Predictors of violence against others in non-affective psychosis: A systematic review and meta-analysis

Introduction The risk of committing violence against others is increased in patients with non-affective psychosis compared to the background population. However, the profile of patients with non-affective psychosis, who commit violence against others, is not fully known. Prior reviews and meta-analyses have included heterogenous psychosis populations and operated with broad definitions of violence. In this study, we narrowed the focus to patients with non-affective psychosis and restricted the definition of violence to physical violence committed against others. Methods Following the PRISMA guidelines, we searched PubMed, PsycInfo, and Ovid Embase for studies comparing patients with non-affective psychosis with and without a history of violence against others. Results 13 studies with a total of 1446 patients were included. Few predictors of violence were identified across the studies. In the meta-analysis, only higher score on Hare Psychopathy Checklist-Revised (PCL-R) and shorter years of education significantly predicted belonging to the violence group across studies. None of these predictors were robust after Bonferroni correction. No differences were found in psychopathology on the Positive and Negative Syndrome Scale (PANSS), substance use, cognition, duration of illness, marital status, employment status, and age of onset of illness between patients with a history of physical violence against others and those without such a history. Conclusions Predictors of violence previously identified in other studies proved non-significant in our more narrowly focused and stringent review and meta-analysis. More research is needed to improve characterization and identification of patients with non-affective psychosis at risk of committing violence against others.