Transcript: Racial Bias and Pulse Oximeters Part 1—A Problem Hiding in Plain Sight | Johns Hopkins

The following is a transcript of Public Health On Call’s Special Series: Racial Bias and Pulse Oximeters: Part 1—A Problem Hiding in Plain Sight.

Listen to episode →

View the transcripts for the rest of the series:

[Show Intro, Joshua Sharfstein: Welcome to Public Health On Call, a podcast from the Johns Hopkins Bloomberg School of Public Health, where we bring evidence, experience, and perspective to make sense of today’s leading health challenges. If you have questions or ideas for us, please send an email to PublicHealthQuestion@jhu.edu. That’s PublicHealthQuestion@jhu.edu for future podcast episodes.]

Lindsay Smith Rogers: Welcome to Public Health in the Field, a special series of our award-winning Public Health On Call podcast. I’m producer Lindsay Smith Rogers.

In this three-episode series, we explore a longstanding issue that only caught the nation’s attention in recent years. It’s been widely noted that pulse oximeters—devices used to read blood oxygen levels in hospitals and at home—are far less reliable for people of color, and especially for darker-skinned patients, than for white patients. The error results in readings that are falsely normal, creating the potential for clinical staff to miss life-threatening complications.

In this first episode of the series, Annalies Winny and Nicole Jurmo delve into how COVID-19 shone a light on an issue that was known, but largely ignored, and the history that led us to the flawed pulse oximeters that are on the market today. Here’s Annalies Winny.

Annalies Winny: Do you remember being a kid in a pool and seeing how long you could hold your breath under water?

It’s not so hard for 10 seconds, or even 30. But at one minute? Or two? Your heart starts to beat faster, you might get confused, or start panicking and kicking. And the longer you deprive your body of oxygen, the more it fights. It doesn’t take very much time without air for life to start slipping away.

The air we breathe contains a simple molecule made of two atoms of oxygen. It’s colorless, it’s odorless, it’s invisible. But being deprived of oxygen means certain death.

During the early days of the COVID-19 pandemic, when information about the virus was scarce and immunity to it was nonexistent, being hospitalized with the respiratory disease meant battling for every breath in packed hospitals, away from your loved ones, and hooked up to a raft of medical equipment in intensive care units, not knowing the path forward.

And depending on the color of your skin, it could also mean that not even your medical team could see when you were essentially drowning because of the failure of your lungs to function effectively. You may have been surrounded by lifeguards—your health care team—and you might have been saved, but no one knew you were running out of air. Why?

Because chances are one of those pieces of machinery in the ICU—the one meant to sound the alarm when a patient’s oxygen dips so low they are in danger of crashing—wasn't working as well with darker-skinned patients. The device was the pulse oximeter. This device is a tiny sensor usually taped or clipped to a finger or toe that reads a patient’s oxygen levels.

Oxygen saturation in the blood is often referred to as a vital sign—an essential piece of a care team’s understanding of how a patient is doing overall. An oxygen saturation of 100% indicates that your body is carrying as much oxygen as it can. And generally speaking, any level 95% or above is very good. At sea level, oxygen levels in the low 90s or worse reflect an underlying problem, such as pneumonia, heart failure, or other lung diseases.

Pulse oximeters come as reusable plastic clamps or foam velcro straps that attach to a finger or toe, or can be embedded in flexible adhesive tape that wraps around a finger tip for continuous use. For clinical staff in an ICU, these devices allow for more vigilant, but less staff intensive, continuous monitoring of patients’ vital signs and earlier alerts of respiratory complications.

Here’s how they work: The devices rely on light to determine oxygen saturation. Visible red light is emitted from the pulse ox, which travels through the patient’s finger or toe. The probe measures how much light is absorbed by the finger, and the higher the level of oxygen in the blood, the more light at one particular wavelength gets absorbed. If the probe detects that more light is escaping, that means oxygen saturation has fallen. If this happens, it’s time for clinical staff to examine the patient for signs of respiratory distress.

And during the pandemic, care teams in overwhelmed hospitals used this device as they have done for decades, relying on it to make decisions about whether patients could go home, or whether they needed to stay at the hospital. Severe COVID patients can experience something called “happy hypoxia,” seeming to be comfortable despite dangerously low oxygen levels, making reliable oxygen readings all the more important.

While FDA-approved pulse oximeters are used in hospitals and cost hundreds or even thousands of dollars to purchase, versions of the device are commercially available to anyone and can be bought online or at a pharmacy for about $25.

In fact, many COVID patients, especially those with other conditions, were advised to use them at home to monitor their health if they had symptoms. And if their blood oxygen fell below 90%, the advice was to seek emergency care. But what many doctors and their patients didn’t know is just how unreliable these devices can be when it comes to accurately measuring a patient’s blood oxygen levels—particularly for patients of color.

It was because of the COVID emergency that this became clear to many people, including emergency physicians, for the first time.

The situation created a unique case study: During the pandemic’s peak from May 2020 through April 2021, it’s estimated that over three-and-a-half million Americans were hospitalized, all with the same condition.

For Tom Valley, a pulmonary critical care physician at the University of Michigan, there was another big change: during that first spring surge, his patient population flipped.

Tom Valley: Almost overnight, our demographics change from about 80 to 85% white to I would say, probably 80 to 85% Black and Hispanic patients.

Annalies Winny: And data shows that this demographic suffered disproportionately high infection, hospitalization, and death rates from COVID-19 compared to their white counterparts.

[Archive clip, Newscaster: Dr. John, that was a very disturbing and upsetting report. African Americans are being hit disproportionately hard.]

[Archive clip, Woman: She flatlined in my arms…]

[Archive clip, Newscaster: African Americans being hit disproportionately hard by the virus…]

[Archive clip, Newscaster: In this chart, we can see African Americans represent 41% of COVID-19 deaths.]

Annalies Winny: Something else changed, too: During the pandemic, doctors were also measuring blood oxygen with an arterial line, which is a tube placed in an artery. Valley says that testing arterial blood gas known as ABG from the arterial catheter is the “gold standard” for measuring oxygen levels in the bloodstream.

An arterial catheter is a thin hollow tube that is placed into an artery in the wrist and can be taped down. The line gives nurses a way to draw blood without having to prick patients multiple times, and it’s standard care for some patients like birthing moms. The arterial catheter allows doctors to draw blood to measure the oxygen level directly, rather than rely on an indirect measurement such as light absorption through the pulse oximeter. But the arterial catheter had fallen out of favor. For one thing, it was labor-intensive to place and remove the catheter, and if it was only being used for oxygen monitoring, the pulse ox was far easier.

Tom Valley: We've been kind of moving away from it, mainly because people were like, do we really need these devices? Pulse oximeters work so well. But then during COVID, it became standard of care, in the sense that, we wanted to minimize the amount of time that physicians and nurses were spending in the room. And so for every patient who came in, we were putting in these special catheters into their wrist so that we could very easily go in there, draw some blood off the artery and leave the room instead of kind of going in and poking and trying to get that arterial blood. And so every patient in our ICU had this arterial catheter, and had their pulse oximeter values that were being measured continuously.

Annalies Winny: And between the two ways of measuring oxygen—in an ICU full of darker-skinned patients—a major discrepancy emerged…

Tom Valley: We would be getting these arterial oxygen levels. And then we'd be looking at their pulse oximeter value at the same time on the monitor from right outside the room, and we'd say, ‘hey, like, these two things don't match up.’ And that was very unusual to us as we went from room to room to room. Like maybe in the past, it might have come up like once in a while. But it wasn't something that we were seeing. We were seeing this quite frequently in this patient population. And I'll be honest, we didn't understand why, at first, in fact, some members of our team, were sending emails out to colleagues that they knew around the country wondering whether this might have been a COVID phenomenon, maybe there was something unusual about COVID that was leading to this mismatch between someone's arterial oxygen level and what was being seen on the pulse oximeter.

And after we read that article, it was like a light bulb. And we said, oh, man, like, maybe this wasn't a COVID thing. Maybe this was a color of the patient's skin thing.

Annalies Winny: He’s referring to an August 2020 article published in the Boston Review. It was by Amy Moran-Thomas, a cultural anthropologist at MIT who studies medical objects like glucose meters, dialysis devices, and oxygen chambers for diabetic wound care.

And like many Americans during the pandemic, she had used a $20 consumer-grade pulse oximeters at home to keep an eye on her family’s oxygen levels when they had COVID—knowing how important those readings could be for their care. She and her husband are both white, but she also knew from her work that light sensing devices could be problematic for darker skin tones.

As Black Lives Matter demonstrations passed outside her window, she couldn’t help but see the connection between the systemic inequities being protested across the country, and the little device she held in her hand.

Amy Moran-Thomas: I was, you know, seeing the red light glowing inside this object. This question occurred to me: how did whoever made this object deal with the fact that it's a light sensing device, but many light sensing devices are known not to work accurately, for people with the full variety of human skin tones? Because they were designed primarily, you know, to work for white populations. And the physics of absorption really matters when you're building a color sensing device and not thinking about the full range of human skin tones. And so I wondered how, how did the people who made this device deal with that? And the backdrop of the protests happening at that time also made me think like, what if they didn't? And so I Googled it.

Annalies Winny: So Moran-Thomas went to work investigating these questions, resulting in an article that laid out how for decades, studies had concluded that pulse oximeters routinely give unreliable readings when used on patients with darker skin tones.

These studies found that pulse oximeters often overestimate the amount of oxygen a patient with darker skin may actually have in their blood, meaning their blood oxygen can read as normal when the levels are in fact dangerously low. She outlined how these findings had consistently been sidelined in mainstream medicine, and in medical education. Valley, and many of his colleagues around the country, were shocked.

Jack Iwashyna: What the [bleep]!

Tom Valley: It was shocking. It was absolutely shocking to us.

Noha Aboelata: I felt shocked.

Tom Valley: I certainly felt guilt. I thought of the patients that I've potentially caused harm to, because I didn't consider that these devices might not work as well for them compared to others. And so it did make me kind of question my own clinical abilities, my own care that I provided.

Annalies Winny: In December 2020, Valley worked with his colleague Mike Sjoding and others to publish their own data in the New England Journal of Medicine, documenting what they’d seen at their own hospitals during COVID’s first wave.

They found that Black patients were three times as likely to have a significant discrepancy between pulse oximeter and arterial blood gas readings. Twelve percent of the time, when Black patients had a pulse oximeter reading in the “safe range,”—92 to 96%—their actual saturation on an arterial blood gas test was below 88%. In white patients, this discrepancy occurred 4% of the time.

Their findings made headlines…

[Archive clip, Newscaster: But now researchers are saying readings are more likely to be inaccurate in people with darker skin. Doctors became aware of the issue when they were caring for COVID patients from the Detroit area who were more racially diverse.]

Annalies Winny: And they brought fresh attention and scrutiny to the issue—including from the Food and Drug Administration. And studies published since have added more pressure, showing that inaccurate pulse oximeters were linked to delayed or lack of treatment for Black patients with severe COVID.

Another reason the pulse oximeter issue became so much more known during COVID was because many people purchased the devices to monitor symptoms at home. But some people have relied on pulse oximeters to monitor chronic conditions for years. We spoke with Ryan Jolly, a Kansas City nurse who uses pulse oximeters to monitor her two young African American children.

Ryan Jolly: The potential for failure of the device—that it just isn't reading anymore, or that it just suddenly stopped doing its job—for both of my children is death.

Annalies Winny: Her 12-year-old daughter has a rare chromosomal abnormality that has left her nonverbal and dependent on a tracheostomy tube to breathe. And her 6-year-old year old son has dysautonomia, a condition that causes the automatic systems of your body, like respiration, heart rate, and your temperature regulation to simply go offline. His condition has no external manifestation. So loss of oxygen, it’s invisible.

Ryan Jolly: Now, in the same way that if I were to hold my breath, and not tell anybody, I was holding my breath or not making an exaggerated, you know, holding-my-breath face, there would be no outside sign that I'm holding my breath. And that's a daily occurrence for him that he has these episodes.

Annalies Winny: Both of Jolly’s children use a pulse ox when they sleep – so for this family, it’s hard to overestimate its importance. It’s not only lifesaving for them on essentially a daily basis, but being able to use one at home… it gives them private time as a family. Without it, they’d need to have medical staff in the house overnight to watch over them while they sleep.

But Jolly has gone through trial and error on numerous pulse ox devices. The readings for her son, who is lighter-skinned than her daughter, are pretty consistent. Her daughter, who is darker-skinned, her pulse oximeter alarm goes off two to three times per week for no reason. And these are hospital grade devices she’s using, which she can afford thanks to a Medicaid waiver from the state of Kansas. She uses two different brands, Masimo and Convaid, and has the same issue with both.

Ryan Jolly: I'm a nurse by training and so you always check the patient first. We turn off the monitor, count to 10, and turn it on. Don't move the probe, don't disconnect the probe, and when it comes back on, her numbers are in perfect range.

Annalies Winny: But what if her daughter was in distress and not even her pulse oximeter raised the alarm?

Ryan Jolly: She can't recognize the need to seek assistance, nor can she call out to draw assistance to her. So if she were to have a airway crisis in the night, she would die.

We spent years of, I don't trust the machine, I'm going to take you out of your crib and put you in bed with me on the machine with my hand on your chest, because I have to sleep—for my daughter, I have to sleep— but I don't trust that this machine is going to do a good job.’ So it was years and years of trial and error and frustration and, and then I got mad. Because it's 2020 and it's 2021, and we should know better, we should do better. Home pulse oximetry has been around a really long time, and we're just now starting to figure out the deficiencies in this equipment, made me angry.

And then through this COVID rabbit hole of sleepless nights and anxiety off the chart, like everybody else,. started to see those first nuggets of research that said, maybe it's the melanin that's prompting the machine to stop receiving the signal.

Annalies Winny: How did we get here? How is it possible that this everyday medical device hasn’t been working well for so many patients—a fact that has been open knowledge in the medical community for years—and nothing was done about it? In researching this series, we’ve been told this is everything from a market failure, to a regulatory failure, to a medical education failure, and even a textbook case of medical racism. We’ll go into all of that.

But to help answer those questions, we wanted to first understand where the pulse oximeter in its current form came from. And that takes us back to the 1940s. The products that led to the pulse oximeters used today were not actually intended as medical devices.

Simar Bajaj studies the history of science at Harvard University and is a research fellow in cardiothoracic surgery at Stanford University School of Medicine. He explained that during World War II…

Simar Bajaj: You had the American and German militaries trying to figure out a solution for their fighter pilots who are going to these heights and blacking out and not having a good sense of their oxygen saturation. So they had these very clunky devices that would clip onto these pilots’ ears, let them know when they might need supplemental oxygen, lest they black out. So that was the origin of these devices. Not as medical technology, but a military one.

Annalies Winny: Then came the space race.

[Archive clip, NASA commentator: 3, 2, 1, zero, all engine running, liftoff, we have a liftoff, 32 minutes past the hour, liftoff on Apollo 11.]

Annalies Winny: Hewlett Packard—the computing giant now known for making computers, printers, and ink—was partnering with NASA engineers to build devices to help astronauts. They wanted to improve oxygen measurements on spacecrafts after the 1960s explosion aboard Apollo 1, which was caused by too much oxygen in the space capsule.

At the time, the company was thriving in the biomedical space, and they sought to parlay that NASA technology into a game-changing medical device. So, they developed another ear-clip oximeter, this one designed to allow patients’ oxygen levels to be continuously measured during surgery. There was something else special about the device:

Simar Bajaj: It was remarkable because HP had a real focus on equity at a time when it wasn't in the news, it wasn't such a broad focus. They were testing this ear oximeter in almost 250 Black patients and created a way to calibrate this oximeter for someone’s blood, giving a way to really make sure the readings were accurate –

Annalies Winny: …regardless of skin tone. Amy Moran-Thomas teaches a class on the social lives of medical objects at MIT. She got her hands on one of HP’s old oximeters on eBay, and recently unboxed it with a group of students. She described it as looking like an old-timey record player.

Amy Moran-Thomas: This is actually the oldest object I've ever been part of like opening the casement of. And yeah, I really didn't know what we would find, you know, it could have been, like corroded wires and broken glass or, you know, but it was all still in, like, quite good shape.

Annalies Winny: The students broke into groups and looked into various components of the device to see how it stood out. For Moran-Thomas…

Amy Moran-Thomas: I guess the overall feeling I had was that, you know, another future was possible.

Annalies Winny: That’s because this device had several things that future versions didn’t. Its designers had gone to great lengths to account for differences in skin tone and other factors like whether a patient is sick when they used it. Knowing that a range of brightness settings would be needed to accurately capture the full range of skin tones, the HP device incorporated 8 wavelengths of light. Most of today’s pulse oximeters use two.

Another difference: To create their baseline calibrations, HP included many Black volunteers. We'll hear more about this in later episodes, but the FDA’s guidelines, which were set decades later in 2013, recommended a minimum of just 2 patients of color be included in pre-market testing for the pulse oximeters used today. And, unlike today’s devices, HP didn’t just test it on healthy patients, but on sick ones too.

Crucially, the readings could be individualized by squeezing a drop of blood from the ear and scanning it using spectro-photometry to figure out just how much light was being absorbed by the patient. This allowed the physician to personalize light level calibrations and achieve more accurate readings.

Plus, clipping the device to the ear, which wouldn't lose circulation as quickly as a finger or toe, would provide more accurate readings for longer periods of time. Another major perk of using the ear: unlike today's devices which are designed for larger, male fingers, the ear clip offered a more universal fit, and thus, better readings. HP’s device was hailed for taking continuous, personalized, equitable, and noninvasive readings that were comparable to those “gold standard” arterial blood gas tests.

But it had some problems: It was very expensive and today a single device would cost the equivalent of over $100,000. It was also cumbersome to use, and it required the patient to remain relatively still. But we’ve all seen how quickly devices can improve and miniaturize, and get much, much cheaper as they are mass produced and production is scaled up. Just look at your cell phone or laptop. But this one did not take off. In the 1980s, HP actually abandoned the product. Its shortcomings meant it wasn’t the commercial success they had dreamed of.

Instead, American companies opted to commercialize a device first designed in Tokyo in the early 70s—the Oximet 1471—which was developed and patented by Japanese engineers at Minolta. The device, which used fiber-optic cables to send light to and from a finger or toe clip, had some of the same problems as its predecessor from Hewlett Packard. It was also clunky, expensive, and extremely sensitive to motion, making readings hard to capture. Commercially, it also failed, only selling a couple hundred devices in Japan. But nevertheless, this design was given a second chance in the U.S. when William New, a Stanford doctor and former HP electrical engineer, thought he could fix those problems, and set out to make the device more clinically practical and commercially viable in America.

He tapped into the rising popularity of LED lights and used a disposable sensor on the device, which made it both cheaper to produce and easier to sanitize. And they also fixed the motion-sensitivity issue. Under a new company, Nellcor, the N-100 pulse oximeter was launched in 1981—and it sold like gangbusters. That demand was driven in part by anesthesiologists, who had come under fire for basically guessing patients’ oxygen levels.

Simar Bajaj: A lot of doctors were just eyeballing it, frankly, a lot of anesthesiologists were looking at a patient, ‘Oh, they look a little bit blue. Looks like they don't have enough oxygen.’ But that is sort of why the Nellcor device became so popular. There was this increasing anxiety and increasing malpractice lawsuits against anesthesiologists who were eyeballing the oxygen levels. People were dying because of it, and there was a sense that there needs to be a better way. And that created the market success for some of the later pulse oximeters.

Annalies Winny: One thing that device didn’t address: skin color. After all, its original design came from Japan, where the population has relatively little variation in skin tone.

Simar Bajaj: In this quest for efficiency, making something that's ready to go to market, there was no time or effort to test this in people with different skin colors. And that I think was something where you had this device transplanted from Japan, without a real understanding of the racial diversity, the ethnic diversity of the United States.

It takes time, money and effort to test your device on people of different skin colors. It's a technological problem to address. How are you going to adjust for skin color? How are you going to adjust for the different pigments? And for many companies, it just wasn't worth it, or they thought it wasn't a big deal. That I mean, what's gonna be the problem, right? A little bit here, there being off, right? Who cares?

Annalies Winny: But there was a reason to care. So, what exactly were they ignoring?

Here’s Joseph Wright, a pediatric emergency physician and Chief Health Equity Officer at the American Academy of Pediatrics.

Joseph Wright: The transmission of bivalent wavelengths of infrared light through the skin is the technology behind the way that pulse oximeters work. And the challenge for melanated individuals is that the presence of melanocytes disrupts the transmission of that infrared light through the skin, the melanocytes absorb some of the light or disrupt the transmission.

Annalies Winny: And by the way, melanocytes are a skin cell that produces melanin, the pigment that darkens skin.

Joseph Wright: And as a result, the reading that the pulse oximeter generates can be spuriously higher than the actual presence of oxygen in the individual's blood, and that difference that—what we call “occult hypoxemia”—is the clinical challenge. In other words, being misled, that a pulse oximeter reading in the normal range is actually not detecting a dangerous situation with regard to the patient.

Annalies Winny: This was all-too-easy to ignore in a society that has historically treated white-skinned people as the baseline for defining whether a product is viable. And where were the regulators in all this? The U.S. Food and Drug Administration was just setting up a device regulatory program. The agency did not stop the marketing and proliferation of a biased device. In fact, it would be decades before the FDA considered more stringent guidelines around testing the devices on a range of skin tones.

Following the success of the Nellcor N-100, many copycat products emerged—some FDA-approved for clinical use, as well as unregulated consumer versions that now sell for cheap on Amazon. And their popularity hasn’t slowed down. In 2022, the global pulse oximeter market was valued at $2.4 billion and is expected to reach $5.4 billion by 2033. And the devices themselves remain largely unchanged from that 1980s Nellcor model.

One analysis of data from more than 200 hospitals by Nicholas Bosch and Anica C. Law of Boston University found that the vast majority of pulse oximeter usage in hospitals—almost 80%—is outside of operating rooms and ICUs. That is, they are being used where there is no other continuous patient monitoring. According to the soon-to-be published analysis, in these contexts, there are fewer opportunities to detect and correct for false readings.

Pulse oximeters are everywhere and clinicians are more and more dependent on them to monitor their patients despite their inaccuracies. But changing course on the design of products would be a massive undertaking. And no one knows entirely what the new design might look like. Is it as simple as adding more wavelengths of light? Or increasing oxygen thresholds for certain people? Or recalibrating the devices for a more diverse population? More sophisticated algorithms to interpret outputs from the devices? Or scrapping the current design entirely? Maybe focusing on the design isn’t asking large enough questions. Here’s Amy Moran-Thomas again.

Amy Moran-Thomas: It's actually really difficult to fix this device without dealing with the institutions around it. Market institutions, regulatory institutions, professional education in medicine. You know, all these other things actually have to also be worked on in order for the object to change.

Annalies Winny: Calls are getting louder for that to happen, and more questions arising about the best way to do it. For Moran-Thomas, imagining a different future for pulse oximeters means remembering that however ubiquitous an object is, it was not inevitable. At a recent gathering…

Amy Moran-Thomas: …Someone brought up the Hewlett-Packard oximeter and called it mythic. It's not a myth that these things could be made differently. They actually were made differently, and they still could be made differently. And just like the Hewlett-Packard device tells a story about the society that made it, the pulse oximeters in hospitals right now do, too.

Annalies Winny: In the next episode, we’ll look deeper at the society that brought us the biased pulse oximeters used today. We will also explore the systemic factors that have kept reform and innovation from happening with these products—from manufacturers to regulators to medical education—and why there is still so little movement on the problem.

Then, we’ll look into the efforts that are being made to fix the bias in pulse oximeters, including the push for new or better products, tighter regulations, and advocacy and education. We hope you’ll keep listening.

Lindsay Smith Rogers: Thanks for listening to part one of our three-part series on pulse oximeters and racial bias, and stay tuned for parts two and three. A special thanks to Annalies Winny and Nicole Jurmo for co-producing this series.

[Podcast Credits, Joshua Sharfstein: Public Health On Call is a podcast from the Johns Hopkins Bloomberg School of Public Health, produced by Joshua Sharfstein, Lindsay Smith Rogers, Stephanie Desmon, and Grace Fernandez Cecere. Audio production by JB Arbogast, Holly Cardinell, Spencer Greer, Matthew Martin, and Phillip Porter, with support from Chip Hickey. Distribution by Nick Moran. Production management by Catherine Ricardo. Social media run by Grace Fernandez Cecere. Analytics by Aliza Rosen. If you have questions or ideas for us, please send us an email to PublicHealthQuestion@jhu.edu. That's PublicHealthQuestion@jhu.edu for future podcast episodes. Thank you for listening.]