Violence in severe mental illness : risk factors, risk assessment, and prevention

Individuals with severe mental illness are at increased risk of violence compared with the general population. Risk factors for violence need clarification as a basis for risk assessment, and to inform evidence-based risk management. The aim of this thesis is to examine the risk factors for violence in schizophrenia-spectrum disorders, create a simple, scalable risk assessment tool to predict violence in forensic psychiatric populations, and review the evidence for violence prevention interventions in psychiatry. The first study reports risk factors for violence in a Swedish total population cohort of patients with schizophrenia-spectrum disorders (n=24,297), using register data. Substance use disorders, previous criminality, and self-harm were strong predictors of violence in patients, but also in unaffected sibling and general population controls. The second study used pre-specified criminal, sociodemographic, and clinical risk factors to derive a clinical prediction rule for violent crime within 12 and 24 months of discharge from forensic psychiatric hospitals in Sweden (n=2,248). The model was used to develop a 12-item online tool (FoVOx) with good measures of calibration and discrimination (area under the curve = 0.77 at 12 and 24 months). The third study reports an umbrella review of violence prevention interventions in psychiatry, through searches in ten bibliographic databases, and assesses the strength of evidence for each review using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach. Out of five included reviews, only one received a GRADE rating of 'moderate', and reported that therapeutic community interventions may reduce reincarceration in drug-using offenders with mental illness. Given the lack of intervention research in violence prevention in psychiatry, and considering the shared risk factors between populations, interventions from non-psychiatric populations may need to be relied upon. Research recommendations include the validation of FoVOx, and routine inclusion of violence outcomes in primary research.