Guide

Nutrition and longevity: the levers that hold, and the ones that do not

What nutrition actually changes, from the best documented lever to the promises trials do not support.

9 analyses · updated 2026-10-30

In nutrition, only one lever has a volume of data comparable to a drug: dietary fiber, documented in seventeen million people. Intermittent fasting does no better than an equivalent calorie restriction. Supplements have a demonstrated effect in only two situations, correcting a deficiency and supporting resistance training.

This guide brings together the analyses the School devotes to nutrition. They all start from the same question: what is the measured outcome, in what population, and by what type of study.

Fiber is the best documented nutritional lever

Few nutrients have such a volume of data: seventeen million people in a review of reviews, three and a half million in a meta-analysis of cohorts. The association with all cause mortality is consistent, and it survives adjustment. This is one of the rare places in nutrition where a practitioner can speak without excessive hedging.

The detail contradicts the shortcuts. Sources are not equivalent, and in people already ill the effect has been measured in trials, not only in cohorts. What translates into a coaching program is not a dose of powder, it is a volume of food.

the orders of magnitude on mortality

Intermittent fasting does no better than equivalent restriction

Randomized trials converge: at equal calorie deficit, intermittent fasting brings no superior benefit on weight. Fasting glucose changes little and inconsistently. What sets it apart lies elsewhere, in tolerance and in the simplicity of a clock rule.

The 14 to 16 hour eating window remains the best tolerated modality. Protocols of 24 hours and more are a matter of medical supervision. And reading a meta-analysis means looking at who was excluded from the trials: treated diabetes, pregnancy, a history of eating disorder.

what randomized trials actually show

Multi-day fasting is still at a preliminary stage

Prolonged fasting gets a lot of media attention and little study in humans. The fasting mimicking diet is the best described protocol, and it was designed to avoid complete deprivation. The available data concern markers, not clinical outcomes.

The risks, on the other hand, are documented, and so are the precise contraindications. Refeeding requires as much attention as the fast itself. This ground belongs to medical follow-up, from the prior workup to the exit from the fast.

the contraindications and the frame of a supervised trial

A supplement is judged on the outcome measured, not on the mechanism

Creatine is the best documented supplement on the market, and one of the most poorly used. Its effect exists, it is modest, and it depends on two conditions: the presence of resistance training and consistency of intake. Analyses looking at intake limited to training days do not find the benefit seen with daily intake.

Only one form is supported by trials, monohydrate. On safety, a double blind trial measured 41 blood and urine parameters, including kidney and liver function, and found no change. Known kidney disease, nephrotoxic treatment, pregnancy or breastfeeding require prior medical advice.

the doses, the loading phase and what safety covers

Between cohorts and trials, the gap is not a methodological detail

On omega 3, observational studies describe a reduction in dementia risk of about twenty percent. Randomized trials find a modest improvement, limited to certain functions. Understanding this gap is worth more than picking the number that suits.

A cohort compares people who eat fish with people who do not, with everything else that this covers. A trial administers a capsule. The positive trials have one thing in common, and that is what tells you whom supplementation can serve.

what the positive trials have in common

The microbiome is fed, it is not fixed by a test

The gut microbiome changes with age, diet, medication and health status, with no single identifiable cause. In prospective data, plant food dominates, and fiber supplies the substrate for colonic fermentation. The loop comes back to the first lever in this guide.

Microbial diversity reads less from a test sold online than from a set of habits held over time. Probiotics have narrow indications, established strain by strain, and they do not transfer from one situation to another.

what changes with age and the real place of probiotics

Magnesium, or how form replaced the question

The debate is about bisglycinate, citrate or oxide, and it misses the essential. The only well documented difference in form contrasts oxide, which did not separate from placebo on the markers followed in the reference trial, with the organic forms. Between bisglycinate and citrate, no trial settles the question on a clinical outcome.

The useful question is whether supplementation itself matters. A meta-analysis of nineteen publications and more than a million participants associates dietary magnesium intake with lower mortality, while intake from supplements is associated with nothing. On cramps, a Cochrane review concludes that there is no effect.

what form changes, and what supplementation changes

Genetics personalizes almost nothing today

Genetic tests sold in nutrition overinterpret weak associations. A variant associated with a difference of a few percent in a population says nothing useful about a person's plate. Personalized nutrition starts from real, measured habits, before any consideration of genotype.

Three clinical situations are exceptions, and in those cases the result changes management. For the rest, a genetic report is assessed on verifiable arguments: which population, what effect, what outcome.

the seven checks before accepting a genetic report

Level of evidence for the main nutritional levers, by type of study and outcome measured.
LeverWhat was measuredLevel of evidence
Dietary fiberAll cause mortalityMassive cohorts and converging trials
Protein 1.0 to 1.2 g/kg/dayMuscle mass and strengthRandomized trials
Plant foodMicrobiome compositionProspective data
Creatine monohydrateStrength, with resistance trainingRandomized trials, modest effect
Intermittent fastingWeight, at equal calorie deficitNo demonstrated superiority
Omega 3Cognitive declineGap between cohorts and trials
ProbioticsClinical outcomesNarrow indications, strain by strain
Nutritional genetic testsEating decisionWeak associations, not transferable

Level of evidence for the main nutritional levers, by type of study and outcome measured.

The analyses in this guide

Questions

Which dietary change has the largest effect on longevity?

Increasing fiber. It is the only nutritional lever documented in millions of people with a consistent association with all cause mortality, and confirmed by trials in people already ill. Sources matter: legumes, whole grains, vegetables, nuts.

Does intermittent fasting make you lose more weight?

No, not at equal calorie deficit. Randomized trials find no benefit superior to equivalent restriction. Its interest is practical: a clock rule is simpler to keep than counting, and the 14 to 16 hour window is the best tolerated.

Which supplements are worth it after 50?

Two cases are documented. Correcting an established deficiency, vitamin D first, and creatine monohydrate in a person who does resistance training. Outside these two situations, trials show no benefit on a clinical outcome.

Should you get a microbiome test?

In routine coaching, no. The result does not change management: the documented levers are fiber and the plant share of the diet, which you would recommend anyway. A costly test moves attention to a number instead of to habits.

Can a non-physician practitioner advise a multi-day fast?

No. Deprivation of more than 24 hours carries documented risks and precise contraindications, and it interacts with ongoing treatment. The frame is medical, from the prior workup to refeeding.