Multiple performance measures are needed to evaluate triage systems in the emergency department

Area of Science:

• Emergency medicine
• Health services research
• Clinical decision making

Background:

• Emergency department (ED) triage systems function as prediction rules with ordinal outcomes.
• Misclassification in triage has varying clinical implications, necessitating careful performance evaluation.

Purpose of the Study:

• To evaluate strategies for comparing the performance of triage systems.
• To propose a set of recommended performance measures for future triage studies.

Main Methods:

• Literature review and expert knowledge were used to identify performance measures.
• A case study of 14,485 pediatric ED visits evaluated two triage modifications.
• Systematic appraisal of the strengths and weaknesses of various performance measures.

Main Results:

• Commonly reported measures include statistical association (34/60 studies) and diagnostic accuracy (17/60 studies).
• No single measure perfectly evaluates triage; decision curves offer advantages but require dichotomization.
• Paired diagnostic accuracy for dichotomized analysis and triage-weighted kappa/Nagelkerke's R² for ordinal analysis are recommended.

Conclusions:

• For comparing triage modifications, use decision curves and diagnostic accuracy in dichotomized analyses.
• Employ triage-weighted kappa and Nagelkerke's R² for ordinal analyses of triage systems.