Pulse oximeter devices routinely overestimate blood oxygen levels in darker-skinned patients—a racial bias that can trigger downstream health harms for Black individuals, compounding well beyond any single inaccurate reading.

A Black patient with a pulse oximeter reading of 94%, for example, may actually have an oxygen level of 91% or lower—a clinically meaningful difference that could warrant intervention. But the clip-on-finger device gives no indication that anything is wrong for that Black individual, so neither the patient nor the clinician is alerted.

In a study published in the Proceedings of the National Academy of Sciences (PNAS) coauthored by Marcella Alsan, MD, PhD, the Annie and Ned Lamont professor of economics and of health policy (by courtesy) at Stanford, researchers investigated how racial bias embedded in pulse oximetry contributes to disparities in follow-up care between Black and white patients. 

“Our research on pulse oximetry is an example of how bias in technical design can propagate through the system and affect health-care decisions,” said Alsan, who is also a senior fellow at the Stanford Institute for Economic Policy Research (SIEPR) and director of the Health Inequality Lab.

The researchers note the device’s measurement error doesn’t just create a one-time disparity; it gets carried forward in subsequent clinical decisions and Black patients systematically receive lower rates of follow-up care than white patients.

The widely used devices give artificially high blood oxygen readings for people with darker skin because the melanin in their skin absorbs light differently than lighter skin, and the devices are typically calibrated using mostly light-skinned people.

Pulse-Ox Problem Persists

There have been many studies over the last two decades showing that pulse oximeters give less accurate readings for darker-skinned patients—with more than half of those studies having come out since 2020. This is likely due to an upsurge of attention from the increased use of pulse oximetry during the COVID-19 pandemic, Alsan notes, as well as the heightened awareness of racial disparities in health care that followed the pandemic's disproportionate toll on Black Americans.

The FDA has acknowledged this problem and has pushed for updated performance standards, but the devices remain largely unchanged because they are deeply embedded in standard medical workflows and no universally adopted, bias-free alternative exists. Some clinical researchers, however, are testing LED-based devices and green-light sensors that measure light reflection rather than transmission—an approach less influenced by melanin.

Earlier studies of this problem shared a common weakness, according to the researchers: They only include patients who happened to get both a pulse-ox reading and a follow-up arterial blood gas (ABG) test—which measures how well your lungs and kidneys are functioning. ABGs require a blood draw and are not routine, so if Black patients are less likely to be sent for one than white patients with the same pulse-ox reading, then the studies were drawn from data that was itself shaped by bias.

Compared to the many papers documenting bias in paired measures of oxygen saturation from pulse oximetry and ABG, there is relatively less work on the consequences of this bias for follow-up care or downstream health outcomes. 

That is what Alsan focused on.

“We complement these existing studies by conditioning only on a pulse oximetry reading so that we can examine the differences in follow-up care among Black and white patients with the same pulse oximetry reading,” said Alsan, a MacArthur Foundation “genius grant” recipient for her work on health inequities.

Using Veterans’ Data

The researchers used data from electronic health records at the Veterans Health Administration (VA), the largest integrated health-care system in the United States. Its comprehensive records and high proportion of Black patients make it well-suited for this analysis. Prior VA research has shown that among inpatients who received follow-up arterial blood gas (ABG) tests, Black patients had lower actual blood oxygen levels than white patients with identical pulse oximetry readings.

The research team used patients in VA emergency departments (ED) where a pulse oximeter is routinely applied to all patients as part of collecting vital signs at intake. They began with all ED visits to 105 VA facilities over the five-year period from 2014 to 2018.

“Using data from 3.5 million ED visits in the VA, we find that Black patients with the same pulse oximetry readings as white patients receive significantly less follow-up care, including an intervention listed by the World Health Organization as being essential: supplemental oxygen,” the researchers wrote.

The other coauthors of the study were the Charles R. Schwab Professor of Economics at Stanford Liran Einav, and health economists Amy Finkelstein of MIT and Jonathan Zhang, of and Duke University.