What are the absolute and relative risks of perpetrating violence toward others in individuals with schizophrenia spectrum disorders compared with the general population?
In this systematic review of 24 studies, the absolute risk of perpetrating violence in a subgroup of register-based studies was less than 1 in 20 in women with schizophrenia spectrum disorders and less than 1 in 4 in men with schizophrenia spectrum disorders over a 35-year period. The elevated relative risk for all violence-perpetration outcomes was higher for women with schizophrenia spectrum disorders than for men with schizophrenia spectrum disorders, but with substantial heterogeneity in the findings.
Violence perpetration outcomes may be an important target for prevention and to reduce stigma in people with schizophrenia spectrum disorders.
Violence perpetration outcomes in individuals with schizophrenia spectrum disorders contribute to morbidity and mortality at a population level, disrupt care, and lead to stigma.
To conduct a systematic review and meta-analysis of the risk of perpetrating interpersonal violence in individuals with schizophrenia spectrum disorders compared with general population control individuals.
Multiple databases were searched for studies in any language from January 1970 to March 2021 using the terms violen* or homicid* and psychosis or psychoses or psychotic or schizophren* or schizoaffective or delusional and terms for mental disorders. Bibliographies of included articles were hand searched.
The study included case-control and cohort studies that allowed risks of interpersonal violence perpetration and/or violent criminality in individuals with schizophrenia spectrum disorders to be compared with a general population group without these disorders.
Data Extraction and Synthesis
The study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines and the Meta-analyses of Observational Studies in Epidemiology (MOOSE) proposal. Two reviewers extracted data. Quality was assessed using the Newcastle-Ottawa Quality Assessment Scale. Data were pooled using a random-effects model.
Main Outcomes and Measures
The main outcome was violence to others obtained either through official records, self-report and/or collateral-report, or medical file review and included any physical assault, robbery, sexual offenses, illegal threats or intimidation, and arson.
The meta-analysis included 24 studies of violence perpetration outcomes in 15 countries over 4 decades (N = 51 309 individuals with schizophrenia spectrum disorders; reported mean age of 21 to 54 years at follow-up; of those studies that reported outcomes separately by sex, there were 19 976 male individuals and 14 275 female individuals). There was an increase in risk of violence perpetration in men with schizophrenia and other psychoses (pooled odds ratio [OR], 4.5; 95% CI, 3.6-5.6) with substantial heterogeneity (I2 = 85%; 95% CI, 77-91). The risk was also elevated in women (pooled OR, 10.2; 95% CI, 7.1-14.6), with substantial heterogeneity (I2 = 66%; 95% CI, 31-83). Odds of perpetrating sexual offenses (OR, 5.1; 95% CI, 3.8-6.8) and homicide (OR, 17.7; 95% CI, 13.9-22.6) were also investigated. Three studies found increased relative risks of arson but data were not pooled for this analysis owing to heterogeneity of outcomes. Absolute risks of violence perpetration in register-based studies were less than 1 in 20 in women with schizophrenia spectrum disorders and less than 1 in 4 in men over a 35-year period.
Conclusions and Relevance
This systematic review and meta-analysis found that the risk of perpetrating violent outcomes was increased in individuals with schizophrenia spectrum disorders compared with community control individuals, which has been confirmed in new population-based longitudinal studies and sibling comparison designs.
Schizophrenia spectrum disorders are associated with a range of adverse outcomes. These include suicide,1 premature mortality,2 experiencing violence3 and contact with the criminal justice system,4 and deficits across broader social domains.5 Perpetration of interpersonal violence is also associated with these diagnoses, an outcome with considerable implications for individuals, families, caregivers, those who experience interpersonal violence, and public services. Examining such associations poses particular challenges. Public perceptions around dangerousness in mental illness can contribute to stigma,6 and a focus on violence risk may inadvertently increase this. The contribution of mental illness to violent crime is often overstated, such as in media reports on mass shootings,7 which can bias public perceptions, and such issues have been integral to the appraisal of epidemiological evidence.8 However, failure to acknowledge associations with violence perpetration in turn risks leaving clinical needs unmet in psychiatric services. An accurate and balanced understanding of the nuanced association between mental illness and violence perpetration, development of effective prevention, and balanced communication and contextualization of research evidence may be one approach to reducing stigma.
At a population level, violence is a leading cause of premature mortality and morbidity. Public health approaches to tackling violence focus on prevention and delineating modifiable risk factors, such as mental illness, to inform interventions.9 In schizophrenia spectrum disorders, the public health importance of violence perpetration has been reinforced by recent estimates of its economic impact.10 Updated and precise estimates of violence perpetration risk are needed to inform prevention in health services.
A previous systematic review11 identified 20 primary studies from 11 countries that reported a substantial increase in risk of violence perpetration in individuals with schizophrenia compared with people without schizophrenia, and a doubling of risk with substance misuse comorbidity. However, the search period for this review ended in February 2009. Since then, several large population investigations have been conducted.2,12 These newer studies provide improved approaches to examine confounding, test comorbidities, and allow for examination of additional outcomes. Although other systematic reviews have been conducted,13,14 to our knowledge, no recent quantitative analyses have integrated these new data. This longer-term perspective provides more precision for risk estimates, subgroup analyses, and meta-regressions. The current study therefore provides an updated systematic review and meta-analysis of the risk of violence perpetration in individuals with schizophrenia spectrum disorders compared with general population control individuals.
This study followed the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline (eFigure 1 in the Supplement)15 and the Meta-analysis of Observational Studies in Epidemiology (MOOSE) reporting guideline16 (eTable 1 in the Supplement). The review was registered on PROSPERO (CRD42020224942). Risk of bias was assessed using the Newcastle-Ottawa Quality Assessment Scale.17
In a previous review,11 MEDLINE, Embase, and PsycInfo databases were searched from January 1970 to February 2009. For the update, we searched PubMed, PsycInfo, and Global Health from March 1, 2009, to March 17, 2021. No limits were imposed on country or age range.
Identification of schizophrenia spectrum disorders in included studies was from clinical diagnosis or based on diagnostic criteria. See the eMethods in the Supplement for full eligibility criteria. Applying these criteria, 416 relevant articles were identified from updated searches, 25 of which (reporting on 24 studies) met eligibility criteria.
A standardized form was used to extract data on study design, location, last year of follow-up (study period), diagnostic definition, comparison group (including any use of sibling controls), definition and ascertainment of violence, violence perpetration outcomes in individuals with schizophrenia spectrum disorders and control individuals, outcome period, sample size, mean age, and information on comorbid substance misuse, personality disorder, experiencing violence, and treatment. If separate data were available for a range of psychoses, we chose schizophrenia spectrum disorders where possible, including schizophrenia and nonschizophrenia psychoses (ICD-10 codes F20-F29), and excluding affective, drug-induced, and organic psychoses. Where separate data were available for male and female individuals, these were disaggregated for analysis. See the eMethods in the Supplement for details of absolute risk data extraction. Where applicable, authors were contacted for additional data. Two authors extracted data (D.W., G.G.), and discrepancies were resolved by consensus.
We combined eligible studies from this update with data from the original review11 for analysis to ensure that the current review presents the most comprehensive synthesis of available information.18,19 To avoid duplication, where studies included overlapping populations or where articles were superseded by updated investigations in the same population, we included the study that used a longitudinal design or the study with the largest sample of individuals with schizophrenia spectrum disorders. Applying these methods, 10 studies were excluded from pooled analyses (eTable 2 in the Supplement).3,20-28
Data for analysis were considered across 4 violence perpetration outcome classifications. First, a broad interpersonal violence perpetration definition (any physical assault, robbery, sexual offenses, illegal threats or intimidation, and arson). Within this classification male, female, and data combined by sex were grouped separately using nonoverlapping samples, and subgroup analyses (not in protocol but added based on reviewer suggestion) were undertaken to examine for sources of heterogeneity (substance misuse comorbidity, geographical region of study, study quality, study type, nature of control group, patient sample, and outcome measure). While data from studies that included homicide, sexual offenses, or arson within their broad violent outcome definition were included within this classification, if a study considered only 1 of those specific offenses in isolation, it was considered separately. Data included within separate classifications for homicide, sexual offenses, and arson were either disaggregated from studies that used a broad definition but also reported separated offenses or from studies that considered only these outcomes (ie, no broad violence category). A total of 12 studies of homicide,21,29-32 arson,33-35 and sexual offenses36-39 were included in these separate classifications that were not otherwise included in the broader violence perpetration classification owing either to reporting only that specific offense or to issues of population overlap that did not apply within the separate offense classification. Within the arson classification, meta-analysis was not undertaken owing to the clinical heterogeneity of fire-setting definitions.
We generated unadjusted pooled odds ratios (ORs) with 95% CIs for risk of violence perpetration outcomes using random-effects meta-analysis. Heterogeneity was examined using Cochran Q and the I2 statistic. Analyses were undertaken using R version 3.6.1 (the R Foundation)40 and the R package meta.41 The Mantel-Haenszel method was used to combine odds ratios from 2 × 2 tables, with continuity correction of 0.5 in studies with zero cell frequencies.
Interpersonal Violence Perpetration as Outcome
Twenty-four studies were included (N = 51 309 individuals with schizophrenia spectrum disorders; reported mean age of 21 to 54 years at follow-up; of those studies that reported outcomes separately by sex, there were 19 976 male individuals and 14 275 female individuals), 16 of which were from the original review (n = 9954 individuals with schizophrenia spectrum disorders [19.4%]), and 8 of which were identified from updated searches (n = 41 355 [80.6%]). Studies were based in 15 countries: 5 from the US,42-46 3 from Sweden,2,47,48 2 from Israel reported in 3 publications,49-51 2 from Denmark,52,53 2 from Finland reported in 3 articles,54-56 1 from Switzerland reported in 3 publications,57-59 and 1 each from Australia,12 Austria,60 Canada,61 Czech Republic,62 Egypt,63 Ethiopia,64 Germany,65 New Zealand66 and the UK.67 See the Table for a summary of included studies and eTable 3 in the Supplement for additional details.
Characteristics of Included Studies
For outcomes, 15 studies used register-based measures of violent criminality2,12,46-48,51-53,55,56,59-61,65,67 (n = 50 250 [97.9%]) and 9 used self- or informant-report measures (n = 1059 [2.1%]) either solely42,44,45,50,62-64 or in combination with register-based data.43,66 Ten studies included only individuals with schizophrenia spectrum disorders who had at least 1 inpatient hospitalization43,46,47,51-53,59,62,65,67 (n = 11 011 [21.5%]), whereas 14 studies (n = 40 298 [78.5%]) identified individuals with schizophrenia spectrum disorders in the community or who had been in contact with a broader range of inpatient and outpatient care settings.2,12,42,44,45,48,50,55,56,60,61,63,64,66 Thirteen reports were considered overall good quality (scoring 7 to 9 of 9 overall and evenly across domains on the Newcastle-Ottawa Quality Assessment Scale),2,12,43,45,47,51-53,55,56,61,62,66 and 11 were considered poor to fair (eTable 4 in the Supplement).42,44,46,48,50,59,60,63-65,67 The mean (SD; interquartile range) quality score was 7.0 (1.5; 6-8).
There were 19 976 male individuals with schizophrenia spectrum disorders. The random-effects pooled OR for the risk of violence perpetration in individuals with schizophrenia spectrum disorders compared with general population control individuals in this group was 4.5 (95% CI, 3.6-5.6), with substantial heterogeneity (I2 = 85%; 95% CI, 77-91) (Figure 1).
Among the 14 275 female individuals with schizophrenia spectrum disorders, the random-effects pooled OR for the risk of violence perpetration in individuals with schizophrenia spectrum disorders compared with general population control individuals was 10.2 (95% CI, 7.1-14.6), with substantial heterogeneity (I2 = 66%; 95% CI, 31-83) (Figure 1). Excluding 3 studies of female individuals with schizophrenia55,57,67 (n = 256) with no events in either individuals with schizophrenia spectrum disorders or control individuals made no material difference to risk estimates (pooled OR, 10.3; 95% CI, 7.0-15.1).
Studies That Did Not Report Outcomes Separately by Participant Sex
Data from 11 studies with 17 058 individuals with schizophrenia spectrum disorders were not available separately by sex (Figure 1). Risk of violence perpetration was increased compared with general population control individuals (random-effects pooled OR, 4.6; 95% CI, 3.7-5.7), with substantial heterogeneity (I2 = 72%; 95% CI, 48-85).
Register-Based Outcomes and Study Quality
When restricting analysis to studies that used exclusively register-based violence perpetration outcomes (n = 50 250) (eFigure 2 in the Supplement), the pooled OR for violence perpetration in male individuals was 4.6 (95% CI, 3.6-5.9; I2 = 88%; 95% CI, 79-93); in female individuals, the OR was 12.8 (95% CI, 9.5-17.4; I2 = 48%; 95% CI, 0-78); and in studies where sex was not reported separately, it was 4.3 (95% CI, 3.4-5.4; I2 = 78%; 95% CI, 47-91). When studies were grouped by quality, pooled OR for studies rated good (n = 46 425) was 5.7 (95% CI, 4.3-7.5; I2 = 96%; 95% CI, 95-97) and for studies rated poor to fair (n = 4884) was 4.6 (95% CI, 3.7-5.9; I2 = 48%; 95% CI, 6-71) (eFigure 3 in the Supplement). Subgroup analysis by study design showed that findings were comparable (eFigure 4 in the Supplement).
Absolute rates of violence perpetration in included studies of register-based violent outcomes ranged from 2.3% to 24.7% in men with schizophrenia spectrum disorders and from 0% to 5.4% in women with schizophrenia spectrum disorders, over outcome periods that ranged from 1 year to lifetime measures spanning up to 35 years (Table; Figure 2).
Substance Misuse and Other Comorbidity
Seventeen studies2,12,42-45,47,50,51,53-55,59,61-65 included data on substance comorbidity among individuals with schizophrenia spectrum disorders, either under a broad substance misuse definition12,42,43,45,50,51,53,61,63,65 or separately for drug use2,44,47,59,62,64 and alcohol use.2,44,47,54,55,59,62,64 A variety of descriptors were used to define the comorbidity; in studies that described misuse (without or without dependence),42-45,50,51,53-55,59 or use disorders,2,12,61 prevalence ranges were reported for men and women (eTable 3 in the Supplement). In 13 studies,12,42-45,50,51,53,54,59,63,65 data allowed violence perpetration in individuals with schizophrenia spectrum disorders with or without substance misuse comorbidity to be compared with general population groups separately. The pooled odds of violence perpetration in individuals with schizophrenia spectrum disorders without substance misuse comorbidity (n = 11 079) was 3.5 (95% CI, 2.6-4.6; I2 = 81%; 95% CI, 68-89) (eFigure 5 in the Supplement), and in individuals with schizophrenia spectrum disorders with substance misuse comorbidity (n = 3586), it was 9.9 (95% CI, 7.2-13.5; I2 = 82%; 95% CI, 69-89) (eFigure 6 in the Supplement).
Information on other comorbidities was limited. One study reported on comorbid antisocial personality disorder.62 No included studies provided treatment information.
Subgroup analyses were undertaken for studies undertaken in Europe (n = 33 836), North America (n = 9856), Australia and New Zealand (n = 4207), and Africa (n = 122). Relative risk estimates for violence perpetration among individuals with schizophrenia spectrum disorders compared with general population control individuals did not significantly vary by region (eFigure 7 in the Supplement).
Two included studies, conducted in Sweden2 and Israel,51 used population registers to examine risk of violence perpetration in individuals with schizophrenia spectrum disorders and their unaffected siblings (n = 33 892 unaffected siblings) in addition to comparing individuals with schizophrenia spectrum disorders with general population control individuals. Both studies found relative risks of violence perpetration in unaffected siblings remained significantly increased compared with population control individuals, but to a lesser degree than in the corresponding patient samples (eFigure 8 in the Supplement), indicating familial confounding. When comparing violence perpetration in individuals with schizophrenia spectrum disorders with their unaffected siblings, the study in Sweden2 reported a ratio of ORs of 4.2 (95% CI, 3.8-4.5) and the study in Israel51 an OR of 2.8 (95% CI, 2.5-3.3).
Inpatient and Community Samples
When data were grouped according to samples identified by a history of psychiatric hospital admission or using a wider definition of outpatient or inpatient psychiatric settings, there were no significant differences in relative risk estimates (inpatient pooled OR, 5.4; 95% CI, 4.1-7.0; combined community and inpatient pooled OR, 5.5; 95% CI, 4.0-7.4) (eFigure 9 in the Supplement).
Homicide Perpetration as Outcome
Eight studies were identified with homicide perpetration as an outcome (Figure 3; eTable 5 in the Supplement). All used official sources for outcomes. There was an increase in risk of homicide perpetration in individuals with schizophrenia spectrum disorders compared with those without (Figure 3), leading to a random-effects pooled OR of 17.7 (95% CI, 13.9-22.6) with substantial heterogeneity (I2 = 70%; 95% CI, 39-85). Excluding 1 study (n = 282 [0.4%])59 with no events in the control group did not materially change the pooled risk estimate (OR, 17.9; 95% CI, 14.1-22.8). Four studies were rated good quality21,29,30,51 and 4 were rated poor to fair31,32,59,60 (eTable 4 in the Supplement).
Sexual Offense Perpetration as Outcome
Nine studies22,36-39,51,52,59,64 were included in a meta-analysis of sexual offense perpetration outcomes (Figure 4; eTables 4 and 6 in the Supplement). All but 2 studies38,64 used a register-based outcome. Risk was increased in individuals with schizophrenia spectrum disorders compared with control individuals, with a random-effects pooled OR of 5.1 (95% CI, 3.8-6.8) with moderate heterogeneity (I2 = 54%; 95% CI, 7-78) (Figure 4).
Four studies presented data on risk of arson. Findings ranged from no association between schizophrenia and self-reported fire-setting in 1 study64 to 2 register-based studies of arson and related convictions that showed a significant positive association with schizophrenia spectrum disorder diagnosis, with ORs of 18 (95% CI, 12-26) in male individuals and 23 (95% CI, 13-41) in female individuals in a study in Sweden,33 and 7 (95% CI, 4-13) in a study in Australia that did not report outcomes separately by sex of participants34 (eResults in the Supplement). Meta-analysis was not considered owing to the heterogeneity of arson outcomes.
This systematic review and meta-analysis included 51 309 individuals with schizophrenia spectrum disorders from 24 studies in 15 countries over 4 decades and investigated risk of violence perpetration outcomes. Additionally, we examined data on homicide perpetration and studies reporting sexual offense perpetration and arson as outcomes. Our findings update and extend a previous meta-analysis11 in 4 main ways.
First, the study included more than double the number of individuals with schizophrenia spectrum disorders in the meta-analysis of violence perpetration outcomes than the previous meta-analysis. This is despite deploying strict exclusion criteria to avoid double-counting of individuals in pooled analyses that excluded otherwise eligible studies. The improved precision of the current study is principally owing to the inclusion of 4 large, high-quality register-based studies conducted since the previous review.2,12,51,61 Despite studies from the last decade contributing most of the data, estimates for the relative risk of violence perpetration have remained broadly stable based on visual inspection of the forest plots (Figure 1).
Second, we pooled risks of sexual offense perpetration. Primary studies varied as to whether they included sexual offenses in an overall definition of violent offending or provided separate data on these offenses. Although there have been descriptive reviews of associations with sexual offending,68 to our knowledge this is the first quantitative synthesis. Data on arson and fire-setting were more limited, although 2 register-based studies reported a significant increased association with schizophrenia spectrum disorders.
Third, we summarized information on absolute risks of violent crime perpetration and on the time period over which these outcomes occurred. Time periods ranged from 1 to 35 years in register-based studies, and absolute risks varied considerably over time and between sexes. Absolute risks were less than 1 in 20 (or 5%) in female individuals with schizophrenia spectrum disorders and less than 1 in 4 men.
Fourth, large register-based studies undertaken since the original review11 provided more power to investigate confounding. We replicated in pooled analyses the finding that violence perpetration risk was reduced but remained increased relative to control samples when individuals with substance misuse comorbidity were excluded.12,51 In individuals with schizophrenia spectrum disorders and substance misuse comorbidity, there was increased odds of around 10-fold compared with general population control individuals, whereas in individuals with noncomorbid schizophrenia spectrum disorders, this was around 3-fold compared with general population control individuals. Additionally, the contribution of familial factors has now been estimated in studies that have compared risks of violence perpetration in individuals with schizophrenia spectrum disorders, their unaffected siblings, and general population control individuals. These investigations reported residual risk among unaffected siblings and therefore some familial confounding, but risk remains elevated among affected individuals compared with their unaffected siblings.2,51 The latter finding further confirms the independent association between schizophrenia and violence perpetration.
There are several implications of our findings for clinical services and future research. In clinical settings, risks in absolute terms, particularly in young men, are typically high enough to be considered of routine clinical importance to prevent. This is reflected to a limited extent in clinical guidelines, but guidance on assessing and managing risk could be improved.69 Likewise, the importance of comorbid substance misuse has been further highlighted, reinforcing the need to prioritize the assessment and treatment of comorbid substance misuse disorders in clinical services. The applicability of findings to a particular clinical setting is a limitation of pooling data from individuals at different stages of illness and treatment in a variety of geographical and health care settings. However, the amount of data included has allowed for exploration of some relevant factors, such as variation by region, patient samples, and substance misuse comorbidity. This improves the ability of clinicians to consider the extent to which findings may apply to their practice.
This replication of relative risk estimates for violence perpetration with the inclusion of larger data sets suggests that further small studies of the relative risk of violence in schizophrenia are not priorities. Future epidemiological research could focus on specific offenses or outcomes, such as sexual offenses and arson. Violence perpetration as an outcome is necessarily broad, but there may be important differences in clinical and risk profiles between more specific violence perpetration outcomes.
This meta-analysis found robust associations between schizophrenia spectrum disorders and violence perpetration, but did not answer the question of the nature of causal pathways and whether they are present. Empirical studies have suggested that symptoms such as persecutory delusions may be causally linked to violence.70 The current study found that the association between schizophrenia spectrum disorders and violence perpetration remained, albeit with reduced estimates, after adjusting for substance misuse and familial confounders, which can exist on the causal pathway and explain a proportion of the association. However, other background, psychiatric, and socioeconomic factors unaccounted for in these pooled analyses can contribute to this pathway, and there is research in support of early conduct problems,71 adverse childhood experiences,72 and economic disadvantage.73 Large prospective studies are needed to further evaluate causality.74 For clinicians, the multifactorial association means that a key priority should be to refine approaches to individualized risk assessment and management. Scalable prediction tools offer one way to translate contemporary knowledge on risk factors (including sociodemographic, familial, and treatment factors) into clinical practice and achieve a more stratified approach to assessing and managing clinical needs around violence risk.75
There was little information on treatment effects. There is evidence from pharmacoepidemiological studies that antipsychotics can reduce violence perpetration outcomes.76-79 This needs confirmation in randomized trials, and violence perpetration should be a routine outcome in effectiveness trials of pharmacological and nonpharmacological interventions. One question is how risk of violence perpetration in individuals with schizophrenia spectrum disorders treated with antipsychotic medication compares with that risk in general population samples. One population study in Sweden found rates of violent offenses of 33 per 1000 person-years during treatment with olanzapine compared with 96 offenses per 1000 person-years prior to treatment.77 To what extent these treatment effects bring rates of violence perpetration down toward baseline population levels requires comparison with matched groups.
This study has limitations. Most data in pooled analyses used register-based violence perpetration outcomes, which offer advantages over self- or informant-rated measures in that they objectively cross a severity threshold for clinical relevance and resource implications, do not rely on ongoing contact with health services, and avoid interrater variability and recall bias. A limitation is to what extent these measures may be confounded by differential treatment of individuals with mental disorders by the justice system. However, a study examining the Dunedin cohort found that associations with mental disorders were similar when using self- or informant-report and official court records, suggesting that the degree of underreporting is similar in individuals with schizophrenia spectrum disorders and control individuals when using either outcome.66 In terms of other limitations, only 3 studies presented information on experiencing interpersonal violence61,62,64 and 1 on the contribution of childhood trauma.62 The importance of the overlap between risk of experiencing violence and perpetrating violence is increasingly recognized.3,61,80 Further, although analyses did not reveal significant differences in risk estimates when studies were grouped by quality, variation between studies in the level of matching of comparison groups is an additional consideration to explain heterogeneity, as this was only partially captured by the quality rating. A study of more than half a million individuals using population registers in Denmark replicated findings of increased risk of violence perpetration in individuals with schizophrenia spectrum disorders (with incident rate ratios of 6 in men and 18 in women),81 but could not be included in pooled analyses because the study lacked the necessary tabular data.
This review has quantitatively summarized the association between schizophrenia spectrum disorders and violence perpetration outcomes. Prevention of violence perpetration should be a target for clinical services that assess and treat individuals with these disorders. New work should consider preventive approaches, including the improvement of the clinical assessment of risk and targeted resource allocation to treat modifiable factors. Violence perpetration outcomes should be routinely collected in studies of the effectiveness of clinical services and therapeutic interventions.
Accepted for Publication: November 1, 2021.
Published Online: December 22, 2021. doi:10.1001/jamapsychiatry.2021.3721
Corresponding Author: Seena Fazel, MD, Department of Psychiatry, University of Oxford, Warneford Hospital, Oxford, England OX3 7JX (firstname.lastname@example.org).
Author Contributions: Dr Whiting had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: All authors.
Acquisition, analysis, or interpretation of data: Whiting, Gulati, Fazel.
Drafting of the manuscript: All authors.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Whiting.
Obtained funding: Fazel.
Conflict of Interest Disclosures: None reported.
Funding/Support: Dr Whiting is funded by the National Institute for Health Research (DRF-2018-11-ST2-069). Dr Geddes is supported by the National Institute for Health Research Oxford Health Biomedical Research Centre. Dr Fazel is funded by the Wellcome Trust (202836/Z/16/Z).
Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The views expressed in this article are those of the authors and not necessarily those of the National Institute for Health Research, or the Department of Health and Social Care.
Additional Contributions: We thank Hygiea Casiano, MD (Department of Psychiatry, University of Manitoba, Winnipeg, Manitoba, Canada), Charlotte Hanlon, PhD (Department of Psychiatry, Addis Ababa University, Addis Ababa, Ethiopia), Ruth Tsigebrhan, MD (Department of Psychiatry, Addis Ababa University), and Nomi Werbeloff-Becker, PhD (Department of Psychiatry, Sheba Medical Center, Tel-Hashomer, Israel), for providing additional information from their studies. None of the contributors were compensated for their work.
J. Age at onset and the outcomes of schizophrenia: a systematic review and meta-analysis. Early Interv Psychiatry
. 2017;11(6):453-460. doi:10.1111/eip.12412PubMedGoogle ScholarCrossref
JR. Comparing violence in schizophrenia patients with and without comorbid substance-use disorders to community controls. Acta Psychiatr Scand
. 2013;128(4):306-313. doi:10.1111/acps.12066PubMedGoogle Scholar
JE. The epidemiology of violent behavior in patients with a psychotic disorder: a systematic review of studies since 1980. Aggress Violent Behav
. 2013;18(1):183-189. doi:10.1016/j.avb.2012.11.018Google ScholarCrossref
et al; Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. Meta-analysis of observational studies in epidemiology: a proposal for reporting. JAMA
. 2000;283(15):2008-2012. doi:10.1001/jama.283.15.2008PubMedGoogle ScholarCrossref
P. Criminal offending in schizophrenia over a 25-year period marked by deinstitutionalization and increasing prevalence of comorbid substance use disorders. Am J Psychiatry
. 2004;161(4):716-727. doi:10.1176/appi.ajp.161.4.716PubMedGoogle ScholarCrossref
S. Genetic and environmental determinants of violence risk in psychotic disorders: a multivariate quantitative genetic study of 1.8 million Swedish twins and siblings. Mol Psychiatry
. 2016;21(9):1251-1256. doi:10.1038/mp.2015.184PubMedGoogle ScholarCrossref
C. Correlates of intimate partner violence perpetration: results from a national epidemiologic survey. J Trauma Stress
. 2015;28(1):49-56. doi:10.1002/jts.21986PubMedGoogle ScholarCrossref
B. Killing the mother of one’s child: psychiatric risk factors among male perpetrators and offspring health consequences. J Clin Psychiatry
. 2016;77(3):342-347. doi:10.4088/JCP.15m09564PubMedGoogle ScholarCrossref
S. Is arson the crime most strongly associated with psychosis? a national case-control study of arson risk in schizophrenia and other psychoses. Schizophr Bull
. 2011;37(3):580-586. doi:10.1093/schbul/sbp098PubMedGoogle ScholarCrossref
M. Severe mental illness and risk of sexual offending in men: a case-control study based on Swedish national registers. J Clin Psychiatry
. 2007;68(4):588-596. doi:10.4088/JCP.v68n0415PubMedGoogle ScholarCrossref
RT. Violence and psychiatric disorder in the community: evidence from the Epidemiologic Catchment Area surveys. Hosp Community Psychiatry
. 1990;41(7):761-770. doi:10.1176/ps.41.7.761PubMedGoogle Scholar
BG. Gender differences in the relationship between mental illness and violence: evidence from a community-based epidemiological study in Israel. Soc Psychiatry Psychiatr Epidemiol
. 1998;33(1)(suppl 1):S61-S67. doi:10.1007/s001270050211PubMedGoogle Scholar
J. Psykisk Afvigelse Og Kriminel Adfærd: En Undersøgelse Af 11.533 Mænd Født I 1953 I Det Metropolitane Område København. Justitsministerie; 1981.
J. Specific major mental disorders and criminality: a 26-year prospective study of the 1966 northern Finland birth cohort. Am J Psychiatry
. 1997;154(6):840-845. doi:10.1176/ajp.154.6.840PubMedGoogle Scholar
et al. Police-registered offenses and psychiatric disorders among young males: the Finnish “from a boy to a man” birth cohort study. Soc Psychiatry Psychiatr Epidemiol
. 2007;42(6):477-484. doi:10.1007/s00127-007-0192-1PubMedGoogle ScholarCrossref
H. Schizophrenia and violence. results of a comprehensive inquiry in an Austrian providence. Article in German. Nervenarzt
. 2001;72(11):859-866. doi:10.1007/s001150170020PubMedGoogle ScholarCrossref
A. Frequency, severity and types of aggressive behavior among patients with schizophrenia at psychiatry outpatients clinic in comparison to normal controls. AJP
. 2015;26(2):164-171.doi:10.12816/0014483Google ScholarCrossref
C. Violence and violent victimization in people with severe mental illness in a rural low-income country setting: a comparative cross-sectional community study. Schizophr Res
. 2014;152(1):275-282. doi:10.1016/j.schres.2013.10.032PubMedGoogle ScholarCrossref
et al. Associations between conduct problems in childhood and adverse outcomes in emerging adulthood: a longitudinal Swedish nationwide twin cohort. J Child Psychol Psychiatry
. 2020;61(7):798-806. doi:10.1111/jcpp.13169PubMedGoogle ScholarCrossref
TR. Identification of low risk of violent crime in severe mental illness with a clinical prediction tool (Oxford Mental Illness and Violence tool [OxMIV]): a derivation and validation study. Lancet Psychiatry
. 2017;4(6):461-468. doi:10.1016/S2215-0366(17)30109-8PubMedGoogle ScholarCrossref
et al. Clozapine treatment and offending: a within-subject study of patients with psychotic disorders in Sweden. Schizophr Bull
. 2020;46(2):303-310.PubMedGoogle Scholar
S. Associations between individual antipsychotics and the risk of arrests and convictions of violent and other crime: a nationwide within-individual study of 74 925 persons. Psychol Med
. 2021;1-9. doi:10.1017/S0033291721000556PubMedGoogle Scholar