Pulse oximeters are everywhere these days—not just in the ICU, on the floor, and at outpatient practices, but even on the wrists of people walking down the street.
“These devices have become really mainstream. In addition to the pulse oximeters that patients could buy off the internet or from a pharmacy or any store, really, these devices are being integrated into wearable devices,” said Ashraf Fawzy, MD, MPH, a pulmonology and critical care specialist and assistant professor of medicine at Johns Hopkins University in Baltimore.
That prevalence makes researchers such as Dr. Fawzy particularly concerned about the growing evidence that pulse oximeters don't work as well in patients with darker skin tones.
“For a long time now we have taken whatever the pulse oximeter says at face value and made decisions day in and day out based on it,” said Valeria Valbuena, MD, MSc, a general surgery resident at Michigan Medicine in Ann Arbor who has also studied pulse oximeters. “We really need to rely less on single pulse oximeter measurements as a profession.”
Evidence that pulse oximeters may overestimate oxygen saturation in those with darker skin dates back to at least 1990, when a small study published in CHEST compared pulse oximetry in White and Black patients on ventilators, finding that in the latter group, a pulse oximetry reading of 92% was commonly associated with significant hypoxemia.
The issue got little attention, however, until the pandemic brought both oxygen saturation and racial equity to the forefront of health care. “It's telling that there was literature that existed describing the inaccuracy of the pulse oximeter by skin pigment, but we weren't even aware of it,” said Michael Sjoding, MD, MSc, an assistant professor of pulmonary and critical care and hospital medicine at Michigan Medicine in Ann Arbor. “It wasn't in any of my educational curriculums.”
He first got an inkling of the problem early in the pandemic, when Ann Arbor starting receiving COVID-19 patients transferred from Detroit.
“I and other colleagues were caring for a high number of racially diverse patients, especially in the first wave. And anecdotally, several of us noticed that we were seeing patients with pulse oximetry on monitors that were normal and then when, just by chance, an arterial blood gas was performed, it showed the actual saturation was lower,” he said. “None of us could figure out what was going on at the time.”
Dr. Sjoding and his colleagues investigated and published their results as a letter to the editor of the New England Journal of Medicine in December 2020. The study compared pulse oximetry and arterial blood gases from thousands of White and Black patients at multiple hospitals before and during the pandemic and found that pulse oximetry failed to detect hypoxemia almost three times as frequently in Black patients as in White patients.
Soon, a number of other researchers had built on their findings. A database analysis, published by JAMA Network Open in November 2021, compared pulse oximetry and arterial oxygen saturation in more than 80,000 patients. It found that 6.9% of Black patients had oxygen saturation below 88% on blood gas but not pulse oximeter, as did 6% of Hispanic patients, compared to 4.9% of Asian and White patients.
The findings highlighted key questions about pulse oximeters, according to lead author A. Ian Wong, MD, PhD, a critical care specialist and pulmonologist at Duke University in Durham, N.C. “Myself, I'm neither White nor Black. How does this work in people who don't fit either of those categories?”
It was also important to determine if the inaccuracies had any effect on patient outcomes. “We were trying to figure out, ‘OK, does this affect mortality?’” he said. It appeared so in the study, with patients in whom the discrepancy, or “hidden hypoxemia,” was found having higher mean sequential organ failure assessment (SOFA) scores and higher in-hospital mortality.
“The third thing that was really interesting and that we did not expect to find was that there appeared to be different rates at which different races got blood gases,” said Dr. Wong.
Potential connections between the shortcomings of pulse oximeters and already well-established racial disparities in outcomes, including from COVID-19, were further elucidated by a study Dr. Fawzy and colleagues published in JAMA Internal Medicine on May 31.
“The renewed attention to the inaccuracies of pulse oximetry led us to ask whether or not this discrepancy in pulse oximetry may translate into changes in how we manage patients with COVID,” said co-lead author Tianshi David Wu, MD, MHS, an assistant professor at Baylor College of Medicine in Houston.
The study found that Black patients were again more likely to have their oxygen saturation overestimated by pulse oximetry and that this led to them being less likely to be treated with oxygen when needed. Among the COVID-19 patients who eventually got oxygen, there was a median treatment delay of an hour for Black patients compared to White patients.
These recent studies may answer the question of whether pulse oximetry inaccuracies are clinically important, according to Dr. Wong. “I've actually heard some physicians say, ‘I know that this is an interesting finding, but how much does it really make a difference?’” he said.
“On an absolute scale, the accuracy difference is small,” Dr. Sjoding noted. “We're talking one or two percentage points, on average, but the problem is that we, in clinical medicine, make decisions—to treat or not to treat, to admit to the hospital or not to admit to the hospital—often based on cutoffs.”
The search for fixes
In February 2021, the FDA issued a safety communication about pulse oximeters. It warned clinicians to “be aware that multiple factors can affect the accuracy of a pulse oximeter reading, such as poor circulation, skin pigmentation, skin thickness, skin temperature, current tobacco use, and use of fingernail polish.”
A collaborative of critical care societies, including the Society of Critical Care Medicine, American Thoracic Society, American College of Chest Physicians, and American Association of Critical-Care Nurses, has pushed the agency to take more action. In a February 2022 letter, the group urged the FDA “to directly engage the developers and manufacturers of all pulse oximeters to rectify this urgent situation in a timely manner.”
On June 21, the FDA updated its safety communication with plans to convene a public meeting of its medical device advisory committee later this year to discuss evidence about the accuracy of pulse oximeters, recommendations for patients and clinicians, and the amount and type of data that should be provided by manufacturers to assess the devices' accuracy.
The agency does already require some evidence that pulse oximeters are accurate in patients of different skin colors, but some experts find the criteria insufficient. “Right now, I would say that the standards aren't rigorous enough to ensure that these devices are performing the same in all people,” said Dr. Sjoding. “It requires a sample size of 10 patients, two of whom are darkly pigmented.”
“There's nothing in science that I could be like, ‘I tested it on three people,’” said Dr. Valbuena. “We have had [pulse oximeters] around for over four decades. When you think about the things that we do with other devices—ventilators, machines to support your heart and lungs—we have constant innovation.”
At the University of California, San Francisco's Hypoxia Lab, Michael Lipnick, MD, has had a close eye on innovation in pulse oximetry. “We're further behind than we should be,” he said. “We don't have solutions in the public domain, so it's not clear for all manufacturers how to address this. Some devices perform better than others, but in almost all the devices that we've tested, we find some bias, some inaccuracy, that we think is attributable to skin pigment.”
The lab tests pulse oximeters for manufacturers as well as large purchasers of the devices, explained Dr. Lipnick, who is an anesthesiologist and intensivist at Zuckerberg San Francisco General Hospital.
They're also currently working on research that could lead to regulatory changes. “We have received funding from the FDA to do a clinical trial to better understand exactly how big of a problem this is and some of the reasons why this may be so,” he said, noting that the inaccuracies likely involve not only skin color, but also other factors such as perfusion.
For anyone outside of his lab, getting valid data about the accuracy of any one pulse oximeter is difficult, according to Dr. Lipnick. “It's not always so clear what the performance of these devices is, and it may not always match up with what some of the manufacturers are claiming,” he said.
Dr. Lipnick and his colleagues are hoping to remedy that by building a website to share oximeter research. “We think that we can build what would be the largest data set for this type of performance data, both for clinicians and users, and also for researchers and manufacturers,” he said. Their site, openoximetry.org, is currently still in the pilot stage.
Of course, today's desaturating patient can't wait for data sharing, regulatory changes, and technological innovation.
“The message that I think many researchers, including us, are trying to promote is not that you can't trust a pulse oximeter or that you shouldn't use it, or that these are not extremely valuable and essential tools for diagnosis, but rather just to be aware that there are some devices out there that may be overestimating oxygen saturation, particularly in those with darker skin pigment,” said Dr. Lipnick.
He noted that FDA-approved devices are likely more accurate than over-the-counter alternatives. “I've observed a lot of clinicians who have their own fingertip pulse oximeter or their own low-cost device that they've procured for a variety of different reasons,” said Dr. Lipnick. “Be careful with any device that hasn't at least gone through FDA certification.”
Unfortunately, there's no “easy bedside conversion” for overestimates due to skin color, “like, oh, it seems like it's five points lower for every single measurement,” said Dr. Valbuena. Instead, she recommends considering multiple readings from the device to get a more accurate picture. “We really need to be looking at trends rather than single pulse oximeter measurements until we have more equitable machines.”
Hospitalists should also trust their instincts, recommended Dr. Fawzy. “If the pulse oximeter is giving you a different answer than what you're seeing clinically, it might be worthwhile to question it in certain populations,” he said.
Dr. Sjoding agreed. “If I have a patient, particularly someone who has darkly pigmented skin, and I have a borderline pulse oximetry, I will be much less likely now to weigh that information heavily in my decisions about how to care for that patient,” he said.
There are other biomarkers available to help identify hypoxia, Dr. Valbuena noted. “I look at things like lactate and like whether or not there are any derangements in urine output or creatinine,” she said.
And when you really need to know the patient's oxygen saturation, go to the gold standard test, the experts agreed. “I hate to say it, but when you're in doubt, especially in hospitalized patients or in patients in the clinic that are not looking good, I just think adding more blood gases,” said Dr. Valbuena. “There are consequences to getting a painful procedure and it's not done for free, but there are bigger consequences to not having enough oxygen.”
Clinicians should be honest with patients about the problem, Dr. Wong suggested. “‘I'm not choosing to subject you to this painful test just because I'm highly curious. I'm trying to make sure you get the care you need,’” he said. “It's a delicate conversation.”
Conversing with colleagues about pulse oximeters can also be part of the solution. “I am still surprised when I have conversations with trainees and physicians that are not aware of it, because everybody's using this device. Everybody,” said Dr. Valbuena.
She would like to see clinicians join the push for better solutions. Remaking pulse oximeters to be accurate across the range of skin colors and then replacing all those in use in health care will be a large undertaking, she noted. “I really think that every single provider needs to be aware and actively engaged,” Dr. Valbuena concluded.