Skip to main content

NHS Pathways: Designing with clinical equity

Siama Latif, Senior Clinical Informatician at NHS Digital, looks at the process behind how we triage COVID Oximetry @home patients using NHS 111 or 999 services.

NHS Pathways is a clinical tool used for assessing, triaging and directing the public to urgent and emergency care services. There is no ‘guidebook’ for remote clinical triage – the clinical team at NHS Digital created the tool using our clinical expertise and by extrapolating information from as many sources as possible. We are constantly updating and refining the tool with new evidence.

Older woman measuring her oxygen saturation with a pulse oximeter

We have many dilemmas while doing this, resulting from a variety of moral, clinical and operational uncertainties, and we face many unique ethical and legal considerations within NHS Pathways.

I recently completed a master’s in Medical Ethics and Law at King’s College London that has helped me think about these decisions. What is the “right thing” to do? And how do we go about answering that seemingly simple question in a complex environment where there are knock-on consequences for many different people, whichever direction we choose to go in?

The challenge

Our triage for COVID Oximetry @home patients is an interesting example of this. We were asked to implement a line of questioning for health advisors answering NHS 111 and 999 calls to ensure that patients who were known to the COVID Oximetry @home programme would receive the agreed level of care if silent hypoxia occurred.  

Silent hypoxia is often associated with late and severe COVID presentations and can be deadly. It is termed ‘silent’ as it occurs where oxygen levels become very low without obvious symptoms. It is impossible to identify without the use of an oxygen saturation monitor (pulse oximeter), prompting the development of the national COVID Oximetry @home programme. 

COVID-19 patients were identified by health care professionals to be at risk from silent hypoxia. They were provided with a monitor to use at home and given advice on when to call for medical assistance.

Our challenge was to ensure the safety of those calling, but we also had to think about the impact of decisions on the population as a whole.

Those who have an oxygen saturation reading of 92% or less should receive an emergency ambulance (‘category 2’) disposition. For those with higher readings (93-94%), additional triaging was necessary. Some may still need emergency care while others might be better managed at home. But how should we develop triage questions to distinguish between these patients?

Our challenge was, obviously, to absolutely ensure the safety of those calling, but we also had to think about the impact of decisions on the population as a whole, as well as the 111 and 999 providers. Too many inappropriate ambulances would divert care from other callers who had a clinical need, and not sending enough could cause harm to those oximetry patients that needed help.

We worked with the NHS England COVID Oximetry @home team and other senior clinicians involved with silent hypoxia and reviewed recent evidence and literature. Among us, there was initial concern that the oxygen saturation 'number' was being treated and not the patient, which could lead to both over- and under-referral, especially as oxygen saturation readings are not always accurate.

With this in mind, we considered the parameters which might affect the accuracy of the reading. That includes skin colour.

We needed to get beyond a one-size-fits-all approach.

There were concerns at the time that people with darker skin tones did not always have accurate oxygen saturation readings, which could have led to a double inequity:  Not only do they have more chance of being seriously ill than those with lighter skin tones, but the tools we were using, and the references we had, might not always have identified them as accurately as other groups. We needed additional questions and safety nets to make sure these patients weren’t harmed. Since then, our knowledge of oxygen saturations in people with darker skin tone has greatly increased.

An 'abnormal reading' might also be normal to a subset of the population due to underlying medical problems or user error. We needed to get beyond a one-size-fits-all approach and consider how we could manage callers with equity. For example, those who happened to have an oxygen saturation monitor and had a low reading should not be disadvantaged (as they would not be part of the Oximetry @home population we were focussed on).

We worked through these considerations carefully as we developed the lines of questioning. The changes had to be simple enough for a non-clinically trained call handler to understand easily.

What happened next?

We managed to identify early outcome data based on Hospital Episode Statistics (HES) metric data. About 90% of patients who called 111 and about 80% of those who called 999 can be followed through the health care system using this anonymised data.

I will focus on 2 populations: those with a very low oxygen saturation (less than 90%), and those who had an oxygen saturation of 91-92% and were unwell.

Headline figures from NHS 111

For those with oxygen saturations less than 90%:

Approximately half who called were admitted to hospital.

For those with oxygen saturations of 91-92% and unwell:

Just under a third were admitted.

For 999, the numbers were much smaller and the admission rates were higher.

Did we improve outcomes?

It is unclear whether any of these patients would have been identified at the ‘category 2’ ambulance level by other NHS Pathways questions, but it is likely that most would not have been without the use of pulse oximeters. Therefore, it is likely we have improved outcomes for the individual patients. The outcome data helped confirm this.

Were ambulances dispatched proportionately?

The numbers for this population are quite low and they retain a high degree of specificity for all populations concerned, so it is likely they were dispatched proportionately (for NHS 111, half of patients with oxygen saturations 90% or less and a third of patients with oxygen saturations of 91-92% who were unwell were admitted at ‘category 2’ ambulance level). This is important, because if we over-refer or have a low degree of specificity, others who are truly in need may be impacted adversely by having to wait longer.

Did we benefit patients, the 111 and 999 providers and urgent and emergency care?

Patients benefitted with a good patient journey. As we tried to be equitable, we asked everyone whether they had an oxygen saturation reading, not just those who might be silently hypoxic.

This new line of questioning allowed the health advisors taking the calls to be able to manage them more succinctly and without needing support from clinicians to interpret medical data. We therefore assume the changes have helped both the 111 and 999 services as well as patients.

Did we follow what is expected of us by the population, medical experts and urgent and emergency care?

The changes made were in line with expectations about how callers should be managed by urgent and emergency care. There is a general expectation that very sick people will be managed as a priority. Two of our 111/999 providers have requested that this line of questioning be extended into other Pathways triage algorithms.

Did we manage people who are in the protected characteristic groups and others with equity and equality of opportunity?

This is harder to answer as currently, the picture we’re getting from our equality data is not complete and does not give us a solid quantitative evidence base in some areas. For example, 3.4% of the population in the UK is comprised of those who are in black communities (including Black African, Black Caribbean, Black American or Black European). This percentage is also lower in the older age groups, where complications of COVID-19 are also more likely to present. We do not know the ethnic make-up of our callers at specific questions or how the outcome data corresponds to them.

We can say that we put in safety catches and backstops to account for any possible inaccuracy due to skin colour – for example, those with a darker skin tone that had a reading between 92 and 95% needed to speak with a clinician immediately. The focus is on serving the individual not the oximeter reading.

Evaluating success

The ethical judgments involved in a nationwide service like NHS Pathways are complex. Traditional approaches founded on direct relationships between an individual clinician and an individual patient only take us so far. Indeed, the complex relationship between balancing the autonomy of individual patients, doing good, preventing harm and considering justice is a well-known ethical framework in clinical medicine defined by Beauchamp and Childress.

A simplistic consequentialist approach – maximising care for the most people – doesn’t capture the complexity of the moral environment we are working in. The people most likely to come to harm in the pandemic have been people with learning disabilities, people from some ethnic minorities and the very old. These are minority populations, often with different lived experiences and perspectives from the mass of the population.

We are not in a situation where all populations, let alone conditions, are equally visible or understood.

Within a pandemic, Professor Julian Savulescu from the University of Oxford argued a more complex consequentialist approach can be beneficial, but this is not the everyday reality of telephone triage outside of a pandemic, and, for this to be successful, data and knowledge is key. We must study all data and evidence and look at the knock-on consequences of lines of questioning because they can have system-wide consequences.

However, we are not in a situation where all populations, let alone conditions, are equally visible or understood. If we have not lived the experiences of a minority population group, such as people in the transgender community or people with learning disabilities, are we in a good position to apply generic values to decisions about the service they receive? If we don’t know what people consider to be right for themselves, how can we do the right thing for or with them, when needed? How do we know what we do is meaningful for them, respects their autonomy and treats them with dignity?

More information and more analysis, particularly in relation to different ethnic groups and their experience, is always needed.

In the context of Pathways, that means being as open as we can possibly be to the data, considering as many different populations as possible through all mechanisms, collaborating closely with our partners, specialist groups and patients, and thinking through the “what if” scenarios as deeply as we can to ensure we manage people equitably and to provide a meaningful patient journey.  The early feedback and initial outcome data on the Oximetry @home lines of questioning looks very positive, but more information and more analysis, particularly in relation to different ethnic groups and their experience, is always needed.

In July 2022, NHS Digital started a managed closure of the COVID Oximetry @home data services it provides to clinical commissioning groups (CCGs). This was to align with the end of the COVID-19 COPI notices and to reduce data collection burden. The closure is now complete. A national supply of oximeters remains available by request to support COVID Oximetry @home services.

If you are a patient or member of the public, please visit the NHS website for more information about looking after yourself at home, including information on pulse oximeters. If you have coronavirus and think you are eligible for the service, please contact your GP.

Related subjects

Find out about NHS Pathways, a clinical tool used for assessing, triaging and directing the public to urgent and emergency care services. Learn more about the about the benefits of the service, how it works and how to use it.


Last edited: 29 September 2022 1:31 pm