Risk assessment of imminent violence in acute psychiatry: a step towards an extended model

Aggression occurs frequently in mental health care settings, and studies have reported that 17% to 31% of patients admitted to acute psychiatric wards commit violence. Inpatients' fluctuating mental states and behaviour patterns reinforce the need for an assessment instrument to predict potential violence in a timely manner. This naturalistic prospective inpatient study investigated whether an extended short-term risk assessment model that combines (a) short-term risk assessment with the Broset Violence Checklist (BVC), (b) patient's own prediction of violence with the Self-Report Risk Scale (SRS) and (c) single items from the Violence Risk Screening 10 (V-RISK-10) provides better short-term predictive accuracy for violence than the BVC alone. All patients admitted to a psychiatric emergency hospital in Norway during one year were included (N = 508). Stepwise multivariate generalised linear mixed model analyses were conducted. When adjusting for repeated measurements, the results indicated that an extended model for short-term risk assessment, consisting of the BVC, SRS and Item 2 Previous and/or current threats from the V-RISK-10 explained more variance of imminent violence, compared to the BVC alone. Further studies are recommended to investigate whether the extended model provides a clinically better short-term risk prediction of imminent violence, compared to the BVC alone.