Jun 4, 2021 9:00 AM

Oximeters Used to Be Designed for Equity. What Happened?

The pandemic drew attention to the racial bias built into pulse oxes. But calls to create a fairer device are missing one thing: It once existed.

A collage of images including a photo of an oximeter.

The gray oximeter sitting on my kitchen table looks like a record player. A product of the 1970s, its alarm for low blood oxygen levels is set by analog dial. I bought it on eBay late last year, after writing a story about the racial bias built into oximetry for the Boston Review. A professor of medicine at Yale, Meir Kryger, reached out afterward with a suggestion: I should also look into the models predating contemporary pulse oximeters, specifically one made by Hewlett-Packard. It’s a technological dinosaur, but in certain ways, its inner workings are more advanced than many devices that measure blood oxygen in hospitals today.

For decades, researchers have documented that many pulse oximeters commonly used in hospitals do not meet FDA safety thresholds for people of color. Since these devices assess oxygen in the blood through optical color-sensing, they can be riddled with errors for people with darker skin, due to racial biases in the calibration process. But when Covid-19 first hit, pulse oximeter readings were nonetheless hailed as a “biomarker” for early hospitalization and during triage. Some patients of color who told ER doctors they couldn’t breathe well were actually sent home when the device indicated they didn’t need oxygen.

It wasn’t until a team of physicians at the University of Michigan reinvestigated the device last December that the broader medical community began to pay more attention. “When a pulse oximeter says 91 percent [oxygen saturation], more than 50 percent of Black patients actually had a value less than 88 percent,” notes study coauthor Tom Valley. The issue has since been taken up by senators and the FDA, drawing intense interest from doctors and engineers, as well as confused patients. There are now widespread calls for redesigning the “pulse ox”—as well as rethinking the review systems that failed to catch or prevent these errors for decades.

But these crucial debates about what may be possible for future models are often still missing an important fact: Oximeters designed to work equitably regardless of skin color, gender, and disability status actually existed back in the '70s. Yet somehow, the history of the device sitting in my kitchen has been mostly erased.

Photograph: Amy Moran-Thomas

Back in the ’60s and ’70s, Hewlett-Packard was collaborating closely with NASA engineers to create devices for the health of astronauts, where the precise measurement of oxygen played a vital role. When the company expanded into hospital markets, it carefully designed an oximeter that could similarly be calibrated for individual patients. It was based on a transmittance approach (shining light through tissues), incorporating fiber optics and eight wavelengths of light. (Many current pulse oximeters use only two wavelengths.) Hewlett-Packard's engineers knew the device would need different brightness settings to work consistently across a range of skin colors. The company offered a transparent and thoughtful discussion—still accessible today in its archive of newsletters—of how it went about building a light-sensing technology that would work equally well on all skin tones.

As steps toward mitigating racial bias, Hewlett-Packard’s engineers marshaled a range of more inclusive approaches to oximetry. The instrument’s baseline calibrations were set by working with a “carefully selected” group, including 248 Black volunteers—which is, notably, 246 more nonwhite people than the FDA currently suggests for pre-market testing of the oximeters in hospitals today. Most importantly, the device could be adjusted for each individual. There was an option to squeeze a small droplet of blood from the wearer’s ear to scan the blood using spectrophotometry. This measurement, which helped discern exactly how much light was being absorbed by an individual’s skin and tissues, allowed the physician to personalize light-level calibrations and optimize the device’s accuracy.

The oximeter could also account for circulation idiosyncrasies. Unlike modern pulse oxes that are tested only on healthy people, Hewlett-Packard’s device was designed to work for people who may be sick. The sensor was not made for the fingertip, for instance, because then the device wouldn’t work as well for patients with common health conditions such as shock, sepsis, and certain chronic illnesses. Instead, Hewlett-Packard placed its sensor on the top curve of the ear, one of the last parts of the body to be impacted by circulation issues during illness. This choice helped prevent building ableism into oxygen measures, while also avoiding gender disparities due to poor device fit. While ear oximeters still exist in specialty niches, by far the most common models in ERs and homes today are nonadjustable and built to fit the “average” geometry of a man’s finger, at times producing suboptimal readings for all others that may well compound with other errors.

Despite these achievements, when the personal computing market exploded in the 1980s, Hewlett-Packard shifted its focus and stepped back from medical equipment shortly before it released a long-planned miniature version of its oximeter. But Kryger still describes its larger device as “the best oximeter ever made.” His lab’s publications from that time show that the HP oximeters were in a number of ways more accurate than the pulse oximeters that soon came to replace them. They were referred to in clinical studies as the noninvasive "gold standard" by which early pulse oxes were tested, because Hewlett-Packard oximeter readings more closely matched the invasive arterial blood gas tests.

As the pandemic has painfully reminded us, the consequences of such inaccuracies can be devastating. Because today’s hospital oximeters are not made with the capacity for personalization, they can inadvertently feed flawed data not only to doctors but also to other machines. Oximeter numbers provide key inputs to a range of computing systems, including the algorithms guiding ICU triage and certain insurance reimbursements. They are also on closed-loop algorithms with many ventilators—and when fed error-ridden inputs, such devices may not be able to optimize as effectively.

Having these conversations now is crucial: As part of AI’s growing role in health care, a wide range of noninvasive sensors are being developed with the pulse oximeter as their model. Some, like certain optical sensors for sepsis or blood glucose, may already be at your local hospital or present in your home. Without care, a coming generation of optical color sensors could easily reproduce the unequal errors for which pulse oximetry is now known across many other areas of medicine.

We tend to assume that technology will unfold with a kind of linear progress, and that useful features or key questions will be built into future models. The history of devices often gets written later as if this had always been the case—that alternative approaches didn't succeed because they were inferior. But like any history, it is useful to ask who wrote it and what’s left out.

You could tell the story of these lost oximeter features as a case of happenstance: When Hewlett-Packard shifted its focus away from medical devices in the '80s, most of the smaller firms that came to fill the hospital device niche didn't have the kind of broadly applied, multidisciplinary experience that years of working with NASA had brought. So when American companies began commercializing the Japanese bioengineer Takuo Aoyagi’s “pulse” addition to oximetry models, they adopted his insights without giving equity considerations a public-facing accounting the way their predecessors had for US markets. Many hospitals that purchased these devices for the first time didn't even realize the newer oximeters were missing features that had previously existed in other models, because the "pulse ox" was their first exposure to noninvasive oxygen measurement at all.

This technological loss could also be told as a story about shifting historical and societal norms over time. Hewlett-Packard's 1970s model included a transparent discussion of equitable design—but this also took place against the backdrop of the hard-fought gains of the Civil Rights movement, when issues of racial equity were being more publicly discussed across sectors. Meanwhile, the later pulse ox models (the first in the US was patented by Biox in 1980) became one more face of the corporate enclosures of the era. When they first came to market, Kryger recalls, he tried to ask the engineers about calibration data, as was once considered standard safety practice. But they would no longer share it. “They were black boxes hidden with proprietary algorithms,” Kryger says. “The engineers at the time would not give any technical information such as accuracy in people with dark skin pigment.”

When black-boxing began further encroaching into medical technology in the ’80s and ’90s, there was a flurry of studies by doctors concerned about pulse ox accuracy. Eventually, though, doctors got used to not knowing and stopped asking certain questions about the bias built into devices. Blind spots grew. “Today coded inequity is perpetuated precisely because those who design and adopt such tools are not thinking carefully about systemic racism,” Princeton professor Ruha Benjamin observes in Science, in examining the range of ways racial biases are built into hospital algorithms. People often refer to today as the age of precision medicine. But the customizable Hewlett-Packard counter-model stands as evidence of a more uneasy and uneven story about what happens when data and design decisions are put in a black box, severed from public accountability despite insidious errors.

However you frame it, relearning these histories is a chance to envision different futures. Beyond this pandemic and others that may follow, oximeters are used everyday in crucial and delicate moments, such as childbirth, that are already notorious for amplifying racial inequities. That the pulse ox’s longstanding design problems have become harder to ignore during Covid is a reminder of the ongoing stakes of remaking it—and the root causes that have made its biases possible. Today a rising generation of physicians such as health policy specialist Onyeka Otugo and the transnational critical care trainees who recently wrote in The Lancet join with the multidisciplinary engineers and social scholars of health like Kadija Ferryman and Mikaela Pitcan, who are asking hard questions about hospital technologies anew, with fresh insights for design policy.

And a few, like Kryger, still remember what got written out of later histories of oximetry: It is possible to engineer color-sensing devices with a foundational commitment to equitable design. It is possible to ask companies to openly share full calibration and accuracy data as a matter of medical safety and fairness. It is possible to design hospital devices that work optimally for all patients. Hewlett-Packard’s oximeter accomplished all of these things nearly half a century ago. The particular one sitting on my kitchen table doesn’t have all the proper retro cables, so it’s not really functioning. But it’s become an important reminder for me: Equitable device design has been achieved before. It needs to happen again.