Predicting aggressive behaviour in acute forensic mental health units: A re-examination of the dynamic appraisal of situational aggression's predictive validity.

In the present study, we explored the predictive validity of the Dynamic Appraisal of Situational Aggression (DASA) assessment tool in male (n = 30) and female (n = 30) patients admitted to the acute units of a forensic mental health hospital. We also tested the psychometric properties of the original DASA bands and novel risk bands. The first 60 days of each patient's file was reviewed to identify daily DASA scores and subsequent risk-related nursing interventions and aggressive behaviour within the following 24 hours. Risk assessments, followed by documented nursing interventions, were removed to preserve the integrity of the risk-assessment analysis. Receiver-operator characteristics were used to test the predictive accuracy of the DASA, and generalized estimating equations (GEE) were used to account for repeated risk assessments, which occurs when analysing short-term risk-assessment data. The results revealed modest predictive validity for males and females. GEE analyses suggested the need to adjust the DASA risk bands to the following (with associated odds ratios (OR) for aggressive behaviour): 0 = low risk; 1, 2, 3 = moderate-risk OR, 4.70 (95% confidence interval (CI): 2.84-7.80); and 4, 5, 6, 7 = high-risk OR, 16.13 (95% CI: 9.71-26.78). The adjusted DASA risk bands could assist nurses by prompting violence-prevention interventions when the level of risk is elevated.