.15

The battle during World War I between organic and psychogenic models of post-traumatic stress was misguided. PTSD is a multifaceted disorder, in which biological, psychological and social components are entwined in its aetiology, and must therefore be considered in any recovery strategy. An effective way to honour the suffering of those who fought in World War I is to ensure that our care for the current generation of veterans is diligent and informed by independent and adequately funded high-quality science.