Two Wavelengths of Light, One Number That Matters at Night

The volunteer breathes, and the air is not quite air. In the validation protocol, the gas mixture is adjusted, oxygen fraction lowered step by step, inducing a controlled hypoxia — the deliberate kind, the kind with an arterial line attached. From the artery, blood is drawn and run through co-oximetry, the reference method, the ground truth. On the wrist, a small device presses light-emitting diodes and a photodetector against the same patch of skin, reading back the pulsatile signal of blood volume: reflective photoplethysmography. Two wavelengths, absorbed differently by oxygenated and deoxygenated hemoglobin, converted by an algorithm into one number — SpO2. The protocol is patient. Step down, stabilize, sample, record. Step down again.

Across that induced descent, one wrist-worn model held a bias of about 1 percent against the arterial standard. Other validation work has been less kind: a 2023 Respiratory Care study compared a smartwatch against a pulse oximeter during pulmonary rehabilitation, where patients actually need the number to be right, and the broader literature finds performance varies — some consumer devices simply fail formal accuracy thresholds.

Which sets up the real question. The device is only a means; the point is the physiology it watches while its owner sleeps. Does overnight oxygen saturation matter enough to measure? The most defensible answer: yes — the nocturnal signal predicts serious outcomes in large cohorts — provided the measurement is accurate, which is not a given.

The signal that predicts

Researchers have learned to quantify how much oxygen a sleeper loses across the night: the sleep apnea-specific hypoxic burden, a measure of the depth and duration of desaturation events. In large cohorts such as the Sleep Heart Health Study and HypnoLaus, greater hypoxic burden tracks with higher risks of hypertension, cardiovascular events, all-cause mortality and cognitive decline. A 2022 study in the American Journal of Respiratory and Critical Care Medicine tied that burden to cardiovascular events and death; a 2023 analysis in the Annals of the American Thoracic Society showed automatically identified desaturations carry the same associations; a 2022 review in Archivos de Bronconeumología lays out where the metric stands and where it is going. These are cohort associations — consistent, prospective, but not proof that fixing the number fixes the risk. And overnight oximetry desaturation indices screen for obstructive sleep apnea with high diagnostic accuracy, which is the actionable part: a bad night curve is a reason to get a formal sleep study, not a diagnosis.

The shadow on the reading

The limitation here is not statistical; it is physical, and it falls unevenly on people. Pulse oximetry is less accurate in darker skin pigmentation. In a large multi-cohort analysis, occult hypoxemia — real hypoxemia hidden behind a reassuring reading — occurred about three times more often in Black than White patients: 11.7 percent versus 3.7 percent of measurements. The bias is more pronounced in reflectance mode, the very mode wrist-worn consumer devices use, and it has prompted updated FDA guidance. A number that is trustworthy for one user can be quietly wrong for another, and a false reassurance is worse than no number at all.

What the wrist can and cannot do

As a screening and trend tool — flagging desaturation patterns worth bringing to a clinician, tracking respiratory status over time — a validated wearable oximeter earns its place. As a verdict, it does not. It remains a monitoring tool: any benefit depends on what the user, and their clinician, do with the data. This evidence concerns SpO2, photoplethysmography and nocturnal hypoxemia as measured phenomena — not this specific commercial product.

In the laboratory, the gas mixture returns to normal, the arterial line comes out, and the device on the wrist keeps glowing against the skin, still reading. It will read all night, in the dark, honestly within its limits — which is exactly why those limits are worth knowing. Educational, not medical advice.