SpO2 and Sleep Apnea: What the Numbers on Your Device Mean

The SpO2 number on your wristwatch or finger clip estimates the percentage of hemoglobin in your blood that is carrying oxygen. For most healthy adults, that figure hovers between about 95% and 100% while awake. During sleep, brief dips are normal, but repeated or prolonged drops below 90% are a hallmark of sleep apnea and a signal worth taking seriously. Understanding what those numbers mean, where consumer devices get it right, and where they fall short can help you decide whether a conversation with a doctor is overdue.

How a Pulse Oximeter Gets Its Number

Pulse oximeters work by shining two wavelengths of light through your skin and measuring how much each wavelength is absorbed. Oxygenated blood absorbs light differently from deoxygenated blood across the visible and near-infrared spectrum, and the device uses that difference to calculate your oxygen saturation.¹ Traditional medical-grade oximeters clip onto a fingertip and use transmission-based sensing, meaning the light passes all the way through the finger to a detector on the other side. Smartwatches, by contrast, shine light into the wrist and read what bounces back, which is inherently noisier because the signal has to contend with more tissue, motion, and variable blood flow.

What Happens to Oxygen Levels During Sleep Apnea

In obstructive sleep apnea, the soft tissue at the back of the throat collapses repeatedly during sleep, partially or fully blocking the airway. Each blockage chokes off airflow for seconds to over a minute, and your blood oxygen drops until the brain rouses you just enough to restore muscle tone and reopen the passage. In central sleep apnea, the problem is different: the brain temporarily stops sending the signal to breathe, producing similar cycles of insufficient ventilation and compromised gas exchange.² In either case, the oxygen-desaturation-and-recovery cycle can repeat dozens or even hundreds of times per night.

The pattern of oxygen drops also shifts depending on your sleep stage. In people with severe sleep apnea, oxygen saturation tends to be lower during REM sleep than during non-REM sleep. One study found average SpO2 during REM was about 92% versus about 93% during non-REM in the severe group, a statistically significant gap driven by the greater muscle relaxation that characterizes REM.³ People who snore but don’t have apnea showed the opposite pattern, with slightly higher saturation during REM. If your device logs timestamps, you may notice your worst dips cluster in the second half of the night, when REM periods are longest.

The Metrics That Matter Beyond a Single Number

A single SpO2 snapshot tells you very little. Sleep specialists care about patterns over the full night, and two metrics in particular have become central to understanding how much damage intermittent hypoxia is doing.

The oxygen desaturation index (ODI) counts how many times per hour your blood oxygen drops by a set amount, usually 3% or 4% from baseline. ODI correlates strongly with the apnea-hypopnea index (AHI), the traditional measure of sleep apnea severity. Research has found the two metrics match in classifying severity about 87% of the time, and an ODI above 20 per hour has roughly 97% sensitivity for detecting severe obstructive sleep apnea.⁴ A separate study found a correlation coefficient of 0.9 between AHI and ODI across all severity groups, leading researchers to argue ODI is as valuable as AHI for diagnosing and grading the condition.⁵ This matters for you because some consumer devices report an ODI-like number, even if they can’t give you a formal AHI.

The other metric is T90, shorthand for the percentage of total sleep time your SpO2 spends below 90%. This captures cumulative oxygen burden rather than individual dip counts. In patients with severe obstructive sleep apnea, spending more than 10% of sleep time below 90% roughly doubled the odds of having high blood pressure compared to spending 5% or less, and the risk remained elevated after accounting for other factors like weight and age.⁶ Longitudinal data from two large cohorts also found that higher T90 was associated with increased all-cause mortality in both clinical and community settings.⁷ So when your device shows you a graph of your overnight SpO2, the area of that graph spent below the 90% line is arguably the single most important thing to look at.

How Trustworthy Is Your Smartwatch?

The good news is that current leading smartwatches have gotten meaningfully better at measuring SpO2. A comparison study testing three popular models against a medical-grade reference oximeter found that all three met the accepted accuracy threshold of no more than 4% root-mean-square deviation, even during induced low-oxygen episodes below 90%. The Apple Watch 8 performed best in that comparison, with the lowest bias and highest correlation to the reference device.⁸

The bad news is that “meets a 4% threshold” still leaves a lot of room for error when the numbers that matter most are in a narrow range. The difference between an SpO2 of 92% and 88% is clinically enormous, but it falls within the acceptable error margin. Add in wrist motion during sleep, a loose band, or cold hands, and accuracy degrades further. Cold fingers alone have been shown to cut pulse signal strength by roughly half compared to warm-hand conditions, and the resulting SpO2 estimates drifted further from reference values.⁸ A smartwatch is useful for spotting a pattern of repeated dips that warrants investigation. It should not be used to make treatment decisions or rule out a problem.

The Skin Tone Problem

Pulse oximeters have a documented bias related to skin pigmentation, and it runs in a direction that can mask real danger. In people with darker skin tones, SpO2 readings tend to read higher than the true arterial oxygen level. A large NHS study found readings were 0.6 to 1.5 percentage points higher in patients with darker skin compared to lighter skin, and the rate of false negatives, meaning the device says oxygen is fine when it actually isn’t, was substantially higher in darker-skinned patients.⁹ At a clinical threshold of 94%, the proportion of reassuring readings masking genuinely low oxygen was two to seven times higher in people with darker skin.

A systematic review confirmed this overestimation pattern and noted that the bias grows worse precisely when oxygen levels are already low, which is when accurate readings matter most.¹⁰ If you have darker skin and your smartwatch or bedside oximeter shows borderline numbers, the real values may be worse than what the screen displays. This doesn’t mean pulse oximetry is useless for you, but it does mean trends matter more than any single reading, and a professional sleep study with arterial blood gas confirmation is especially important if sleep apnea is suspected.

Why Repeated Oxygen Drops Are Harmful

The danger of sleep apnea isn’t just lousy sleep. The repeated cycle of oxygen dropping and then surging back when you gasp awake creates what researchers call intermittent hypoxia, and it sets off a chain of problems that extend well beyond the bedroom. The re-oxygenation phase is in some ways worse than the low-oxygen phase, because the rapid return of oxygen generates a flood of reactive oxygen species. Over time, this oxidative stress damages the inner lining of blood vessels.¹¹

This endothelial damage reduces the availability of nitric oxide, a molecule that keeps blood vessels flexible and dilated, while ramping up inflammation and clotting tendency.¹² At the same time, the repair capacity of the vessel lining is diminished, meaning the body loses its ability to fix the damage as fast as it’s being created.¹³ The downstream consequences include sustained high blood pressure, atrial fibrillation, and accelerated atherosclerosis.¹⁴

The cardiovascular effects get most of the attention, but the brain takes a hit too. Research has found a moderate correlation between sleep apnea severity and cognitive impairment, with more severe apnea linked to worse performance on tests measuring executive function. Depression and excessive daytime sleepiness also track together in these patients.¹⁵ The cognitive fog many people with untreated sleep apnea describe isn’t just about fragmented sleep; it’s also about nightly oxygen deprivation to the organ most sensitive to it.

What Treatment Does to the Numbers

The standard treatment for obstructive sleep apnea is continuous positive airway pressure (CPAP), which splints the airway open with a steady flow of pressurized air. CPAP does more than just raise average SpO2; it fundamentally changes the pattern. The desaturation-recovery cycling largely stops, T90 drops toward zero, and the sympathetic nervous system activation that drives blood pressure spikes quiets down.

A meta-analysis comparing CPAP to supplemental oxygen therapy found that both approaches raised overnight SpO2 levels by similar amounts. However, CPAP was significantly better at reducing AHI, lowering T90, improving sleep quality, and bringing down blood pressure.¹⁶ In other words, simply pumping in more oxygen doesn’t fix the underlying airway collapse and its consequences, even if the SpO2 number looks better. This is a useful reminder that chasing a single number on your device can be misleading: treating the cause matters more than treating the metric.

Data from titration studies show just how dramatic the oxygen improvement can be. In patients undergoing positive airway pressure titration, average SpO2 rose from roughly 80% before treatment to about 89% after, a clinically massive shift.¹⁷ Those numbers reflect patients with very severe disease during a titration study, not typical home readings, but they illustrate the scale of the problem CPAP can address.

When Other Conditions Muddy the Picture

Sleep apnea doesn’t exist in isolation, and the most important confounding condition for SpO2 interpretation is chronic obstructive pulmonary disease (COPD). People with both COPD and OSA, sometimes called overlap syndrome, have two separate reasons for nocturnal oxygen drops: airway collapse from apnea and impaired gas exchange from damaged lungs. The Sleep Heart Health Study found that patients with both conditions had more prolonged nocturnal oxygen desaturation than those with OSA alone, and the degree of airway obstruction measured by lung function testing correlated with the risk of extended hypoxemia.¹⁸

If you have COPD and your overnight SpO2 readings look terrible, sleep apnea may be contributing more than you think, or the lung disease alone may explain the numbers. Separating the two usually requires a formal sleep study, because treatment differs. CPAP addresses airway collapse but doesn’t fix emphysematous lung tissue; supplemental oxygen addresses the gas exchange problem but doesn’t prevent apneas. Many overlap patients end up needing both.

Altitude and Overnight Oxygen

If you’ve ever checked your SpO2 while camping in the mountains and panicked at the number, context matters. A systematic review found a strong inverse correlation between altitude and mean sleep SpO2, meaning the higher you go, the lower your overnight readings drop.¹⁹ At the same time, altitude increases central apneas, where the brain temporarily stops commanding the lungs to breathe, as part of the body’s unstable ventilatory response to low oxygen. AHI climbs with altitude, driven primarily by these central events.

Interestingly, the periodic breathing that develops at high altitude doesn’t seem to make mean SpO2 worse beyond what the altitude itself already does. A study on healthy young men at simulated high altitude found no within-subject difference in average SpO2 between periods of normal breathing and periods of periodic breathing during sleep. The researchers concluded that sleep SpO2 at altitude is primarily determined by baseline waking saturation rather than the breathing pattern that develops during sleep.²⁰ The practical takeaway: if you live at sea level and travel to altitude, expect your device to show lower overnight readings without that necessarily meaning you’ve developed sleep apnea. If you already have sleep apnea, altitude will likely make your numbers worse and may require adjustments to your CPAP pressure or supplemental oxygen.

Home Sleep Tests Versus the Lab

A full in-lab polysomnography (PSG) remains the gold standard for diagnosing sleep apnea, but it involves spending a night in a sleep lab wired with sensors for brain waves, eye movements, muscle activity, airflow, chest effort, and SpO2. Home sleep apnea tests have become far more accessible and increasingly comparable in accuracy. Newer Type 2 home testing devices have shown no statistically significant differences from in-lab PSG across key respiratory metrics.²¹ Patch-based systems have also demonstrated equivalent respiratory findings between home and lab settings, with the added benefit that people tend to sleep longer and more efficiently in their own beds.²²

One lingering issue is how home devices score hypopneas, the partial breathing reductions that count toward AHI. Standard home test scoring criteria have been shown to underestimate AHI compared to in-lab PSG by an average of about 7 events per hour. Modified scoring approaches narrowed that gap to under 2 events per hour and eliminated the statistical significance of the difference.²³ This scoring nuance matters because it can shift someone from a “mild” to “moderate” classification, potentially changing whether insurance covers treatment. If a home test gives you a borderline result and you have strong symptoms, it may be worth pushing for a lab study or asking your sleep specialist about the scoring method used.

Cold Hands and Other Practical Accuracy Killers

Beyond skin tone, a few mundane factors reliably degrade pulse oximeter accuracy. Cold extremities are the most common culprit. When your fingers or wrists are cold, blood flow to the surface shrinks, and the optical signal the device depends on weakens dramatically. Research measuring pulse signal strength found that cold conditions cut the infrared signal amplitude by about 60% and the red-light signal by roughly 46% compared to normal temperature. Warming the hands reversed this and improved signal quality by up to four times, narrowing the gap between experimental and reference SpO2 readings to a median of about 1.5 percentage points.⁸ If you sleep in a cold room or tend to have cold hands, warming up before reading or wearing your watch snugly on a warm wrist will give you better data.

Nail polish, especially dark or blue shades, can absorb light at the same wavelengths the sensor uses and throw off fingertip devices. Motion is the other big enemy: tossing and turning, restless legs, or simply shifting position creates artifacts in the signal that the device may interpret as oxygen changes. Most clinical-grade devices filter out some motion noise, but consumer wearables vary widely in how well they handle it. If your overnight graph has brief, isolated spikes downward that don’t cluster with others, motion artifact is a more likely explanation than a real apnea event.