Roughly one in four adult men and one in ten adult women have moderate-to-severe obstructive sleep apnea, and most of them do not know it. The Wisconsin Sleep Cohort prevalence work (Peppard et al., American Journal of Epidemiology, 2013) estimated that 80 to 90 percent of moderate-to-severe cases remain undiagnosed in the general population. The gold-standard diagnostic is in-lab polysomnography, which is expensive, requires a referral, and books out months in advance in most countries. The good news: two metrics derived from continuous overnight pulse oximetry — T90 and the Oxygen Desaturation Index — are validated screening tools that wearables with continuous SpO2 can compute. This article explains what they are, what thresholds matter, which devices can produce them, and what to do if your numbers are concerning.
What sleep apnea is, briefly
Obstructive sleep apnea is the repeated collapse of the upper airway during sleep, causing partial (hypopnea) or complete (apnea) blockage of airflow. Each event lasts at least 10 seconds and typically ends with a brief arousal as the brain registers falling oxygen and elevated carbon dioxide. The biological consequence is fragmented sleep, repeated nocturnal oxygen desaturation, sustained sympathetic activation, and over years, elevated risk for hypertension, atrial fibrillation, type 2 diabetes, stroke, and cognitive decline (Marin et al., Lancet, 2005, n=1,651 with 10-year follow-up).
The diagnostic standard is the Apnea-Hypopnea Index (AHI), the number of apnea plus hypopnea events per hour of sleep, measured via polysomnography. AHI under 5 is normal, 5 to 14 is mild, 15 to 29 is moderate, and 30 or higher is severe (American Academy of Sleep Medicine scoring manual, 2017).
The reason wearables matter for screening: although wrist or finger devices cannot directly measure airflow or respiratory effort, they can measure SpO2 continuously, and the pattern of repeated oxygen desaturations that accompanies obstructive apnea is detectable from SpO2 data alone in most clinically meaningful cases.
T90: time below 90 percent oxygen saturation
T90 is the total minutes during sleep that your blood oxygen saturation falls below 90 percent. The metric is straightforward, well-validated, and clinically interpretable.
The thresholds, drawn from the cardiovascular outcome literature:
- T90 under 1 percent of total sleep time: normal in healthy adults.
- T90 between 1 and 5 percent of sleep time: mild, often associated with positional or mild OSA.
- T90 above 5 percent: meaningful nocturnal hypoxemia, associated in cohort studies with elevated cardiovascular mortality independent of AHI.
- T90 above 10 percent: substantial nocturnal hypoxemia, strongly suggestive of moderate-to-severe OSA or other respiratory pathology.
Punjabi et al. (American Journal of Respiratory and Critical Care Medicine, 2009, Sleep Heart Health Study n=6,294) found that T90 was a stronger predictor of all-cause and cardiovascular mortality than AHI itself. The hazard ratio for cardiovascular mortality at T90 above 3.5 percent was 1.7 to 2.1 across age-sex strata, after adjustment for AHI and standard covariates. This is part of why the field is shifting toward hypoxic burden metrics as primary outcomes rather than treating AHI as the sole reference.
A practical caveat: pulse oximetry from consumer wearables is calibrated for higher saturation ranges and tends to be less accurate below 88 to 90 percent. If your wearable shows substantial time below 85 percent, the absolute number may be noisy, but a sustained pattern of desaturation is still meaningful and worth investigating with a clinical pulse oximeter or sleep study.
ODI: Oxygen Desaturation Index
The Oxygen Desaturation Index counts the number of times per hour of sleep that your SpO2 drops by a defined amount from baseline (typically 3 percent or 4 percent) and recovers. The 3 percent ODI is the most commonly reported metric and approximates the AHI well in patients without comorbid hypoventilation.
The thresholds:
- ODI under 5 per hour: normal.
- ODI 5 to 15 per hour: suggestive of mild OSA; warrants further evaluation.
- ODI 15 to 30 per hour: suggestive of moderate OSA.
- ODI over 30 per hour: suggestive of severe OSA.
Chung et al. (Anesthesiology, 2012) validated ODI from overnight pulse oximetry against polysomnography in 475 surgical patients and found that ODI ≥ 5 had a sensitivity of 87 percent and specificity of 66 percent for detecting OSA at AHI ≥ 15. Sharma et al. (Indian Journal of Medical Research, 2017) showed similar performance using portable home oximetry across a wider patient population.
The ODI is not a perfect surrogate for AHI. Central apneas without significant desaturation will be missed. Apneas during REM sleep, when oxygen reserves are depleted faster, are weighted more heavily. The metric tends to under-detect mild OSA and over-detect in patients with chronic hypoxemia from other causes. As a screen, however, it is well-validated and easily computed from data that a continuous-SpO2 wearable already collects.
Which devices can actually compute these
Not all wearables that show "SpO2" produce continuous overnight data suitable for T90 and ODI computation. The distinction matters.
Oura Ring (Generation 3 and 4): Continuous overnight SpO2 sampled at intervals through the night. The Oura app reports a "Breathing Regularity" metric and flags potential breathing disturbances. The underlying minute-level SpO2 data is exportable via the API and can be used to compute T90 and ODI directly.
WHOOP 4.0 and 5.0: Continuous overnight SpO2 introduced in the 4.0 hardware and refined in 5.0. WHOOP reports a "SpO2" tile but does not natively surface T90 or ODI.
Withings Sleep Analyzer: Mat-based device under the mattress. Detects breathing disturbances via micro-movements and provides an apnea risk indicator. Validated against polysomnography in published work (Edouard et al., Sleep, 2021, n=118), with sensitivity around 80 percent for moderate-to-severe OSA. This is the most ready-to-use consumer screen for adults who suspect apnea.
Apple Watch: SpO2 spot readings only, with sampling triggered manually or at irregular intervals. Not currently suitable for T90 or ODI computation. Apple introduced a "Sleep Apnea Notifications" feature in 2024 based on breathing-pattern movement rather than continuous SpO2; this is a separate methodology and useful as a flag but does not provide T90 or ODI numbers.
Garmin Venu, Forerunner, and Fenix lines: Continuous overnight SpO2 available on most current models. Garmin reports overnight saturation summary but does not natively compute T90 or ODI.
Dedicated screening pulse oximeters (Wellue O2Ring, CMS50F, etc.): Designed specifically for overnight monitoring, with built-in T90 and ODI calculations. Cost is typically $100 to $200. The accuracy is generally higher than wrist or finger wearables because the sensor is purpose-built for low-light skin contact and recording is at higher frequency.
How to interpret your data, and what to do next
If you have continuous overnight SpO2 from a device that exports the data, computing T90 and ODI is straightforward in a spreadsheet or any data analysis tool. T90 is the total minutes with SpO2 under 90 percent divided by total sleep minutes. ODI is the count of desaturations of 3 percent (or 4 percent, depending on definition) per hour, where each event must drop and then recover within a defined window (typically under 2 minutes).
A reasonable decision flow for an adult with continuous SpO2 data over 14 or more nights:
If your T90 is consistently under 1 percent and your ODI is consistently under 5 per hour, your risk of clinically significant sleep apnea is low. The metrics do not rule it out entirely (a polysomnogram remains the diagnostic standard) but the screen is reassuring.
If your T90 is consistently between 1 and 5 percent or your ODI is between 5 and 15 per hour, you are in the "mild concern" range. This is the case where adding the STOP-BANG questionnaire (Snoring, Tired, Observed apnea, blood Pressure, BMI over 35, Age over 50, Neck circumference over 40 cm, male sex) is useful. A STOP-BANG score of 3 or higher in this range argues for clinical follow-up. The STOP-BANG was validated by Chung et al. (Anesthesiology, 2008) and has become the most widely used clinical screen.
If your T90 is above 5 percent or your ODI is above 15 per hour, even on a single well-recorded night, you should pursue clinical evaluation. Home sleep apnea testing (HSAT) is the most common next step in many health systems and is substantially cheaper than in-lab polysomnography. WatchPAT, ApneaLink, and similar devices are commonly used and well-validated for moderate-to-severe OSA.
A note on positional apnea: many people have meaningfully different OSA severity sleeping on their back versus their side. If your wearable also tracks sleep position, look for patterns where desaturations cluster during supine sleep. Positional therapy alone can resolve mild positional OSA in a meaningful subset of patients.
What treatment changes if apnea is confirmed
CPAP remains the first-line treatment for moderate-to-severe OSA and the intervention with the most consistent randomized evidence for improving daytime symptoms, blood pressure, and quality of life. The recent SAVE trial (McEvoy et al., NEJM, 2016, n=2,717) generated debate about whether CPAP reduces cardiovascular events specifically; the trial was negative for the primary cardiovascular endpoint but had adherence issues and a relatively short follow-up. Subsequent analyses have suggested benefit in adherent users with more severe disease.
For mild OSA, mandibular advancement devices, positional therapy, weight loss (typically 5 to 10 percent body weight produces measurable AHI reduction), and treatment of nasal obstruction are first-line options. For severe OSA, CPAP adherence is the most important variable; recent CPAP devices have substantially improved comfort and adherence rates compared to a decade ago.
Key takeaways
- 80 to 90 percent of moderate-to-severe sleep apnea cases are undiagnosed in the general population. Untreated OSA elevates risk for hypertension, atrial fibrillation, stroke, and cognitive decline.
- T90 (time below 90 percent SpO2) and ODI (oxygen desaturations per hour) are validated screening metrics computable from continuous overnight pulse oximetry.
- T90 above 5 percent or ODI above 15 per hour warrants clinical evaluation.
- Oura, WHOOP, Withings Sleep Analyzer, Garmin, and dedicated overnight pulse oximeters provide suitable data. Apple Watch currently does not, except for its movement-based apnea flag.
- If your screen is positive, the next step is a home sleep apnea test or polysomnography. CPAP remains the first-line treatment for moderate-to-severe disease.
Sources
1. Peppard PE, et al. Increased prevalence of sleep-disordered breathing in adults. American Journal of Epidemiology. 2013;177(9):1006-1014. 2. Marin JM, et al. Long-term cardiovascular outcomes in men with obstructive sleep apnoea-hypopnoea with or without treatment with continuous positive airway pressure. Lancet. 2005;365:1046-1053. 3. American Academy of Sleep Medicine. The AASM Manual for the Scoring of Sleep and Associated Events. Version 2.4. 2017. 4. Punjabi NM, et al. Sleep-disordered breathing and mortality: a prospective cohort study. American Journal of Respiratory and Critical Care Medicine. 2009;181:507-513. (Sleep Heart Health Study.) 5. Chung F, et al. Oxygen desaturation index from nocturnal oximetry: a sensitive and specific tool to detect sleep-disordered breathing in surgical patients. Anesthesiology. 2012;114(5):993-1000. 6. Chung F, et al. STOP questionnaire: a tool to screen patients for obstructive sleep apnea. Anesthesiology. 2008;108(5):812-821. 7. Edouard P, et al. Validation of the Withings Sleep Analyzer, an under-the-mattress device for the detection of moderate-severe sleep apnea syndrome. Sleep. 2021;44(8):zsab070. 8. McEvoy RD, et al. CPAP for prevention of cardiovascular events in obstructive sleep apnea. NEJM. 2016;375:919-931. (SAVE trial.)
---
Want VITA to do this for you automatically? Join the waitlist