You know the mental math. It is 11:40 at night, the alarm is set for 5:40, and you run the numbers: six hours. That is fine. I function on six.
And you do function. You get up, get through the day, and the number in your head says the debt is paid. But that arithmetic treats every hour of sleep as interchangeable. Sleep does not work that way. A study published September 17 in PLOS Medicine offers a reason to look at the shape of your night, as well as its length.[1]
What the researchers actually did
A team led by scientists at Capital Medical University studied 95,559 adults in the UK Biobank who had worn wrist accelerometers for a week. A deep-learning model estimated their time in REM, deep, and light sleep. Researchers then followed their medical records for a median of 8.9 years, looking across more than 1,000 health outcomes.[1]
The headline finding: a greater estimated amount of REM sleep was associated with lower risk across 83 conditions. For an increase of about 48 minutes of estimated REM, the reported associations included heart failure (hazard ratio 0.74), dementias (0.54), and parkinsonism (an umbrella term used for brain conditions that cause movement problems like slowed action, stiffness, and shaking; 0.20). These are group-level associations, not predictions for an individual. Deep sleep was associated with lower risk across seven conditions. Sleep duration of six to eight hours was the lowest-risk range for many, though not all, of the outcomes studied.[1]
The part that should change your evening
REM sleep is not distributed evenly. The National Heart, Lung, and Blood Institute says you usually have more REM sleep later in the night.[2] The earliest REM periods tend to be shorter; later ones tend to lengthen. That makes the final stretch of sleep especially easy to underestimate when you are tempted to set an earlier alarm.
It does not mean every early alarm removes a known number of REM minutes. Nights vary, and the wrist-based estimates in the new study cannot tell us exactly what happened during each individual’s night. But it does make a practical case for giving yourself enough time in bed instead of assuming you can repeatedly trim the end of your night’s sleep without consequence.
The evening glass of wine is another consideration. A 2025 systematic review and meta-analysis of 27 studies found that alcohol before bed delayed REM onset and reduced REM duration, with greater disruption at higher doses.[3] Don’t panic—this doesn’t mean you need to turn every relaxing night into a laboratory experiment. However, it may be useful to notice whether drinking near bedtime changes how rested you feel the next day.
A week of observation can be more informative than one unusually good or bad night. Keep the notes simple: bedtime, wake time, any middle-of-the-night waking, and how you feel in the morning. You may discover a scheduling problem, a bedroom issue, or a pattern worth discussing with a clinician. The aim is to understand your own routine, not to assign yourself a grade.
What this does not prove
The UK Biobank study was observational. The authors say its design precludes causal inference: more REM could reflect a healthier body, contribute to better outcomes, or both. The sleep stages were estimated from wrist movement rather than measured with laboratory brain and eye sensors, so they may not perfectly match biological sleep stages.[1] Nothing here establishes that increasing REM prevents dementia or any other disease.
There is another reason to resist turning your tracker into a nightly report card. A single number cannot capture the full context of your sleep. Look for patterns over time and pay attention to how you feel, especially if sleep problems persist or affect your days.
Four things to try this week
1. Make room for the morning. If an earlier alarm is optional, consider keeping your usual wake time. If it is necessary, move bedtime earlier when you can.
2. Work backward from a realistic wake time. Choose the time your life allows, then protect a bedtime that gives you a sufficient sleep opportunity. Consistency is more useful than chasing a perfect night.
3. Notice the timing of your last drink. Alcohol can alter sleep stages.[3] If you drink, compare nights when you finish earlier with nights when you drink close to bedtime, without treating one evening as a diagnosis.
4. Make the room easier to sleep in. A cool, dark room and a phone charging outside your bedroom can make your routine easier to keep.
If you are in your 40s or 50s and finding that deep sleep is harder to achieve, none of this is a referendum on your willpower. SWAN reports that sleep problems commonly increase during perimenopause and often stabilize or improve afterward.[4] If interrupted sleep is frequent, talk with your clinician about what is driving it.
Tonight, defend the morning.