Pulse oximeters fail Black patients. The science, the FDA's response, and what to ask in the ER.

The anchoring study is Sjoding et al., Racial Bias in Pulse Oximetry Measurement, published December 17, 2020 in the New England Journal of Medicine (PMID 33326721). Researchers at the University of Michigan compared paired pulse oximeter readings against arterial blood gas measurements (the clinical gold standard for measuring oxygen in the blood) in two cohorts: 1,333 white patients and 276 Black patients in the Michigan Medicine ICU, then a multicenter cohort of 7,342 white and 1,050 Black ICU patients. They asked a precise question: how often does a pulse oximeter reading in the 'safe' range (SpO2 between 92% and 96%) coexist with an actual blood-oxygen level that is dangerously low (SaO2 below 88%), the situation the literature calls 'occult hypoxemia'?

The answer, in both cohorts: occult hypoxemia occurred in Black patients at roughly three times the rate of white patients. The Michigan ICU sample showed 11.7% of Black patients vs. 3.6% of white patients had hidden low oxygen. The multicenter sample showed 17.0% of Black patients vs. 6.2% of white patients. The bias was not subtle and it was not explained by other clinical factors.

Pulse oximetry is built on a physical principle. The device shines two wavelengths of light, red (around 660 nanometers) and near-infrared (around 940 nanometers), through tissue and measures how much of each wavelength comes out the other side. Oxygenated hemoglobin and deoxygenated hemoglobin absorb the two wavelengths in different ratios, and that ratio is what the device translates into a percentage on the screen.

Melanin (the pigment that gives skin its color) also absorbs light, including in the wavelengths the device is measuring. The more melanin in the path between the LED and the photodetector, the more attenuation of the signal, and the noisier the calculation that produces the SpO2 reading. The clinical algorithms inside commercial pulse oximeters were originally calibrated against volunteer cohorts that were predominantly white, which compounds the problem: not only does melanin attenuate the signal, but the calibration data the device uses to translate the signal into a number did not include enough darker-skinned subjects to model that attenuation correctly.

The result is what Sjoding's team measured: the device reads systematically high in patients with more melanin. A true SaO2 of 86% (genuine hypoxemia, supplemental oxygen indicated) gets reported as 92% on the screen (within the safe range, no action triggered).

Missed hypoxemia is not a paperwork problem. Hospital decisions about supplemental oxygen, escalation of respiratory support, ICU admission, and (during COVID) eligibility for monoclonal antibody and antiviral therapy were all routinely triggered by SpO2 thresholds. A patient whose true oxygen was 86% but whose pulse oximeter read 92% did not get the care they needed.

Subsequent peer-reviewed research replicated and extended the finding. A study in JAMA Network Open in 2021 by Wong and colleagues, using a multi-site ICU cohort, documented that Black, Hispanic, and Asian patients all had higher rates of occult hypoxemia than white patients, and that those readings were associated with delays in recognition of clinical deterioration. Other studies have examined whether the bias affects whether and when patients receive supplemental oxygen. The work is ongoing, and the weight of the evidence so far is that the bias has real downstream effects on care.

The FDA acknowledged the problem in a public Safety Communication on February 19, 2021. The communication advised clinicians and patients that pulse oximeter readings can be affected by 'poor circulation, skin pigmentation, skin thickness, skin temperature, current tobacco use, and use of fingernail polish,' and to interpret them in the context of patients' overall clinical presentation rather than as standalone numbers. The FDA's pulse oximeter overview page hosts the current consumer guidance.

On November 1, 2022, the FDA convened a meeting of its Anesthesiology and Respiratory Therapy Devices Panel of the Medical Devices Advisory Committee to specifically address pulse oximeter performance across skin pigmentation. The panel reviewed Sjoding's data, the broader literature, and industry submissions. The FDA's own briefing materials for that meeting acknowledged that the existing 2013 guidance for industry on pulse oximeter clearance, which required device validation against arterial blood gas in at least ten subjects spanning a range of skin tones, was almost certainly not enough to detect the kind of bias Sjoding documented.

In late 2023, the FDA published a Discussion Paper describing proposed approaches to improving how pulse oximeters are evaluated for accuracy across skin pigmentation. The draft included larger required subject panels with explicit pigmentation distribution, performance metrics that account for the bias type Sjoding identified, and labeling that would surface a device's skin-tone validation data to clinicians and consumers. Public comment opened. The FDA's ongoing prospective clinical study on pulse oximeter errors across varying skin pigmentation is the current research program backing that effort. As of this writing, the final guidance has not been issued. In practice, this means consumer pulse oximeters sold in 2026 are still cleared under the older standard.

This part is service journalism. If you or a family member is in the emergency department or being assessed for respiratory illness, the SpO2 reading on the screen is one data point. It is not the only one available, and you have the right to ask for confirmation when it matters.

Sentences worth saying out loud:

• 'I've read about racial bias in pulse oximeters. I'd like to know the exact reading and whether it should be confirmed with an arterial blood gas.'
• 'Can you document my pulse oximeter reading and that I asked about confirmation?'
• 'My symptoms are X, Y, Z. Please consider those alongside the device reading, not instead of them.'
• 'If my reading is borderline, can we err on the side of more diagnostic testing rather than less?'

An arterial blood gas is the clinical gold standard. It involves a small blood draw from an artery (usually in the wrist) and measures the oxygen content of the blood directly, without going through the indirect light-based calculation that pulse oximetry uses. It is more uncomfortable than a finger clip, but it is the test that is not affected by skin pigmentation, perfusion, or nail polish.

The honest framing: as of 2026, no consumer-grade pulse oximeter is fully solved on this. Even devices that publish validation data on darker skin tones do so against the older FDA standard, which the FDA itself has acknowledged is inadequate to detect the bias Sjoding measured. Some practical guidance:

• Look for FDA-cleared finger pulse oximeters (not 'wellness' or 'fitness' devices). Cleared devices have at least submitted accuracy data; wellness devices have not.
• Look for manufacturer-published validation data that explicitly includes darker skin tones. Masimo, the dominant clinical-grade pulse oximeter manufacturer, has published more on this than most. Their MightySat consumer product is FDA-cleared for prescription use and uses the same signal-processing platform Masimo sells to hospitals.
• Treat smartwatch SpO2 sensors as wellness data, not clinical data. Apple Watch, Fitbit, and Garmin oximetry features are not FDA-cleared for medical decisions and have not been independently validated across skin tones at the rigor a clinical decision warrants.
• If your child has a respiratory condition (asthma, post-RSV, post-bronchiolitis), ask the pediatrician directly about pulse-oximeter use at home, including which specific device and how to interpret it given the racial-bias literature.

One more honest caveat: home pulse oximeters became popular during COVID and many studies have found that patients use them without context, which produces either false reassurance or unnecessary anxiety. Whatever device you buy, talk to your clinician about when a reading should prompt a call, not just a record.

It is not a finding that pulse oximeters are useless, and it is not a finding that Black patients should refuse them. The device remains the dominant non-invasive tool for monitoring oxygen at the bedside, and for most patients most of the time, the reading is in the right ballpark. The finding is more specific: when the actual oxygen level falls into the dangerous range, the device is meaningfully more likely to miss it on darker skin than on lighter skin. That gap is what the FDA process is trying to close, and it is what the clinical reader should keep in mind when the number on the screen is borderline.

The fix, when it comes, will be a combination of better signal processing, larger and more representative calibration datasets, and labeling that makes a device's skin-tone performance visible to the clinician at the bedside. Until then, awareness is the patch.

• Sjoding MW, Dickson RP, Iwashyna TJ, Gay SE, Valley TS. Racial Bias in Pulse Oximetry Measurement. New England Journal of Medicine. 2020 Dec 17;383(25):2477-2478. PMID 33326721. DOI 10.1056/NEJMc2029240.
• FDA Safety Communication: Pulse Oximeter Accuracy and Limitations. February 19, 2021. Current consumer guidance at fda.gov/medical-devices/.../pulse-oximeters.
• FDA Anesthesiology and Respiratory Therapy Devices Panel Meeting on Pulse Oximetry, November 1, 2022. FDA Advisory Committee record.
• FDA Pulse Oximeter Research Program (Office of Science and Engineering Laboratories). Current prospective clinical study on pulse oximeter errors across varying skin pigmentation at fda.gov/science-research.
• Wong AKI, Charpignon M, Kim H, et al. Analysis of Discrepancies Between Pulse Oximetry and Arterial Oxygen Saturation Measurements by Race and Ethnicity. JAMA Network Open. 2021;4(11):e2131674. PMID 34730820.