Abstract
IMPORTANCE: The Emergency Severity Index (ESI) is the most widely used triage system in US emergency departments (EDs), but its performance in triaging children presenting with behavioral health symptoms is not well studied. OBJECTIVE: To assess the frequency of overtriage and undertriage and to identify characteristics associated with both among children presenting to the ED with behavioral health symptoms. DESIGN, SETTING, AND PARTICIPANTS: This retrospective cross-sectional study was performed in 15 US EDs participating in the Pediatric Emergency Care Applied Research Network (PECARN) Registry. Participants included children and adolescents aged 5 to 17 years presenting to EDs within the PECARN Registry with a behavioral health chief concern from January 1, 2021, to December 31, 2023. Data were analyzed from July 1, 2024, to January 15, 2026. EXPOSURES: Sociodemographic and clinical characteristics. MAIN OUTCOMES AND MEASURES: Appropriate triage, overtriage, and undertriage were defined using combinations of first-obtained vital signs, Glasgow Coma Scale, pain score, receipt of emergency medications, distinct resource types used (eg, laboratory tests, imaging studies), and disposition. Multivariable logistic regression assessed characteristics associated with overtriage and undertriage, compared with appropriate triage, adjusting for year and site effects. RESULTS: A total of 78 411 ED visits by children and adolescents with a behavioral health chief concern (37 328 [47.6%] aged 10-14 years; median age, 14.4 [IQR, 12.4-16.1] years; 47 496 [60.6%] female) were included in the analysis. Of 74 564 visits with nonmissing data, 25 668 (34.4%) were appropriately triaged, 42 589 (57.1%) were overtriaged, and 6307 (8.5%) were undertriaged. The adjusted odds of overtriage were higher for visits by children aged 5 to 9 years (adjusted odds ratio [AOR], 4.43; 95% CI, 4.13-4.76) compared with those aged 10 to 14 years and for visits by non-Hispanic Black (AOR, 1.17; 95% CI, 1.12-1.22) compared with non-Hispanic White patients. The adjusted odds of undertriage were higher for visits by Hispanic patients (AOR, 1.46; 95% CI, 1.31-1.63) and non-Hispanic Black patients (AOR, 1.28; 95% CI, 1.19-1.37) compared with non-Hispanic White patients and for those with Spanish language preference (AOR, 1.31; 95% CI, 1.11-1.54) compared with those preferring English. CONCLUSIONS AND RELEVANCE: In this cross-sectional study of children presenting to the ED with behavioral health symptoms, overtriage was common, and the likelihood of overtriage and undertriage differed by sociodemographic characteristics. Prospective studies are needed to assess behavioral health triage practices and to design triage systems that allocate resources accurately and equitably.