Witt, van Dorn, and Fazel (2013) noted many inconsistencies and varying emphases in the current literature on the association of violence and psychosis. This led the researchers to perform a meta-analyses of the current literature base, essentially combining all current studies on violence risk and psychosis into one helpful summary. The authors noted this task is important to the field for several reasons. First, combining and analyzing this information would hopefully help to develop evidence-based approaches to risk assessment. Next, this information can help focus treatment with relevant populations to the most pertinent risk factors, while simultaneously enhancing protective factors. Finally, consolidating this information can help clinicians and researchers better understand why certain individuals with psychosis have a higher risk of violence.
Six major databases were searched from their inception until December 2011. For some databases, this meant going back as far as 1960. Non-English articles were translated by qualified post-graduate students. For inclusion, diagnoses had to be assigned based on Diagnostic and Statistical Manual of Mental Disorders (DSM) or International Classification of Diseases (ICD) criteria, and more than 95% of study participants were aged 18 or older and diagnosed with either schizophrenia, schizophreniform disorder, schizoaffective disorder, delusional disorder, schizotypal disorder, psychosis not otherwise specified, and bipolar disorder. It is important to note that psychoses as the result of medical conditions, substance intoxication, or substance withdrawal were excluded from the collected data. Studies were excluded if the focus was on genetic or epigenetic associations with violence, childhood violence, or offender populations. Furthermore, items were only included in the data collection process if the risk factor was included in three or more separate studies, which helped improve the validity of risk estimates. Risk factors were separated by time in that “recent” factors were those that occurred within the past year from the time of the original study, while “history of” factors were those that occurred at some point in the past, more than one year from the time of the original study. Data collected from each study may have been reported in different measurements; therefore, all collected data was converted to an odds ratio (ORs). For each factor identified, ORs, 95% confidence intervals, number of studies, the z score, number of violent participants, and total number of participants were reported.
A total of 110 studies that included 73 independent samples met inclusion criteria. This equated to a large number of participants (n=45,533) of whom 18.5% (8,439) were reported to be violent. Just over 85% of participants were diagnosed with schizophrenia, just under 12% were diagnosed with other psychoses, and 0.4% were diagnosed with bipolar disorder. The age of participants ranged from 21.1 to 54.3 years, with the average age of 35.8 years. The data included studies conducted in 27 countries.
Overall, the strongest domains associated with violence include the criminal history, substance misuse, demographic, and premorbid factors. When analysis was restricted to inpatient samples, the substance misuse domain was significantly associated with violence, but less so compared to the findings in the overall analysis. Additionally, analysis restricted to inpatient samples found the psychopathology and positive symptom domains were more strongly associated with violence, while the negative symptom, neuropsychological, demographic, premorbid, suicidality, and treatment-related domains were not significantly associated with risk of violence when compared to the overall analysis. The finding of differences in factors associated with violence among inpatient samples versus community samples could lend itself to the field developing different violence risk assessment approaches depending on whether the individual is in inpatient or outpatient treatment currently. A rather interesting finding was the association of previous suicide attempts with violence, especially considering most current and commonly used violence risk assessments do not usually include assessment of suicide. The authors speculate that history of previous suicide attempts was associated with violence, while experiencing suicidal ideation was not, because impulsivity may be a contributing factor to violence toward self and violence toward others. The authors close by identifying the most important factors to attend to during violence risk assessments: hostile behavior, poor impulse control, lack of insight, general symptom scores, recent alcohol and/or drug misuse, psychotherapy non-compliance, and medication non-compliance.
The major findings are described below in outline format for easy reference.
- Demographic Factors
- Strongly associated with violence:
- History of being violently victimized
- Moderately associated with violence:
- Recent homelessness or history of homelessness
- Male
- Weakly associated with violence:
- Member of ethnic minorities
- Currently having a lower socioeconomic status
- NOT significantly associated with violence:
- Received no more than a primary school education
- Received no more than a high school education
- Lower family socioeconomic status during childhood
- Shorter duration of education in years
- Lacking any formal education qualifications
- Currently living in an urban environment
- Currently living alone
- Unmarried
- Widowed or divorced
- Currently unemployed
- Having children
- Younger age at study enrollment in years
- Premorbid Factors
- Moderately associated with violence:
- History of childhood physical or sexual abuse
- Parental history of criminal involvement
- Parental history of alcohol misuse
- NOT significantly associated with violence:
- Experienced the death of one parent during childhood
- Experienced divorce or separation of parents during childhood
- Raised by a single parent
- Criminal History Factors
- Significantly associated with violence:
- History of assault
- History of imprisonment for any offense
- Recent arrest or history of arrest for any offense
- History of conviction for a violent offense
- History of violent behavior
- Hostility during the study period
- Psychopathological Factors
- Strongly associated with violence:
- Lack of insight
- Poor impulse control
- Moderately associated with violence:
- Diagnosis of comorbid antisocial personality disorder
- Higher total Positive and Negative Symptom Scale (PANSS) scores
- NOT significantly associated with violence:
- Diagnosed with bipolar disorder
- Diagnosed with any subtype of schizophrenia
- Diagnosed with schizoaffective disorder
- Diagnosed with psychotic disorder not otherwise specified
- Younger age of onset in years
- Positive Symptom Factors
- Associated with violence:
- Higher positive symptom scores
- NOT significantly associated with violence:
- Experienced paranoid thoughts
- Experienced delusions of any type
- Experienced auditory hallucinations, including command auditory hallucinations
- Acutely symptomatic
- Negative Symptom Factors
- NOT significantly associated with violence:
- Higher poor attention span scores
- Diagnosed with comorbid depression
- Neuropsychological Factors
- NOT significantly associated with violence:
- Lower Full Scale IQ scores on the Wechsler Adult Intelligence Scale (WAIS)
- Lower Performance IQ scores on the WAIS
- Lower Verbal IQ scores on the WAIS
- Lower scores on the Picture Completion subtest of the WAIS
- Lower total scores on the National Adult Reading Test (NART)
- Higher perseverative errors on the Wisconsin Card Sorting Test
- Substance Misuse Factors
- Strongly associated with violence:
- History of polysubstance misuse
- Diagnosis of comorbid substance use disorder
- Recent substance misuse
- Moderately associated with violence:
- Recent or history of alcohol misuse
- History of substance misuse
- Recent or history of drug misuse
- Treatment-Related Factors
- Strongly associated with violence:
- Psychotherapy treatment non-compliance
- Moderately associated with violence:
- Medication non-compliance
- NOT significantly associated with violence:
- Not having a prescription of antipsychotic medication of any type
- Higher antipsychotic dosage
- Shorter duration of antipsychotic treatment in months
- Shorter duration of current inpatient admission in months
- Shorter duration of current outpatient treatment in months
- Younger age at first psychiatric inpatient admission in years
- Greater number of previous psychiatric admissions
- Longer duration of untreated illness in years
- Suicide Factors
- Moderately associated with violence:
- History of previous suicide attempts
- NOT significantly associated with violence:
- History of experiencing suicidal ideation
- History of self-harm
Witt, K., van Dorn, R., & Fazel, S. (2013). Risk factors for violence in psychosis: Systematic review and meta-regression analysis of 110 studies. PLOS One, 8(2), 1-15.
Danielle M. McNeill, M.S., M.A.
Doctoral Intern