A national study of violent behavior in persons with schizophrenia
Swanson JW, Swartz MS, Van Dorn RA, Elbogen EB, Wagner HR, Rosenheck RA, et al.;
Commented by , 23 May 2006
Aim
It is well established that violent episodes occur more frequently among people with schizophrenia than among normal controls. Although the absolute increase of risk is small, this situation is still problematic. The risk factors that contribute to violent behaviour (e.g. psychopathology, interpersonal, and social-environmental factors) have, however, not been sufficiently studied.
The aim of the present study, therefore, was to examine the prevalence of violence among patients with schizophrenia who are living in the community and to find risk factors that contribute to its occurrence.
Methods
The study used the population of the National Institute of Mental Health Clinical Antipsychotic Trials of Intervention Effectiveness study (CATIE) as a basis. It was a retrospective investigation using clinical interviews to evaluate violent behaviour in the past 6 months. 1410 patients with schizophrenia from 56 sites in 24 American states were included.
The sites included a variety of settings – university clinics, state mental health agencies, Veteran Affairs Centers, and private sites – to allow for a generalisability of the findings. Two severity levels of violence were coded for:
-
one was minor violence which meant “simple” assault without injury or weapon use
-
the other one was serious violence meaning an assault that led to injury or that involved the use of a dangerous weapon, or sexual assault.
In addition, the authors applied a composite measure of “any violence”. The MacArthur Violence Risk Assessment Interview was used for this purpose. The following risk factors were analysed: social contact and support, overall disease severity (CGI), PANSS psychotic symptoms, PANSS negative symptoms, depression, insight, substance abuse, quality of life and daily functioning. The statistical analysis involved multivariable models using mixed-model logistic regression.
Results
Within 6 months before the assessment 19.1% of the patients reported any violence and 3.6% of the participants reported serious violent behaviour. A number of risk factors which partly overlapped were statistically significantly associated with minor and serious violence: Minor violence was associated with younger age, female sex, lack of vocational activity, restrictive housing, residing with family, not feeling “listened to” by family members, and recent history of police contact.
In terms of psychopathology minor violence was associated with positive symptoms, duration of treatment and substance abuse.
Serious violence was associated with younger age, childhood conduct problems, arrest history, and positive symptoms. High levels of negative symptoms, however, decreased the risk.
Dr Leucht's comments
A strength of the study certainly is its large sample size, the fact that it was conducted on patients in the community and in multiple centers. I, however, feel that the term “national” study in the title is somewhat misleading. It suggests that the study was population based, but it was not and the patients included were the population of a clinical antipsychotic drug trial.
Furthermore, the retrospective design using interviews is also not perfect. Nevertheless, the occurrence of any violent behaviour of 19,1% within 6 months is high. Some of the risk factors identified such as high level of positive symptoms, recent history of police contact, younger age, restrictive housing and substance abuse can be understood intuitively. Others such as female gender or residing with family are less intuitive.
The authors explain that the female patients were more likely to reside with their families so that there might have been a link between the two factors. The (unusual) increased risk in women maybe a selection effect, since the vast majority of the patients was male. The effect of positive symptoms on the occurrence of violence stresses the crucial role of symptom management of the disorder.