AAny homicide is a tragedy, with devastating consequences for all involved, including perpetrators, their families and the community. People with a psychotic illness make a small, steady contribution to homicide figures. Strangers are rarely victims of such events, with family members and others known to the individual at much higher risk.1,2 Other than the impact of psychotic symptoms themselves, the risk factors for violence by people with psychotic illness are similar to those for the rest of the population, and include being young and male, having a background of crime and/or violence, substance misuse, personality disorder, brain injury, and socioeconomic deprivation.3

This issue of the Journal features a case series by Nielssen and colleagues of 10 years of homicides committed by people whose acute psychotic illnesses were considered by the authors to be largely responsible for their crimes (→ Homicide during psychotic illness in New South Wales between 1993 and 2002).4 The authors make two key observations. Firstly, they note that evolving persecutory symptoms in which the subject perceives an immediate threat were highly correlated with homicidal behaviour in their sample. While caution must be exercised in generalising the findings from an uncontrolled series of homicides, selected because their illness was felt to be responsible for their crime, persecutory psychotic symptoms have been repeatedly but not consistently implicated in severe violence in other more rigorous studies.5-7 At the same time, it is clear that most people who experience such symptoms do not act on them.5 It is likely to be a complex and largely unpredictable interaction of risk factors and circumstances that bring about the final tragic outcome.

Secondly, in this series, 69% of all homicides occured during the first year of the perpetrator’s illness. The authors raise the possibility that this may reflect inadequate detection and treatment. Although highly plausible, it is not clear from their study how many patients were in treatment at the time of their crime, nor whether that treatment was considered adequate. Other explanations are possible. From a population perspective, the peak age at which people commit homicide and have their first psychotic episode largely overlap.8,9 Therefore, lower homicide rates would be expected in those who are older. Also, family members are likely to be most vulnerable to becoming victims of homicide during a first psychotic episode because they are less attuned to indicators of risk. Further research is needed to clarify the nature of the observed elevated early risk of homicide, but, as the authors conclude, those in the first year of illness are worthy of particular attention regarding risk of violence.

Is it possible to predict or prevent homicide by people with a psychotic illness? The accurate prediction of rare events is inherently problematic, with unacceptably low specificity and sensitivity.10 Nielssen and colleagues suggest that: “. . . many of the deaths might have been prevented if the dangerous symptoms had been identified and there had been assertive intervention”.4 Such a statement may greatly overestimate the capacity of mental health services to predict and prevent homicide in the context of psychosis, described by one prominent author as just “part of the human condition”.10 In retrospect, it is often clear where opportunities for intervention have been missed, but this does not guarantee that prevention was possible. This need not, however, be seen as an excuse for nihilism. Careful monitoring and appropriate intervention remain important.

As well as preventing homicide, it may be rewarding to focus on the reduction of violence. While accurately predicting violence by individuals is statistically almost impossible, with high rates of false positives, it is possible to identify groups of individuals at high risk of committing violence and by extension, homicide.11,12 Paul Mullen, Director of Forensic Mental Health Services for Victoria, believes that only by intervening intensively and assertively with this often very difficult-to-engage group, and addressing the many biopsychosocial factors that mediate violence in addition to acute symptoms, might it be possible to significantly reduce violence by people with schizophrenia.12 His approach would require intensive, multidisciplinary input and a change in attitude by mental health and other services towards this very challenging group. With luck, such a pathway might reduce violence and occasionally even prevent homicide, while also improving the quality of life of many highly vulnerable people and their families and associates along the way.

How then should clinicians proceed when attempting to develop a management plan for a person with a psychotic illness? Clearly, risk assessment necessitates careful enquiry about persecutory symptoms, with an emphasis on the patient’s experience of fear, and consideration of any associated risk to others. Such symptoms need to be considered in the context of other risk factors for violence when predicting risk and developing a management plan.

The burden of risk borne by family members and close associates reinforces the need for close and ongoing family consultation. Perceived risk by family members must be taken very seriously and be embedded in the ensuing management plan, which should include direct advice as to how they should respond to actual threat.

Because threatening and violent behaviours are rare and complex, predicting and preventing serious violence and, by extension, homicide will always be problematic. Despite its limitations, the study by Nielssen and colleagues reminds us of the need to place grave emphasis on persecutory delusions in the mental state examination, particularly during the early stages of psychotic illness, and to listen carefully to the concerns of family, who remain most at risk of violence. Improved multidisciplinary community mental health services, as are advocated in the early intervention psychosis model, if applied uniformly and rigorously, might go some way to improving the chances of preventing homicide.