Peer-reviewed article

Evaluating Triage Acuity and Injury Mechanisms as Drivers of Emergency Department Length of Stay in Pediatric Forensic Cases.

Özer MR, Jaradat Ö, Küçükceran K, Avcı A · 13 Sept 2026

Abstract

Background/Objectives: Pediatric forensic cases represent a growing public health concern and impose substantial burdens on emergency departments (EDs). However, data on operational predictors of ED length of stay (ED-LOS) in this specific population remain limited. We aimed to identify predictors of increased ED length of stay (ED-LOS) and evaluate clinical profiles of pediatric forensic cases. Methods: This retrospective observational study included 1113 pediatric forensic cases. Cox proportional hazards regression was used as the primary method to identify predictors of time-to-discharge. Firth's penalized logistic regression and Random Forest (RF) were used for binary prediction based on a data-derived 45 min high-acuity threshold established via Reciever Operating Characteristic (ROC) analysis. Performance was assessed via AUC, Brier score, bootstrap validation, and Decision Curve Analysis (DCA). Results: The cohort was predominantly adolescent (71.5%) and male (65.3%), with interpersonal assault (36.7%) and standard forensic examinations (26.7%) being the most frequent etiologies. In the Cox regression, lower clinical acuity (higher Emergency Severity Index [ESI] scores) significantly increased the hazard of discharge (Hazard Ratio [HR]: 1.954). Conversely, specific trauma mechanisms notably delayed discharge, particularly intoxication (HR: 0.126), burns (HR: 0.127), and out-of-vehicle traffic accidents (HR: 0.237). Parent-accompanied presentation was independently associated with faster ED processing (HR: 1.303), while ambulance arrival was not a significant independent driver of overall time-to-discharge (p = 0.176). RF confirmed that the ESI was the dominant operational predictor (37.0% importance; test AUC 0.906), and the Firth regression demonstrated that these same mechanisms were the strongest predictors. The model showed high Negative Predictive Value (96.9%) and DCA benefit. Conclusions: This study proposes a framework for predicting increased ED-LOS in pediatric forensic cases. Cox regression suggested that ESI and trauma mechanisms modulate discharge speed, reflecting operational patterns. The findings may support triage stratification and resource allocation, but external validation is needed.