Research Sets Total Sleep Targets, Not a Deep Sleep Number

There is no established number of deep sleep hours. The evidence attaches targets to total sleep. A 2016 CDC report used at least 7 hours for adults, and 65.2% of 444,306 US adults reported it. An overview of 11 reviews found 7 to 8 hours most favorably associated with health. Deep sleep has been linked to outcomes: in 6,785 wearable users, more deep sleep went with lower odds of atrial fibrillation. Those are associations, not thresholds.

Is there a target for hours of deep sleep?

Not in the research found here. The targets attach to total sleep. A 2016 CDC report noted that adults aged 18 to 60 are recommended at least 7 hours a night. In a 2014 survey of 444,306 US adults, 65.2% reported that much.

A sleep hygiene review recommends 7 to 9 hours. An overview of 11 systematic reviews covering 4,437,101 adults found 7 to 8 hours most favorably associated with health, with no clear change by age.

None of these sets a minutes-per-night figure for deep sleep. The deep sleep number on a tracker has no published target to compare against. On an Apple Watch, deep sleep sits beside a much larger block labeled Core, which Core vs Deep sleep explains.

What does deep sleep do?

It is the stage most tied to memory consolidation. A review describes neuron firing patterns replaying in hippocampal groups during slow-wave sleep. This goes together with ripples, spindles and slow oscillations.

The authors say replay likely favors turning hippocampus-dependent episodic memory into longer-lasting cortical memory. REM sleep following slow-wave sleep may balance the process. This is a mechanism review with no minutes target.

Does more deep sleep go with better health?

Some studies link them, and none sets a threshold. A wearable cohort of 6,785 people tracked for a median of 4.5 years found deep sleep and REM sleep inversely associated with the odds of incident atrial fibrillation. Median age was 50.2, and 71% of participants were female.

In 492 Framingham Offspring participants with a mean age of 58.8, less slow-wave sleep went with lower cortical brain volume and higher white matter hyperintensity volume. It did not go with hippocampal volume or covert brain infarcts. The authors note the link could run the other way, because subtle brain injury might reduce slow-wave sleep.

In 129 people with Parkinson's disease, higher slow-wave power went with slower motor progression over a mean of 4.6 years. The authors call the finding preliminary.

All three are associations. See whether poor sleep speeds up aging for related evidence.

Does deep sleep fall with age?

The two meta-analyses disagree. A 2019 analysis of 169 polysomnography studies in 5,273 healthy adults found no significant change in the percentage of N3 sleep per decade of age (-0.1%). Total sleep time fell 10.1 minutes per decade.

A 2004 analysis of 65 studies, published between 1960 and 2003 and covering ages 5 to 102, found the percentage of slow-wave sleep decreased with age in adults. It also found effect sizes depended on how well participants were screened for disease, apnea and drug use, and that poor screening could mask or shrink age associations.

The difference may reflect who was studied. It is not settled. See whether sleep need changes with age and how wearables track sleep stages.

What reduces deep sleep?

Caffeine did in one meta-analysis. Across 24 studies, caffeine reduced total sleep time by 45 minutes and sleep efficiency by 7%. Deep sleep fell by 11.4 minutes, or 1.4 percentage points, and light sleep rose by 6.1 minutes.

Short total sleep has its own costs. In 48 healthy adults held to 4, 6 or 8 hours in bed for 14 days, 4 and 6 hours brought cumulative, dose-dependent deficits on every cognitive task. In 11 young men held to 4 hours for 6 nights, glucose tolerance fell and evening cortisol rose. Both are total sleep results. See caffeine timing and sleep.

What does this not settle?

It does not settle how many minutes of deep sleep are healthy. The wearable cohort compared odds across a natural range and did not test a specific amount.

Device estimates also vary against a sleep lab, as covered in how wearables track sleep stages. Reverse causation is possible in every association above.

Common questions