Evaluating Clinical Decision Tools for Intensive Care Unit Admission Prediction in Patients With Coronavirus Disease 2019.

Abstract

Introduction: During the coronavirus disease 2019 pandemic, several clinical decision rules were promulgated to support triage and optimize resource allocation, particularly regarding intensive care unit admissions. This retrospective cohort study aimed to validate and compare the predictive performance of 3 such systems: the National Early Warning Score 2, the Pandemic Respiratory Infection Emergency System Triage score, and the American College of Emergency Physicians coronavirus disease 2019 severity classification. The primary outcome was intensive care unit admission within 24 hours after ED presentation.

Methods: Data were collected from unvaccinated patients with coronavirus disease 2019 presenting to the emergency department of Antwerp University Hospital between March 1, 2020 and June 27, 2021. The National Early Warning Score 2 was calculated during routine care, and the Pandemic Respiratory Infection Emergency System Triage and American College of Emergency Physicians coronavirus disease 2019 severity classification scores were calculated retrospectively. Diagnostic performance was assessed using the area under the receiver operating characteristic curve (AUC), positive likelihood ratio, and negative likelihood ratio.

Results: Among 490 patients, optimal cutoff values were National Early Warning Score 2 of ≥3, Pandemic Respiratory Infection Emergency System Triage score of ≥8, and American College of Emergency Physicians coronavirus disease 2019 severity classification of moderate severity. All scores demonstrated good discrimination: areas under the receiver operating characteristic curve were 0.885 (National Early Warning Score 2), 0.851 (Pandemic Respiratory Infection Emergency System Triage), and 0.892 (American College of Emergency Physicians coronavirus disease 2019 severity classification), with no significant differences. The American College of Emergency Physicians coronavirus disease 2019 severity classification had the highest positive likelihood ratio (3.79), significantly higher than the National Early Warning Score 2 (2.48; P<.001). The National Early Warning Score 2 had the lowest negative likelihood ratio (0.04), although differences were not statistically significant.

Discussion: Although the American College of Emergency Physicians coronavirus disease 2019 severity classification demonstrated superior rule-in characteristics, the National Early Warning Score 2 offered the best rule-out characteristics. Given its simplicity, widespread use, and rule-out potential, the National Early Warning Score 2 may offer the greatest practical value to predict intensive care unit admission 24 hours after ED presentation of unvaccinated patients with coronavirus disease 2019, particularly in a pandemic context.