in principlemonths of pandemic,Ashraf Fawzy, then a pulmonary colleague at Johns Hopkins University, noticed something strange. The black patients he treated, all admitted with Covid-19, had oxygen levels scattered throughout the facility, often out of sync with their severe respiratory symptoms.
“I remember one encounter in particular… a relatively young African-American woman with asthma came in withCOVID 19[and] we were removing her supplemental oxygen,” says Fawzy, now an assistant professor of medicine at Johns Hopkins.reverse. But the patient's pulse oximeter, a non-invasive medical device for quickly measuring blood oxygen levels, was giving higher readings than the real thing.oxygen levelsThey were on blood tests.
“We realized that we weren't treating her well with oxygen because we were dependent on these inaccurate pulse oximeters,” he says.
An inaccurate machine may have been an isolated case, but Fawzy wasn't the only one who noticed something odd withpulse oximeters. like himPandemic of covid-1continued and the simple fingertip device became an increasingly recommended way of monitoring oxygen dips, aStudy December 2020noNew England Journal of Medicinecaught national attention: Black patients were nearly three times more likely than white patients to have occult hypoxemia, or dangerously low oxygen levels that were completely missed by pulse oximeters.
"We realized we weren't treating her with enough oxygen because we were relying on these inaccurate pulse oximeters."
The study caused considerable alarm in the medical community, as pulse oximeters are highly reliable medical devices. In 2021, patients and policymakers sent acartato the Food and Drug Administration (FDA) expressing concern about possible racial disparities, calling the pulse oximeter disparity "a matter of life and death".
“The pulse [oximeter] has become a poster child for [health inequality], in part because of the pandemic. It really highlights the need for diversity at all stages of technology development, from the early stages of creating a design to how these things are tested, verified and distributed.”kimani toussaint, a professor of engineering at Brown University, tellsreverse.
Pulse oximeters are just one example of a persistent, frequent, and overlooked problem that thePandemic of covid-1brought up:medicine and healththey harbored racial prejudice from the beginning. But this long-known problem has finally reached the federal level. In November 2022, the FDA discussed plans to improve regulation of pulse oximeters and provide for clearer labeling and more testing of these devices. Now the question is: Can scientists correct these biases, starting with the ubiquitous pulse oximeter, and prevent them from occurring in the future?
The History of the Pulse Oximeter
Long before reaching hospitals and consumer markets, the pulse oximeter was originally developed for the military by American and German scientists during World War II, according toa 1986 article in theClinical Follow-up Diary. The purpose was to prevent fighter pilots from losing consciousness due to lack of oxygen while flying at high altitudes. The first iteration of the device, according tothe report, was more of a pressure earring worn by pilots that warned them to take supplemental oxygen when their oxygen levels dropped below a critical level, making them more prone to blackout.
Pulse oximeters are just one example of a persistent, frequent, and overlooked problem brought to the forefront by the Covid-19 pandemic: medicine and health care has harbored racial bias from the beginning.
Between the 1960s and 1970s, while working with NASA to develophealth devices for astronauts, the computer company Hewlett-Packard (HP) tried to expand the in-ear pulse oximeter into the hospital market.
Interestingly, this would be the first attempt at a fair device, though it probably wasn't just for the sake of fairness.
At that time, and still today, pulse oximeters worked like this: A light source emits two wavelengths of light (red light at 600 nanometers and infrared light at 940 nanometers) onto a finger, passing through the nail, skin, tissue and blood. Hemoglobin, the molecule that carries oxygen throughout the body, absorbs any wavelength, depending on whether it is oxygenated (absorbing more infrared light) or not (absorbing more red light). The amount of each wavelength that is transmitted to the pulse oximeter is used to provide a percentage of the oxygen level in the blood.
HP said it wanted something that worked for all skin types; at theOctober 1976 issue of their magazine., calledO Jornal da Hewlett-Packard, the company acknowledged how oxygen level readings were affected by "skin and blood pigments and skin surface characteristics". HP engineers borrowed the ear clip design and connected it with fiber optic cables that transmitted eight different wavelengths of light to the skin – different brightness settings for different skin types. This construction allowed the oxygen reader to be specifically tuned and calibrated for skin color. In fact, it seemed to work in experimental tests that HP conducted with black patients.Additional HP device studiesshow that this initial pulse oximeter was more in line with invasive arterial blood gas tests, which take a blood sample to measure oxygen levels along with carbon dioxide and pH.
Despite its initial promise, the HP pulse oximeter did not take off and had some limitations,Felipe Bickler, director of the Hypoxia Laboratory at the University of California at San Francisco, tellsreverse.
“It was a clunky clip you put in your ears that didn't really explain the differences in the size and shape of [someone's] ears,” says Bickler.
In addition, it was very expensive in$13,000 a pop in 1970and is mostly found in a few select research labs. With the rise of the personal computing market in the 1980s, HP changed its approach and abandoned plans to miniaturize its pulse oximeter.
Layers upon layers of bias
Later devices would continue to use light as a means of measuring oxygen levels. Given the versatility of light with its spectrum of wavelengths and intensities, it would not have been difficult for pulse oximeter manufacturers to calibrate their devices for darker skin tones, as HP had done. But at the time, the nascent market was largely dominated by people with lighter skin, as in Europe, the United States and Japan, says Bickler. The bias persisted simply because differences in skin color were not considered to be important.
"Any additional complexity [like skin color] was met with resistance and a lack of interest," says Bickler, whoem 2005he co-authored several other early studies investigating how skin color affects pulse oximeter readings. “[Our study] was largely ignored. It was inconvenient for pulse oximeter manufacturers to accept that the way they were testing and calibrating their devices [with lighter skin tones] was flawed.”
The indifference speaks to a broader and persistent crisis of bias in medicine, healthcare, and pharmaceutical or medical device development.Achuta Kadambi, professor of engineering at the University of California, Los Angeles, tellsreverse.
“Essentially, there are three types of bias: physical layer bias, which is what we often talk about with pulse oximeters,” he says. “Then there is a bias in the [artificial intelligence] or computational layer, and a third layer is the interpretation bias, which means that even if the rest of the system is fair, it is possible for a human factor to adjust the output of agreement." .
An excellent example of interpretation bias, says Kadambi, is lung function tests, where results are filtered through equations that compare or correct the data with healthy values based on gender, height, age, and race. For blacks and other ethnic minorities, these adjustments tend to result in lower lung function. This doesn't necessarily reflect someone's actual innate biology. Instead, it normalizes the disparities that lead to poor lung health and exacerbate chronic disease.How did the studies find out?.
The bias extends beyond what is typically used in a clinical setting to our everyday wearable devices that transmit a constant stream of data that we use to monitor and guide our own health.
Other medical-grade devices that harbor bias include the ubiquitous infrared thermometers, which are just as reliable as pulse oximeters for making clinical decisions. FOR2022 Study outside of Emory Universityfound that these forehead thermometers, which infer body temperature through infrared radiation produced by the body, were up to 26% less accurate in detecting fever in black patients compared to oral thermometers.
The bias extends beyond what is typically used in a clinical setting to our everyday wearable devices that transmit a constant stream of data that we use to monitor and guide our own health.
Devices like Fitbits and Apple Watch work similarly to pulse oximeters, using light sensors to capture information like oxygen levels and heart rate. Most of these wearable devices use green light (a cheaper option than the red and infrared light used in a hospital-grade device), which numerous studies over the years have found can provide inaccurate results for those ofdarker skins and people with obesity. This bias did not go unnoticed, judgingonline complaintsand the fact that last month,Apple was hit by a class action lawsuitfor allegedly failing to warn consumers that their watch could not accurately measure blood oxygen levels for darker-skinned people.
In the case of pulse oximeters, subsequent studies since the December 2020 report inNew England Journal of Medicinefound inaccurate readings from pulse oximeters that delay timely Covid-19 treatments for patients with darker skin or who do not receive adequate treatment, according to Fawzy, in a study he co-authored.em 2022, which was published inJAMA. (It's important to note that pulse oximeters aren't the only culprits in the high Covid-19 death rate in blacks and other ethnic minorities.) And this is problematic not just for any darker-skinned person hospitalized with Covid-19, but for any darker-skinned person with breathing problems monitored by pulse oximeters.
“As a pulmonologist, my main research interest is COPD [chronic obstructive pulmonary disease], and in COPD, we prescribe oxygen to patients based on pulse oximeter readings,” says Fawzy. “We've relied on [pulse oximeters] for decades to tell if someone has sleep apnea. So it's really problematic that there's potential underdiagnosis in lung disease due to prescription oxygen [and] potential underdiagnosis of sleep apnea with these devices."
What's the solution?
In February 2021, the FDAissued a warningthat skin pigmentation and other factors can affect pulse oximeter readings. And last November, a medical device advisory committee met to review current clinical data on pulse oximeters and recommended that the agency update its regulation and place labels warning of potentially inaccurate readings. The FDA requested further studies to evaluate pulse oximeter accuracy in hospital settings for adults and children.
For black engineers likeValencia Koomson, an associate professor of electrical and computer engineering at Tufts University, the challenge in compensating for racial bias is not so much about completely overhauling the pulse oximeter as it is about finding ways to improve its performance.
“We are dealing with very weak optical signals that need to pass through fabrics with many [other] elements that absorb and scatter light,” he explains.reverse. “It's a lot like when you get into a car and go through a tunnel. You lose signal due to absorbing materials in the tunnel, so the signal transmitted from the cell tower is too weak for your phone to process.
To remedy this, Koomson and his fellow scientists are working on a pulse oximeter that uses the same light as devices currently on the market, but includes technology that measures a person's skin tone. So, if you have darker skin (i.e. more melanin), the pulse oximeter will emit more light.
However, skin color is not the factor that can lead to inaccurate pulse oximeter readings. At the Hypoxia Lab, Bickler and his colleagues seek to comprehensively understand how these other factors, such as blood flow and body temperature, as well as skin color, affect pulse oximeter readings. In a recent studywhich is currently in prepress, researchers found that poor perfusion, or blood flow through the body's vessels and tissues, exacerbated inaccurate pulse oximeter oxygen readings, along with skin color.
“Poor perfusion is very common in people who are sick,” says Bickler, due to a combination of problems like low blood oxygen levels, dehydration, or medications that cause blood vessels to constrict.
Koomson says that while additional research and innovation is all well and good, it doesn't change the fact that the heritage project needs to happen at the federal regulatory level, with how the FDA sets guidelines for drug approval. pulse oximetry, which are usually relatively neglectful.
“We know that if you have a [federal] guideline that says a device must be tested on a minimum of 10 people and at least 15% must have dark skin pigmentation, that opens the door to a lot of loopholes because 'dark' is so subjective, says Koomson. "This subjectivity will affect the type of product you launch."
Other black engineers, such as Brown's Kimani Toussaint, are taking a slightly eccentric approach in investigating how electromagnetic light waves interact and behave with matter, such as the body's tissues, the skin pigment melanin, and the skin's blood. .
“We're trying to explore [these] properties to see if we can use this to differentiate between the oxygenated hemoglobin and the deoxygenated hemoglobin response,” says Toussaint. "While we strive to completely get rid of bias, it remains an open question how effective our approach is... and how small the effect is of having melanin contribute."
Toussaint's lab also has a prototype that has undergone some initial testing on healthy volunteers. An upcoming clinical trial began earlier this year in the intensive care unit at Miriam Hospital in Providence, Rhode Island. “We're trying to look at this [device] now, compare it to [the] gold standard arterial blood that doctors typically use,” he says.
Bickler, Fawzy and Toussaint are optimistic that highlighting the pulse oximeter will be the clarion call within the healthcare community, encouraging physicians and other healthcare professionals to advocate for a better understanding of the inequalities inherent in these devices. , as well as a search for solutions, especially since we don't have more equitable and non-invasive technologies or alternatives in the foreseeable future.
“The FDA is working very hard to improve [pulse oximeters]. We received funding from the FDA to study pulse oximeters on patients in the hospital and in real-world clinical conditions in a very controlled manner,” says Bickler. “It's been almost 20 years since this was actually defined as an issue, and here we are 20 years later. Yeah, it's too late, but at least it's on the radar."
It may also encourage talk about bias beyond the confines of the pulse oximeter, firmly entrenched in other medical devices and in health in general.
Part of that need for diversity is essential in a workforce that can inform and guide equity during the development of a technology.According to a 2021 Pew Research Center studyWhile science, technology, engineering, and math (STEM) jobs have grown considerably in recent years, black and Hispanic workers remain underrepresented.
“Now we have to think about how technologies are not as agnostic as we thought in terms of their impact on communities,” says Toussaint, “It's part of the human condition to think about how prejudice can arise. ".and be perpetuated, deliberately or not".
THE FUTURE OF YOUexplore tantalizing advances in personal health, from a period-free future to computers in our brains. Read the rest of the storieshere.
this has beenHORIZONS, a newsletter that explores today's innovations shaping tomorrow's world. Sign up for free.