Why Pulse Oximeters Can Miss Low Oxygen in Darker Skin

Pulse oximeters can read falsely high in people with darker skin, and the gap matters most exactly when oxygen is running low. In a widely cited 2020 New England Journal of Medicine analysis, Black patients had roughly three times the rate of "occult hypoxemia," meaning genuinely low arterial oxygen that the fingertip reading missed, compared with White patients. Melanin absorbs some of the light these devices depend on, nudging the estimate upward. The FDA's January 2025 draft guidance tightens how new devices must prove their accuracy across skin tones, but a draft is not a fix, and it does not recall a single meter already clipped to a patient's finger.

Why the light gets fooled

A pulse oximeter is an educated guess dressed up as a hard number. It shines red and infrared light (typically near 660 and 940 nanometers) through a fingertip and infers oxygen saturation from how much of each color the pulsing blood absorbs. Oxygenated and deoxygenated hemoglobin absorb those two wavelengths differently, and the ratio yields the SpO2 you see on the display.

The trouble is that melanin also absorbs light in this range. In skin with more pigment, that extra absorption can distort the ratio the algorithm relies on, and the calibration curves baked into many devices were historically built on lightly pigmented volunteers. The result is a systematic lean, not random scatter. The reading tends to overestimate true saturation, and the overestimate is largest at the low end, where a clinician most needs the truth.

What the evidence actually shows

The signal here is consistent and clinically pointed. In the NEJM correspondence by Sjoding and colleagues, the authors compared paired pulse oximeter readings against arterial blood gas, the direct measurement of oxygen in the blood. They defined occult hypoxemia as an arterial saturation below 88 percent despite a reassuring oximeter reading of 92 to 96 percent. In the University of Michigan cohort, that hidden hypoxemia appeared in 11.7 percent of Black patients versus 3.6 percent of White patients. In a second, multicenter cohort, the figures were 17.0 percent versus 6.2 percent. The pattern held across tens of thousands of paired measurements.

Those percentages describe a decision trap. Oxygen thresholds gate real actions: whether to escalate care, titrate supplemental oxygen, admit a patient, or, during the COVID-19 surge, whether someone even qualified for certain treatments. If the number reads 94 when the artery says 86, the trap closes silently.

What the FDA's 2025 draft changes

On January 7, 2025, the FDA released draft guidance titled "Pulse Oximeters for Medical Purposes," with a public comment period that ran to March 10, 2025. It rewrites how manufacturers should demonstrate accuracy across pigmentation for devices seeking clearance.

Two changes stand out. First, sample size and diversity: the draft recommends clinical validation in at least 150 healthy participants spanning the full range of skin tones, with substantial representation in the darkest categories. That is a large step up from the prior expectation, which could be satisfied with as few as ten subjects and only a token number of darkly pigmented participants. Second, and arguably more important, the draft asks sponsors to characterize skin tone with both a subjective scale (the Monk Skin Tone Scale) and an objective optical measure (individual typology angle, derived from a colorimeter reading) rather than a vague label like "dark." Measuring the variable properly is the precondition for detecting bias at all.

What it does not fix

Here the appraisal has to be honest. A draft guidance is a proposal, not a rule. Until it is finalized through notice-and-comment and applied to submissions, it changes nothing a manufacturer is legally required to do, and even once final, FDA guidance describes the agency's current thinking rather than binding law.

Its reach is also forward-looking. The recommendations govern how new devices are tested and labeled. They do not retroactively recalibrate the millions of oximeters already in hospitals, clinics, and nightstands, and they say little about the unregulated consumer and smartwatch oximeters sold for "wellness," which are not cleared as medical devices and were never held to the clinical standard in the first place.

There is a deeper limit. Better test design detects the bias more reliably; it does not, by itself, redesign the optics or the algorithm so the bias disappears. Validation still happens in healthy volunteers under controlled desaturation, a cleaner setting than a hypotensive, cold, or moving patient in an actual bed. And the guidance cannot legislate interpretation at the bedside, where the reflex to trust a round green number remains the real hazard.

What this means for reading the number

The practical posture that follows from the evidence is skepticism calibrated to context. A single SpO2 value carries an error bar that widens with skin pigmentation and with lower true saturation, so a borderline reading in a patient with darker skin deserves more corroboration, not less: the trend over time, the clinical picture, and, when the stakes are high, a direct arterial measurement. This article is educational and not medical advice, and it is not a verdict on any specific device.

The larger lesson is about measurement humility. A number that looks precise can still be biased, and the correction begins with testing that is honest about who was in the room when the device was calibrated.