The word “Triage” comes from French roots, the origin word “trier” meaning “to sort”.1 This is a process vital to emergency medicine. The idea was born on the battlefields by Baron Dominique Jean Larrey2, a chief surgeon in Napoleon’s Imperial guard, because the number of casualties demand exceeded resources available. The need for a way of prioritising patients was described beautifully by Larrey in his memoirs:
“Those who are dangerously wounded should receive attention first, without regard to rank or distinction. They who are injured in a less degree may wait until their brethren in arms, who are badly mutilated, have been operated on and dressed; otherwise the latter would not survive many hours; rarely, until the succeeding day”2
From the first systems created on Napoleon’s battlefield to our modern day emergency departments, we still encounter situations when demand exceeds resources. In the UK, demand for acute services almost always outstrips supply and thus we frequently have to make priority decisions. There is no doubt that we require a way of sorting out our patients so that we can then direct our resources to those who are in most in need or to those who will benefit the most from early intervention.
There is inevitably some variability in how this is achieved. Each Emergency Department will have subtle differences in their individual ways of sorting, but underpinning local practice it’s likely that they will be using some sort of ‘system’. Worldwide there are four main triage systems that differentiate patients into 5 tiers of priority; ATS 3, CTAS4 MTS5 and ESI. They all allocate patients into one of five time related categories, each determining how long a patient can wait to be seen by a trained physician6.
Not all presentation and diseases are equally time critical to the emergency physician. To a
cardiologist, the patient having a myocardial infarction is clearly the most important in the department, as ‘time is myocardium’; for the medic, the patient with sepsis or suspected neutropenia might demand the most urgent attention in order to give timely, targeted antibiotics; to the trauma surgeon, the incoming trauma patient may take priority as they have the potential to have life-threatening injuries. Within this, let us not forget the patient themselves – what is most important to them? In order to prioritise emergencies, one will be seen first, the others will wait. Who should that be? How should that be judged in a manner that is fair, honest, reproducible and correct? Perhaps it should be the patient who is deemed to be suffering the most, but what do we mean by ‘suffering’ and who determines that scale?7–9
Triage systems are used on every patient registering in an emergency department, and they are used to determine who gets seen first (and last). We should expect that systems are based on high quality evidence that can validate and justify how one patient with chest pain gets seen before another with an ankle injury. The tools that we use should have face and evidential validity if they are to be acceptable to clinicians and patients alike. Such acceptability implies that they must be correct, that there is some standard to which we can measure them and upon which we might all agree. Thus to validate any tool, we require a standard – ideally a gold standard – to measure it against. Therin lies the rub; what is the standard of a triage score? Stop and think at this point: ask yourself what you want an ED triage score to deliver.
For us and many others interested in triage research, a major stumbling block is that no gold standard has been set, or that researchers into triage have effectively chosen their own to suit their personal agendas (Ed – bit controversial but I do agree). Looking at the literature it does rather appear that much of the research out there uses some form of surrogate marker of triage ‘success’
What is the measure of a well performing triage system?
Some research, such as the paper recently published in the EMJ 10,11, looks at whether triage makes an accurate diagnosis (in that paper the authors examine whether it can predict sepsis). Should triage be judge by its ability to correctly diagnose? Triage systems were never set up for this; accurate diagnosis is a process that requires at least a focused and accurate history, together with a correctly interpreted examination of a patient. We do not require a diagnosis in order to treat a patient who is acutely unwell and arguably emergency medicine is one area where we must master the art of undifferentiated resuscitation.
It’s also worth considering that diagnosis doesn’t always equate to severity of illness. For example, a pneumothorax is a diagnosis that represents a spectrum of severity, from those who require life saving to treatment to others who require nothing more than some simple advice and follow up. So can triage accurately diagnose? If it does then it’s purely an association rather than a specified aim. I am unconvinced this is a marker of success, or indeed what we require of our triage system.
What about mortality and serious illness? Several triage research papers use mortality as a means to validate triage scores. Unsurprisingly there is an association between a triage score and mortality: the more urgent the triage category, the higher the inpatient mortality.12 This may seem appropriate, but from the perspective of the emergency physician it’s still not a great measure. Surely we want to identify those patients in whom earlier intervention will makes the most difference and that’s not the same as eventual outcome. Are the patients who die in hospital the ones who are the highest priority at the front door? Mortality clearly has an association with disease severity, and it would make sense that patients with higher priority are the patients with greater disease severity and therefore this cohort would have a higher mortality. However should the prime function of triage be to highlight the patients with a high mortality?
Arguably there is a paradox with scores seeking mortality. If a score is very good at identifying patients for whom there is no available intervention and for whom early intervention will make no difference it’s not really of very much help at all to the emergency physician and might arguably be a distraction, or at the very least an opportunity cost to another patient who has a beter chance of survival. In a system where we need to identify priorities then surely we need to identify patients who would benefit from early intervention. We should be seeking those patients for whom there is something that we can do which will influence and improve the clinical trajectory of that patient.
Inpatient admission and admission to intensive care are other surrogate markers commonly used by researchers. There is an association between increasing frequency of hospital or intensive care admission and increasing triage urgency.13 This isn’t quite the right marker of triage success either; thinking back to the patient with a pneumothorax, a tension pneumothorax needs urgent assessment and treatment however not all these patients go to intensive care and some may even be discharged.
In a system that only highlights patients who have the most severe diseases, are there things we may miss? The answer to this question is clearly yes. The obvious example is pain. Patients in pain do not necessarily have a disease that is presenting as a medical emergency, but I would argue they need rapid prioritisation and treatment in order to prevent or reduce their suffering.
Measuring patients’ urgency based on need for life-saving interventions in the department seems logical, however is this a measure to judge triage validity by? For example, a patient with a presenting complaint of breathlessness may have a tension pneumothorax, or they may be having a panic attack. This leads us back to the discussion around whether triage should be diagnostic of specific pathologies, in order for the system to determine accurately and reliably which patients require immediate life saving intervention.
Major incident researchers have tackled these challenges to some extent14,15, using Delphi based studies to determine patient priorities against clinical interventions.16–18 These studies go some way towards understanding a link between illness, injury and intervention, but the major incident setting is different from the front door of the ED. The categories are simpler, the outputs understandably imprecise and they are linked to organisational priorities as well as clinical ones.
I don’t believe that judging triage against diagnosis, mortality, hospital admission or patient suffering alone are quite the right markers of a successful triage system. We haven’t found the perfect maker of triage success yet, the “just right” bowl of porridge is still awaiting us – or should we just settle for the current scores? These are essential questions if we are to improve the way we assess patients and in how we approach the validation of triage scores. What is clear in the literature at the moment is that the ‘gold standard’ applied in many studies may be a fool’s gold and the true treasure is yet to be determined or found.
Are we the only people who think this?
Well no. We’ve already alluded to the work that our South African and military friends have put together on this, but co-incidentally (and pointed out to us after publication of this blog) there are several excellent articles in this months EMJ on similar subjects. Kuriyama et al note that few studies of triage have a coherent approach to triage gold standards.19 . In paediatric practice similar problems exist with difficulties in comparing scores across studies and populations 20. Finally, you really need to read the excellent commentary by Kirsty Challen that explains why triage is not quite what it seems and why we have a long way to go in understanding what it is that we really want from a triage score 21.
So what do we want from a triage score?
Frustratingly, there is no simple answer for this. Perhaps the key word in the question above is ‘we’ as it is the perspective of the questionner that currently determines success. Speaking as an emergency physician, my belief is that the initial ED triage score should be able to identify patients for whom an early intervention will change the clinical course of the patient. All of us here a in #Virchester would love to know your perspective, so add some comments below or comment on social media. We look forward to hearing from you.