A biopsychosocial approach to violence risk assessment in acute psychiatry

Violence risk assessment methods in psychiatry seem to have reached an upper limit of accuracy. Our main aim was therefore to explore predictive accuracy of a biopsychosocial approach (BPS model) to violence risk assessments in an acute psychiatric ward, compared to that of an existing psychosocial risk screen. Sub-aims were to examine predictive accuracy of single elements of the model and gender differences. This was a prospective naturalistic study (N = 528 admitted during one year from March 2012). The BPS model (study variables) included an existing risk screen: Violence risk screening – 10 (V-RISK-10) and a Self-Report Risk Scale (SRS), both recorded after admission and before discharge, and two potential biological risk markers of violence: total cholesterol (TC) and high density lipoprotein cholesterol (HDL) (from routine blood samples after admission). Violence (outcome variable) was recorded during hospital stay and the first three months after discharge. Results indicated good predictive validity of the BPS model (inpatient AUC = 0.8), although improvements compared to V-RISK-10 alone were not significant. SRS seemed to contribute to the model after discharge (OR = 3.1). Some potential gender differences in single items in V-RISK-10 were also indicated. For men, HDL seemed to contribute significantly to the model during hospital stay (OR = 0.26) and as a risk marker of violence after discharge (OR = 0.1). Despite indications of usefulness of a biopsychosocial approach to violence risk assessments in acute psychiatry and also findings of some possible gender differences, results should be interpreted with caution due to low statistical power in some analyses