The Predictive Properties of Violence Risk Instruments May Increase by Adding Items Assessing Sleep

Area of Science:

• Psychiatry
• Sleep Medicine
• Clinical Psychology

Background:

• Current violence prediction tools for psychiatric inpatients often omit sleep assessment.
• Sleep disturbances may serve as early indicators of aggressive behavior in this population.
• The Brøset Violence Checklist (BVC) is a tool for assessing aggression risk.

Purpose of the Study:

• To evaluate if incorporating sleep data enhances the predictive accuracy of the Brøset Violence Checklist (BVC).
• To investigate the role of sleep duration and variability in predicting aggression among psychiatric inpatients.

Main Methods:

• A 6-month study involving 50 patients in a psychiatric intensive care unit (PICU).
• Data collected included staff-observed sleep patterns (duration, night-to-night variation), BVC behavior assessments, and aggressive incidents via the Staff Observation Scale-Revised.
• Statistical analysis focused on changes in predictive ability (AUC-ROC) with the addition of sleep variables to the BVC.

Main Results:

• The Brøset Violence Checklist's (BVC) predictive accuracy for aggressive incidents improved from an AUC-ROC of 0.757 to 0.873 when a combined sleep variable was added.
• The combined sleep variable alone showed a trend towards predicting aggressive incidents (AUC-ROC 0.653, p=0.051).
• The inclusion of sleep data significantly enhanced the overall predictive model.

Conclusions:

• A variable assessing sleep disturbances demonstrably improves the predictive power of the Brøset Violence Checklist (BVC) in psychiatric intensive care units (PICUs).
• Sleep duration and its nightly variations are important factors to consider in predicting patient aggression.
• Further research is warranted to explore the nuanced relationship between sleep patterns and aggressive behavior in psychiatric settings.