Skip to Content

Electronic Triage System Risk-Stratifies ER Patients

MONDAY, Oct. 2, 2017 -- An electronic triage (e-triage) system based on machine learning can predict the likelihood of acute outcomes, enabling improved patient differentiation, according to a study published online Sept. 6 in the Annals of Emergency Medicine.

Scott Levin, Ph.D., from Johns Hopkins University in Baltimore, and colleagues retrospectively analyzed 172,726 emergency department visits from multiple urban and community hospitals. Outcomes predicted by e-triage for elevated troponin and lactate levels were compared with the Emergency Severity Index (ESI).

The researchers found that e-triage predictions had an area under the curve ranging from 0.73 to 0.92 and demonstrated equivalent or improved identification of clinical patient outcomes versus ESI. For the ≥65 percent of visits triaged to ESI level 3, e-triage provided rationale for risk-based differentiation. E-triage identified more than 10 percent of ESI level 3 patients requiring up triage who had substantially increased risk of critical care or emergency procedure (1.7 percent ESI level 3 versus 6.2 percent up triaged) and hospitalization (18.9 percent versus 45.4 percent, respectively) across emergency departments.

"E-triage more accurately classifies ESI level 3 patients and highlights opportunities to use predictive analytics to support triage decision making. Further prospective validation is needed," conclude the authors.

Several authors have been supported by an award to commercialize e-triage, and Johns Hopkins University has filed a patent application for e-triage.

Abstract/Full Text (subscription or payment may be required)

© 2019 HealthDay. All rights reserved.

Posted: October 2017