Development and Validation of a Novel Triage Tool for Predicting Cardiac Arrest in the Emergency Department

Abstract

Background: Early recognition and prevention of in-hospital cardiac arrest (IHCA) have played an increasingly important role in the chain of survival. However, clinical tools for predicting IHCA are scarce, particularly in the emergency department (ED). We sought to estimate the incidence of ED-based IHCA and to develop and validate a novel triage tool, the Emergency Department In-hospital Cardiac Arrest Score (EDICAS), for predicting ED-based IHCA. Methods: In this retrospective cohort study we used electronic clinical warehouse data from a tertiary medical center with approximately 100,000 ED visits per year. We extracted data from 733,398 ED visits over a seven-year period. We selected one ED visit per person and excluded out-of-hospital cardiac arrest or children. Patient demographics and computerized triage information were included as potential predictors. Results: A total of 325,502 adult ED patients were included. Of these patients, 623 (0.2%) developed ED-based IHCA. The EDICAS, which includes age and arrival mode and categorizes vital signs with simple cut-offs, showed excellent discrimination (area under the receiver operating characteristic [AUROC] curve, 0.87) and maintained its discriminatory ability (AUROC, 0.86) in cross-validation. Previously developed early warning scores showed lower AUROC (0.77 for the Modified Early Warning Score and 0.83 for the National Early Warning Score) when applied to our ED population. Conclusion: In-hospital cardiac arrest in the ED is relatively uncommon. We developed and internally validated a novel tool for predicting imminent IHCA in the ED. Future studies are warranted to determine whether this tool could gain lead time to identify high-risk patients and potentially reduce ED-based IHCA.

Development of a machine-learning algorithm to predict in-hospital cardiac arrest for emergency department patients using a nationwide database

“…Based on the SHAP analysis, we found that clinical variables associated with the occurrence of IHCA (i.e. oxygen supply and saturation, vital signs, and mental status) in previous studies were in line with variables that had a high influence on the output of the present model 32 – 34 . In addition to these key factors, our model included more clinical data such as medical history, symptoms, first aid, and hospital factors, increasing the predictive power over traditional statistical methods.…”

Physician gestalt for emergency department triage: A prospective videotaped study

“…Based on our previous data, 7 the AUROC of the model using the TTAS triage alone to predict hospital admission was 0.65. We further anticipated a difference of 0.075 in AUROC between the two triage methods.…”

“…Additionally, an electronic device is required to assist in triage evaluation. The TTAS also suffered from the same problem as the Emergency Severity Index (ESI), i.e., a large portion of undifferentiated Level 3 patients 7,8 . By contrast, physician gestalt perception, if carefully combined with some structured tools or data, might permit the defoliation of the many branches built into diagnostic trees 9 .…”

A New Vital Sign in Determining the Triage Category in Emergency Department Presentations: End-Tidal Carbon Dioxide

“…The importance of VSs monitoring in clinical practice is indisputable; however, it remains unclear which VS is more associated with the clinical risk of patients, at what interval VSs should be evaluated, and the best way to monitor them [6]. There are many studies in the literature examining the relationship between VSs measured at the time of presentation to the emergency department and mortality [23][24][25]. However, only a few studies have explored the relationship between VSs alone and hospitalization indications.…”