SpO2 During Sleep: Using a Wearable to Screen for Sleep Apnea

Sleep apnea is one of the most common serious health conditions in adults, and one of the most under-diagnosed. Roughly one in four adults between 30 and 70 has at least moderate obstructive sleep apnea, and most of them do not know it. This piece explains how consumer wearables can help flag the pattern — repeated overnight drops in blood oxygen — that suggests apnea, what the underlying number (the oxygen desaturation index) actually measures, why a wrist reading is never a diagnosis, and when you should walk into a sleep clinic.

How common is this actually?

The single most-cited paper on modern sleep apnea prevalence is Peppard and colleagues in the American Journal of Respiratory and Critical Care Medicine, 2013. Their analysis of the Wisconsin Sleep Cohort — one of the longest-running community-based sleep studies in the world — estimated that about 26 percent of adults aged 30 to 70 have at least mild obstructive sleep apnea, and about 10 percent have moderate to severe disease.

That is roughly one in four adults with the mild form, and one in ten with the form that clinically matters. And a large majority of them — some estimates put it above 80 percent — are undiagnosed. Sleep apnea does not usually announce itself. The person who has it is asleep for the interesting part.

What sleep apnea does, mechanically, is repeatedly collapse the airway during sleep. Each collapse — an apnea (full closure) or a hypopnea (partial closure) — causes airflow to stop or drop for at least ten seconds. Blood oxygen falls. The brain briefly wakes to reopen the airway. Sleep is fragmented, cardiovascular strain accumulates over years, and the person wakes up unrefreshed without knowing why.

What clinicians measure

The gold-standard test for sleep apnea is a full polysomnography (PSG) study — an overnight recording in a sleep lab or at home with a full sensor kit. From that recording, technicians count two things:

• Apnea-hypopnea index (AHI). The number of apneas and hypopneas per hour of sleep. Under 5 is normal. 5 to 14 is mild sleep apnea. 15 to 29 is moderate. 30 or more is severe.
• Oxygen desaturation index (ODI). The number of times per hour that SpO2 drops by three or four percentage points. Modern sleep labs typically use a 3 percent threshold. ODI closely tracks AHI in most patients but is calculated only from the oxygen signal, which is why consumer wearables can approximate it.

ODI is the important concept for wearable screening. If a consumer device can measure SpO2 continuously through the night, it can count the number of times per hour saturation dips meaningfully. That count is a real physiological proxy for how disrupted your breathing is overnight.

Why 3 to 4 percent desaturations matter

The threshold is not arbitrary. A healthy adult's SpO2 during sleep is stable — a nearly flat line at 96 to 98 percent through the night. Small dips (one percentage point) happen constantly and mean nothing. A drop of three to four percentage points from a stable baseline, sustained for at least ten seconds, almost always reflects a real physiological event: an airway obstruction, a central respiratory pause, or a hypopnea.

An occasional three-percent dip through the night is normal. Five or more per hour, especially clustered in REM sleep and accompanied by snoring, is the pattern of sleep-disordered breathing. Fifteen or more per hour is the pattern of moderate obstructive sleep apnea. Consumer wearables that report overnight SpO2 variability are counting exactly this.

What consumer wearables can honestly detect

The Withings ScanWatch 2 is the wrist device we would point people to first for overnight breathing screening. Its breathing disturbance detection is FDA 510(k) cleared, which means Withings has submitted validation data to the FDA showing the algorithm can reliably identify a pattern consistent with sleep-disordered breathing. Withings has also published peer-reviewed work on the algorithm.

Important word: screening, not diagnosis. The device flags a pattern. It does not tell you that you have moderate obstructive sleep apnea with an AHI of 21. That determination requires a sleep study. What the ScanWatch does — well — is give you a nightly and monthly record of whether the pattern is present, so you can walk into a doctor's office with data rather than a hunch.

The Withings Sleep Tracking Mat is the highest-scoring sleep device in our catalog and the one we would consider strongest for detecting breathing disturbance patterns overnight. The mat records ballistocardiographic and respiratory signals directly from the mattress — cleaner than reflective wrist optics for both heart and breath rate — and Withings has published validation work specifically for its snore and breathing-disturbance detection. It does not measure SpO2 directly, but its breathing-pattern analysis complements a wrist device that does.

The Fitbit Sense and Sense 2 report continuous overnight SpO2 and a variability metric intended to help users notice patterns of disrupted breathing. Fitbit's approach is less regulated than Withings' — it is marketed as a general wellness feature, not as a specific screening tool — but the underlying signal is comparable and the SpO2 variability graph does flag the same kind of pattern in practice.

The false positives you need to know about

Anyone using a wearable to look at overnight SpO2 should know when a scary-looking trace is not what it seems.

• Cold fingers or wrist. Peripheral vasoconstriction reduces pulsatile blood flow at the wrist, which weakens the optical signal. The device sees a low or unstable saturation because it cannot see the pulse well, not because your saturation dropped.
• Loose wristband. If the sensor is not in firm contact with skin, ambient light contaminates the reading. Result: erratic, low, jittery SpO2. Tighten the band before drawing conclusions.
• Sleeping at altitude. Overnight desaturation is normal at altitude — everyone drops. A trace from a night at 8,000 feet does not mean apnea. Compare within similar altitude bands.
• Arrhythmia. Atrial fibrillation and frequent ectopy disrupt pulsatile flow enough to spoil the pulse oximetry algorithm. Reported SpO2 will look artifactual — noisy, low, or missing.
• Skin pigmentation. The FDA's 2021 safety communication and Sjoding et al. in the New England Journal of Medicine (2020) established that pulse oximeters tend to overestimate saturation in darker-skinned users. In this context that matters in the opposite direction: a wearable may under-report the size of overnight drops.

The clinical picture you should not ignore

Consumer wearables are useful when they push people who really do have apnea toward a proper evaluation. The clinical picture that most reliably means "get a sleep study" is a combination:

• Loud snoring, witnessed pauses in breathing, or gasping arousals (partner's report).
• Excessive daytime sleepiness — falling asleep watching TV, in a meeting, driving.
• Morning headaches, dry mouth, or the sense of never feeling rested.
• High blood pressure that is hard to control, especially with early morning surges.
• Repeated overnight desaturations on a wearable — a pattern, not a one-off night.

Any two of the above are worth a doctor's conversation. Three or more is essentially a sleep study waiting to happen.

Home sleep test vs full polysomnography

If your primary-care doctor agrees a sleep study is warranted, there are two main paths.

• Home sleep apnea test (HSAT). A simplified overnight recording you do in your own bed — usually pulse oximetry, nasal airflow, respiratory effort belts, and sometimes body position. Adequate for diagnosing moderate to severe obstructive sleep apnea in patients whose clinical picture is straightforward. Insurance often prefers this path first because it costs a fraction of an in-lab study. Limits: cannot reliably diagnose central sleep apnea, cannot measure sleep stages, will underestimate AHI in some cases because it divides events by total recording time rather than total sleep time.
• In-lab polysomnography (PSG). The full study — EEG, EOG, EMG, ECG, respiratory effort, airflow, pulse oximetry, body position, and video. Overnight in a sleep lab with a technician. Required when the picture is complex (suspected central apnea, comorbid sleep disorders, mild disease with atypical presentation, treatment titration for CPAP).

Consumer wearables sit upstream of both. Their job is to raise the question, not to answer it.

A 30-day self-screening protocol

Here is how to use a wearable well for sleep-breathing screening.

1. Wear the device every night for 30 nights. Same wrist, same tightness, on a snug fit that does not slide during sleep. If your device has a specific sleep-mode SpO2 setting, turn it on.
2. Note your context each night. Alcohol within four hours of bed? Slept on your back? Congested or sick? Altitude? These matter for interpreting the trace.
3. Look at the pattern, not each night. If you have five or more nights in the month showing clustered SpO2 dips, or if your device's proprietary breathing-disturbance metric is elevated multiple times, that is a pattern.
4. Add the clinical questions. Do you snore? Is there daytime sleepiness? Do you have a bed partner who has ever mentioned you stop breathing?
5. Bring the data to a doctor. Print or export a month of nights. Frame the conversation as "here is a pattern I want to understand," not as "I have sleep apnea." A primary-care doctor can refer you for an HSAT, which is where the real work happens.

If a study confirms apnea — the treatment landscape briefly

This piece is about screening, not treatment, but it is worth naming what usually happens after a positive study, because it changes how people think about screening.

• Continuous positive airway pressure (CPAP). Still the first-line treatment for moderate to severe obstructive sleep apnea. A mask connected to a small bedside blower keeps a positive pressure in the airway that prevents collapse. Modern machines are quiet, self-titrating, and cellular-connected. Adherence — actually wearing the mask nightly — is the main barrier, and it improves substantially with good mask fit and patient support.
• Oral appliances. Custom dental devices that hold the lower jaw forward to open the airway. Effective for mild to moderate apnea and for CPAP-intolerant patients. Fitted by a dentist trained in dental sleep medicine.
• Positional therapy. A meaningful fraction of apneas occur only when sleeping on the back. Positional devices (a bump on a chest strap, a vibrating patch) prevent supine sleep. Useful adjunct or standalone treatment for positional apnea.
• Weight loss. Not a fast fix, but a 10 percent body weight reduction typically drops AHI by roughly 25 percent in patients with overweight-related apnea.
• Surgical and implantable options. Hypoglossal nerve stimulators (Inspire) and upper-airway surgeries exist for CPAP-intolerant patients with the right anatomy. Specialist territory.

The point of listing these is straightforward: the reason to screen with a wearable is that if the pattern is there, effective treatments exist. Untreated apnea accumulates cardiovascular, metabolic, and cognitive damage over years. Treated apnea is a manageable condition.

The bottom line

Consumer wearables are not diagnostic. They will not tell you your AHI, they will not stage your sleep against a hypnogram, and they will occasionally throw false-positive traces because your wrist got cold or your band slipped. What they will do — and this is genuinely valuable — is give millions of unaware adults a reason to walk into a sleep clinic before the cardiovascular damage of untreated apnea accumulates for another decade.

If you want the best-supported wrist device for overnight breathing screening, the Withings ScanWatch 2 is the answer — its FDA 510(k) clearance for breathing disturbance detection puts it in a small class. If you want the cleanest passive signal and never want to wear anything to bed, the Withings Sleep Tracking Mat is our highest-scoring sleep device. If you already live in the Fitbit ecosystem, the Fitbit Sense family has continuous overnight SpO2 and a variability metric that will surface the same pattern.

Any of them is a screening tool, not a stethoscope. Use them to notice. Then get the study.