Are We Asking the Right Triage Questions?

Abstract

The century-old definition of “triage” describes battlefield sorting of injured patients and allocation of treatment resources to maximize survival. However, that triage now bears little resemblance to emergency department (ED) triage, an act that occurs hundreds of thousands of times each day in modern EDs. Present-day ED triage not only helps sort patients by perceived need but also frequently is used as a broad term for all kinds of assessments unrelated to matching acute need and resources to improve outcomes. For example, triage personnel assess medicine and allergy use, ask about vaccinations and perceptions of safety, and seek information on potentially actionable behavior such as alcohol and drug use. Like others, we want to separate those latter functions—important but not necessarily best done on arrival despite the allure—and focus on our ability in an ED to improve outcomes for the greatest number. Setting aside the confusion created by conflating traditional triage with assessment, the existing body of ED triage research aims to validate the scoring tools used to help “sort” patients. Any tool used should create a consistent assessment across users absent clinical change between assessments, something achieved by common tools used. Consistency alone does not define utility; the best triage tool or process will optimize outcomes for a patient or group of patients. In this issue of Annals, Gravel et al report on an assessment of an ED triage scale. This observational cohort examined 550,940 children triaged with the Canadian Triage and Acuity Scale during a 1-year period. In addition to its size, the study is strengthened by the “real-world” setting of 12 pediatric EDs after implementation of the scale. Although each ED was university affiliated, the sites varied in volume, presence or absence of a fast-track short-stay unit, and proportions of patients admitted. Nurses applying the Canadian Triage and Acuity Scale score had differing clinical experience and training according to local practice. Additionally, the authors sought to minimize observational biases during analyses. All of these features mitigate many potential threats to study validity and generalizability. Gravel et al acknowledge the absence of standard reference criteria in validating triage scales, relying instead on construct p