Missed signals: the limits of the pulse oximeter

I

n December, it emerged that pulse oximetry – a commonly used technology for measuring blood oxygen levels – may not be as accurate in Black patients as it is in white ones.

Writing in the New England Journal of Medicine (NEJM), researchers at the University of Michigan stated that Black patients in their sample were far more likely to suffer hypoxemia (low oxygen in the blood) that went undetected by pulse oximeters.

The researchers looked at patients who, according to pulse oximeter readings, had a blood oxygen saturation of 92%-96%. They re-tested these patients via an arterial blood gas test – a more accurate tool used in hospital settings.

A shocking 11.7% of the Black patients had levels under 88%, a level at which medical professionals would likely take action. This compared to just 3.6% of white patients.

“What we did in our study was ask a fairly simple question, which is when the pulse oximeter is reading in an acceptable range, how often is the real oxygen saturation in the blood at a lower level?” says Dr Michael Sjoding, assistant professor at the University of Michigan and lead study author.

“We found it to be the case a lot a lot more frequently in patients who are self reported as Black rather than white. It’s quite concerning that the pulse oximeter does not perform as well in this patient population.”

The probable reason for the discrepancy has to do with the way skin pigmentation affects light transmission. It’s a bit like adding static to your radio signal.

“The pulse oximeter measures how much light is transmitted through the finger versus absorbed, meaning the pulsar camera can calculate the oxygen level in the blood,” says Sjoding. “The problem is that melanin, which is what causes skin pigmentation, is likely also reflecting that light, and interfering with the pulse oximeter’s ability to figure out the oxygen level.”

Especially since the start of the pandemic, pulse oximetry has proven to be a valuable tool.

Small, handheld units that slip over the middle finger and shine a light into the body, these devices are cheap and can even be found sold on Amazon.

In the case of Covid-19, they can flag up ‘silent hypoxia’ – dangerously low oxygen levels in the body, which the patient themselves may not have noticed.

In the UK, the NHS has purchased around 200,000 pulse oximeters for use during the pandemic. Covid-19 patients who don’t need immediate hospital attention – but who are nonetheless at risk of deterioration – have been given these devices to monitor their symptoms at home. If their oxygen levels fall too low, the patient will likely be admitted to hospital.

In the US, consumer interest in these devices soared at the start of the pandemic. Sales spiked by 527% when the first US case of Covid-19 was identified, even though the American Lung Association has warned against buying one unnecessarily.

Any inaccuracies, then, could have far-reaching implications – particularly those pertaining to skin colour. Throughout the pandemic, Black and Asian populations have consistently suffered worse than white populations, with a disproportionate number of cases and deaths occurring among those from BAME backgrounds.

“If we're basing our medical decisions on a pulse oximeter that's potentially missing on low oxygen in Black patients, that could have a significant impact on their care and their health,” says Sjoding. “It’s easy to imagine how if I knew that the patient had low oxygen, I would treat them differently.”

In February 2021, the US Food and Drug Administration (FDA) issued a warning about the limitations of pulse oximetry, pointing out a number of factors (including skin temperature and nail polish) that might skew a reading. It also mentioned skin pigmentation as a potentially limiting factor, although it did not go into details, or mention the NEJM study directly.

“While pulse oximeters may be useful for estimating blood oxygen levels, these devices have limitations that can result in inaccurate readings,” said Dr William Maisel, director of the Office of Product Evaluation and Quality in the FDA’s Center for Devices and Radiological Health.

“Patients with conditions such as Covid-19 should not rely solely on pulse oximeter measurements to monitor their health at home as they are not a substitute for a medical diagnosis by a health care provider.”

As to how this situation arose in the first place, Sjoding remarks that the device was evaluated and studied mostly in lighter skinned patients and was probably calibrated using their data.

“Even now, current guidance for evaluating pulse oximeters doesn't really have what I would consider robust requirements for ensuring that the device works the same in all patients,” he says.

“There are recommendations to ensure that the test population has a variety of patients in it, and there is increasing awareness that that’s important, but this is a really good example of why that’s important.”

There is another twist in this story too, which is that the racial disparities in pulse oximetry have actually been recognised for some time. As early as the late 1980s, studies began to show that pulse oximeters might be affected by skin tone. This finding was confirmed by a series of publications throughout the 1990s and 2000s. However, the issue received scant attention.

“Even as a practicing pulmonary doctor, I was unaware of previous research showing that the pulse oximeter is less accurate,” says Sjoding. “When we were caring for Black patients in the hospital with Covid-19, there were a few occasions where I noticed that there was this discrepancy, which I couldn't explain. That was what motivated our study.”

He comments that his research differs from previous work in the field. The earlier studies focused on bringing healthy subjects into a laboratory to test an oximeter device, whereas the NEJM paper was based on real-world data from a hospital setting.

What’s more, the widespread use of pulse oximeters – not to mention the racial politics of patient care in the pandemic – means this issue can no longer be swept under the carpet.

While Sjoding himself is not an expert on the technology behind pulse oximeters, from talking to people who have more expertise he believes there could be engineering solutions to the problem.

“Pulse oximeters could potentially be re-designed in such a way that they are less biased, and long-term I'm hopeful that will happen,” he says. “But I do think it's going to take a concerted effort by the public health care providers, and there’ll need to be some additional regulatory oversight to push companies towards redesigning their devices.”

Current FDA guidance requires that, when a device of this kind is being tested, 15% of the patient cohort should have a darker skin colour. However, the FDA doesn’t stipulate that the device accuracy should be calculated and reported in each group individually. Sjoding thinks that needs to change if racial biases are to be eliminated.

“I think it's disheartening that this was known for as long as it has been known and hasn't really been addressed,” he says. “I do hope that it will, but it's going require a push from patients, physicians and regulators.”