The Food and Drug Administration should move more quickly to ensure pulse oximeters — the ubiquitous devices used to measure blood oxygen — work well in all patients, should better inform clinicians about the devices’ shortcomings in patients with darker skin, and should recall versions that are particularly problematic, patients, researchers, and consumer advocates testified during a public meeting on Friday.
The testimony occurred during an advisory panel meeting, held virtually to consider how to improve the safety of pulse oximeters, It comes more than three years after a study published at the height of the Covid-19 pandemic found that the devices didn’t detect low oxygen levels in some patients with darker skin, and after other studies showed the devices were tied to no or delayed treatment for Black patients with severe Covid.
The lack of swift action by the FDA has been frustrating to many health equity advocates, consumer advocates, and a group of state attorneys general, who sent a letter to the FDA last November urging action. “It is imperative that the FDA act now to prevent additional severe illness and mortalities among darker skinned people resulting from inaccurate or misleading pulse oximeter readings,” the letter said.
The slow progress is also frustrating to physicians and researchers. A number of devices on the market which were approved by the FDA based on data submitted by manufacturers ”would not meet the current standard for clearance,” Ashraf Fawzy, an assistant professor of medicine at Johns Hopkins University, told the panel.
Fawzy and colleagues found in a 2022 study that the devices were far less accurate in Black patients, who accounted for more than half of patients whose need for Covid therapy was never recognized. “We have fallen short because our tools have failed us,” he testified.
Many speakers expressed concern that the larger clinical community is not receiving word that the devices don’t perform as well in darker-skinned patients. Edward McClure, a Black patient with COPD, said he has many physicians due to multiple health issues but “none of them except my African American primary care physician had ever heard of this problem. That’s a big, big problem.”
McClure told the panel he put on two different pulse oximeters during the meeting and the readings differed by six points. Such inaccuracy, he said, has contributed to panic attacks when he gets low readings. “Do I need my rescue inhaler or to call EMS?,” he asked. “These panic attacks are real.”
A Kansas City mother who uses pulse oximeters to monitor her two children, one who is Black and one who is white, told the panel the devices are problematic when used on her Black daughter but not on her white son. “His is much more consistent,” she said. The inconsistencies led her to repeatedly replace probes on the pulse oximeter device she used, which was a hardship because the additional probes were not always covered by insurance. “We’re allowed four probes a month,” said Ryan Jolly. “We were going through four probes a week.” Jolly said she learned to deal with the errors through trial and error or by turning off and resetting the machine or moving where she placed the probe.
One panelist who uses a pulse oximeter because she has long Covid, said the average consumer was not aware of issues with the devices, causing a danger to people who buy the devices for use at home. Over-the-counter devices “need to be pulled off the market, or we should put this behind the counter where a pharmacist can explain this,” said the panelist, Rachel Brummert, communications lead for the American Society of Pharmacovigilance.
Others said the line between use for medical needs and wellness in direct-to-consumer devices was becoming blurry and that any devices used for medical decision-making should be subject to the same testing and performance standards regardless of where they are sold.
Without clear guidance from the FDA, many physicians are left unsure what to do when they get low blood oxygen readings while using the device on darker-skinned patients. “As someone who’s intimately aware of this issue, I still don’t know what to do for my patients,” Tom Valley, an associate professor in the division of pulmonary and critical care medicine at the University of Michigan who was part of the group who conducted the pivotal 2020 study, told STAT. “It shouldn’t just be, ‘Hey, when you remember, consider the limitations of pulse oximeters and draw arterial blood gasses.’”
While he appreciates action the FDA took early on, issuing a safety communication in February 2021, Valley is frustrated that more has not been done since then. “It’s been more than three years since our paper came out,” he said. “These are devices used every day, all the time that continue to place Black patients at risk of harm.”
Friday’s meeting was the second the agency has held on the issue. The FDA also released a discussion paper and request for feedback in November.
The panel, a group of experts in critical care medicine, anesthesiology, neonatology, statistics, and other fields, spent most of the 10-hour meeting considering a host of highly technical and statistical questions posed by the FDA on how the agency can improve testing required to approve the devices, including how to better assess skin tone in studies and how many darker-skinned patients to include in studies to ensure that devices work across a range of skin tones.
The FDA has suggested using the Monk skin tone scale, a way to measure skin tones that is more inclusive than ones used previously because it includes 10 skin tones across both light and dark hues, as well as devices called spectrophotometers, which provide quantitative measures of skin tone.
The panel also discussed whether it was OK, as is standard, for manufacturers to test pulse oximeters on healthy volunteers, who can undergo testing that involves lowering the oxygen they breathe through tubes with little risk, as opposed to sicker patients with clinically low oxygen levels where the devices work less well.
They wrangled with how to separate out issues other than skin tone that lead to lower oxygen levels, such as poor blood perfusion. In some studies, different oxygen levels were reported from the same patient in readings taken at different times. But overall, error rates were larger for Black and Hispanic patients, said Fawzy. For patients with light skin, they were near zero, he said.
The FDA is also proposing increasing the number of patients included in studies from a minimum of 10 patients, with at least two, or 15% of all patients being “darkly pigmented,” to a minimum of 24 patients divided into three cohorts of light, medium, and dark skin based on the Monk skin tone scale.
A number of both panelists and speakers questioned whether this sample size was large enough to detect differences between groups of patients with different skin tones. Representatives from device manufacturers also suggested it might be hard to recruit patients from both the lightest and darkest ends of the skin tone scale.
Some on the panel appeared to lean toward recommending a larger number of subjects for pre-market studies required for FDA approval, while others called for overrepresenting the number of subjects with darker skin in studies.
The panel also discussed how large of an error to allow for devices to be approved, particularly for readings at lower oxygen levels that are more prone to error.
FDA officials will consider the advisory panel’s views as they work on new regulations for the devices. No date was given for when those rules will be issued.
While the slow progress is frustrating many, both FDA officials and many panelists said they were committed to making the devices more equitable. But they said it was important to move deliberately and carefully to institute changes to make sure they solve the problem and do not achieve only incremental improvements or introduce additional harms or reduce access.
Many panelists were frustrated by the lack of data on how the devices perform, particularly on actual patients, and said they were frustrated they did not understand exactly what was causing the errors and why errors did not occur in all dark-skinned patients or in all readings in the same patients. Some said they wanted these issues better understood before adopting new rules.
Some of that data may come from two FDA-funded studies on hospitalized adults and children now underway at the University of California, San Francisco and Stanford that have recruited hundreds of patients.
Researchers urged more transparency and data sharing from device manufacturers — such as raw data from the devices — so they could better analyze performance. A collaborative group called Open Oximetry is working to create a shared data repository and work with researchers in Africa and Asia to increase the amount of data on patients with a wider range of skin tones, said Michael Lipnick, an anesthesiologist and intensivist at UCSF who helps run the group.
Michael Abrams, a senior health researcher at the consumer watchdog group Public Citizen, told the panel that more than a dozen new pulse oximeter devices were cleared by the FDA last year, even after these issues became public, and that very few submissions discussed the issue of skin pigment in their applications. Those that did simply declared whether patients were light- or dark-skinned without more specificity, he said. He also called for an immediate change of labeling of devices to educate clinicians that they’re less accurate in people with darker skin.
Kimani Toussaint, an engineer and optics expert at Brown University working on creating more equitable pulse oximeter technology, earlier this week called for more transparency in studies conducted by industry. Toussaint stressed he was not implying any misrepresentation occurring in data submitted by pulse oximetry manufacturers and believed they were acting in good faith, but said, in the absence of transparency, it comes down to “Trust me.”
Toussaint made his comments Wednesday at a National Academies of Science panel discussing the use of race in biomedical science where pulse oximeter issues arose as a major concern. “If I were to say, ‘Trust me, this works,’ the scientific community would want to see evidence,” he said. “That should be true across the board.”