REM-Related Sleep Apnea: Mild on Paper, Not in Practice
In 2,240 patients, 18.3% had apnea concentrated in REM sleep. Their AHI and oxygen numbers looked better than everyone else's. Their comorbidity burden was worse, and they were disproportionately younger women.
During REM sleep your postural muscles are deliberately paralysed. That atonia is what stops you acting out dreams, and it includes the muscles holding your upper airway open.
So for some people apnea is overwhelmingly a REM phenomenon. REM is only about a fifth of the night, which creates a specific and consequential problem: the standard severity measure averages their events across the whole night and makes them look fine.
What the data shows
A 2026 clinical cohort study analysed 2,240 patients with obstructive sleep apnea — 1,541 men and 699 women, mean age 49.9, mean BMI 31.6 — after excluding people with poor sleep efficiency, insufficient REM, prior airway surgery, sedative use or existing treatment.
REM-related apnea was identified in 18.3% — roughly one patient in five.
On every standard measure, that group looked better:
- lower apnea-hypopnea index
- lower oxygen desaturation index
- less time with oxygen saturation under 90%
- higher mean and lowest overnight oxygen saturation
All at p<0.001. If you judged these patients by their numbers, you would treat them less aggressively than everyone else.
Then the comorbidities. REM-related patients were younger, more often female, and had more comorbidities. In a multivariable model adjusting for age, sex, Epworth score, apnea severity and comorbidity, REM-related apnea was independently associated with hypertension (p=0.001), coronary artery disease (p=0.003), hypothyroidism (p<0.001) and asthma (p<0.001).
Independently of severity. The association survived adjustment for the AHI that made them look mild.
The authors' recommendation is unusually direct: clinicians should consider alternative screening and lower treatment thresholds for high-risk groups, including young women and those with comorbidities.
Why the AHI understates it
Arithmetic. If you have 40 events per hour during REM and almost none otherwise, and REM is 20% of your night, your overall AHI lands around 8 — technically mild.
But the events you are having are concentrated in the sleep stage where obstruction lasts longer and desaturations tend to run deeper, because the muscles cannot compensate. The number is diluted; the physiology is not.
This is a specific instance of a general problem with the AHI, which treats a 10-second hypopnea and a 60-second obstruction as one event each and averages across stages and positions. How to read your sleep study properly.
Evidence tier 2. One large single-cohort study with consistent internal findings and a plausible mechanism. It is observational, and the comorbidity associations are associations.
Why this matters most for women
Women are over-represented in REM-related apnea, and this compounds an existing diagnostic problem. Women with apnea more often present with fatigue, insomnia and mood symptoms than with classic loud snoring and witnessed pauses, and are more likely to be worked up for something else first.
Add a diluted AHI and you get a familiar outcome: a woman reporting significant daytime symptoms, a sleep study reading "mild," and no treatment offered. Sleep apnea in women.
What to do with this
If your AHI is mild and your symptoms are not, ask about the REM breakdown. Sleep studies record REM AHI separately. It is in the data; it is often just not in the summary you were handed. Ask for REM AHI versus non-REM AHI, the same way you would ask for the supine breakdown.
Treat a mismatch as a question, not an answer. A mild overall AHI with REM clustering, significant sleepiness, or hypertension that is hard to control is a reasonable basis for discussing treatment despite the category.
Know that it is treatable the same way. CPAP does not care which stage your events occur in. There is also active interest in drug approaches aimed specifically at the ventilatory-drive problem in REM, including a randomised trial of acetazolamide in REM sleep apnoea — early, and a sign the phenotype is being taken seriously.
And do not let "mild" close the conversation. Mild on the index is not the same as mild in effect, which is the entire point of this literature. How apnea traits differ between people, and what separates snoring from apnea.