Seven hours is a nadir reading, not a night to cut or stretch
This is a reading, not a dial. Long nights are more likely illness than a lever to cut — Yin et al. 2017 has not shown that assigning 7 hours changes death.
A self-reported hours-per-night number, not a sleep-extension trial. The nadir is ≈7 h. Long sleep is the steeper side. Short sleep is the disputed one.
Dose
≈7 h/night is HR 1.0 (Yin). The engine’s long side is Table 2: 8 / 9 / 10 h = RR 1.04 / 1.15 / 1.32. The short side is Yin’s linear 1.06 per hour below 7 h (6 / 5 / 4 h = 1.06 / 1.12 / 1.19), not Table 2’s flatter 1.01 / 1.04 / 1.08. PCHIP interiors are constructed.
PCHIP interiors are constructed — the fill between the knots, not a table from any cited paper.
Dose-response shape: j-curve
How to get there
Protect the sleep opportunity window — the short side is the one you can actually act on.
Self-reported hours per night, tracked over time — the same measure the dose curve above is built on.
Do not sleep longer to score better. Yin's Table 2, Kwok, Ungvari, and Cappuccio all point the same way: the long side is the steeper risk, and most likely illness rather than a lever.
Evidence, by endpoint
| Endpoint | Grade | Citation |
|---|---|---|
| All-cause mortality, dose-response vs 7 h | B | Yin et al. 2017, Journal of the American Heart Association Engine paper. Prospective cohorts in generally healthy adults, search to December 2016. Nadir ≈7 h. Table 2 (nonlinear, 40 ACM risk estimates, reference 7 h): 8 h RR 1.04 (1.04–1.05), 9 h 1.15 (1.14–1.16), 10 h 1.32 (1.29–1.35). Linear short-sleep slope RR 1.06 (1.04–1.07) per hour below 7 h (P for nonlinearity on the short side = 0.12); linear long-sleep slope 1.13 (1.11–1.15) per extra hour. Extreme categorical pooling is 57 reports (shortest vs ref RR 1.13, longest 1.35) — not Table 2. |
| All-cause mortality, short and long categorical | C | Ungvari et al. 2025, GeroScience Newest categorical ACM pooling (search to October 2024): 79 cohort studies. Short sleep (<7 h vs 7–8 h) HR 1.14 (1.10–1.18); long sleep (≥9 h) HR 1.34 (1.26–1.42). Not a dose–response and not the engine curve. First author is editor-in-chief of the publishing journal. |
| All-cause mortality below 7 h (competing dose-response) | B | Kwok et al. 2018, Journal of the American Heart Association 74-study review; the reported dose–response is 30 studies (>1 million people). Linear model: 9 h RR 1.14 (1.05–1.25), 10 h 1.30 (1.19–1.42), 11 h 1.47 (1.33–1.64). No significant increase below 7 h — that is the short-side disagreement with the engine’s 1.06/h tax. Long side sits next to Yin Table 2. |
| All-cause mortality, nighttime hours vs 7 h | B | Shen et al. 2016, Scientific Reports 35 articles. Nighttime Table 1 vs 7 h (36 results, 146,830 deaths / 1,526,609 people): 4 h RR 1.07 (1.03–1.13), 6 h 1.01, 11 h 1.55 (1.47–1.63). 24-hour sleep is a different column (11 h RR 1.84). Short side is closer to Yin Table 2 than to the engine’s 1.19 at 4 h. |
| All-cause mortality, 24-hour sleep vs 7 h | B | Liu et al. 2016, Sleep Medicine Reviews 40 cohorts, 2,200,425 people, 271,507 deaths. J-shaped vs 7 h (24-hour sleep, n=29): 4 h RR 1.05 (1.02–1.07), 11 h 1.38 (1.33–1.44). Female short-sleep signal; male short sleep not significant in the subgroup. Not the engine paper. |
| All-cause mortality, short and long categorical (foundational) | C | Cappuccio et al. 2010, Sleep Foundational categorical pairing, not a dose–response: 16 studies / 27 cohorts, 1,382,999 people, 112,566 deaths. Short sleep RR 1.12 (1.06–1.18); long sleep RR 1.30 (1.22–1.38). Not the engine paper and not a sleep-extension trial. |
| All-cause mortality from a randomized sleep-duration intervention | X | No trial has randomized adults to a hours-per-night target and measured all-cause mortality. Cappuccio’s pairing is a marker, not a treatment effect. |
What the dose is
About seven hours a night. Yin et al. 2017 (Journal of the American Heart Association) put the all-cause nadir at ≈7 h in generally healthy adults. The calculator scores that as HR 1.0. Eight, nine, and ten hours print Yin’s Table 2: RR 1.04, 1.15, 1.32. Six, five, and four hours print the same paper’s linear short-sleep slope (1.06 per hour below 7 h), not Table 2’s nonlinear short side. The fill between those knots is constructed.
What the evidence shows
Yin’s long side is a dose-response (40 ACM risk estimates), grade B here. Ungvari et al. 2025 (GeroScience, 79 cohorts) is the newer categorical companion — short <7 h HR 1.14 (1.10–1.18), long ≥9 h HR 1.34 (1.26–1.42) versus 7–8 h — grade C, and not a curve. Cappuccio et al. 2010 (Sleep) is the old 1.12 / 1.30 pairing this page used to lead with. Shen et al. 2016 (Scientific Reports) and Liu et al. 2016 (Sleep Medicine Reviews) are other 7-hour-referenced dose-responses; Shen’s short side is closer to Yin Table 2 than to the engine’s 1.19 at 4 h.
What the number is not
It is not a demonstration that sleeping more, or less, cuts death. No trial in this set assigned a hours-per-night target and counted deaths, which is why the intervention endpoint is grade X. It is not Kwok et al. 2018’s short side: that 30-study dose-response found no significant increase below 7 h, so the engine’s 1.06/h tax is a choice, not a consensus. It is not Ungvari’s two bins, and it is not Yin’s 57-report extremes.
Why seven is not the whole sleep story
Duration is U-shaped. Regularity is scored on its own page, off Windred et al. 2024, and in that cohort regularity was the stronger all-cause predictor. Hitting 7 h on a wildly different clock each night is not this protocol, and this protocol does not print Windred’s 0.70.
What argues against this
Kwok et al. 2018 found no significant increase in all-cause mortality below 7 hours — the short-side tax this target prints (6 / 5 / 4 h = 1.06 / 1.12 / 1.19) is Yin’s linear 1.06 per hour, not a result every matching-construct meta shares. Yin’s own Table 2 is flatter on that side (1.01 / 1.04 / 1.08). Liu et al. 2016’s short-sleep signal held in women and was not significant in men. Long sleep is the more consistent side (Yin Table 2, Kwok, Ungvari, Cappuccio), and long sleep is also the side most likely to be illness rather than a lever. Nobody in this set assigned a 7-hour target and counted deaths.
Gotchas
- 57 reports is Yin’s extreme categorical pooling (shortest 1.13, longest 1.35). Table 2 is 40 ACM risk estimates. Do not hang the curve on 57.
- The engine welds two analyses from the same paper: nonlinear Table 2 on the long side, linear 1.06/h on the short side. Table 2 at 4 h is 1.08, not 1.19.
- Ungvari’s 1.14 / 1.34 is categorical vs 7–8 h, not Yin’s 7-point curve. The first author is editor-in-chief of GeroScience.
- Shen’s 1.55 at 11 h is nighttime sleep. The same paper’s 24-hour column at 11 h is 1.84. Those are not interchangeable with Yin’s 10 h of 1.32.
- The slider is self-reported hours. These pools are self-reported hours. A wearable’s time-in-bed is a different quantity.
Safety
- Long sleep can be disease, not a choice. Do not cut a 9-hour night to chase HR 1.0 without asking why the night is long. This set does not prescribe a sedative.
Educational content only — not medical advice, diagnosis, or treatment. Not a medical device; not FDA evaluated. Consult a qualified healthcare professional about your own health, and call emergency services for urgent symptoms. Full medical disclaimer →