Modification of the Emergency Severity Index Improves Mortality Prediction in Older Patients

Abstract

Introduction Older patients frequently present to the emergency department (ED) with nonspecific complaints (NSC), such as generalized weakness. They are at risk of adverse outcomes, and early risk stratification is crucial. Triage using Emergency Severity Index (ESI) is reliable and valid, but older patients are prone to undertriage, most often at decision point D. The aim of this study was to assess the predictive power of additional clinical parameters in NSC patients. Methods Baseline demographics, vital signs, and deterioration of activity of daily living (ADL) in patients with NSC were prospectively assessed at four EDs. Physicians scored the coherence of history and their first impression. For prediction of 30-day mortality, we combined vital signs at decision point D (heart rate, respiratory rate, oxygen saturation) as "ESI vital," and added "ADL deterioration," "incoherence of history," or "first impression," using logistic regression models. Results We included 948 patients with a median age of 81 years, 62% of whom were female. The baseline parameters at decision point D (ESI vital) showed an area under the curve (AUC) of 0.64 for predicting 30-day mortality in NSC patients. AUCs increased to 0.67 by adding ADL deterioration to 0.66 by adding incoherence of history, and to 0.71 by adding first impression. Maximal AUC was 0.73, combining all parameters. Conclusion Adding the physicians' first impressions to vital signs at decision point D increases predictive power of 30-day mortality significantly. Therefore, a modified ESI could improve predictive power of triage in older patients presenting with NSCs.