Sleep Duration and Mortality

The relationship between how long people sleep and how healthy they are — including all-cause mortality — is U-shaped: both short and long sleep are associated with worse outcomes than a middle band. The "eight hours" rule is not where that evidence comes from; it descends from Robert Owen's 1817 labour slogan ("eight hours labour, eight hours recreation, eight hours rest").

Where the optimal band sits

Source	Design	Population	Optimal band
Sleep Health 2015 (National Sleep Foundation panel)	Review of 575 studies, 2004-2014	Adults 18-64 / over-65	7-9 h / 7-8 h
Sleep 2015 (AASM/SRS consensus)	Analysis of 311 studies	Adults	~7 h; uncertain harm above 9 h
Nature 2026, Wen et al. (doi 10.1038/s41586-026-10524-5)	Cohort, ~500,000 UK Biobank adults aged 37-84, self-reported sleep	Middle and late life	6.5-7.8 h women, 6.4-7.7 h men

The 2026 study is the refinement: instead of disease endpoints alone it uses 23 biological-ageing clocks across 17 organ systems (imaging, plasma proteomics, metabolomics) and finds the U-shape in nine body systems including brain, lungs, liver and skin. The sleep duration with the smallest "biological age gap" is the optimum above.

Disease and mortality associations (2026 Nature study)

• Sleep <6 h: depressive episodes, anxiety disorders, obesity, hypertension, type 2 diabetes, heart disease.
• Sleep <6 h or >8 h: COPD, asthma, gastrointestinal conditions (gastritis, GERD), and higher all-cause mortality.
• Mediation: ageing clocks appear to partly mediate the path from long sleep to late-life depression; short sleep's link is more direct.

What is not established

• Causation. The data are observational. Long sleep may be a marker of existing illness, pain, depression or poor sleep quality rather than a cause of harm; the study's Mendelian-randomisation analysis did not resolve direction. Whether moving an individual's sleep into the band slows ageing is untested.
• Individual applicability. The band is a population minimum. Experts quoted by The Economist (Michael Grandner, Arizona) stress that quality, regularity and continuity are separate dimensions, and that only a "very unusual number" deserves concern.
• Measurement. Exposure is self-reported duration, not device- or lab-measured sleep.
• The long-sleep arm of the U is the less certain one: the 2015 Sleep consensus explicitly flagged uncertainty about harm above 9 h.

Why it matters for the rest of this topic

The band is what Sleep Tracking devices are sold to deliver, and chasing it is the mechanism behind Orthosomnia — the 2025 AASM survey found 76% of US adults had lost sleep worrying about sleep.

Related

• Is Too Much Sleep as Bad as Too Little (Economist) — 2026-07-04, the primary source on the Nature study.
• Should You Use a Sleep Tracker (Economist) — 2026-06-06, cites the same study as the motivation for tracking.
• Sleep Tracking · Orthosomnia