DASA score performance for predicting safety interventions among patients with and without emergency behavioral...

Violence in emergency departments (EDs) is increasing, threatening patient and clinician safety. The Dynamic Appraisal of Situation Aggression (DASA) score, implemented at ED triage, is associated with subsequent behavioral interventions, but its performance in patients with vs without behavioral health consultation remains unclear. This study evaluates DASA's performance for predicting safety interventions in ED patients with and without BH consultation.

At five affiliated EDs, nurses assigned DASA scores at triage for all adult patients as low (0), moderate (1-3), high (4-5), or imminent (>5). Scores, demographics, BH consultation status, and use of safety interventions were extracted from the electronic health record. Sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and positive likelihood ratios (LR+) were calculated for patients with and without BH consultation across DASA categories. Discrimination was assessed using receiver operating characteristic curves, and predicted probabilities of restraint were estimated using logistic regression marginal effects.

Over 34 months, 480,896 adult ED encounters were recorded (mean age 51.8, 53.2% female), including 21,170 (4.4%) requiring BH consultation. There were 7861 restraints and 22,914 medication interventions. Higher DASA thresholds increased specificity and PPV but reduced sensitivity, especially in BH-consulted patients, where DASA outperformed non-consulted patients. AUC was 0.567 for all patients, 0.543 for non-consulted patients, and 0.718 for BH-consulted patients.

DASA demonstrates modest discriminative performance overall but stronger discrimination and predictive value in patients receiving BH consultation. Higher DASA scores were associated with an increased likelihood of requiring safety interventions during the ED encounter.