Re-Examining the Predictive Validity and Establishing Risk Levels for the Dynamic Appraisal of Situational Aggression: Youth Version

Abstract

The Dynamic Appraisal of Situational Aggression: Youth Version (DASA:YV) is a brief instrument, most often used by nurses and was specifically designed to assess risk of imminent violence in youth settings. To date, it has been recommended that DASA:YV scores are interpreted in a linear manner, with high scores indicating a greater level of risk and therefore need more assertive and immediate intervention. This study re-analyses an existing data set using contemporary robust data analytic procedures to examine the predictive validity of the DASA:YV, and to determine appropriate risk bands. Mixed effect logistic regression models were used to determine whether the DASA:YV predicted aggression when the observations are correlated. Two approaches were employed to identify and test novel DASA:YV risk bands, where (1) three risk bands as previously generated for the adult DASA were used as a starting point to consider recategorising the DASA:YV into three risk bands, and (2) using a decision tree analysis method known as Chi-square automated interaction detection to produce risk bands. There was no statistically significant difference between a four and three category of risk band. AUC values were 0.85 for the four- and three-category options. A three-category approach is recommended for the DASA:YV. The new risk bands may assist nursing staff by providing more accurate categorisation of risk state. Identification of escalation in risk state may prompt early intervention, which may also prevent reliance on the use of restrictive practices when young people are at risk of acting aggressively.