Schizophrenia Research Forum - A Catalyst for Creative Thinking

What Drives Violence During Psychosis?

25 June 2013. Two recent studies from the United Kingdom have delved into the relationship between violence and psychosis, examining factors that influence the association. The most recent, led by Simone Ullrich of Queen Mary University of London and published March 6 in JAMA Psychiatry, examined how specific symptoms—delusions—elevate the risk for violence in first-episode psychosis. Anger due to delusions—particularly those concerning persecution, being spied on, and conspiracy—was significantly associated with both minor and serious violence.

An earlier study, published February 13 in PLoS One and led by Seena Fazel of the University of Oxford, performed a meta-analysis of broader risk factors for violence in individuals with psychosis, finding that several criminal history factors were the most strongly associated.

Accurately determining the link between violence and schizophrenia and other psychotic disorders has proved challenging. The majority of studies have found an elevated risk of violence in schizophrenia, although just how much higher the risk is varies widely among studies—from seven times higher to no greater than that of the general population (Fazel et al., 2009).

In addition, many studies suggest that the increased risk of violence in schizophrenia may be due to other factors besides the illness itself. For example, a large Swedish study conducted by Fazel and colleagues reported that the risk of violence in schizophrenia is minimal unless a comorbid diagnosis of substance abuse is also present (see SRF related news story). A meta-analysis by the same group later supported this finding (Fazel et al., 2009). The belief that people with schizophrenia are dangerous is a large contributor to the stigma that patients face (Torrey, 2011). Therefore, identification of the factors that may contribute to the elevated risk for violence among the schizophrenia population has important implications not only for violence prevention, but also for stigma reduction.

Delving into delusions
Many studies have investigated the link between delusions and violence in schizophrenia, finding evidence it may result, at least partially, from feelings of anger (Cheung et al., 1997). The current JAMA Psychiatry study further examined the causal pathway from delusions to violence by searching for mediating factors. Based on evidence that psychosis prior to first treatment is more strongly associated with a risk for committing serious violence than that after treatment (Nielssen and Large, 2010), first author Jeremy Coid and colleagues examined participants with first-episode psychosis. They found that 38 percent of study participants had a history of minor violence within the preceding year, of which 12 percent had engaged in serious violence.

Delusions were assessed using the Maudsley Assessment of Delusions Schedule, which included a measurement of the subject’s affect resulting from delusions. Violence during the preceding year was measured using the MacArthur Community Violence Interview, in which raters judged whether subjects were psychotic at the time of violence. Violent episodes in which the participant was not psychotic, or those in which the subject was judged to be responding to violence against him or her, were excluded.

Coid and colleagues first looked for a relationship between affect related to delusions and violence. They found no associations between violence and elation, fear, or anxiety related to delusions, and report that depressed affect was associated with less violence, both minor and serious, consistent with an earlier report (Dean et al., 2007). In contrast, anger due to delusions was positively associated with both levels of violence and could not be explained by mania symptoms or trait anger. This effect remained significant after adjusting for the effect of variables known to increase risk for violence such as gender and drug use. The risk of violence attributed to anger due to delusions was 31 percent for minor violence and 56 percent for serious violence.

To delve more deeply into the association between anger due to delusions and violence, the researchers next examined a total of 32 specific delusions and four delusional characteristics (bizarreness, systematization, monothematic, and conviction). Of these, six delusions and one delusional characteristic resulted in anger. In addition, six of the 32 delusions were associated with minor violence, but after incorporating the effect of anger due to delusions, one of these (delusions of familiar people being impersonated) became less significant, indicating that it was partially mediated by anger. The authors concluded that the relationship between minor violence and some delusions (such as delusions of pregnancy or a delusional lover) is direct. In contrast, the relationship between delusions of familiar people being impersonated and violence was indirect, a consequence of the subject’s anger resulting from the delusion.

Three delusions were significantly associated with major violence—delusions of being spied on, persecution, and conspiracy—but these were no longer significant when anger due to delusions was taken into account. Thus, the relationship between major violence and delusions appears be indirect and entirely mediated by anger. The results have important treatment implications, say the authors: “If anger due to delusions could be identified and treated, a substantial number of violent incidents could potentially be prevented.”

Meta-analyzing violence risk
In the PLOS One study, Fazel and colleagues took a broader approach to assessing risk factors for violence among individuals with psychosis. First author Katrina Witt and colleagues combed through the literature to meta-analyze 110 studies in over 45,000 adults diagnosed with schizophrenia and other psychoses. Nearly 20 percent of participants were violent, and the majority of all participants (88 percent) had a diagnosis of schizophrenia. The researchers examined 146 risk and protective factors that fell into 10 psychosocial and clinical domains: demographic, premorbid, substance misuse, treatment related, criminal history, suicidality, psychopathology, neuropsychological, positive symptoms, and negative symptoms.

The negative symptom and neuropsychology domains were the only ones not significantly associated with an increased risk of violence. The criminal history domain had the highest odds ratio and was most strongly associated with an increased violence risk. Within this domain, the highest odds ratio was found for a history of assault. Substance abuse was the next highest domain, with a history of abuse of multiple substances being the factor most strongly associated with violence. The demographic and premorbid domains followed, with the top factors in these categories being history of violent victimization during adulthood and physical abuse during childhood, respectively. Notably, many of the risk factors identified (such as hostile behavior, recent drug and alcohol abuse, poor impulse control, lack of insight, and treatment non-adherence) were “dynamic,” or modifiable, and may prove important for violence prevention.—Allison A. Curley.

Coid JW, Ullrich S, Kallis C, Keers R, Barker D, Cowden F, Stamps R. The Relationship Between Delusions and Violence: Findings From the East London First Episode Psychosis Study. JAMA Psychiatry . 2013 Mar 6 ; :1-7. Abstract

Witt K, van Dorn R, Fazel S. Risk factors for violence in psychosis: systematic review and meta-regression analysis of 110 studies. PLoS One . 2013 ; 8(2):e55942. Abstract

Comments on Related News

Related News: Violence in Schizophrenia: Other Risk Factors Matter More Than the Disease

Comment by:  Jan Volavka
Submitted 24 June 2009
Posted 24 June 2009
  I recommend the Primary Papers

The recent study by Fazel et al. (2009) is based on a sample comprising what amounts to be the entire population of schizophrenia patients in Sweden. This is a major strength of the study. The principal outcome of interest was conviction for violent crime. The authors reported odds ratios for violent crime in patients with and without substance abuse comorbidity, compared to a control group.

Schizophrenia patients commit acts of violence in the community and in the hospital. In the United States, arrests of psychiatric inpatients for assaults and other aggressive behaviors that they commit are rare, and prosecution leading to conviction is extremely unlikely. I suspect that the situation in Sweden is not dramatically different in this respect. If this is so, I have some comments.

First, schizophrenia patients generally spend more time in hospitals than members of the general population do. A minimum of two hospitalizations was an eligibility criterion for inclusion in the group of schizophrenia patients in this study. During hospitalization, they were not available to be arrested and convicted. Given these facts, it is likely that the patients had, on average, less time to commit convictable violent crime than members of the comparison group. It appears that the authors did not account for this difference in time exposure in their analyses. If so, the odds ratios they computed were underestimates.

Second, we need to keep in mind that violence by schizophrenia patients in the hospital is common (Volavka, 2002). This is not surprising, since violence is a leading cause prompting hospitalization, and patients are not discharged until the risk for violence is deemed acceptably low. Therefore, if violent patients were not hospitalized, violent behavior by schizophrenia patients in the community would occur at a substantially higher rate. Thus, the odds ratio for conviction for violent crime is kept low by the constant removal from the community to hospitals of the patients who are the most violent at the moment. Furthermore, if violent acts of hospitalized patients, such as causing injuries to others (which would be considered criminal if the patient were in the community), were included in the Fazel et al. analyses, the odds ratio would increase.


Fazel S, Långström N, Hjern A, Grann M, Lichtenstein P. Schizophrenia, substance abuse, and violent crime. JAMA. 2009, May 20; 301(19):2016-2023. Abstract

Volavka J. Neurobiology of Violence. 2002. Washington, DC: American Psychiatric Publishing, Inc.

View all comments by Jan Volavka

Related News: Violence in Schizophrenia: Other Risk Factors Matter More Than the Disease

Comment by:  Mark Serper
Submitted 30 June 2009
Posted 1 July 2009
  I recommend the Primary Papers

Related News: Violence in Schizophrenia: Other Risk Factors Matter More Than the Disease

Comment by:  Seena Fazel
Submitted 16 July 2009
Posted 16 July 2009

Comment by Seena Fazel, Martin Grann, and Niklas Långström
We are grateful for the interest and comments that our paper has attracted. We would like to take this opportunity to respond to some of these comments.

First, Dr. Volavka raises the potentially important issue of time at risk and whether the patients had less time at risk than the controls. In Sweden, any serious offences by hospital inpatients are reported to the police and lead to convictions. This is partly a consequence of the insurance system, which requires police involvement if insurance claims are made by health care workers or other patients. It is possible that minor offending, such as theft, by hospital patients does not lead to conviction, but our study focused on violent offending.

Second, we agree with Dr. Torrey that our comments about treatment for substance abuse comorbidity in patients should not detract from the need for treatment of the underlying schizophrenia. A recent review reinforces Dr. Torrey’s point and suggests that consideration should be given to using second-generation antipsychotics (Wobrock and Soyka, 2008).

Third, Dr. Hodgins states that our study “failed to provide information that could be relevant to improving treatments and services to reduce violence in this population.” However, it is possible to argue that our study does clearly endorse the case for good-quality randomized controlled trials for the treatment of comorbidity in schizophrenia, which have been lacking, according to a recent review (Wobrock and Soyka, 2008). In addition, our study underlines the importance of the assessment and management of comorbidity to reduce violence risk.

Fourth, Dr. Hodgins states that we were “inaccurate” to state that our study was larger than previous studies combined. In fact, we stated that it was larger than all longitudinal studies combined. A systematic review of all studies examining the risk of violence in schizophrenia that clearly demonstrates this is coming out in the August issue of PLoS Medicine (Fazel et al., 2009). Even if you assume (incorrectly) that we said that our study was larger than all previous studies combined, then the number of individuals with schizophrenia in the other 19 studies that we identified from 1980-2009 (longitudinal and other) included 10,420 patients. Our study included 8,003 patients with schizophrenia.

Fifth, Dr. Hodgins makes an interesting point about gender differences. This was not the focus of our paper, which examined primarily familial factors and substance abuse comorbidity over 30 years in patients with schizophrenia in Sweden. However, the review in PLoS Medicine finds an overall non-significant association toward higher risk estimates in female compared with male patients, which may not be quite as strong an effect as Dr. Hodgins suggests.

Finally, we agree with, and acknowledged in our paper as limitations, the comments about lack of information about treatment and a history of conduct disorder. However, including reliable and valid information on these and other more detailed potential risk factors in large population-based studies will remain an ongoing challenge to all research in the field.


Wobrock T, Soyka M. Pharmacotherapy of schizophrenia with comorbid substance use disorder--reviewing the evidence and clinical recommendations. Prog Neuropsychopharmacol Biol Psychiatry. 2008 Aug 1; 32(6):1375-85. Abstract

Fazel S, Gulati G, Linsell L, Geddes JR, Grann M (2009) Schizophrenia and Violence: Systematic Review and Meta-Analysis, PLoS Med 6 (8). Abstract

View all comments by Seena Fazel