Pulse Oximeters Don't Work as Well for People With Dark Skin

The problem with these medical devices, essential to treat COVID-19 and other respiratory illnesses, lays bare how racial bias can affect healthcare in unexpected, harmful ways

Early on in the COVID-19 pandemic, Kimani Toussant's wife, an epidemiologist, brought him an article about how pulse oximeters appeared to be less accurate in people with darker skin. The small devices, which can instantly and painlessly measure a person's blood oxygen level by transmitting light, usually through a fingertip, are an essential tool in medicine and have been especially indispensable during the pandemic.

Toussaint, a professor in the school of engineering at Brown University who studies optical imaging, says he was familiar with pulse oximeters but hadn't closely followed the studies over the years that raised concerns about their uneven accuracy. "I was kind of surprised that this was something that was not addressed such a long time ago," he says. Eventually, Toussaint and one of his PhD students, Rutendo Jakachira, took on the project of designing a better pulse oximeter, one that would perform equally well for people of all skin colors.

Toussaint and Jakachira are just two of the many researchers tackling the now well-documented problem of pulse oximeters and their poor performance in Black people and others with darker skin. It's a critical issue of inequity in the medical system with one of its most widely used tools. (For more on racial bias in healthcare, watch "Bad Input: Medical Devices," below.)

But fixing pulse oximetry isn't easy. First, there are regulatory roadblocks. The Food and Drug Administration's current recommendations for testing pulse oximeters don't adequately ensure that the devices are accurate for people with a wide range of skin tones, experts say. And that only applies to medical pulse oximeters used in professional health settings. The FDA doesn't require any testing at all for over-the-counter pulse ox devices.

Despite the new focus on the problem, there are still gaps in understanding why pulse oximeters don't work as well in darker-skinned people. In the meantime, healthcare providers—who must still rely on the devices they know are flawed—need better guidance on what they can do right now to account for those deficiencies.

Here, we’ll explain what scientists are up against when it comes to solving the problem of pulse oximetry, some of the many solutions already in the works, and what clinicians and patients need to know now.

A pulse oximeter transmits light through a part of the body, often a fingertip. It can determine how much oxygen is in blood based on the extent to which certain wavelengths of light are absorbed by hemoglobin, a protein in blood. The measurements it produces are critical because a low oxygen reading can be an emergency, requiring lifesaving action like supplemental oxygen or sometimes even a ventilator.

But as far back as the late 1980s, researchers noticed that pulse oximeters seem to provide less accurate readings in people with darker skin. And the COVID-19 pandemic brought renewed attention to the problem, kicked off in December 2020 by a study in The New England Journal of Medicine. It found that Black patients in a Michigan hospital were nearly three times as likely as white patients to have hidden hypoxemia, or low blood oxygen saturation, that wasn't detected by a pulse oximeter.

Watch a short film about how biases affect the development of medical technologies.

Even after a flurry of studies documenting the problem, scientists weren't sure to what extent inaccurate readings were harming people. Some experts wondered whether pulse oximetry readings off by 1 or 2 percentage points were significant enough to make a difference in medical outcomes, says Ashraf Fawzy, MD, a pulmonary and critical care physician and an assistant professor of medicine at Johns Hopkins.

But the COVID-19 pandemic provided plenty of opportunity to investigate that question. Some of the protocols developed for treating the disease relied on specific blood oxygen cutoffs. That meant that researchers could see who was eligible for COVID-19 medications and when. And, Fawzy says, research using that data showed that Black and Hispanic patients were 29 percent and 23 percent, respectively, more likely to experience a delay in being diagnosed with low enough blood oxygen that they could benefit from targeted COVID-19 medications. Fawzy says that research, which he co-wrote and was published in JAMA Internal Medicine in 2022, demonstrated that relying on pulse oximetry "delayed care for Black and Hispanic patients compared with white patients."

One problem with pulse oximeter technology, Toussaint says, is that hemoglobin isn't the only thing that absorbs light. Melanin, which gives skin its pigment, does too. Failing to factor that into the design of pulse oximeters means that the devices aren't as accurate in people with more melanin in their skin, he says.

The algorithms powering oximeters are also potentially problematic, says Valencia Koomson, an associate professor in the department of electrical and computer engineering at Tufts University, who's working on developing a better pulse oximetry device.

Pulse oximeters are benchmarked against a more invasive measurement called an arterial blood gas test, in which blood is drawn from an artery. (Most routine blood tests draw from veins; arteries are harder to reach but carry blood that's richer in oxygen.) Device makers compare the results from a pulse oximeter to the results from a blood gas test in a group of people at varying levels of oxygen saturation. Then they use that data to create an algorithm to predict someone's blood oxygen saturation from the oximeter reading alone. But if a manufacturer doesn't collect enough data—say, if tests aren't done on people with a wide enough variety of darker skin tones—then the algorithm developed probably won't be as accurate for people with those darker skin tones, Koomson says.

Those algorithms, however, are considered trade secrets, making it difficult to know exactly what goes into them or how to improve them.

While the evidence that pulse oximeters are less accurate for people with darker skin is compelling, there are still significant gaps in basic research and in how they’re regulated. "We are working on solutions while we’re also simultaneously trying to figure out what the problem is here," says Michael Lipnick, an associate professor at the University of California at San Francisco who co-leads the Open Oximetry project, a collaboration aimed at improving pulse oximeters.

A major one: Basic research on pulse oximetry performance generally doesn't measure skin tone directly but instead uses self-identified race or ethnicity as a proxy. More precise measures are needed, but the tools now used to measure skin, including one called the Fitzpatrick scale, have shortcomings. "It was never actually intended to be used in the way that it's been adopted," says Ellis Monk, an associate professor of sociology at Harvard University.

Originally, the Fitzpatrick scale was used to measure how different skin tones reacted to UV light, which means the scale skews toward skin tones (notably lighter ones) that are more sensitive to UV. To try to address that problem, Monk has developed a new scale of measuring skin tones that's been adopted by Google to inform its machine learning research and development.

The need for such a scale stretches beyond pulse oximetry, Monk notes, encompassing facial recognition, self-driving cars, and more. "Any kind of light-sensing computer vision application is going to run into this issue of, ‘How do we make sure everything works well for everyone?’ " he says. "Well, you need a measure that's agreed upon so that you can construct a representative sample of people to test."

Some scientists are also trying to see if other variables besides skin tone may contribute to pulse oximetry inaccuracies. For example, research from the UCSF Hypoxia Research Laboratory suggests that pulse oximeters make more mistakes in people with darker skin who also experience "low perfusion," or have less blood reaching the tips of their fingers.

Changes are also needed in how the FDA regulates the development of pulse oximeters. To bring new ones to market, the FDA guidance suggests that manufacturers test their devices on at least 10 subjects, two of whom (or 15 percent, if the group is larger) need to have darker skin pigmentation. "Concerningly small" and "woefully inadequate" are a few of the ways experts described this guidance during a November 2022 meeting the agency held to discuss the issue.

Based in part on input from that meeting, the FDA is expected to issue new guidelines for manufacturers looking to bring new pulse oximeters to market.

The UCSF Hypoxia Lab, one of the Open Oximetry Project's collaborators, has a long history of testing pulse oximetry devices. Lipnick says it plans to test devices with new FDA guidance once it's finalized, and help people who are developing new pulse oximeters test their prototypes.

And there's a final wrinkle that makes this all especially confusing for consumers. You might be one of many people who bought an over-the-counter pulse oximeter during the pandemic to help monitor your oxygen levels at home. But unlike the devices used in medical settings, manufacturers of those sold online and in stores don't need to show performance data. That means users of home devices have even less assurance of their accuracy.

While the FDA works on new regulatory standards, diverse teams of engineers are already tackling the challenge of creating a better pulse oximeter.

Students at Johns Hopkins University have been working on a new device that employs cheap, off-the-shelf components and a novel algorithm factoring skin tone into its calculations. Their goal is to make the device accessible in low-income countries with fewer medical resources.

Toussaint's team at Brown is taking a different approach. Instead of relying on the different wavelengths of light to differentiate oxygen-rich and deoxygenated blood, they aim to harness another property of light: its polarization. The hope is that employing a different property of light might help better mitigate some of the confounding factors that lead to inaccuracies in pulse oximetry readings, including melanin.

Koomson, meanwhile, is taking lessons learned from improving radio signals in the face of interference from, for example, fog, and applying them to pulse oximeters. The idea is to create a device that does a better job of making sure the light signals of the device get where they need to go without the interference of factors like skin tone that could affect performance.

These devices are all in the early stages of development, just beginning to be tested on people in medical settings, so it will be some time before they’re widely available.

What can clinicians do until these new devices become available?

Relying much more on the arterial blood gas test isn't a realistic option, in part because that test is painful, has more potential complications, requires special training, and doesn't provide instantaneous results (as pulse oximeters do).

Tom Valley, an assistant professor in the department of internal medicine in the division of pulmonary and critical care at the University of Michigan, says one easy adjustment isn't to rely solely on the reading of the pulse oximeter but to pay more attention to the symptoms a person reports and the clues from a physical examination. "I can certainly think of times in my past where perhaps I discounted what someone was saying because I had an objective number that said everything's okay," says Valley, who co-authored the 2020 New England Journal of Medicine study that helped spark renewed attention to the problem of pulse oximeters. "Maybe I was misled by a number that was inaccurate, and I trusted that device too much."

Another proposed alternative is to change the oxygen level cutoff that prompts doctors to provide supplemental oxygen. Typically, they prescribe supplemental oxygen when oxygen saturation dips below 92 percent. But some experts say that raising that level for darker-skinned patients makes sense. Valley says that this workaround is one he's used with Black patients, although he worries it might create another disparity: Some might receive supplemental oxygen treatment or even be required to stay longer in a hospital unnecessarily.

That possibility could be partly mitigated by using higher oxygen targets for all patients measured with a pulse oximeter, not just those with darker skin, as proposed by the authors of a December 2022 study published in the journal Respiratory Care. The researchers noted this might result in a slight increase in people getting too much oxygen, though experts suspect that the benefits of getting oxygen to people who need it outweigh the potential risks (which are unclear) of providing too much to some people who don't.

Despite their shortcomings, pulse oximeters are still vital. "We don't want people to lose faith" in them altogether, Lipnick says.

"It's also important to realize that in all the studies that have been done, we’ve shown that this doesn't happen all the time. It only happens in a small minority of cases," says Fawzy of Johns Hopkins. "But the important thing is that it potentially happens in people who are sicker, where the numbers matter even more."

There's not a lot that an individual person with darker skin can do: This is a systemic problem. Still, Fawzy suggests that if you’re in a situation in which a pulse oximeter reading says your oxygen levels are normal but you still feel short of breath, it's worth gently pushing back with your provider.

"So for someone who has a respiratory illness, whether it's chronic or acute," Fawzy says, "and their symptoms aren't really lining up with what they’re seeing on the pulse oximeter, it's probably a good idea to seek further care if they’re concerned that their oxygen levels aren't being been measured correctly."

Editor's Note: The video included in this article was made possible with the support of the Kapor Foundation, which works toward rebuilding a more equitable tech sector, economy, and society.

Catherine Roberts

Catherine Roberts is a health and science journalist at Consumer Reports. She has been at CR since 2016, covering infectious diseases, bugs and bug sprays, consumer medical devices like hearing aids and blood pressure monitors, health privacy, and more. As a civilian, her passions include bike rides, horror films and fiction, and research rabbit holes. Follow her on Twitter @catharob.