A long-documented, under-the-radar disparity is garnering new attention as the Covid-19 pandemic stretches into another year: Some pulse oximeters, which measure oxygen levels and are critical to making decisions about patient care, aren’t as accurate in Black patients and other people of color.
New research on the issue has drawn the attention of lawmakers and clinicians alike, who say they want to see the Food and Drug Administration launch sweeping studies to analyze how well pulse oximeters used in the hospital and at home work in a diverse pool of patients. Democratic Sens. Elizabeth Warren, Cory Booker, and Ron Wyden urged the agency late last month to review the devices, calling the issue “a matter of life or death” as the pandemic continues to take a disproportionate toll on communities of color.
“I hope for a pulse oximeter that I can use every day in the ICU that I don’t need to worry about if it doesn’t work as well for Black patients as for white patients,” said physician Thomas Valley, an author of a December study published in the New England Journal of Medicine that suggested that pulse oximeters produce misleading measurements in patients with darker skin. “And I hope for improved FDA regulations to prevent something like this from ever happening again.”
But there appears to be an emerging gap between what some clinicians are calling for — including stricter requirements for the diversity of patients a device must be tested on and more oversight of oximeters sold over the counter — and the agency’s current plans.
William Maisel, director of the Office of Product Evaluation and Quality at the FDA’s Center for Devices and Radiological Health, told STAT the agency is “reviewing available data to determine whether additional studies or new guidance is needed.” He said the FDA is broadly confident that hospital-based oximeters are highly accurate, but added the same might not be true for devices sold online and in drugstores, which the agency doesn’t review or approve.
“We agree that individuals with dark pigmentation — particularly if they’re relying on oximeters in the home — may get false or inaccurate readings, and that’s important for them to be aware of,” Maisel said.
But researchers and clinicians argue that neither they nor the public have enough information to know how accurate a given pulse oximeter is in patients with darker skin. Several told STAT that a comprehensive study of the issue is long overdue.
“We don’t want this to be a problem anymore,” said Steven Gay, a pulmonologist and NEJM study co-author who is also an associate professor of internal medicine at the University of Michigan Medical School. “The best way is to begin to conduct clinical research and look at our devices and how they work, how this may have happened, and how to correct it.”
In and outside of hospitals, clinicians use pulse oximeters to make critical decisions about caring not just for patients with Covid-19, but also those with pneumonia, asthma, and countless other conditions. Amid the pandemic, the devices have taken on a more critical role.
“When beds are scarce, we’re making decisions on whether to admit a patient based on this datapoint,” said Utibe Essien, a physician and an assistant professor of medicine at the University of Pittsburgh School of Medicine.
Hospital pulse oximeters, the kind typically clamped onto a finger, work by beaming infrared and red light through the skin. When oxygen levels are low, the blood absorbs more red light than infrared light, while blood that is full of oxygen absorbs more infrared light. The device compares the amount of red and infrared light that it detects to determine a patient’s blood-oxygen saturation level.
But research suggests that because of how darker skin absorbs light, the measurement may be inaccurate in patients with non-white skin. (Nail polish may similarly affect pulse oximeter readings.)
In most healthy people, a blood-oxygen saturation reading of roughly 96% is considered normal. Conversely, a reading that is 90% or below is considered hypoxic, or potentially dangerous. Since Covid-19 impacts the body’s ability to oxygenate blood, even small differences in percentage points matter a great deal.
“These numbers that we use are incredibly important,” said Valley, who is also a pulmonary critical care physician at Michigan Medicine. “They’re essentially the difference between being admitted to the hospital or not, in starting oxygen or not, having your symptoms treated or not.”
After the researchers came across an article in the Boston Review on the issue by Amy Moran-Thomas, a professor of anthropology at MIT, they scoured the medical literature and found references to racial disparities in pulse oximeters as far back as the late 1980s and 1990s.
That body of literature included a 2005 study, which found that three pulse oximeters commonly used in hospitals produced inaccurate readings in people with dark skin when their oxygen levels dropped below 90%. That finding should have raised alarms everywhere, Essien said.
“For centuries, white males have been the base for medical research, from cardiovascular studies all the way to pulse oximetry,” said Essien. “Going back to the data and saying, ‘where did we go wrong?’ and addressing that is going to be critical.”
At the time, however, it seemed not enough people were listening. There was little, if any, public reaction, according to Philip Bickler, one of the study’s authors and the director of the hypoxia laboratory at the University of California, San Francisco.
Valley said he and the doctors on his team were unaware until they started their own research that the issue had been previously documented.
“I think that’s probably true for most physicians, most nurses, most clinicians out there,” said Valley. “We didn’t recognize that it was this big of a problem, and that was a failure of medical education and dissemination of medical knowledge.”
That awareness appears to be changing after the paper from Valley and his colleagues, which found more inaccurate pulse oximeter measurements in Black patients than white patients. They had compared pulse oximeter readings with arterial blood gas measurements, which are collected by drawing blood with a needle and are considered the gold standard for measuring blood-oxygen saturation. The researchers analyzed more than 10,000 pairings of pulse oximeter readings and arterial blood gas measurements from their hospital that year, as well as from datasets from other hospitals in previous years.
Oximeters, they found, were nearly three times as likely to miss hypoxemia in Black patients compared with white patients.
“When a pulse oximeter says 91%, more than 50% of Black patients actually had a value less than 88%,” said Valley, an author of the 2020 paper.
The paper from Valley and his colleagues had limitations — the authors used self-reported race data, for example, rather than skin color identified using a standard measure like the Fitzpatrick scale, and took blood gas and pulse oximeter readings up to 10 minutes apart rather than simultaneously. While those caveats could have affected the accuracy of the oximeter readings, Valley and his colleagues said the limitations could not fully explain the racial disparities documented by the study.
The FDA’s Maisel said that when it comes to hospital-based oximeters, the agency has “confidence in the accuracy of the products that we have reviewed.”
But the authors of the December paper and other clinicians said they don’t feel that same sense of trust.
“To not be confident in the accuracy of that number, and to depend on a device that may have bias toward a certain population is really troubling,” Essien said.
Along with the majority of medical devices on the market, pulse oximeters are approved by the FDA through a pathway known as the 510(k) process, which primarily requires a company to prove that a device is equivalent to one the agency has previously reviewed.
In 2013 guidance, the FDA encouraged — but did not require — developers of pulse oximeters to test their devices on participant groups that include a range of skin tones. “Your study should have subjects with a range of skin pigmentations,” the guideline notes, “including at least 2 darkly pigmented subjects or 15% of your subject pool, whichever is larger.” That figure does not track with the country’s broader demographic makeup, which is far more diverse.
“Two patients? This just seems so glaringly low for a device like a pulse oximeter that is so widely used,” said Michael Sjoding, a pulmonary physician at Michigan Medicine and an author on the 2020 paper. “The current guidance on ensuring these new technologies work adequately across these populations is inadequate.”
As the FDA assesses the situation with hospital-based oximeters, the agency and some researchers also called attention to another facet of the problem: the thousands of oximeters sold online, in drugstores, and embedded on smartwatches. These devices do not undergo FDA review and have not published data showing their devices work across a diverse population. For their part, spokespeople for Apple and Fitbit told STAT they had validated their pulse oximetry features across diverse populations that included a variety of skin tones, but neither company has made the data public.
Sjoding, for example, pointed out that multiple medical groups have recommended pulse oximeters to monitor Covid-19 patients at home.
“If the FDA feels these are inappropriate uses of a pulse oximeter, I think they should do a better job of raising awareness — like including a black box warning label on these devices, just like for certain medications,” he said.