Incidence and predictors of delirium-associated violence towards healthcare workers in a Swiss ICU: results from the CAVE study

Abstract Background To assess the frequency of delirium coinciding with aggressive and/or violent patient behavior in intensive care unit (ICU) patients as documented by nurses and to identify clinical associations or risk factors. Methods This was a retrospective observational cohort study conducted at University Hospital Basel in Switzerland including data from 2023-2024. Data collected included demographics, treatment characteristics, clinical scores (including the Richmond Agitation-Sedation Score [RASS] and the Intensive Care Delirium Screening Checklist [ICDSC]) and interventions. The primary objective was to assess the prevalence of delirium in patients with aggressive/violent behavior within ±72 hours of incidents. Secondary objectives included identifying independent associations with violence via uni- and multivariable logistic regression. Results Of 10,817 ICU patients, 300 (2.8%) were reportedly aggressive or violent. Delirium occurred in 89.3% of patients within 72 hours of the event and 76.7% were delirious at event. Violence was mostly physical (80.4%), with 22.2% preceded by aggression within a median of 21.6 hours. Multivariable analyses identified an increasing RASS (OR = 1.46 per increasing unit ; p = 0.001) and ICDSC (OR = 1.14 per increasing unit ; p = 0.039) before the violent incident as independent associations. Management shifted from verbal de-escalation for aggression to more invasive measures for violence; sedation rose from 45.3% to 69.8% (p < 0.001), restraint use from 20.0% to 43.1% (p < 0.001) and sedation-related complications reached 64.9%. Unchanged one-on-one nursing rates indicate a reliance on restrictive measures over staffing increases. Conclusions While ICU-related aggression and violence is rarely reported, the independent association of increasing RASS/ICDSC scores before violence and a prodromal phase of aggression in many cases provide a warning pattern. Identifying these early signs could allow for proactive de-escalating intervention, potentially reducing the need for physical restraints and heavy sedation.