Of all the numbers a wearable produces, blood oxygen is the one where knowing the limits matters most. Read well, it's a useful pattern-spotter. Read as a clinical measurement, it will mislead you in both directions.
What the pattern can show
What repeated overnight readings can do is show you a pattern. A saturation that sits steadily in the healthy range, week after week, is mildly reassuring. A trend that has drifted lower across several weeks, or a cluster of low nights that keeps recurring, is a genuine observation worth taking seriously - especially if it arrives alongside loud snoring, unrefreshing sleep or daytime fatigue, the company that disrupted overnight breathing tends to keep.
The device cannot tell you why. A wearable can't distinguish sleep apnoea from a chest infection from a sensor that lost contact at 3am. It flags that something may be worth looking at. Establishing what, if anything, is a job for a doctor with proper equipment.
Why readings can be wrong
Wrist-based SpO2 is the most error-prone estimate your wearable makes, because everything depends on how well light passes between the sensor and your blood. Readings can come out low for reasons that have nothing to do with your lungs:
- Fit. A watch worn loose, or too close to the wrist bone, is the single most common cause of stray low readings.
- Movement. Rolling over mid-reading corrupts it. Devices discard what they can catch, but not everything.
- Cold skin. Poor circulation to a cold wrist weakens the signal the sensor depends on.
- Skin characteristics. Tattoos under the sensor interfere with the light, and accuracy can vary with skin tone - a documented limitation of the underlying technique rather than of any one brand.
- Altitude. Saturation drops in thinner air, so a trip to the mountains will read lower without anything being wrong.
- Sparse sampling. Most wearables measure periodically overnight rather than continuously, so a handful of poor-quality readings can drag down one night's average.
A single low night is far more likely to be one of these than a change in your body.
A wearable is not a medical pulse oximeter
The fingertip pulse oximeter a clinician clips on is a regulated medical device, applied to a site chosen because the signal there is strong, while you sit still. Your watch estimates the same quantity through your wrist, unsupervised, while you roll around asleep. The two share a measurement principle and little else.
This cuts both ways. A run of low readings doesn't establish that something is wrong. A run of normal readings doesn't establish that nothing is - wearable data is not grounds for dismissing real symptoms, your own or a doctor's concern about them.
When to get proper advice
Symptoms come first. Waking gasping or choking, loud persistent snoring with pauses in breathing, breathlessness at rest, or fatigue that doesn't track your sleep duration are each worth raising with a GP on their own - whatever your app shows.
If the readings themselves concern you - a sustained downward trend, or low nights that keep recurring after you've ruled out fit and the other measurement explanations - that's also a fair thing to bring to a doctor. Bring the trend rather than the worst single night; a pattern over weeks tells a clinician far more than one number. One reading from one night is not a diagnosis, in either direction.
