The racial bias in pulse oximeters

Pulse oximeters are little devices you clip on your finger, which shoot light into your skin to measure the percentage of oxygen in your blood.

They're crucial tools for doctors and patients in many situations, but most notably right now in treating COVID-19 — because it's critical to know how well lungs are getting oxygen into the blood.

But a new study documents a racial problem: Pulse oximeters don't work as well with Black patients as with white ones. In a letter to the New England Journal of Medicine, a trio of doctors described how they analyzed data from both Black and white patients, in which oxygen levels were read using both pulse oximeters and arterial blood-gas measurements, which detects oxygen levels directly from an actual blood test.

It turns out Black patients were three times more likely to have discrepancies in the readings. As the doctors note in a post at the University of Michigan:

"We wanted to know: when the pulse oximeter is measuring between 92 and 96%, how often is the oxygen saturation actually less than 88% on arterial blood gas?" says Sjoding, adding that a reading of below 88% is generally where most physicians start to take notice and intervene.

Black patients were three times as likely to have a significant discrepancy in these readings. Twelve percent of the time, when Black patients had a pulse oximeter reading in this safe range, their actual saturation on an arterial blood gas was below 88%. In white patients, this discrepancy occurred 4% of the time. The study appears as a correspondence in the New England Journal of Medicine.

What a mess. It looks like a textbook case of designing a piece of technology without thinking about the full range of people who'll be affected by it. I'd guess these devices were primarily created, and tested, with white skin in mind — with serious consequences for Black patients, and apparently anyone with darker skin.

Worse, the device manufacturers have had plenty of opportunity to learn about this problem and fix it. As that Michigan U post notes …

The findings aren't entirely new, the team reports. A paper published in an anesthesia journal in 2005 also reported that skin pigmentation led to inaccuracy in pulse oximeter readings. 

… and people like med student Max Jordan note that they've been talking about it too; his Twitter thread here is how I first learned of this pulse-oximeter study …

In the short run, there are big implications here for how we treat COVID-19. As noted by the doctors who did this study, health-care professionals should be extra careful when using pulse oximeters on Black patients, or any patients with darker-than-white skin.

Plus, folks using these devices at home to monitor their oxygen levels should be similarly cautious, as Dr. Jewel Mullen, the associate dean for health equity at the University of Texas's med school in Austin, tells NPR:

She's concerned about the impact the finding could have on people who may rely on consumer-grade pulse oximeter devices at home, sometimes in lieu of medical care. The devices are still a valuable tool, she says, but it's important to look at the results in context.

"If somebody has coronavirus and they're feeling really short of breath, but they're getting [an acceptable reading] — say an oxygen saturation of 92% to 96% or more — they should pay more attention to their shortness of breath, more than that oxygen that's being measured," Mullen says.

People should also keep an eye on trends, rather than absolute readings, she says. If a person's oxygen level drops significantly, that should prompt a call to a medical provider, even if the number is still in the "normal" zone.

(That CC-3.0-licensed photo of a pulse oximeter courtesy Thinkpaul and Wikimedia)