A multifaceted model for risk assessment of violent behaviour in acutely admitted psychiatric patients.

Violence risk assessment instruments are mainly based on historical and clinical risk factors. Biological factors have been related to violent behaviour but hardly used in risk assessment. A recent study indicated that both low total cholesterol (TC) and the patients' own estimates about their risk of future violent behaviour, the Self-report Risk Scale (SRS), significantly predicted violent behaviour during hospital stay and the first 3 months after discharge from an acute psychiatric department. Our aim was to investigate whether combining three methods, a screening instrument for risk of violence (V-RISK-10), low TC and the SRS, into one multifaceted risk assessment model may enhance predictive validity. This prospective naturalistic study involved 134 of all (489) acutely admitted patients to a general psychiatric hospital during 1 year. Low TC, SRS and V-RISK-10 at admission were prospectively compared with recorded violence during hospital stay and the first 3 months after discharge. The multifaceted risk assessment model yielded a significant increase in explained variance beyond that of the V-RISK-10. AUC values were higher, but differences were not significant. The biopsychosocial approach to violence risk assessment seems promising, but further studies are needed to test the feasibility and predictive validity of multifaceted models.