Predicting aggressive or violent patient behavior on acute care medical-surgical inpatient units

Workplace violence against nurses is a growing concern with significant emotional and financial consequences. Despite the prevalence of such incidents, validated tools for predicting violent or aggressive behavior in general medical-surgical units remain limited. The purpose of this study was to validate the Aggressive Behavior Risk Tool (ABRAT) for predicting violent or aggressive behaviors among adult inpatients in medical-surgical units. A retrospective cohort study was conducted using medical records from six hospitals. Patients aged ≥ 18 admitted to 13 pilot medical-surgical units between May and December 2023 were included (N = 6862). Nurses completed the ABRAT within 24 h of admission. Violent events were documented throughout hospitalization. Outcome measures included the occurrence of ≥ 1 violent event and ABRAT scores. Violent events occurred in 1.6 % of patients, with verbal abuse being most common (85.6 %). The ABRAT demonstrated moderate predictive ability (AUC=0.77). With a cutoff score of 1 + , the tool had poor sensitivity (47.7 %) and positive predictive value (6.1 %), but high specificity (87.9 %) and negative predictive value (99.0 %). The ABRAT showed limited sensitivity in identifying high-risk patients but was effective in ruling out risk. These findings suggest the tool may have limited utility in medical-surgical settings for proactive violence prevention. A revised version of ABRAT is available and should be further tested for validity. While brief and easy to administer, the ABRAT’s limited predictive accuracy and documentation burden may reduce its practical value. Revised tools with improved sensitivity and integration into workflow are needed to support nurse safety.