Development of an emergency department triage tool to predict admission or discharge for older adults International Journal of Emergency Medicine Int J Emerg Med 18, 26 (2025). https://doi.org/10.1186/s12245-025-00825-3
Date
02/14/2025Abstract
Background
Older adults present to Emergency Departments (ED) with complex conditions, requiring triage models that support effective disposition decisions. While existing models perform well in the general population, they often fall short for older patients. This study introduces a triage model aimed at improving early risk stratification and disposition planning in this population.
Methods
We analyzed the National Hospital Ambulatory Medical Care Survey data (2015–2019) for ED patients aged ≥ 60 years, excluding those who died in the ED or left against medical advice. Key predictors were identified using a two-step process combining LASSO and backward stepwise selection. Model performance was evaluated using AUC and calibration plots, while clinical utility was assessed through decision curve analysis. Risk thresholds (< 0.1, 0.1–0.5, > 0.5) stratified patients into low, moderate, and high-risk groups, optimizing the balance between sensitivity and specificity.
Results
Of 13,431 patients, 3,180 (23.7%) were admitted. Key predictors for admission included ambulance arrival, chronic conditions, gastrointestinal bleeding, and abnormal vital signs. The model showed strong discrimination (AUC 0.73) and good calibration, validated by 10-fold cross-validation (mean AUC 0.73, SD 0.02). Decision curve analysis highlighted net benefit across clinically relevant thresholds. At thresholds of 0.1 and 0.5, the model identified 18.9% as low-risk (91.2% accuracy) and 7.9% as high-risk (57.7%). Adjusting thresholds to 0.2 and 0.4 expanded low-risk (55.4%, 87.9% accuracy) and high-risk (14.1%, 53.7% accuracy) groups.
Conclusions
This older adult–focused risk score uses readily available data to enhance early discharge, prioritize admissions for high-risk patients, and enhance ED care delivery.
Author List
Abugroun A, Awadalla S, Singh SS, Fang MCAuthor
Sanjay Singh MD, MBBS Assistant Professor in the Medicine department at Medical College of WisconsinView Online
