Every claim, its source. Clickable.

390,124 healthy adults followed for more than 20 years. People who take a multivitamin every day do not die less than the others — not overall, not of heart disease, not of cancer, not of stroke.

More than 120,000 patients: the fitter the heart and lungs, the fewer deaths — from any cause at all. And the benefit never stops: even the fittest still gain from getting fitter.

The origin of the “30 plants a week” rule: observational data, not a demonstration of causation.

Lists the modifiable risk factors for dementia. Uncorrected hearing loss is among the most important.

Eight systematic reviews analysed: valerian does not have sufficient empirical data for insomnia. It appears safe, but its effectiveness is not demonstrated by objective measures.

Seven randomised trials: an improvement in wrinkles and in the marks left by the sun in all of them, visible from one month and maintained beyond two years. A later meta-analysis of eight trials and 1,361 patients confirms the effect on fine and deep wrinkles.

Five randomised trials: elderberry probably does not reduce the risk of catching a cold. It might shorten and soften the symptoms, but the level of certainty remains low.

In around seventy North Carolina households: an activated carbon jug removes on average about half the PFAS, with large differences between models. Reverse osmosis is markedly more effective. An American study, to be transposed with caution.

No over-the-counter supplement — ginkgo, vitamins, omega-3 — has sufficient evidence to prevent or slow cognitive decline in healthy adults.

Describes orthosomnia: chasing the “right” sleep score degrades the very sleep it claims to measure.

No human study shows that BCAAs alone increase muscle protein synthesis. Two intravenous infusion trials even measured a fall. Building muscle requires all nine essential amino acids, not three.

Biotin is everywhere in “hair and nails” supplements. Yet when you go looking for the evidence, you find only eighteen published cases — and in every one, the person had a real illness that explained their hair problem. In someone healthy, the only properly run trial, with a placebo group, found no difference. It is the textbook example of a supplement sold on a hunch that was never checked.

B12 deficiency is common on a vegetarian diet and almost systematic on an unsupplemented vegan one. It is not optional.

The reference framework on prebiotic fibres: fermentation mechanisms, expected effects and tolerance limits.

In most European countries, women who are pregnant or planning a pregnancy should get an intake of about 150 µg of iodine a day. Seaweed and kelp products are to be avoided: their iodine content varies unacceptably from batch to batch.

The basis of tapering: a marked reduction in volume while keeping intensity, over 7 to 21 days before the target event.

Low-sulphur 304 stainless steel releases less than 0.03 µg of nickel per cm² per week in artificial acidic sweat, with no reaction in already-sensitised people. Other grades release enough to trigger symptoms in an allergic person.

Nature’s investigation into the Benveniste affair, the scientific origin of the idea of “water memory”. Repeated under strict control, the experiment produced no result. It is the reference that buried the concept in the literature.

Eleven randomised trials, about 3,200 healthy adults. The fall in stress lasts only about twelve hours, the quality-of-life gains have gone by three months, and no significant effect on immune function is found. The authors stress the small number of trials and their limited sample sizes.

21 studies, 1,610 participants: a fall in anti-thyroid peroxidase antibodies at three and six months. But what the trials measure is a number in a blood test, not how people actually fare: no demonstrated improvement in thyroid function, nor in the need for treatment. An upper safety limit of 400 µg a day, beyond which selenosis becomes possible.

Eight randomised trials, 426 patients. The main analysis on insomnia does NOT reach significance (mean difference −1.92; interval −3.92 to 0.09). A weak effect found on anxiety. Four trials out of eight at high risk of bias, mainly because you cannot blind the weight of a blanket.

Eight studies: cold water immersion AFTER a weights session reduces muscle gain compared with weights alone. A parallel meta-analysis finds the same on strength gains. The effect is specific to strength and hypertrophy; endurance is unaffected.

Five placebo-controlled trials, 1,213 patients: a fall on the Hamilton anxiety scale about 3.8 points greater than placebo over ten weeks, with side effects comparable to placebo. An important limit: all five trials are funded by the manufacturer and share several authors.

37 studies, 3,048 patients. Glycated haemoglobin — the average of your blood sugar over the past three months — falls by about 0.63 points, and blood sugar measured on an empty stomach by about 0.82 mmol/L. But the trials are short, 8 to 16 weeks, their quality is rated weak by the authors themselves, and almost all were run in China.

This is the best summary we have on magnesium and sleep, and it remains modest. Three trials, 151 older people with insomnia. Those taking magnesium fell asleep about 17 minutes faster than those given a capsule with nothing in it. They also slept 16 minutes longer in total, but that gap is small enough to have come from chance. The authors themselves judge their three trials fragile and describe their confidence in the result as “low to very low”. That is why magnesium stays rated as emerging evidence on this site: the effect probably exists, but it was measured in older people who already had insomnia — not in a young adult who sleeps fine.

About 3 extra reps (+6 %) before failure, for 6 to 8 g taken 40 to 60 min before the session. A clear effect on the lower body, non-significant on the upper. The authors flag possible publication bias: the real effect is probably smaller.

14 studies: a weak effect on muscle soreness and a moderate one on strength recovery, in very small samples. A more recent meta-analysis does not find the effect on soreness.

The first serious global estimate of sleep apnoea: around 936 million adults aged 30 to 69 are thought to be affected, including about 425 million with a moderate to severe form justifying treatment. For scale: that is roughly as widespread as diabetes. The vast majority of those people do not know, because the main sign — stopping breathing for a few seconds — happens while they are asleep, and only someone sharing their bed can notice it.

A short-term gain in range of motion and a fall in perceived soreness; no demonstrated effect on next-day performance.

Seven trials pooled: people on peppermint oil saw their irritable bowel symptoms improve about 2.4 times more often than on placebo. Allowing for uncertainty, the true figure lies somewhere between 1.9 and 3 times. All seven trials point the same way, which is rare and strengthens the result. The American College of Gastroenterology recommends it, with reservations.

Twenty-three studies, with large effects reported on depression and anxiety. But Egger’s test detects publication bias, and the authors themselves call for further trials while noting the lack of geographical diversity in the studies.

This meta-regression is where the target of about 1.6 g/kg/day comes from, beyond which the additional benefit becomes limited.

A reproducible fall in LDL — the bad cholesterol — on psyllium.

13 trials, 1,472 women: a level of evidence rated low to very low. The authors remain uncertain about myo-inositol’s effect on live births and pregnancies. A more recent pooling, in 2022 across 17 trials, finds no improvement either in insulin or in blood sugar on an empty stomach.

The relationship between weekly set volume and hypertrophy, with diminishing returns.

A performance effect concentrated on efforts of roughly 60 to 240 seconds.

The benefit is concentrated in people who are genuinely deficient: vitamin D corrects a shortfall, it does not boost a normal status.

An effect on joint pain, across trials of uneven quality and with very different formulations.

Six randomised trials pooled: no significant improvement in statin-related muscle pain on coenzyme Q10. The 2015 European guidelines advise against this use.

Cochrane review: reduced plaque and gingivitis with an electric toothbrush.

124 studies pooled, average dose 2.1 g a day. A fall in bad cholesterol — LDL — of 6 to 12 % depending on dose, and the more you take the more it falls: about 7.6 % at 1.5-2 g and 10.3 % at 2.5-3 g.

24 trials, 4,631 participants, with nobody knowing who got what: no effect on a cold you already have. On prevention, the small benefit seen disappears as soon as all the trials are put end to end — in other words, it could well be chance. On top of that: from one brand to the next, what is in the bottle bears almost no relation.

28 trials: 3 g a day of oat beta-glucan lowers bad cholesterol — LDL — by about 0.25 mmol/L and total cholesterol by about 0.30 mmol/L. The effect is found in Europe, North America, Asia and Australia, in lean and overweight people alike.

A reduction in time to fall asleep and an increase in total sleep time, both modest in size.

A modest but reproducible fall in systolic blood pressure on dietary nitrates.

In the general population, supplementation does NOT PREVENT colds. It slightly shortens their duration, and the preventive benefit only appears in people under intense physical exertion or cold stress.

A modest reduction in symptom duration, provided dosing starts very early.

Ten trials of more than 200 patients, about 3,800 people. The reduction in pain reaches 0.3 to 0.5 cm on a 10 cm scale, below the 0.9 cm threshold set in advance as clinically relevant. The apparent effect is weaker in trials without industry funding.

The basis of 10,000 lux light therapy in the morning for seasonal affective disorder.

Across 627 tap water samples, 9 exceed the 100 ng/L threshold for the sum of the 20 PFAS monitored. TFA is found in 92 % of samples. Monitoring of 20 PFAS has been mandatory since January 2026.

The European agency concludes that no dose of monacolins can be considered safe, and links intakes from 3 mg a day to cases of rhabdomyolysis and liver injury. A European regulation banning them in supplements was being adopted in 2026.

Lets you check, model by model, whether a blood pressure monitor has passed a clinical validation protocol. A binary, checkable criterion.

The tolerable daily intake lowered to 0.2 ng/kg of body weight per day, 20,000 times less than the previous reference, on the basis of immune effects. BPA has been banned in food contact materials in the EU under regulation 2024/3190.

This is the top grade this American body awards, and almost no supplement gets it. The reason is a matter of timing: folate prevents serious malformations of the baby’s nervous system, but the part of the body concerned closes before most women even know they are pregnant. Taking it afterwards is no longer any use. Hence the recommendation: 0.4 to 0.8 mg a day for any woman who might start a pregnancy, before she even plans one.

Evidence judged insufficient to recommend multivitamins for cardiovascular or cancer prevention (grade I). A recommendation AGAINST beta-carotene and vitamin E for this purpose (grade D).

The same conclusion on the American side: CBT-I first line.

A floor of 150 to 300 min of moderate endurance activity a week, plus two resistance sessions.

The groups the French food safety agency advises against melatonin for: pregnancy, breastfeeding, children and adolescents, epilepsy, asthma, inflammatory or autoimmune disease, anticoagulant or antidiabetic medication.

Night-time noise fragments sleep even below the threshold of conscious waking.

The reference position stand on creatine monohydrate: effectiveness on strength and lean mass, tolerability at usual doses.

Cognitive behavioural therapy for insomnia is the first-line treatment, before any medication.

No proven health danger, but no demonstrated effectiveness either. Possible release of silver, sodium, potassium or ammonium ions; acidification of the water; possible degradation of microbiological quality if the jug sits unused. The jug serves taste, not drinkability.

Seven reported cases, including two of congenital hypothyroidism, likely linked to supplements taken by the mother. The agency recommends avoiding simultaneous exposure to several iodine sources during pregnancy: one controlled source, not a stack.

The European Union officially allows the claim that 1.5 to 3 g a day lowers bad cholesterol — LDL — by 7 to 11 %, with the full effect reached in two to three weeks. It is one of the rare quantified nutrition claims validated by EFSA.

Solar radiation and artificial UV are classified group 1: proven human carcinogens. UVA, UVB and UVC included. The risk rises with the dose, with no threshold below which exposure would be safe.

The list of so-called “functional” claims (article 13.1): they state that a nutrient is NECESSARY for a normal bodily function — magnesium for muscle function, for example. CAREFUL: that is not the same as a risk reduction claim (article 14), which requires demonstrating a quantified effect on a risk factor, as for phytosterols on cholesterol. A 13.1 claim does not say that taking more when you are not short brings anything. That is why products citing this regulation stay rated as emerging evidence on this site.

A cause-and-effect relationship recognised between oat beta-glucan and lower LDL — the bad cholesterol — from 3 g a day. It is one of the rare quantified health claims authorised in the European Union.

It is one of the rare fibres for which Europe agrees to say it helps with weight loss — but only if you are already eating less, and at a precise dose: 3 g a day, in three servings, with one to two large glasses of water. You have to look at the figure: 1.27 kg of difference from placebo. The researchers themselves had decided in advance that below 2.5 kg, the gap would not be worth it. They came in under their own bar.

Beyond an hour of effort, the position stand recommends about 0.5 to 0.7 g of sodium per litre of drink: better fluid retention, better taste, and protection against hyponatraemia in those drinking large volumes. Below an hour, water is enough. The underlying rule is to drink to thirst, not as much as possible.

The founding document on measuring heart rate variability: measurement conditions, limits of interpretation.

About 21,000 older adults, followed for 3.6 years on average. On a daily multivitamin: no fewer cancers diagnosed, and no fewer deaths. The gaps observed are too small to be told apart from chance. The trial was funded by Mars Edge.

Ten weeks of fermented foods: a rise in microbial diversity and a fall in inflammatory markers, in a modest sample.

The largest study ever done on something this simple: in 600 Chinese villages, 21,000 adults replaced their cooking salt with one where a quarter of the sodium is replaced by potassium. Nothing else changed. The result after almost five years: a sixth fewer strokes, and an eighth fewer deaths from all causes. One important limit: people with a kidney problem had been excluded from the study, because their kidneys clear potassium poorly. If your kidneys are fine, this salt is one of the best documented things on this entire site. If not, ask your doctor first.

47 adults with confirmed lactose intolerance took lactase and then a placebo, each person going through both. Undigested lactose ferments in the gut and produces a gas you then breathe out: that gas fell by 55 %, and symptoms improved. Two caveats: there were only 47 of them, and the main measurement is still that gas, not day-to-day comfort over time.

This study measured what crosses into the blood when you apply sunscreen. Six chemical filters tested in 48 adults: all of them appear in the blood above the level that, under American rules, requires further safety studies. Some remain for several days. The result has to be read properly, and the agency says so itself: crossing into the blood is not the same as being toxic, no danger threshold has been established, and it is not a reason to stop protecting yourself from the sun. It is a reason to prefer mineral filters, which stay on the surface of the skin.

120 adults with insomnia, an 8 kg weighted blanket against a 1.5 kg light one for 4 weeks. The gap between the two groups is one of the largest ever seen in this field: by the end, 42 % of the people under the weighted blanket no longer had insomnia, against 3.6 % in the other group. Read it bearing in mind that every participant also had a psychiatric condition.

A reduction in stress-related symptoms, without sedation.

A trial on a high-dose pharmaceutical form of EPA — not to be extrapolated as-is to fish oil supplements.

33 people listened to music tuned to 432 Hz rather than the standard: their pulse fell very slightly. Neither the participants nor the researchers knew who was listening to what — that is the right method. But there were only 33 of them, and nobody has repeated the experiment since. The authors write themselves that this is a preparatory study, meant to set up real research, not to conclude.

Collagen taken with vitamin C about an hour before exercise: markers of collagen synthesis increased.

400 mg of caffeine six hours before bed still significantly reduces sleep time — without participants always noticing.

A randomised trial on skin ageing caused by the sun.

A small trial on insomnia in older people: limited numbers, hence an “emerging” rather than “solid” level of evidence.

The frequently cited ashwagandha trial: modest numbers, one specific extract — not generalisable to every product on the market.

Iron supplementation reduces fatigue in women with low ferritin, even without anaemia. Hence the mandatory test before taking any.

Follow-up of the Australian Nambour trial: how many melanomas appear among people who use sunscreen regularly, and among those who do not.

The basis of the caffeine + L-theanine pairing: sustained attention preserved, jitteriness reduced.

A vehicle-controlled trial, retinol 0.4 % over 24 weeks: a clear improvement in fine wrinkles on the treated side. Over-the-counter retinol is less potent than prescription tretinoin, but better tolerated.

3 g of glycine before bed: an improvement in subjective sleep quality in a small sample.

A large randomised trial: a modest effect on bone density, a limited fracture benefit in the general population — clearer in older women who actually took it.

The only peer-reviewed publication involving Masaru Emoto. The authors report a difference in crystal aesthetics judged by volunteers. A low impact factor journal, a result never reproduced by an independent team. Emoto’s popular books are not scientific publications.

Shows that a subliminal prime can influence a brand choice — but only in a laboratory, over a few minutes, and only in participants who were already thirsty. The effect is real, tiny, and unrelated to listening for several minutes a day.

The phase response curve: morning light advances the clock, evening light delays it. It is the basis of the whole light protocol.

Genuine work on so-called “structured” water in contact with certain surfaces (the exclusion zone). It is serious physics. But nothing in that work shows that stirring water into a vortex in a glass produces an effect on health: no study has ever bridged the gap.

Melatonin is suppressed by far lower light intensities than was thought: the order of magnitude of domestic lighting is enough.

The direct test of subliminal self-help audio tapes. Participants reported the effect written on the label — even when the labels had been swapped. In other words: the effect came from what they believed they were listening to, not from the subliminal content.