Does Your Wearable's Blood Oxygen Reading Work the Same for Everyone?

This article covers what the research says about skin pigmentation and pulse oximetry accuracy, drawing on clinical pulse oximeter studies and wearable-specific modeling. It does not cover sleep apnea detection, altitude readings, or medical-device-grade oximeters used in hospitals as a point of comparison beyond what's cited.

Research on pulse oximetry, including devices closely related to the reflectance sensors in wearables, has established that skin pigmentation changes how accurately these sensors read blood oxygen, particularly at lower saturation levels. The direction of the error is consistent across multiple independent study designs: darker skin is associated with a higher rate of falsely reassuring readings, not random noise in either direction.

The question people are actually asking

Someone glances at their wrist after a rough night, sees a blood oxygen number that looks fine, and wonders whether that number means the same thing it would mean on someone else's wrist. It's a fair question, and it's not really about the app or the algorithm version. What it's really about is whether the sensor itself sees everyone's skin the same way.

That question has been asked in clinical settings for a lot longer than wearables have existed. The hospital-grade pulse oximeter clipped on a fingertip works on the same basic principle as the light-based sensor in a smartwatch, so the research base is older and larger than most people assume, even though wearables are the newer application of it.

Why the error shows up where it does

The bias isn't evenly spread across all oxygen levels. Bickler et al. found the overestimation was concentrated specifically when true saturation had already dropped into a low range, not at normal, well-oxygenated levels. Which means the reading is least trustworthy exactly when accuracy matters most.

A Monte Carlo modeling study built specifically around reflectance pulse oximeters, the kind used in fitness trackers and smartwatches rather than the clip-on transmission sensors used in hospitals, simulated light-tissue interaction across light, moderate, and dark skin types between 70 and 100% saturation. Error increased substantially as simulated skin pigmentation increased, and the study concluded that a single calibration approach applied uniformly across skin types may not hold up.

I find that distinction between reflectance sensors (wearables) and transmission sensors (most clinical fingertip devices) worth sitting with, because it means findings from clinical pulse oximeter research don't automatically transfer to a wrist-worn device without some translation. For a fuller look at how the two measurement approaches actually differ mechanically, see how wearables actually measure blood oxygen.

What a systematic review of this exact question found

A 2025 systematic review focused narrowly on this question, pulling together studies that compared pulse oximeter readings against arterial blood gas measurements while stratifying results by skin pigmentation or race and ethnicity. Across 42 studies meeting its inclusion criteria, the consistent finding was that pulse oximeters overestimate true arterial oxygen saturation in individuals with darker skin tones, particularly at lower saturation levels, with the review noting this overestimation may delay recognition of low oxygen and the response that follows.

The review also flagged something that complicates clean interpretation: methodological variability across the underlying studies, including inconsistent ways of classifying race and skin tone in the first place. That's a real limitation in the evidence base itself, not just a footnote. If studies can't agree on how to categorize skin tone, comparing their results directly gets harder.

The bigger studies now underway

One reason this question is still being actively studied rather than considered closed is that a lot of the strongest evidence comes from retrospective hospital data, which has its own weaknesses. Gudelunas et al. point out that retrospective studies can mistime when the blood sample was drawn relative to the oximeter reading, mix up functional versus fractional saturation definitions, and lean on self-reported race as a stand-in for actual skin color. That's a rougher measurement than it sounds.

That's part of why newer prospective work exists. A multisite study protocol out of three pediatric cardiac catheterization centers plans to measure skin color objectively with a spectrophotometer rather than by category, and pair it directly with arterial oxygen samples taken during the procedure itself. It's a protocol, not a finished study yet, so it doesn't tell us an answer today. What it tells us is that the field considers the question important enough to fund a large, careful, prospective look at children specifically, a population the existing pigmentation literature leans away from.

Separately, a large real-world diagnostic accuracy study looked at five fingertip pulse oximeters used in an NHS home monitoring program, comparing readings against arterial oxygen in critically ill adults and measuring skin tone objectively with a spectrophotometer rather than by visual category or self-report.

Where this leaves the wrist-worn reading

None of the individual clinical studies cited here tested a specific consumer smartwatch brand against arterial blood samples in a large, diverse group of daily wearers. The evidence connects most directly to pulse oximetry as a technology and to reflectance-sensor modeling, and it's reasonable to expect the underlying optical physics carries over, since wearables use the same light-absorption principle described across this evidence. But the exact scale of error in any particular consumer device on any particular wrist is a narrower question than what these studies were built to answer.

For the mechanics of what a wearable is actually estimating when it shows a number, and how that differs from the arterial measurement used as the reference standard in all of these studies, see what's the difference between SaO2 and SpO2. For how this plays out specifically in overnight readings, where perfusion and stillness both come into play, see how accurate your watch's overnight blood oxygen reading is.

Common questions