There have been many concerns regarding the discriminatory effects of triage frameworks. Are these well-founded?
Is triage discriminatory?
Patients with disabilities have more complex health requirements and a higher propensity to poor health. Despite this, they have less access to healthcare services, and tend to receive substandard care compared to the general population (1). The relationship between low socioeconomic status and poor health is also well-established — social determinants of health, such as income, social status, employment, and working conditions, have been found to be key in determining life expectancy and the development of illness (2).
When hospitals reach their maximum capacity, they implement triage frameworks to allocate resources and maximise survival. Triage frameworks provide a system for assessing a patient’s health and recommend further management options. Triage guidelines may exacerbate these pre-existing health inequalities and social imbalance, selecting against patients whose poorer background health make them less likely to make a full recovery after treatment. Thus, triage guidelines can appear to perpetuate a cycle of widening health inequalities, as vulnerable groups who suffer from poorer health than the general population are “selected against” in favour of “younger, healthier, and wealthier” people, in a phenomena that has even been denounced as modern-day eugenics.
Where is the problem ?
In the UK, clinicians are recommended to use the framework below (3). We will use this example to illustrate a few problems that are inherent in some guidelines, and ultimately lead to discrimination.
1. Selection should depend on demand for resources
It is implicit that if there are enough resources, there is no need to select who gets the treatment, and thus no need for triage. The notion of triage is intricately linked to some kind of comparison between patients. The algorithm above, however, draws no comparison between patients — it merely instructs doctors to manage all patients meeting certain criteria in a given way. What happens if resources are depleted, and not all of the patients who require certain care can access them? The framework offers no guidance to doctors regarding how to select amongst candidates fairly. Conversely, what happens if resources are not fully occupied? The framework advises doctors to deny some patients potentially life-saving critical care, regardless of whether these resources are available or not. The aim should be to strive to provide life-saving treatment to all patients that would have routinely received it, until demand is so high that choices become inevitable.
2. Good triage needs to be clear
These guidelines recommend different management plans depending on the patient’s age and frailty; however, they run into problems when trying to account for factors like disability or co-morbidity. Clinicians are told to ‘consider co-morbidities and underlying health conditions in all cases.’ The indications for this are unspecified, and don’t offer any further decision-making support. Faced with a new condition, as is COVID-19 pneumonia, how do clinicians have the expertise to assess how a patient’s co-morbidities will affect the short-term outcome, and thus make life or death treatment decisions based on this? Triage guidelines need to be clear and applicable throughout, which many guidelines often are not. Otherwise, they fail to offer decision-making support for doctors, who are unable to objectively apply the framework.
3. Good triage does not rely on ‘stereotypes’
Most concerning, however, is the fact that patients are classified according to criteria that may not adequately predict treatment benefit. The guidelines recommend clinicians to typecast patients by age, outward frailty, and co-morbidities, rather than to look at the patient in front of them and make decisions based on a patient’s current clinical status and vital observations — like oxygen saturations or respiratory rate. Age, disabilities, and co-morbidities affect patients in different ways. For example, the impact of diabetes mellitus on a patient’s health varies greatly depending on many factors, like their glycaemic control and the length of time they have had diabetes for. Information garnered through other tests, like a blood test, offers superior prognostic value than merely assigning 1 point or somehow penalising all diabetic patients — as many guidelines do (3, 4, 5, 6).
A front-line NHS consultant interviewed by the Financial Times corroborates this. ‘The framework’s approach is just a guide; we make the judgment taking into account a lot of information about the current ‘nick’ of the patient — oxygenation, kidney function, heart rate, blood pressure — which all adds into the decision making’, they explain (7). It is a respite to find that such triage as is recommended by the published triage guidelines are not implemented by most physicians. Nonetheless, offering such guidance to physicians amounts to offering no guidance at all, and is a great failing towards both the patients and the clinicians it ought to support.
Not all guidelines are made ‘unequal’
Many frameworks rely more heavily on individualised assessment to assess a patient’s prognosis. For example, in Maryland (U.S.A.), guidelines use the Sequential Organ Failure Assessment (SOFA) score to gauge short-term survival. SOFA scores are based on several measurements that indicate a patient’s respiratory, cardiovascular, liver, renal, nervous, and coagulation functions. They are routinely used to determine the extent of organ failure in a critically-ill patient, thus predicting the likely clinical outcome. In the event of several patients scoring the same SOFA score, long-term survival would be taken into account, and, in the event that the priority scores are still the same, age criteria would then be applied (8).
Guidelines like these protect vulnerable patients who had poor background health, as resources are allocated on the basis of a patient’s likelihood of surviving the current disease (eg. COVID-19 pneumonia), as indicated by markers that reflect a patient’s current function. These guidelines are a great leap forward from the NHS NICE guidelines, which recommend denying treatment from the outset on the assumption that a patient is unlikely to do well merely because he or she is old, frail, or co-morbid.
The incorporation of long-term survival and age into the decision-making process is a very contentious topic. The guidelines recognise their contentious nature, and acknowledge the potential for such principles to lead to unfairness. They attempt to remain just and unbiased by clearly indicating that the foremost principle guiding triage is survival — not medium or long term, but surviving COVID-19 pneumonia. Long-term survival and age only serve as further tie-breaking criteria, which would only be taken into consideration if the strain on resources is so great so as to render it necessary. They also seek to mitigate any discriminatory effect by stipulating that long-term survival criteria only affect patients ‘whose co-morbidities are so serious that they are expected to live no more than 12 months even with successful ICU treatment’ (8).
One step forward
Nevertheless, the Maryland guidelines are still not ideal. SOFA scores have not been developed for the purpose of guiding triage decisions, and ‘a growing body of evidence suggests such scoring systems are unlikely to predict critical care outcomes with sufficient accuracy’ (9). The use of an inappropriate scoring system may be unfair towards some patients with poor background health. Guidelines need to be based on relevant evidence specific to the situation — not generic criteria. In order to ensure the fairest allocation possible, research into the specific factors which hold prognostic value needs to become a priority. This may be time-consuming in the midst of the chaos of an emergency; however, in an age abounding with data science methods and patient health data to apply them to, this process could be optimised in an efficient way that would render it possible, even amid chaos and scarce personnel and resources.
There have been many reasonable concerns regarding the discriminatory effects of triage frameworks. Most triage guidelines seek to account for this possible discrimination by giving much leeway for clinicians to implement their individual clinical judgement. We argue, however, that triage guidelines should strive to provide clinicians with practical, applicable advice, which takes into account the latest evidence-based research while also safeguarding patients from unjust discrimination. We argue that guidelines should be primarily based on clinical observations shown to be have an impact on the likelihood of short-term survival. These criteria would better gauge a patient’s state, and offer superior prognostic value than the often discriminatory “clinical stereotypes” that are the mainstay of many guidelines. These modifications would greatly reduce triage’s discriminatory effects, while enabling resources to be allocated in the optimal way — as all triage ought to do.