Racial bias is built into the design of pulse oximeters

Racial bias is built into the design of pulse oximeters
ADVERTISEMENT

How

One of the most indispensable devices of the coronavirus pandemic is the pulse oximeter, which clips onto a person’s finger, shines out a light and reports back a blood oxygen percentage. Patients use pulse oximeters at home to monitor their conditionswhile hospitals use them to identify and prioritize the sickest covid patients. More generally, blood oxygen is known as the fifth vital sign, alongside body temperature, heart rate, breathing rate and blood pressure.

Pulse oximeters, however, don’t work as well in people with darker skin. There’s a risk of “occult hypoxemia,” where the device says that oxygen levels are fine but patients’ actual saturations are dangerously low. Recent medical studies have quantify this bias and the consequences of overestimated oxygen levels. For example, Hispanic and Black patients with the coronavirus were about a fourth less likely to be recognized as eligible for treatment. Obtaining an accurate oxygen reading can literally be a matter of life or death.

Despite the recent surge of attention to this issue, racial bias in pulse oximeters is nothing new; in fact, it was embedded into the very development of this technology. A closer look at the history of oximeters reveals how placing a premium on market expediency over equity allows bias to leach into medicine.

The first oximeters were developed for military, not medical, use. During World War II, fighter pilots were blacking out at high altitudes, so American and German scientists developed oximeters for their respective air forces. These early devices clipped onto pilots’ ears and alerted them when they needed supplemental oxygen.

Hewlett-Packard (HP) went on to develop the ear oximeter for health care in the 1960s and ’70s, with a remarkably liberal, transparent focus on equity. In the October 1976 issue of their journal, for instance, HP acknowledged how oxygen readings were affected by “skin and blood pigments, and the surface characteristics of the skin,” before describing how they designed their own device so that oxygen readings were accurate, irrespective of skin color. HP’s device was also tested on 248 Black patients and could be personally calibrated with a patient’s blood. Yale professor of medicine Meir Kryger — who tested some of the earliest models of the HP oximeter as a pulmonology fellow at the University of Colorado — said that the company “actually took the business about pigments seriously at a time when nobody was.”

But HP’s oximeter was huge and cumbersome to use, not to mention expensive. It cost $13,000 in 1970. The device was thus relegated to a select few research laboratories and understood to be clinically impractical. HP ultimately discontinued its ear oximeter and stopped manufacturing medical devices altogether.

In 1974, however, two Japanese companies took the next leap in oximetry when Nihon Kohden and Minolta independently invented devices that measured oxygen levels via the throbbing of a patient’s arteries. The first “pulse” oximeters had arrived, with both companies filing for patents within a month of each other.

ADVERTISEMENT

Although electrical engineer Takuo Aoyagi won this patent race for Nihon Kohden, the company didn’t pursue the device because it was just a side project for them. “Aoyagi made a prototype,” Katsuyuki Miyasaka — an anesthesiologist at St. Luke’s International University and a close colleague of Aoyagi’s — explained in an interview, “but there wasn’t much interest in developing it further.”

Minolta kept going and, in 1977, released the OXIMET-Met-1471, probably the first fingertip oximeter ever developed. With fiber-optic cables sending light to and from the clip, the device was technologically advanced but, like HP’s, not clinically practical. It was extremely sensitive to motion, too heavy to be used on patients and often overestimated oxygen levels in very sick patients, although it was fairly accurate otherwise. The device was not tested on any people of color because, in a country as ethnically homogeneous as Japan“skin color may not be a problem,” Miyasaka said.

After being frustrated with a lack of success at home (only 200 devices were ever sold), Minolta tried to market their pulse oximeter in the United States, distributing the device for various American hospitals to evaluate. William New, a former HP electrical engineer and anesthesiologist at Stanford University, soon learned about the Minolta device and saw its shortcomings, but also its great potential.

New and two other colleagues founded the company Nellcor and, in 1981, released their own pulse oximeter: the N-100. The device was designed to be clinically practical. With LED lighting and a flexible paper-like sensor, Nellcor’s oximeter was disposable and largely unaffected by motion. One of the device’s most popular features was how its tone changed based on a patient’s oxygen saturation, enabling easier recognition of low oxygen levels.

Nellcor’s timing was extremely fortuitous. LEDs were becoming increasingly available in the early 1980s, just as a series of malpractice lawsuits were brought against anesthesiologists who were eyeballing oxygen levels during surgery. ace Nihon Kohden’s CEO recently wrote, Nellcor “caught the wave of technological innovation and market changes at this time.” The N-100 dominated the market, “selling like hot cakes,” according to Miyasaka. In fact, one Canadian anesthesiologist described how “Nellcor” and “pulse oximeter” became synonymous.

But Nellcor’s device was not an equitable one. The company was so focused on developing an easy-to-use, clinically practical pulse oximeter that it neglected the racial bias built into their devices. In 1987, Kryger compared the N-100 to HP’s oximeter and found that Nellcor’s device was not as accurate or responsive as HP’s.

Racial bias, of course, isn’t unique to Nellcor: Most pulse oximeters have been calibrated in light-skinned individuals alone. And it’s not enough to say that the medical community didn’t know any better: they’ve long understood how yellow skin color from jaundice, blue from sepsis and naturally non-White skin color could lead to “the skin pigmentation effect,” Miyasaka said, “but they thought that, statistically or practically, you can neglect it.” The supposed innocuousness of biased design continues to justify its existence.

HP’s device in the early 1970s was the exception that proves the rule. Although their oximeter was clunky and clinically impractical, it was a paragon for inclusivity because HP engineers made eliminating bias a priority. Racial bias in pulse oximeters, or in any medical device, is never inevitable. Equity requires intentionality.

ADVERTISEMENT

Leave a Reply

Your email address will not be published.