Risk of people with mental illnesses dying by homicide

BMJ 2013; 346 doi: http://dx.doi.org/10.1136/bmj.f1336 (Published 05 March 2013) Cite this as: BMJ 2013;346:f1336
  1. Roger Webb, senior research fellow in psychiatric epidemiology1,
  2. Jenny Shaw, professor of forensic psychiatry1,
  3. Louis Appleby, professor of psychiatry1
  1. 1Centre for Mental Health and Risk, Institute of Brain, Behaviour and Mental Health, University of Manchester, Manchester, UK
  1. roger.webb{at}manchester.ac.uk

Mentally ill people have an increased risk of becoming victims

Crump and colleagues (doi:10.1136/bmj.f557) report on an important topic that has received little attention in the health sciences literature1: the risk that people with mental illness will die by homicide.2 3 4 This contrasts with the large body of published evidence on the risk of homicide and other acts of serious violence perpetrated by people with psychiatric disorders.5 6 Popular media reporting portrays mental illness as posing a threat to the safety of others,7 8 and these continual stigmatising portrayals may make the violent victimisation of an already marginalised section of society more likely.9 These new research findings therefore deserve to be disseminated widely, so that professional groups and agencies working in mental health, as well as the media and general public, are aware that mentally ill people are at increased risk of becoming victims of someone else’s violence.

Scandinavian registers enable investigation of rare adverse events such as homicide. They provide statistically powerful national psychiatric treatment cohorts with linkage to complete cause specific mortality records. Crump and colleagues used the national Swedish registry to add to findings reported more than a decade ago from a Danish registry study, which drew similar conclusions.3

For the first time, the authors assessed people treated both as outpatients and inpatients to generate findings that are more applicable to mental illness in the whole community, and they adjusted for sociodemographic confounders. They found that, compared with people without psychiatric disorders, those diagnosed with any mental illness had an adjusted relative risk of 4.9 of death by homicide. Examining a more representative study cohort enabled the investigators to uncover independent increases in risk in patients with mild to moderate psychiatric diseases, such as depression and anxiety disorders. The effect sizes for these conditions were at least as high as those for schizophrenia.

As with the investigation of Danish data,3 the highest risk was found in people treated for substance misuse disorders (adjusted relative risk 9.4). The authors of that study had suggested that comorbid substance misuse might be a mechanism leading to homicide victimisation among people with mental illness.3 However, by conducting additional analyses among people in the study cohort without recorded substance misuse diagnoses, Crump and colleagues found increased risks with all mental disorders combined (adjusted relative risk 2.1), and an especially high risk with personality disorder (adjusted relative risk 4.6). They were, however, unable to explore other putative mechanisms, including residing in more dangerous neighbourhoods; displaying hostility provoking symptoms such as irritability or paranoia; reduced awareness of personal safety; mixing with people who are prone to perpetrating violence; and being attacked without motive for having an unusual appearance.3 Even so, clinicians should be aware that patients may inadvertently put themselves at risk by the way they behave or the company they keep.

Mental health professionals routinely assess risk for specific adverse events in their patients. Thus, clinicians working in adult mental health services focus mainly on suicide prevention, those in forensic services on reducing violence, and those in child and adolescent services on the impact of poor parenting and dysfunctional families. A key implication of these new findings is that clinicians should assess risk for the full array of adverse outcomes that may befall people with mental health problems. This would include being a victim of violence as well as committing it, abuse and bullying, suicidal behaviour, accidental drug overdoses, and other major adverse events linked with intoxication or impulsivity. These risks go together, and people with mental illness, as well as their families, should receive advice on avoiding various types of harm. National mental health strategies should reflect the broad nature of safety concerns in mental healthcare, while anti-stigma campaigns among the public should aim to counter fear of mentally ill people with sympathy for the risks they face.8 10

Some important questions remain unanswered. Firstly, to our knowledge, no published reports have directly compared patients’ risk of committing homicide with the risk of being a victim of it. Secondly, although Crump and colleagues did not report temporal effects, we know that discharged psychiatric patients and released prisoners are especially likely to take their own lives immediately after their return to the community.11 12 We need to know whether this heightened vulnerability on leaving institutional settings relates to being at risk from other people. Finally, we do not know how much of the risk of being a victim is a consequence of illness itself and therefore the potential for prevention by mental healthcare and treatment.


Cite this as: BMJ 2013;345:f1336


  • Research, doi:10.1136/bmj.f557
  • Competing interests: All authors have completed the ICMJE uniform disclosure form at www.icmje.org/coi_disclosure.pdf (available on request from the corresponding author) and declare: no support from any organisation for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.

  • Provenance and peer review: Commissioned; not externally peer reviewed.