Rather than provide clarity on how to reduce racial bias in pulse oximeter readings, a long-awaited study commissioned by the Food and Drug Administration has muddied the path forward.
The results were meant to help researchers understand why the handheld devices, widely used in hospitals and clinics to measure blood oxygen levels, do not work as well on patients with darker skin tones. But the initial findings, published in a paper that has not yet been peer reviewed, were surprising and did not provide conclusive guidance to clinicians.
They arrive at a time when research on health disparities — caught up in the Trump administration’s hostility toward diversity programs — feels more tenuous than when the study was commissioned in 2021, leaving experts who have been raising the alarm about the devices feeling exasperated.
No one’s quite sure how to fix pulse oximeters. The FDA asked this lab to find answers
“This is a tremendous effort to be able to do something like this. There’s a reason why pulse oximeter companies themselves have never done a real world prospective study — it’s a big lift,” said Thomas Valley, a professor of pulmonary sciences and critical care at the University of Colorado Anschutz who was not affiliated with the new study. “This is what we’ve been waiting for. We need prospective data in real world patients using these pulse oximeters … Having said that, I don’t know what to make of the results.”
The issue of racial bias in the devices catapulted to national attention because of their ubiquity during the Covid pandemic, coupled with data suggesting that pulse oximeters systematically overestimated oxygen levels in patients with darker skin colors, which some studies suggested led to delays in care.
A body of research spanning decades supported concerns that the devices worked worse on patients with darker skin tones, but that research was always caveated by the fact that it was retrospective — the data used was not collected explicitly to study pulse oximeters. So in response to calls to fix the issue — which grew louder in 2020 and 2021 as medicine reckoned with its history of perpetuating racial inequities — the FDA funded a study to generate prospective data in the clinic quantifying the bias.
Researchers at the University of California San Francisco measured blood oxygen of 631 patients using both pulse oximeters and an arterial blood gas test, a more invasive but more accurate measurement, so that the two readings could be compared. But the results, posted recently on a preprint server, were surprising: The pulse oximeters underestimated oxygen levels.
Pulse oximeters and their inaccuracies will get FDA scrutiny. What took so long?
“In the other studies, there was overestimation, that pulse oximeters gave falsely overestimated readings. We found the opposite,” said Michael Lipnick, an anesthesiologist at UCSF and one of the authors. The difference is important, in part because some researchers have called for a universal correction factor when using pulse oximeters on patients with darker skin tones. These new results suggest that wouldn’t be effective in making readings more accurate.
The researchers were not sure what could explain the confusing results. One potential explanation is that to be included in the dataset, patients had to have stable blood oxygen levels, which would exclude patients whose oxygen levels were falling quickly. The average blood oxygenation in the study was 98%, so another potential explanation is that there was little room for the devices to overestimate oxygen levels.
“We got a very clear signal that we do not adequately understand pulse oximeter performances in sick patients, that we do not have a clear shared understanding between clinicians who use them, patients who live or die because of them on one side, and manufacturers and regulators on the other side, as to how these devices should be used and therefore should be tested to ensure that they work like we need them,” said Jack Iwashyna, a critical care physician and professor at Johns Hopkins University who has studied pulse oximeter bias but was not part of the new study.
The study, called EquiOx, did find that pulse oximeter readings were higher in patients with darker skin tones than lighter tones. But clinicians who have studied the issue are not sure how to incorporate the results into their practice. Finding answers may be more difficult going forward, with efforts to understand health disparities being swept up in the Trump administration’s efforts to curb research it deems related to diversity, equity, and inclusion.
STAT Plus: FDA releases long-awaited guidance to improve accuracy of pulse oximeters for all skin tones
“There were a lot of questions still unanswered, and we’re hoping to follow up with some additional studies,” Lipnick said, adding that there were no funding opportunities from the FDA for additional studies. ”The door was not open to try and continue,” he said. The FDA has also yet to finalize guidance it released in draft form a year ago to address the concerns about bias. The guidance proposed a higher bar for testing pulse oximeters before they make it to the market, including requiring they be tested on more patients with darker skin tones.
In response to a question regarding the timeline for when that guidance would be finalized, a spokesperson said, “The guidance is still a draft and not intended for implementation. The FDA is currently assessing comments received and is considering next steps.”
To better study bias in pulse oximeters, Lipnick and his team have set up a lab in Uganda, which would help recruit patients that have darker skin tones than are common in the U.S. He also hopes to convene several research groups across the country, to create a large dataset spanning the country that could be analyzed.
“We’ve been trying to put together a study group so multiple sites collecting data in the same way, a shared protocol, and putting all that together and doing a bigger analysis of good data. We hope to be able to do that still, we just don’t have funding for it right now,” he said. “We have several groups that have agreed to contribute their data. So that will happen. It’s just maybe not happening as quickly as we’d like if we were funded.”
STAT’s coverage of health inequities is supported by a grant from the Commonwealth Fund. Our financial supporters are not involved in any decisions about our journalism.
