Fibre tracks with lower death. No trial has tested the gram target itself.
You can change this. Whether hitting a gram target itself cuts death is untested — no trial has randomized adults to a fibre-gram target and measured all-cause mortality.
Total dietary fibre from food, not a subtype and not a supplement. The 0.85 floor is highest versus lowest in people without chronic disease. Newer pools look larger because they mix in the sick.
Dose
25–29 g/day from food is Reynolds’s greatest-benefit band (6 of 7 critical outcomes). The calculator scores the IOM/USDA Adequate Intake instead: 34/30 g (men ≤50/>50) or 26/21 g (women ≤50/>50). Both are total grams, not fruit fibre or a powder.
The 15 g kink (HR = 1.0) and the extra −0.005/g past target are constructed — Yao’s curve is steeper below 15 g but does not set HR = 1.0 there.
Dose-response shape: monotone
How to get there
Raise grams from mixed plant foods over days, not in one sitting — GI tolerance is the binding limit, not the mortality curve.
Total grams per day, from food.
Reynolds's greatest-benefit band is 25–29 g/day; the calculator's scored target is the higher IOM/USDA Adequate Intake. Either way, a powder is not this evidence.
Evidence, by endpoint
| Endpoint | Grade | Citation |
|---|---|---|
| All-cause mortality, highest vs lowest intake | B | Reynolds et al. 2019, The Lancet WHO-commissioned series. All-cause mortality is a 10-study observational pooling of people without chronic disease — not the 185-publication count: RR 0.85 (0.79–0.91), GRADE moderate. That 0.85 is the engine floor at/above the age/sex target. |
| All-cause mortality per 8 g/day | B | Reynolds et al. 2019, The Lancet Figure 1A, assuming linearity: RR 0.93 (0.90–0.95) per 8 g more fibre/day (68,183 deaths / 11.3 million person-years). Dose-response described as linear with no plateau in the available data. Categorical benefit was greatest at 25–29 g/day. |
| All-cause mortality, newer matching-construct pool | B | Yao et al. 2023, Frontiers in Nutrition Search to August 2023. Highest vs lowest: 16 studies, RR 0.81 (0.77–0.86), I² 71.9%. Per 10 g/day: 14 studies, RR 0.90 (0.86–0.93), I² 86.1% — the 0.90 in the abstract is this increment, not highest-vs-lowest. Non-linear (P=0.0096): steeper below 15 g/day, more gradual after. Engine not moved onto 0.81 / 0.90. |
| All-cause mortality, general vs mixed population | C | Ramezani et al. 2024, Clinical Nutrition Largest pool: 64-study review, 3,512,828 people; ACM meta is 33 reports, highest-vs-lowest only. Overall HR 0.77 (0.73–0.82). General-population subgroup (19 reports) HR 0.82 (0.77–0.86); non-general (cancer survivors, T2D, CKD, …) HR 0.67 (0.61–0.72). Continuous estimates were dropped. |
| All-cause mortality from fruit fibre specifically | X | Mirrafiei et al. 2023, Food & Function 28 prospective studies, 1,613,885 people, general population. Higher total fibre and most subtypes (cereal, vegetable, legume, soluble, insoluble) associated with lower ACM (HRs 0.77 insoluble to 0.93 legume); fruit fibre was not. The abstract does not report a single total-fibre highest-vs-lowest HR. The calculator collects total grams — moving the slider is not “eat fruit fibre.” |
| All-cause mortality from a randomized fibre intervention | X | No trial has randomized adults to a fibre-gram target and measured all-cause mortality. Reynolds’s 58 trials were not an ACM endpoint for this protocol. |
What the dose is
Twenty-five to twenty-nine grams of total dietary fibre a day from food is where Reynolds et al. 2019 (The Lancet) saw the greatest categorical benefit across six of seven critical outcomes. The calculator does not score that band. It scores the IOM/USDA Adequate Intake by sex and age: 34 g or 30 g for men (≤50 / >50) and 26 g or 21 g for women. Both numbers are total grams. Subtype is not collected.
What the evidence shows
Reynolds’s all-cause mortality estimate is RR 0.85 (0.79–0.91) comparing highest versus lowest intake in people without chronic disease — 10 observational studies, GRADE moderate — and, assuming linearity, RR 0.93 (0.90–0.95) per extra 8 g/day. Yao et al. 2023 (Frontiers in Nutrition) is the newer matching-construct meta: highest-vs-lowest RR 0.81 (0.77–0.86) across 16 studies, and RR 0.90 (0.86–0.93) per 10 g/day across 14, with a non-linear shape that steepens below 15 g/day. Both are grade B here. The engine stays on Reynolds’s 0.85 floor; it has not been moved onto Yao’s 0.81 or 0.90.
What the number is not
It is not a licence to quote Ramezani et al. 2024’s overall HR 0.77 as the general-population effect. That paper’s own split is 0.82 in general-population reports and 0.67 in non-general (cancer survivors, T2D, CKD). It is not a subtype prescription: Mirrafiei et al. 2023 found fruit fibre unassociated with all-cause mortality. And it is not a randomized result — no trial has assigned a fibre-gram target and counted deaths — which is why the mortality claim tops out at grade B.
Where the grams come from
Reynolds excluded supplement trials. The evidence this dose sits on is food. Mirrafiei’s subtype range (insoluble HR 0.77 to legume HR 0.93, fruit fibre null) is why the honest instruction is total grams from mixed plant foods, not a fruit-fibre or powder trick. Yang et al. 2015 (American Journal of Epidemiology) already had the same per-10 g point estimate as Yao (RR 0.90) on a smaller tertile pool; it is the older companion, not a second engine.
What argues against this
Ramezani et al. 2024’s headline HR 0.77 mixes diseased cohorts. The general-population subgroup is 0.82 (0.77–0.86), which sits next to Reynolds 0.85 and Yao 0.81 — the 0.77 is the sick-cohort inflation, not a larger benefit for the person this calculator scores. Reynolds’s own series is the other welding trap: 185 prospective publications entered the review; ACM highest-vs-lowest is 10 studies of people without chronic disease, and trials involving supplements were excluded. Mirrafiei et al. 2023 found fruit fibre unassociated with ACM, so “eat more fibre” is not “eat more fruit fibre.” The 25–29 g band is Reynolds’s greatest categorical benefit, not the calculator’s scored target (IOM/USDA AI 34/30/26/21 g), and the engine’s 15 g kink plus the extra −0.005/g past target are constructed, not Reynolds. Yang et al. 2015 is the older tertile pooling (RR 0.84 top vs bottom; RR 0.90 per 10 g) — same per-10 g point as Yao, smaller and superseded as the engine paper.
Gotchas
- 185 publications is the series, not the ACM estimate. Reynolds Table 1 all-cause mortality is 10 observational studies.
- Yao’s abstract 0.90 is per 10 g/day, not highest-vs-lowest. Highest-vs-lowest in that paper is 0.81.
- The calculator’s age/sex target is the IOM/USDA Adequate Intake, not Reynolds’s 25–29 g band. Hitting 29 g is the evidence peak; hitting 34 g is the scored peak for a man 50 or under.
- The 15 g kink (HR = 1.0) and the extra −0.005/g past target are constructed. Yao’s curve is steeper below 15 g but does not set HR = 1.0 there.
- Reynolds excluded people with chronic disease and trials of supplements. A fibre powder is not this evidence.
Safety
- Raise grams from food over days, not in one sitting — GI tolerance is the usual limit, not the mortality curve.
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 →