Predicting violent behavior in clinical settings: a case-control study of a mental health inpatient unit.

In order to assist staff in recognizing patients prone to violence and guide their clinical decision-making, this study summarizes mental health inpatient unit incidents over a one-year period. Results describe demographic and clinical information for patients, and evaluate risk assessment tools currently used to predict risk. A retrospective analysis included data on patients involved in incidents and frequency matched controls. There were a total of 44 incidents, caused by 38 unique patients. A constructed model to estimate patient characteristics and risk of violent incidents included involuntary admittance (OR 2.07, 95% CI 1.05-6.11, p = 0.039), more than one admission at the facility (OR 4.18, 95% CI 1.71-10.22, p = 0.002) and Global Subjective Irritability on day one (OR 4.24, 95% CI 1.77-10.16, p = 0.001). Violent incidents on the mental inpatient unit threaten safety and disrupt the therapeutic environment. The findings may be useful in aiding clinicians to quickly recognize patients that are prone to violence.