Does more REM sleep lower your risk of disease?
In this study, more REM sleep was linked to lower risk of 83 different diseases. Among 95,559 UK Biobank adults who wore a wrist tracker and were then followed for a median of 8.9 years, more deep sleep was linked to lower risk of 7 diseases on top of that.
This is the largest attempt so far to connect measured sleep stages to actual disease diagnoses. Most sleep research asks people how they slept, and those answers match reality poorly. This team instead used wrist accelerometer data and an algorithm called SleepNet. That let the researchers estimate REM sleep, deep sleep, light sleep, total sleep time, sleep irregularity and time spent awake after falling asleep, in ordinary people sleeping in their own beds.
They then checked each of those sleep measures against 1,049 separate disease outcomes. That is a phenome-wide association study, which is a fancy way of saying they tested everything at once instead of picking one disease and hoping.
What the data show
After correcting for the huge number of comparisons, 156 sleep-disease links held up. More REM sleep was tied to lower risk of 83 conditions, and more deep sleep to lower risk of 7. Atrial fibrillation, ischemic heart disease, multiple sclerosis and Alzheimer’s disease also showed up on the list.
Total sleep time did not behave in a straight line. For 86 conditions the relationship curved, with less than a 1 in 20 chance that the curve is coincidence, and for 69 of those the lowest-risk point sat inside a 6 to 8 hour window. When the researchers split people into five sleep-length groups, 41 associations showed higher disease risk compared with the 6 to 8 hour group, and 37 of those 41 came from people sleeping under 5 hours.
Dr. Kumar’s Take
A phenome-wide scan tests everything at once, and rare outcomes in observational data produce big, unstable estimates. What earns my confidence is a pattern that repeats across unrelated organ systems, not any single striking link.
The bigger problem is direction. Dementia, Parkinson’s and heart failure all damage sleep years before anyone gets diagnosed. REM sleep behavior disorder is one of the earliest known warning signs of Parkinson’s. So when a study finds that low REM predicts Parkinson’s, the honest reading is that some of that is the disease showing itself early, not sleep protecting the brain. Observational data cannot separate those two readings, so an unknown share of this atlas is reverse causation rather than sleep driving disease.
What I do take from it is the shape of the curve. The 6 to 8 hour window keeps reappearing across unrelated organ systems, in data that did not depend on anyone’s memory of how they slept. That is the part with the least room for wishful thinking.
How strong is the evidence?
The sample is large and the outcomes came from hospital records rather than questionnaires, which is a real strength. The adjustments were thorough, covering age, sex, deprivation, ethnicity, education, employment, body mass index, smoking, alcohol, physical activity, screen time, air pollution and nighttime noise.
Three things limit it. Sleep was recorded over a single seven-day stretch, which may not capture how someone usually sleeps. Diagnoses came only from inpatient records, so milder illness handled by a family doctor was missed entirely. And sleep stages were estimated by the SleepNet algorithm from wrist movement, not recorded from brain waves, so these should be read as movement-based estimates rather than true sleep stages.
What about broken and irregular sleep?
Two other measures mattered independently of how long people slept. More time awake after falling asleep was linked to elevated risk of 6 diseases.
Sleep irregularity, meaning a bedtime and wake time that swing around night to night, was tied to elevated risk of 3 diseases. These held even after the researchers adjusted for total sleep duration, so a consistent schedule appears to matter on its own, not just as a proxy for getting enough hours.
Practical Takeaways
- Aim for a sleep opportunity that lands you between 6 and 8 hours of actual sleep, since that window held the lowest risk for 69 of the 86 conditions where sleep length mattered at all.
- Keep your bedtime and wake time roughly consistent across the week, because irregular timing was linked to higher disease risk even in people who slept enough hours.
- Treat frequent waking during the night as worth investigating with your doctor rather than accepting it, since time awake after falling asleep tracked with its own set of risks.
- If your consumer sleep tracker shows low REM or deep sleep, use it to watch trends over weeks rather than reacting to one night, since even the research-grade algorithm here was inferring stages from wrist movement rather than brain waves.
Related Studies and Research
- Long daytime naps linked to higher death risk in older adults
- Insomnia plus under 6 hours of sleep raises death risk
- Bright evening light linked to higher risk of eye disease
- Even light exercise lowers dementia risk in older adults
FAQs
Can I increase my REM sleep on purpose?
Nothing in this study tested any way of raising REM sleep, so it cannot answer that. REM tends to cluster in the second half of the night, which means cutting sleep short at the front end takes a bigger bite out of REM than out of deep sleep. Alcohol and several common medications also suppress it. A tracker that reports REM is estimating from wrist movement and heart rate, not measuring brain waves, so treat its nightly number as a rough signal.
Does a smartwatch sleep score mean anything clinically?
Not as a diagnosis. The algorithm used in this research sorts REM from non-REM and wake using wrist movement rather than brain waves, so it is an approximation even at research grade. Consumer devices are generally not validated against sleep-lab recordings at all and rarely publish their accuracy. What these devices do reasonably well is total sleep time and night-to-night consistency, which are exactly the two measures that showed the clearest patterns here.
Is light sleep bad for you?
Light sleep made up roughly half of total sleep time in this group, so a person with more light sleep often has proportionally less REM and deep sleep in the same night. The links this study reported came from REM sleep, deep sleep, sleep irregularity and time spent awake after falling asleep, so light sleep is better understood as what is left over when the other stages run short than as a stage that does harm on its own.
Bottom Line
Across 95,559 people, 1,049 diseases and nearly nine years of follow-up, measured sleep architecture tracked with disease risk in a way that self-reported sleep never has. More REM sleep lined up with lower risk of 83 conditions, more deep sleep with 7, and a 6 to 8 hour night with the lowest risk for 69. This is observational data, and early disease can disturb sleep years before anyone is diagnosed, so none of it proves that fixing your sleep prevents these diseases. The consistent 6 to 8 hour window and the independent cost of an irregular schedule are the two findings sturdy enough to act on.
