Article Text

Download PDFPDF

How good is triage, and what is it good for?
  1. Kirsty Challen
  1. Emergency Department, Lancashire Teaching Hospitals NHS Trust, Sharoe Green Lane, Preston, UK
  1. Correspondence to Dr Kirsty Challen, School of Health and Related Research, The University of Sheffield, Regent Court, 30 Regent Street, Sheffield S1 4DA, UK; kirsty.challen{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Two papers in this month’s issue of EMJ systematically examine the validity of triage systems for adults and children1, while Tsai et al analyse the performance of prehospital triage criteria in Taiwan.2 Few emergency physicians have the luxury of being able to see all patients immediately and thus not needing triage.

Firstly, before we can find a good triage system, we need to consider why we triage. The underlying principle of triage is a utilitarian ethic of maximising the overall good, originally described in the military setting: ‘those who are dangerously wounded should receive the first attention, without regard to rank or distinction’.3 It presupposes that the patient with an ankle sprain does not benefit from waiting while the patient with ST elevation myocardial infarction (STEMI) is seen before him, but the population as a whole does. To achieve this, the ideal triage would identify all those patients who will either benefit from a time-critical intervention or will suffer harm if a time-critical intervention is not provided.

The problem arises in defining this benefit and harm. On the surface, saving life and reducing morbidity is a straightforwardly identifiable benefit—but how many emergency interventions have good evidence of a time-critical benefit? CPR, defibrillation, reperfusion in STEMI, certainly, but few others.

If as evidence-based practitioners we use triage as a diagnostic test or decision tool, this absence of a gold standard is a problem. In Kuriyama et al’s review, a minority of studies address criterion validity, generally using a gold standard of predefined expert consensus on urgency.1 This is pragmatic but possibly not complete; there is little sign of patient and public involvement and as a professional group, we have patchy insight into what our patients might value. Manchester Triage prioritises patients in severe pain, but we know patients suffer in ways beyond pain4; should they be prioritised too? And what of children? The elderly? Those we know are in the last hours of their life? In the absence of a universal understanding of suffering, how is it possible to develop tools to prioritise those whose suffering can be addressed?

Practically speaking, we often use construct validity instead. This involves analysing how closely a triage tool parallels other measures that might feasibly be related to urgency, like admission rates, mortality or use of investigations. Care is needed here too though; we recognise triage cueing as a cognitive bias (where perception of risk in a patient is influenced by the area to which a patient is triaged).5 The risk is of circularity—patients in high triage categories are more likely to receive investigations simply because they are placed in a higher acuity area. In this issue of the journal, Tsai et al conclude that the existing 2-level prehospital triage system in Taiwan is inferior to a 5-level system in the ED in predicting admission (my italics). If, however, I reflect on my own ED, many of our admitted patients are from the elderly frail population who need several days of multidisciplinary care rather than a blue-light arrival to resus. Severity of illness (the risk of it causing harm) is not the same as urgency (the potential to benefit from treatment delivered now).

There is also an apparent desire (at least in the published literature) for the triage process to identify particular diagnostic entities. Gräff et al 6 highlight the limitations of Manchester Triage in septic patients; Magalhaes-Barbosa et al’s 7 review includes prediction of serious bacterial infection in children; a brief foray into PubMed locates publications on the early identification of stroke, STEMI and pulmonary embolism (PE). We need to reflect on this mission creep in triage; for every tweak that increases diagnosis of one condition, there is likely to be a fall in sensitivity for something else.

So where does this leave the clinician or the department seeking a ‘good’ triage system? It needs reflection on where triage fits in your specific model of patient flow and risk management. Ask not what your triage can do for you, but what you are asking it to do.



  • Competing interests None declared.

  • Provenance and peer review Commissioned; internally peer reviewed.

Linked Articles