All analyses
5 min read

Last updated:

Deep sleep duration: what the watch measures, and what the number is worth

A person arrives with a number: 42 minutes of deep sleep last night, against 1 hour 20 the week before. Before drawing anything from it, you need to know where that number comes from. Eleven consumer devices compared with a sleep examination give agreement ranging from near chance to acceptable, depending on the model.

7 studies citedsee the sources

Benjamin Sant

Co-founder of the School of Longevity

Consultation around sleep data displayed on a tablet
Consultation around sleep data displayed on a tablet

How much deep sleep is normal, and why the question is badly framed

The request arrives in minutes: how much should I have. The best available answer comes from a meta-analysis published in 2004, which pooled 65 studies and 3,577 healthy people, aged 5 to 102, recorded in a laboratory. It describes averages by age group, not an individual target.

In adults, the percentage of slow wave sleep decreases significantly with age, along with total sleep duration, sleep efficiency and the percentage of REM sleep. Time spent awake after falling asleep, by contrast, increases. After 60, a single parameter keeps deteriorating significantly: sleep efficiency.

The authors add a caveat that matters as much as the figures. The size of the differences observed depends heavily on how carefully participants were screened: depending on whether psychiatric disorders, organic disease, alcohol, medication and sleep breathing disorders were excluded or not, the effects linked to age weaken, and sometimes disappear. Part of what is attributed to aging therefore comes from what accompanies it.

Eleven devices, one reference test, agreement that varies threefold

A validation published in 2023 compared eleven consumer devices with polysomnography, in 75 participants recruited in two centers. It totals 3,890 hours of device recording against 543 hours of reference examination, and the analysis covers 349,114 thirty second windows compared one by one.

Agreement on stage classification, measured by an overall score, ranges from 0.26 for the worst device to 0.69 for the best. In other words, two watches on two wrists do not tell the same night, and the gap between models far exceeds the gap between two nights of the same person. Performance also varies by stage: some devices are better on wake and REM sleep, others on deep sleep.

The biases are not random either. Devices worn on the wrist show a proportional bias on sleep efficiency, sensors placed next to the bed a bias on sleep onset latency. A number that is wrong systematically is more misleading than a noisy number, because it repeats from one night to the next and gives the impression of a stable measurement.

The position of the professional societies is explicit

The American Academy of Sleep Medicine published a position statement on these technologies in 2018. Its conclusion fits in one sentence: lacking validation and regulatory clearance, these devices cannot be used for the diagnosis or the treatment of a sleep disorder.

The same statement acknowledges a legitimate use. Presented within an appropriate clinical evaluation, these data can enrich the exchange between a person and their clinician. The boundary is therefore clear: the reading is a starting point for conversation, never a conclusion. A repeated reading that suggests sleep apnea justifies referral, it allows neither a conclusion nor an assessment of severity.

The number linked to a risk is not a single night, it is a slope

There is a solid finding linking deep sleep to brain health, and it does not concern an isolated value. In the Framingham cohort, 346 people aged at least 60, without dementia, underwent two complete sleep examinations separated by 5.2 years on average. Slow wave sleep declined by an average of 0.6 percentage points per year.

Over the seventeen years of follow-up after the second examination, 52 cases of dementia were diagnosed. After adjustment for age, sex, APOE ε4 carrier status, smoking and several treatments, each percentage point of deep sleep lost per year was associated with a 27% higher risk of dementia, hazard ratio 1.27, interval 1.06 to 1.54.

Three precautions are needed before transposing. It is a cohort, so an association and not an effect. The measure comes from two polysomnographies five years apart, not from a wristband. And nothing indicates that acting on the percentage of deep sleep changes this risk: the loss of deep sleep could be an early sign of the process rather than its cause. The link with the glymphatic system, often invoked to explain this result, remains a hypothesis in humans.

The levers that increase slow wave sleep, and by how much

Evening exercise was blamed for years for ruining sleep. A 2019 systematic review of 23 studies finds the opposite: compared with control, evening exercise increases slow wave sleep by 1.3 percentage points and lengthens REM sleep onset latency by 7.7 minutes, while reducing the lightest sleep.

The nuance concerns intensity and timing. Vigorous effort finished less than an hour before bedtime can lengthen sleep onset and degrade sleep efficiency. A higher body temperature at bedtime is associated with poorer efficiency and more awakenings. So the problem is not moving in the evening, it is finishing too late and too hard.

A second lever is documented and costs nothing. A 2019 meta-analysis of seventeen studies shows that a bath or shower at 40 to 42.5 degrees, one to two hours before bedtime and for at least ten minutes, improves perceived sleep quality and sleep efficiency, and shortens sleep onset. The proposed mechanism is the drop in core temperature that follows warming of the extremities. Add stable wake times and morning light, and you have the essentials of what is demonstrated.

When tracking becomes the problem

Sleep clinicians described a new reason for consultation as early as 2017, which they named orthosomnia. People come to be treated for a disorder they diagnosed themselves, based on nights their app labels as light or restless.

The most striking point in that description is this: for these people, device data often seem more consistent with their experience than validated measures. Arguing with the number head on therefore does not work. The pursuit of perfect sleep ends up feeding the insomnia it claims to correct, and the tool meant to reassure becomes a maintaining factor.

The reasonable use of a deep sleep number

The first thing to do is to move attention from the value to the trend, and from the trend to lived experience. An isolated night says nothing, the gap between two models exceeds the gap between two nights, and no individual target exists. What is useful to measure is the actual bedtime and wake time over a week, the regularity of waking, and how the day feels.

The second is to look for what can be corrected. Night pain, urges to urinate, evening alcohol, irregular schedules, breathing disorders: these are what explain most degraded nights, and they belong to a medical history, not to a sensor.

The third is to know the limit. Persistently fragmented sleep, sleepiness that interferes with driving, breathing pauses reported by those around all call for a medical evaluation. A practitioner who is not a physician does not make a diagnosis and does not interpret a reading as an examination, and saying so clearly is how they avoid feeding orthosomnia.

Sources

  1. 1.Ohayon MM, et al. Meta-analysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan. Sleep. 2004.
  2. 2.Lee T, et al. Accuracy of 11 wearable, nearable, and airable consumer sleep trackers: prospective multicenter validation study. JMIR Mhealth Uhealth. 2023.
  3. 3.Khosla S, et al. Consumer sleep technology: an American Academy of Sleep Medicine position statement. J Clin Sleep Med. 2018.
  4. 4.Himali JJ, et al. Association between slow-wave sleep loss and incident dementia. JAMA Neurol. 2023.
  5. 5.Stutz J, et al. Effects of evening exercise on sleep in healthy participants: a systematic review and meta-analysis. Sports Med. 2019.
  6. 6.Haghayegh S, et al. Before-bedtime passive body heating by warm shower or bath to improve sleep: a systematic review and meta-analysis. Sleep Med Rev. 2019.
  7. 7.Baron KG, et al. Orthosomnia: are some patients taking the quantified self too far? J Clin Sleep Med. 2017.

Reading studies is one thing, turning them into a plan is another

The Longevity Practitioner certification teaches the method that turns this kind of data into an accompaniment decision: assess, prioritise, plan over 90 days, re-measure, and know when to hand over.

See the certification

11 modules, €2,200. Enrol

Benjamin Sant

Co-founder of the School of Longevity

Cofounder of the School of Longevity with Tristan Maruyama. A former officer trained at Saint-Cyr from 2019 to 2022 and a coach since 2018, he has supported more than 4,000 people. He leads the School's editorial content with one constant principle: pass on only what can be situated, verified and applied seriously.

This article is published for information and education. It does not replace medical advice. For any personal health question, consult a healthcare professional.