What's new

The hour-by-hour rise comes from single doses given in short test sessions, which may not match a regular drinker's morning cup. Only three trials, of about 2 weeks each, looked past the first day, too short to catch a change that builds more slowly. An older review found the rise strongest in people who already have high blood pressure, so it may be smaller in others.

The trial registry lists two melatonin trials in heart failure, one withdrawn and one with no update since 2020. Of the registered melatonin insomnia trials screened, only 13% had posted results, so some safety data may never have been published.

The study measured an inherited tendency to insomnia and never looked at melatonin or any sleep aid. Everyone in it was of European ancestry. About 28% of the people in the insomnia gene data were also in the heart data, which can pull the estimate toward an ordinary association, though the authors judged that bias likely small.

This rests on one trial, and its benefit showed in sleep and tiredness ratings rather than an overall jet lag score. About a quarter of those enrolled were dropped for missing doses or leaving. In the largest jet lag trial, neither 0.5 mg nor 5 mg beat placebo.

The pooled scores come from four small trials of 17 to 52 people, published from 1987 to 1992. Two later trials that found no benefit, one of them the largest at 257 people, were left out of that pool. A re-analysis that counts them finds about half the effect of a pooling that leaves them out. Ratings varied so much between people that their spread often exceeded 90% of the average score, so the 40% is a gap between group averages, not what each traveller felt. Scores were compared after the flight, not as change from how each person felt before it.

The review mixed trials in children and adults, so the rates are not adults' alone. Only 8 of the 37 trials were rated good quality, and 13 were rated poor. No rate was given for palpitations.

Of the 90 reports, 19 had enough detail to judge whether melatonin caused the problem, and the dose taken was often unknown. Reports show what was noticed, not how often it happens. Limits on melatonin supplements differ by country, from 0.3 mg in Belgium and Germany to under 2 mg in France.

The three trials enrolled 201 people in all. The estimate ran from about a 2-point fall to a 6-point rise in ejection fraction, so a small effect either way is not ruled out. Harms were reported inconsistently across the trials. A second review reaching the same result drew on some of the same trials.

A ranking says which treatment the evidence supports better, not how much it helps. Of the five studies a 2014 review accepted, three tested sleeping pills and one tested light. Melatonin's rating rests on the fifth, an earlier review of ten trials, graded low quality because doses and schedules varied. A later reading of the head-to-head found melatonin alone did not clearly lower overall jet lag.

Pieces have been tried alone: non-drug fixes after real flights mostly failed, with 9 of 13 controlled studies finding no change in one review. Three days of earlier bedtimes with morning light moved the clock 1.5 to 2.1 hours in the laboratory, with no flight afterwards. Six mountain bikers combined all three before flying to Tokyo, with no comparison group.

Laboratory studies of melatonin and the body clock continued after 2004, without flights. Newer drugs that act on melatonin's receptors have had jet-lag trials of their own, which this count leaves out. The registry figure comes from the main US trial registry alone.

Benefit after westward flights rests on fewer trials: two in one pooling, four in another. Two trials found benefit after westward trips of twelve or more time zones.

The carvacrol in the mouse study came from a different plant, and the mice were infected through the nose, far from a person swallowing a capsule. A dish holds the oil against the germ directly, and these studies do not show whether a swallowed dose reaches an infection at that strength.

The US trial registry lists 19 studies of oregano and none for a cold, so no cold trial is registered either. The one oregano trial that did turn up was published outside the main medical indexes, so a small trial published the same way could have been missed.

The labelled amounts are the makers' own and none states a standard carvacrol content. Only combination products with black seed oil were checked, so plain oregano softgels were not compared.

The oregano in it is Origanum syriacum, not the Origanum vulgare sold as oil of oregano. It was blended with mint, eucalyptus and rosemary oils, so no one oil's effect can be picked out. The spray went on the throat and was not swallowed.

The data allow anything from about 13 minutes faster to 3 minutes slower, so a modest gain is not ruled out. The review's headline figure pooled jet lag with shift work and sleep deprivation. Two single trials, the first among them, did report faster sleep onset on melatonin.

One review found melatonin did not speed falling asleep and called it unproven for jet lag; another found overall jet-lag ratings lower. They measured different things, so whether it helps sleep onset in particular is still open.

The decline was shown, not tested, and it runs alongside other changes: the later trials were larger, rated jet lag before the flight, or used a new scale. The 257 travellers had spent only five days away, so their body clocks had less to readjust on the way home. The reviewer who added the two null trials to the pool co-wrote one of them.

Every pooling of jet-lag ratings favours melatonin, but how much it helps turns on two later trials that found no benefit and were left out of the bigger estimates. Whether they were fair tests is disputed; no newer trial has settled it.

The crew's body clocks were already scrambled by back-to-back flights, so the early doses landed at varying body-clock times. Only the ratings recalled on day 6 separated the groups, while the daily ratings showed trends. A crew's working days are not an ordinary traveller's.

The only trial to compare start times found that beginning melatonin days before arrival did no better than placebo, while a pooling across trials leaned the other way. Whether a pre-flight start helps ordinary travellers is unsettled.

The oregano nasal-spray trial ran at one clinic with 25 people a group and has not been repeated. Every group also followed the same Persian-medicine lifestyle routine, which may explain part of the placebo group's gain. A 2023 pooled ranking that put oregano first for this condition rests on this same trial.

The randomised evidence applies only to people whose kidneys were healthy at the start, and early kidney disease is common and silent. Diet in the cohort came from food frequency questionnaires, which are imprecise, and its top quartile is a moderate Korean intake rather than a deliberate high-protein diet. In the cohort, a rapid decline meant the estimated filtering rate falling by more than 3 mL a minute each year, a research threshold rather than a diagnosis of kidney disease. In a sixteen-week trial at 3.2 g per kg a day, kidney and liver markers rose within the high-protein groups even though the comparison between groups came out level.

Short trials show filtration rising without declining; a long cohort saw a signal but read kidney function off creatinine, which protein itself raises. Nothing here shows harm, and nothing establishes safety over decades.

The trial reported which way each measure moved, not by how much. Nine people took each treatment, and three published papers all describe the same 27. The caffeine was a slow-release tablet, not coffee.

Both papers were written to explain a single conference report that had not been peer-reviewed. The letter proposing the blood-pressure route also says untreated insomnia could explain the same result.

With no comparison group and no count of people who took melatonin without trouble, the cases cannot give a rate. The link rests on timing and on recovery after stopping, with no formal causality score applied. Heart block is a rhythm problem, not heart failure.

Slow-release caffeine sped the hormones' return just as much in one of the studies. In a trial of 31 people whose days were full of irregular daylight, melatonin did not speed the body-temperature rhythm's shift. The two smallest studies, of 6 and 8 people, described no placebo group.

Oregano was one oil among several, so its own share cannot be told apart. The study was a pilot comparing three blends. No study of oregano taken with metformin or another diabetes drug has reported results.

An untested pairing is not a safe one; a bleeding effect has not been measured either way. A US trial of how oregano changes the body's handling of a set of drugs has been recruiting since 2024 and has posted no results.

The birds ate oregano as 2% of their feed without a break, far more and for far longer than a person on capsules. A change in gene activity is not an illness, and no illness was linked to it in the chickens.

It has never been tried in an animal or a person. The strengths used, 58 to 588 µg per mL, have not been shown to be reachable in the human gut from a supplement.

In the only trial of oregano oil for gut parasites, nobody went untreated, and parasites can come and go in stool on their own, so their clearing may not be the oil's doing. Two of the three parasites were found in 4 people and 1 person. No trial with a placebo group has followed in the 25 years since, and none is registered.

Tertiary referral centres see the worst cases, so 42% is acetaminophen's share of the liver failures that reach those units and never a risk to anyone taking it. Doses in the accidental group were reconstructed from what patients recalled afterwards. The three routes overlap, so one person can sit in all of them and the percentages do not add to a hundred. The narcotic-combination finding is US-specific, since those products are not sold everywhere.

Single doses, in adults, for pain after an operation. That is the setting this drug performs best in, and it is not the setting most people take it in. Adverse-event reporting across the pooled trials was patchy, so the reassuring safety picture is thinner than the efficacy picture. The review dates from 2008 and has not been updated since.

Fourteen days in hospital with every meal provided is not ordinary home use, and 145 people is a small sample for a safety question. The 28-day study used 3 g rather than 4 g, so the two rates are not measuring the same exposure. Its placebo group had an 11% rate of its own, a reminder that this enzyme moves without any drug at all. The 4 g trial ran inside a hydrocodone product development programme, with several industry-affiliated authors.

A retracted trial leaves an absence of evidence, not evidence that the drug fails, and only the first of those is true here. Cochrane's inclusion rules were strict, which is why three trials survived a search of a much larger literature. This covers low back pain with no identified cause, not back pain traced to a known injury or disease.

No drug was tested against another here. These are numbers from separate trials in separate people, lined up side by side. The authors flag 13 of the drug and dose pairs as possibly affected by publication bias. All of it is a single dose for pain after an operation, which says nothing about taking either drug for weeks. And the arithmetic that makes the pair work is the same arithmetic that makes an unnoticed second acetaminophen product dangerous.

Short-term outcomes only, out to twelve weeks, so this says nothing about months of use. A group average does not rule out individuals who feel better on it. A separate network of 98 trials covering 69 medicines for the same condition rated confidence in every analgesic tested as low or very low, which is a different verdict from a confident null and worth holding alongside it. Four authors, lead and senior among them, had GlaxoSmithKline support for one of the pooled trials, a conflict that runs against the finding rather than towards it.

The nine-point line comes from a companion review by the same authors, not from this analysis, and other groups draw it elsewhere. The 192-trial network excluded any trial with fewer than 100 people per group and anything published outside English. Averages hide people, and some individuals in these trials did improve. Patients on paracetamol were also nearly four times as likely to show an abnormal liver function test, a laboratory result the authors say they cannot interpret clinically.

An umbrella review inherits every weakness of what it reviews and adds no data of its own, so it can say the evidence base is poor without settling the question underneath it. Failing to demonstrate a link is not the same as demonstrating there is none, and the authors word their conclusion as the former. It covers autism and attention-deficit disorder only. Its search closed in September 2025, before the sibling-matched work published in 2026.

This is frequent episodic tension-type headache of moderate or severe intensity, not migraine, which the trials generally excluded, and not the chronic form. Ask an easier question and the number improves: for pain-free or only mild pain at two hours it is 10, and for avoiding a second painkiller it is 7.8. No trial was at low risk of bias on every count, though the reviewers put most of that down to thin reporting rather than bad conduct. The senior author declares grants and fees from analgesic manufacturers.

The review gives each study's cortisol only at the end of its course, so how soon any fall begins is unknown. The falls it reports are measured against where people started, not against placebo, and one of the nine had no placebo group. The two studies that measured cortisol in saliva disagreed, and one found no change. Extract and dose differ from study to study, from 60 mg to 5 g a day.

This is a snapshot of one Irish and Northern Irish cohort recruited between 2008 and 2012, not a picture of what happens over time. Anyone already taking a B12 supplement was excluded, which sharpens the dietary signal but leaves out most real supplement users.

Everyone in the trial was on insulin for type 2 diabetes, so the absolute numbers may not carry to metformin alone. Deficiency was called on the blood level alone, not on markers of whether cells were short. The homocysteine rise that would prove a functional cost fell short of significance.

B12 was measured only at years 5 and 13, so nothing is known about the years between. Frank deficiency at 13 years, 7.4 percent against 5.4, did not reach significance. The neuropathy link is an observation within the group, not a demonstration that the low B12 caused it.

These trials counted blood levels, not whether anyone felt better. The Cochrane review behind the comparison had three eligible trials and 153 people, and graded its own certainty low. By one year the oral group had slipped slightly behind the injected one on the stricter analysis.

The two most specific markers, methylmalonic acid and holotranscobalamin, pointed the same way but missed significance at p=0.06 and p=0.09, so the case rests on homocysteine, which folate and kidney function also move. The comparisons are observational, so other diet-linked differences remain live.

An umbrella review inherits the primary trials' limitations and cannot fix them, and moderate certainty still means the true effect may differ substantially from the estimate. No pooled effect sizes appear, only the gradings, so nothing here says how big the reduction is. Searches ran only to May 2022.

Low certainty for the encephalopathy benefit, very low for everything else, which is as much the finding as the numbers. No trial compared these preparations against lactulose or rifaximin, the drugs these patients actually get, so nothing here says whether they add to standard care.

The bigger effects reported in women, in people already diagnosed, and in Asian populations each rest on a handful of trials, few enough that ordinary chance could produce them. Doses and trial lengths varied widely, so 0.4 points is an average across regimens rather than the result of any one of them. Berberine caused more stomach and bowel upset than placebo in the large trial.

Coffee habits were self-reported and chosen, not assigned. People worried about their blood pressure may cut back, which would hide a rise among heavier drinkers. An earlier pooling of six studies found 9% higher risk at 1 to 3 cups a day and no rise above 3. American studies found slightly lower risk and European ones did not, so the overall figure averages a real split.

The trials gave caffeine itself, not coffee, at doses and for lengths that varied from trial to trial. Eight trials is a small base to split by sex, dose and length, so those three patterns are leads rather than settled findings. The clearest dose effect, above 400 mg a day, was in diastolic pressure.

Everyone in these 45 studies was short on sleep, at six hours or under or awake for a full day, so nothing here says what caffeine does for someone who slept well. The benefit kept rising with dose without the review ever finding the point where it stops rising, so the top of that range is unmapped.

This is a precautionary ceiling drawn from a broad safety review, not a figure any trial in pregnant women produced. It describes a population rather than any particular pregnancy.

This is a threshold for a population of healthy, non-pregnant adults rather than a personal limit; the same 400mg lands differently on people who clear caffeine at different speeds. It is regulatory guidance rather than the result of a single trial. The 200mg single-serving figure rests on thinner data than the daily one.

That 9-hour figure is an average across many doses and timings, not a personal cut-off. How fast a person clears caffeine varies widely, partly for genetic reasons, so the same cup at the same hour lands differently on two people. Older adults' sleep tends to be more sensitive to it than younger adults'.

The 11 trials used a spread of doses and designs rather than one fixed ratio, so this is a pattern across mixtures rather than a test of any single blend. The authors call theanine's contribution comparatively modest rather than absent, which leaves room for a larger pool to move it. And every trial pooled here tested combinations, so none of it measures theanine by itself.

One trial of 48 people, and it was designed to study caffeine, with theanine as the modifier rather than the subject. For anyone taking theanine by itself for calm focus, the same trial is a null result: alertness, jitteriness and mood did not move.

Those timings are population ranges, and plenty of people stop and feel nothing at all. The 100mg figure is the lowest daily intake shown to produce withdrawal, not the usual one. This is a critical review of 57 experiments and 9 surveys rather than a pooled statistical analysis, though nothing since has replaced it.

Nobody was randomised to clobazam — patients arrived on it or not — so the two groups differ in ways beyond the drug. The response was much higher in the clobazam group in both arms, placebo included, which the authors flag rather than resolve. Only Dravet and Lennox-Gastaut trials are pooled.

The trials ran one to six months and 64% concerned nerve pain, so this does not cover long-term use or the everyday aches cannabidiol is sold for. The cannabidiol-dominant result is a null across a whole product category, not a tested null for any particular product or dose.

One trial does not fit cleanly: in Lennox-Gastaut syndrome the 20mg/kg/day arm ran about five percentage points ahead of the 10mg/kg/day arm on drop seizures. No trial here powered its two active doses against each other, so every dose comparison is descriptive rather than a tested difference.

Eighty-four separate comparisons across 23 conditions means some significant findings are expected by chance, and most were graded low or very low quality. The missing cannabidiol effect on pain is an absence of evidence inside these trials, not a tested null.

The placebo groups improved too — 21.8% fewer drop seizures in Lennox-Gastaut. In Dravet, the share of patients whose seizures halved did not separate from placebo (43% against 27%), so the finding rests on the seizure count alone. Both trials were manufacturer-funded.

The search closed in April 2018, before several cannabidiol-specific trials were published. The one anxiety signal the review did find came from pharmaceutical THC, in people whose main problem was chronic pain or multiple sclerosis, and was graded very low quality.

The short trials and the long follow-ups measure different things, weight on one side and disease on the other, so neither one checks the other. The long-term half only watched what people already drank, and people already at higher risk switching to diet drinks would produce the same pattern. This is a joint position from the Korean Diabetes Association and Korean Nutrition Society rather than a new analysis.

The 2025 review rated its own certainty very low on every outcome, because the experiments differed widely and most carried design weaknesses that could tilt a result either way. The older review's ranges were wide enough to hide a real effect: on body weight it ran from a 1.6kg loss to an 11.6kg gain. Not finding an effect is not the same as ruling one out.

That 35g went to ten men on one occasion, with no long-term follow-up. All ten were men, so nothing here speaks to anyone else. And no formal ceiling means nobody has set one, which is not the same as evidence that any dose is safe indefinitely.

These are synthetic drugs given deliberately to ripen a baby's lungs, which is a different matter from living with a lot of one's own cortisol. What becomes of those children years later is still being studied. So is the safety of giving more than one course.

Most of the pooled studies took a single snapshot rather than following anyone, so nothing here shows stress moving the surge. The stress measures themselves are a mixed bag, from job strain to burnout, and they do not all point the same way, which is the finding rather than a flaw in it.

Every study pooled here watched people rather than assigning anything, so a flat curve may mark illness already underway rather than cause it. The slope was measured differently from study to study, from two saliva samples a day to six or more, which blurs the pooled figure.

This only tracks a pattern over time, so it cannot show cause. Just 32 of the deaths were from heart disease, which makes the estimate rough. Cortisol was measured on one ordinary weekday. Morning cortisol on its own showed no link at all.

The mechanism is shown far more directly in the animal studies than in the human ones, where nobody has followed it over time. And this is about how cells respond rather than how much hormone is present, so a blood cortisol reading says nothing about whether it is happening.

The lean-mass figure is the solid one; the fat-loss figure ran anywhere from 1.34kg down to 0.11kg, close enough to nothing at one end that the effect there is unsettled. Everyone pooled here was under 50. In a small share of the trials, how people were assigned or assessed left room for the result to tilt.

The pooled training ran about 12.6 weeks on average, so nothing here shows whether the gains hold over a year. The pool of qualifying trials is small, and the authors ask for more work in this age group before the functional gains are treated as durable.

These trials were built to test performance rather than to hunt for harm, so anything rare or slow enough to slip past them would have slipped past. The adverse-event reports are filed voluntarily, which distorts the count in both directions. Average intake across the pooled trials was about 12.5g of creatine a day.

Twenty of the 23 trials in one analysis enrolled only men, and just 40 women appear across its two female-only trials, so the weaker showing in women reflects how little has been measured rather than a settled absence. Gains were most consistent in younger adults. And these are pooled averages, not what any one lifter should expect.

This is an absence of evidence, not evidence that D-chiro-inositol does nothing. Myo-inositol sat outside the review entirely, so nothing here describes the form most people actually take. The search closed in January 2017.

The community data are observational, so doctors steering higher-risk patients towards particular drugs cannot be ruled out. A relative risk says nothing about a given reader's own starting point. In the randomised trials the absolute excess came to about three extra major vascular events per 1,000 people treated for a year, one of them fatal. Those trial doses were high and sustained.

This measures anticholinergic burden as a whole, not diphenhydramine and not antihistamines. Only 8 of the 25 studies found could be pooled at all. The scales used to score that burden agree poorly with one another, so what is being measured is not the same thing from study to study.

The trial was powered on the diary outcome, so a flat recording is a difference not detected rather than a difference shown to be absent. Mild insomnia in adults averaging 44, not the older people the warnings are about. The drug arm ran 14 nights, so nothing here reaches beyond a fortnight.

This comes from the same forty drivers and the same four sessions, so it inherits that setting: a simulator, a daytime dose, driving one hour later. A failure of self-rating to predict performance across one hour-long test is not proof that people can never judge their own impairment.

An average 3.6 percent of body mass is more dehydration than most recreational exercisers ever reach, and the performance tests averaged 22 minutes. The dehydration was deliberately induced, usually by heat or fluid restriction, which brings its own fatigue on top of the water deficit. The confidence interval on the headline runs from 0.8 to 4.0 percent. Nothing here compares an electrolyte drink with water.

The people tested were mostly older adults with night-time leg cramps, not athletes cramping mid-race, so this settles the mineral rather than the sport. The pregnancy trials in the same review could not be pooled at all because their outcomes were reported inconsistently, and a separate Cochrane review graded those same trials as far less biased than this one did. Nothing here tests magnesium during exercise.

Children with gastroenteritis, which is nobody in a gym. The trials were graded poor to moderate quality and mostly unblinded, since a drink cannot be disguised as a drip. Each route carries its own complication: vein inflammation with the drip, a stalled gut with drinking, roughly one in 50 and one in 33. The death counts, six against two across four trials, are far too small to read anything into.

Hospitalised children with acute diarrhoea, compared against another rehydration solution rather than against water, so nothing here reaches a well person drinking one. This is the 2002 version of a review since updated, cited for the finding that changed the formulation. The trials were mostly unblinded and largely in low- and middle-income hospital settings. Low blood sodium was looked for in only six trials with events in three, so that reassurance rests on thin data.

Five years cannot rule out a cancer that takes longer to appear, and the authors confine their conclusion to that window. The interval reaches a rate ratio of 1.13, so a modest increase is not excluded, only a large one. Almost everyone enrolled was a middle-aged or older adult in a heart-disease or bowel-polyp trial, so the reassurance reaches women of reproductive age by argument rather than by measurement.

The 4mg dose tested here is ten times what the general population is now told to take. The trial was stopped early once the result crossed its stopping rule, which tends to make an effect look larger than it is. Twenty-seven defects occurred in the whole trial, so the estimate is rough. Every woman enrolled had already had an affected pregnancy, so this says nothing about a first one.

A plateau is not a threshold. This models how long a level takes to stop rising, not how long until enough folate is present to protect an embryo, and those are different points. The 906 nmol/L target it is anchored to comes from observational Chinese data rather than a trial. Most of the pooled studies were in adults who were not pregnant, and red cell folate assays are notoriously hard to compare between laboratories.

None of this means 400 micrograms is the wrong number. It means nobody ran that trial. The Task Force reaffirmed its position rather than reviewing the evidence from scratch, and it gives a range without saying where in the range to sit. Its advice also addresses a country whose flour has been fortified since 1998, so the benefit of a tablet on top of that is not the benefit the old trials measured.

This is a strong recommendation resting on an absence of benefit rather than on demonstrated harm, and the guideline notes glucosamine's toxicity is low. It hardened from a conditional recommendation against glucosamine in 2012, so a reader may still meet the older advice. The search closed in 2018.

None was designed to detect rare harm, and the participants were people with knee osteoarthritis, not the general population. The chondroitin review's finding of fewer serious events on treatment than on placebo more likely reflects sparse, inconsistently reported data than a real effect.

The 0.9 cm threshold is these authors' own choice rather than an agreed one, and a reader accepting a smaller threshold would read some of these intervals differently. Restricting by trial size excludes most of the older positive literature, so this is not the body of trials other reviews pool.

A placebo response of six in ten leaves very little room for a supplement to show a difference, and every arm could take up to 4,000 mg of paracetamol a day as rescue, narrowing the gaps further. The trial used glucosamine hydrochloride, not the sulfate salt sold in Europe as a medicine.

This is a continuation in the participants who met stricter X-ray criteria, not a fresh randomisation, so it carries the parent trial's high placebo response with it. Two years of dropout leaves less power, not more. The authors call the trends beneficial but not significant, which describes a null.

The result is fragile: its interval reaches to 0.002 mm, its p value is 0.046, and it survives only after adjustment for sex, weight and baseline severity. It is one of several comparisons, so a multiple-testing correction would likely remove it. Whether a tenth of a millimetre is felt is untested.

The subgroup was one of two specified in advance, but was never powered as a test of its own and no independent trial has repeated it. The Spanish trial is small, at 164 people, and its placebo group's larger improvement appears only when everyone randomised is counted, vanishing among completers.

Risk of bias was the second-largest explanation, cutting the disagreement by a further 32%, so brand and how well a trial was run stay entangled. The shrunken estimate rests on three trials, and even after both factors the trials still disagree substantially. Pain only, nothing on joint structure.

Which preparation a trial used and how rigorously it was designed cannot be separated either: the older, weaker trials are also that manufacturer's. On the standard osteoarthritis questionnaire, pain, function and stiffness reached significance for neither preparation. Its search closed in 2005.

Manufacturer involvement, the salt used and whether allocation was concealed are almost perfectly entangled here, so this cannot say which explains the gap. These are comparisons across trials, not a randomised contrast, so sponsorship is associated with the result, not shown to cause it.

All five datasets shared were independently funded, so the form distinction is tested only inside independent trials, and no dataset from the manufacturer whose product the argument is about was contributed. In the published record, form, sponsorship and concealment cannot be separated.

The five contributed trials still cover 55% of everyone ever randomised in a published placebo-controlled glucosamine trial, so the gap is in which trials are represented, not in people. Because no manufacturer-sponsored dataset arrived, the reanalysis cannot answer those trials on their own data.

These figures are reported as background inside a guideline evidence review, not measured by a study designed for the question. That same review rated every comparison of different inositol ratios at the lowest certainty its framework offers.

The cycle-regularity interval's lower bound is 1.01, so one more small trial could take it to nothing. A one-point change on a hair-growth scale may not be visible to the person looking in the mirror. Six trials, 388 women, three to six months, and no fertility or pregnancy outcomes.

An umbrella review inherits the weaknesses of the trials underneath it, and counts the same small trials again each time they turn up in a different meta-analysis. Over half of the 13 meta-analyses it appraised were themselves rated low or very low quality.

A guideline synthesises trials rather than adding any, and only 3 of the 13 comparisons its evidence review assessed had enough studies to pool. The same guideline records that inositol's side effects and safety are not known, and that supplement regulation, dose and quality control differ from those for medicines.

Ovulation and a regular cycle are steps toward pregnancy, not pregnancy. The ovulation pooling was highly inconsistent between trials, and the whole analysis rests on ten small trials totalling 601 women. Neither pooling reported a live birth.

A null result is not proof of equivalence. Eight small trials, with moderate inconsistency on three of the five outcomes, are underpowered to rule out a real difference between the two. No risk-of-bias assessment or certainty grading is reported.

The trial's one favourable repletion number is a secondary outcome at p=0.049 resting on roughly 2 women against 8, which is too few to build on. It ran at a single centre in 150 young Swiss women, none of them anaemic, so it does not establish the same schedule in men, in older adults, or in anyone with iron-deficiency anaemia.

Doses across the pooled trials ran from 7 to 1,800 milligrams of iron a day, schedules that are not clinically interchangeable. 27 of the 38 studies left out first-time donors, who are the people least likely to come back after being turned away and therefore the ones the intervention most needs to reach. The review was produced by a national blood service, which has a direct operational interest in donor retention.

The 30 microgram figure rests on 9 studies and 512 people, all graded low or very low certainty. The studies underneath it proposed thresholds anywhere from 12 to 200 micrograms per litre. And ferritin was far worse at spotting iron overload than deficiency.

Fifty-four young Swiss women, split across three sub-studies, all with low iron stores and none anaemic. What was measured is how much iron crossed the gut over a day or two, not haemoglobin or stores rebuilt over months. The paper argues the hepcidin timing supports alternate-day dosing but never tested such a schedule.

People carrying only one copy, and those with the milder H63D variant, showed no excess of anything — and they far outnumber the two-copy group, which is roughly 1 in 150 to 200 people of northern European ancestry. UK Biobank volunteers are healthier than the general population, so these percentages are not population rates. The one-in-ten lifetime figure is for men; the figure for women is substantially lower.

Only 10 of the 67 trials were judged at low risk of bias. The harm figures rest on far fewer women than the benefit ones — 521 to 604 against 6,861 for haemoglobin — so they are much less precise. The review's composite measure of any side effect at all was not significant, even though the specific ones were.

The iron dose was 100 milligrams every eight hours, high enough that absorption is already saturating and there is little room for an enhancer to help. Everyone in the trial was frankly anaemic, in whom absorption is already turned up. And this tested a vitamin C tablet taken with an iron tablet, not the vitamin C in a meal acting on the iron in food.

The larger figures come from a slice of the data examined after the fact rather than the review's main comparison, which makes them the weaker number even though they are the more striking one. The ten pooled reviews overlap, so some of the same original trials are counted more than once. And blood pressure is a stand-in here: none of these trials counted heart attacks or strokes.

This is magnesium sulphate at monitored clinical doses through a vein; magnesium sold in capsules has never been shown to do anything for eclampsia risk. The trial ran across 33 countries with very different standards of care, so the background risk it was measured against varied from site to site. And the Cochrane review is built around this same trial, so the two are not independent of each other.

The reviewed trials establish the direction but not the size; how far the risk falls is not settled here. This is a different job from preventing a mother's seizures, working a different way, and the two should not be read as one use. And it applies to births that are very preterm, not to pregnancy generally.

That pairing is the peak of a curve drawn across many different trials, not a regimen any single trial actually ran. Nobody has yet put it head to head against the usual 30-minutes-before dose, which is the comparison that would settle it. The pooled trials span 1987 to 2020 and differ widely in whom they enrolled.

The search closed in October 2025, so anything published since is missing. The human trials measured such different things that they could not be pooled, which makes the null conclusion a reading of the literature rather than a calculated number. The review also found no outcomes trial at all of intravenous NAD, which is sold commercially.

Between two and five trials per outcome, so a real effect could still be hiding in an evidence base this thin. What argues against that is how closely the trials agreed with each other: they scattered almost not at all around the null. The nicotinamide riboside results could not be pooled and ran in both directions, longer six-minute walks in peripheral artery disease against worse physical performance scores in mild cognitive impairment.

These were high, sustained, prescription-level doses in people with arthritis, not occasional tablets. Every regimen studied raised upper-gut complications, so this is a ranking inside a class that all carried the risk, not a safe option beside a risky one. Heart failure risk was roughly doubled by every drug in the analysis, naproxen included.

The over-60 systolic figure is a subgroup finding inside a 349-person pool, which generates a hypothesis rather than confirming one. The analysis assumed one true effect across trials spanning 250 to 2,000 mg a day and 14 days to 24 weeks, which narrows the confidence interval by assumption rather than by evidence. No trial in the pool measured a heart attack or a stroke; the reading on the cuff is the entire result.

Eight trials and 342 people in the metabolic pool, so a modest real effect could be missed. Participants were mostly healthy and not diabetic, which is the population least likely to show a metabolic improvement and leaves the question open in people who already have metabolic disease. Several pooled trials ran fourteen days, far too short for HbA1c, which reflects about three months of blood sugar, to move at all.

The longest trial in the pool ran 24 weeks. Ten trials contributed to the safety analysis, too few to catch a harm affecting one user in a thousand, and most were funded by manufacturers, who historically under-report adverse events. The four-week study establishes the highest dose formally tested, not a dose any regulator has certified. No trial measured a cardiovascular event.

Median follow-up was 56 days, far too short to pick up the harms that build with long-term use, and the review says outright that it cannot make firm statements about safety. The reported adverse-event ratio of 1.04 almost certainly understates real-world harm for the same reason. Only English, German and Dutch language trials were eligible. No individual drug in the class outperformed another.

Diclofenac at 150 mg a day, the most effective oral option in the network, also carried a raised risk of people dropping out because of side effects, so the ranking rewards a preparation more people stopped taking. Only English-language trials with at least 100 patients per group were eligible, which leaves the smaller ones out. This covers knee and hip osteoarthritis, not pain in general. Several authors declare consulting relationships with industry.

The review rates most of its own evidence as low quality, chiefly because the trials reported their methods poorly, and the studies disagreed with each other substantially on the main pain comparison. Fifty-nine per cent of the trials were commercially funded and a further 31% did not say who paid. No one drug in the class could be shown safer or more effective than another, because most head-to-head comparisons rested on very few small trials.

Most of the included trials were at high risk of bias or raised concerns, and all of them ran for a year or less, so nothing here bears on years of combined use. The result for ulcer complications, as opposed to ulcers found by endoscopy, had an interval crossing no effect, from 0.10 to 1.07. Comparisons against other stomach medicines rest on single small studies. The trials were published between 1996 and 2014.

Most of the underlying data come from wisdom-tooth extraction in otherwise healthy adults, a standardised research model rather than everyday pain. These are single doses measured over four to six hours and say nothing about taking something repeatedly. The overview ran no statistical test between the drugs, so these numbers sit side by side and were never compared. Some drug and dose combinations were judged unreliable because of susceptibility to publication bias.

The best diclofenac figure rests on two studies, and some of the ketoprofen data come from 1980s trials with loosely defined outcomes. Around 5,900 participants' worth of registered but never fully reported efficacy data was identified and could not be obtained, which is the direction publication bias usually runs. Only 220 people across all 61 trials were given a tablet for comparison, so the comparison with swallowing one is thin.

About half the people using the carrier gel alone also counted as a success, roughly twice the usual response to a dummy pill, so part of what people feel comes from the vehicle and the act of rubbing rather than from the drug. Every trial in the topical review was in osteoarthritis, almost all of it knee osteoarthritis. Up to 6,000 participants' worth of completed but unpublished data was unavailable to the reviewers. Trials with fewer than 100 patients per group were excluded from the network.

For the other oral decongestants pooled in that search, a week or more of daily use changed systolic pressure by 0.91 mmHg on average, at very low certainty. None of those trials used phenylephrine, so the figure cannot be carried over to it.

The trial's own title calls it open-label, so people may have known what they were taking. That usually inflates a self-rated benefit rather than hiding one. Everyone in it had hay fever, not a cold.

All four trials were in hay fever, so none tested a blocked nose from a cold. Two of them gave a single dose and followed congestion for 6 hours. The largest used a 30 mg slow-release tablet, not the 10 mg tablet sold in cold remedies.

In about 80% of the pooled trials people were already eating at least 1.2 g per kg a day, so this measures going from enough to more, not from too little to enough. The 1.62 figure comes with a range running from 1.03 to 2.20, wide enough to be a region rather than a line. Most trials ran six to twelve weeks. Effects shrank with age and were larger in people already training.

A null result is harder to establish than a positive one, and this one carries the weight of the finding. Strength was pooled across one-rep-max tests, isometric tests and handgrip, which do not always move together. The spline curves are smooth in a way the underlying trials are not. The search covered two databases.

The trial cannot separate how much of the benefit came from the weight loss and how much from the drug's other effects, since the treated arm got both. Adverse events forced 16.6% off the drug permanently, against 8.2% on placebo, in a group with strong reason to persist. Novo Nordisk funded it and employed several of the authors.

Both trials were funded by Novo Nordisk. Participants stayed on varying background diabetes medication, which is part of what separates the two groups rather than diabetes on its own. Gastrointestinal side effects reached 63.5% here against 34.3% on placebo.

This was a prespecified secondary analysis, not what the trial was built to answer. The 10.2% is lower than the obesity trials' figures because this group was older, already had heart disease and started at a lower body mass index, not because the effect wears off. People came off the drug more often than off placebo.

Persistence was inferred from gaps between prescription fills, which cannot tell someone who chose to stop from someone who lost their insurance or could not fill during the shortage. Only 53.6% of the initiators found met the continuous-enrolment rule, and people who keep coverage are probably the more persistent ones, so the real figures are likely lower still. The analysis was run by a pharmacy benefit manager. Commercially insured Americans only.

Novo Nordisk funded the trial and the drug is theirs. The cohort was largely female and largely white, so it says less about everyone else. Sixty-eight weeks is what was measured, and nothing here describes what happens after the last injection.

Three hundred and four people is small for a phase 3 trial, and 93% were white and 78% women. Gastrointestinal effects, mostly mild, reached 82.2% against 53.9% on placebo. Nobody was followed after the last injection, so this describes two years on the drug rather than two years of results.

A network meta-analysis compares drugs that were never tested against each other, so this ranking is inferred rather than measured. Fourteen of the 56 trials were semaglutide trials and most came from one company's own programme, which pooling does not correct for. Every figure describes people still taking the drug.

Everyone had already tolerated 20 weeks of dose escalation before being randomised, so the people most likely to quit were gone before the clock started. Lifestyle support carried on in both arms, which is not the same as stopping everything. Four in five participants were women.

These are the frequent and mild ones, gathered over about twelve weeks. The rare and serious events the label warns about, an erection lasting beyond four hours among them, are too uncommon for trials this size to count, so their absence here is not evidence of their absence.

The two figures come from separate trials rather than one comparison, so the size of the gap is indicative rather than measured. Men in the diabetes trial had lived with it twelve years on average, and response is not broken down by how much nerve or vessel damage they had.

Enrolled men skewed towards mild and moderate erectile dysfunction, so these figures sit at the optimistic end of what an unselected clinic sees. Success was recorded by the men themselves, in diaries. Most trials ran about twelve weeks, and almost all were funded by the manufacturer.

Six minutes of walking stands in for exercise capacity, not for living longer: the trial ran twelve weeks, was not powered for mortality, and found no significant difference in clinical worsening. It also enrolled only certain causes of raised lung pressure.

This covers only men who had both nerve bundles spared. Where the nerves were not spared, recovery is substantially worse and these trials say nothing about it. Eight trials, several industry-funded, with dosing and timing after surgery differing between them.

Ninety-six volunteers across eleven groups is about nine per arm, so a null in any one arm means no effect was detected at that size, not that none exists. Digoxin is handled almost entirely by the drug pump rather than by the liver enzyme, so these results speak to one half of the mechanism.

Forty-two men across six arms leaves about seven per arm, so each arm's estimate is imprecise. The preparations were picked to span the hyperforin range rather than to represent what sits on a shelf. Every arm showed some induction, including the near-hyperforin-free one, so this is a slope rather than a safe threshold.

The trials disagree with each other far more than a pooled number suggests, with heterogeneity running to 89 percent, and the reviewers downgraded every headline estimate because of it. Smaller trials reported roughly double the benefit of larger ones. Trials from German-speaking countries, where the extracts are manufactured, reported results more favourable than trials from elsewhere.

The large US trial most often cited as proof that it fails in severe depression enrolled people its own authors described as moderately severe, so it did not test the question either. No trial has been designed with severity as the thing being varied.

Trial size stands in for several things at once, including how carefully a trial was run and who paid for it, so this does not isolate a single cause. The 2026 estimate rests on three trials in 207 people, all of one proprietary extract.

The 35,562-patient figure comes from post-marketing surveillance never designed to detect interactions, which duly reported none. The review's searches stopped in 2003. Its conclusion is explicitly conditional on a doctor knowing the patient is taking it, which is the condition most over-the-counter use fails.

Every trial ran between 4 and 12 weeks, so nothing here describes a year on either treatment. These are comparisons against antidepressants rather than against placebo, so the equivalence inherits whatever each trial's drug arm was worth. None of the trials was built to catch rare side effects.

The shape comes from observational data, so it maps an association rather than a demonstrated causal curve. Nonlinearity cleared significance for recurrence only; for breast-cancer death and death from any cause it did not, and most of the reduction there sat between 20 and 40 mg a day.

Diet was self-reported, and soy eaters may differ from others in weight, diet quality and health behaviour that no adjustment removes. Only recurrence reached significance; both death-rate estimates crossed the line of no effect. Diagnoses ran from 1991 to 2006.

The tamoxifen-user estimate, unlike the non-user one, still includes no effect within its range of uncertainty, so the near-identity is between point estimates rather than between two equally firm results. Every study pooled is observational, with self-reported diet.

Forty-six of the 62 trials, three-quarters of them, carried a high risk of bias in three or more quality domains. Against a typical starting point of several flushes a day, four-fifths of one is a real change and a small one. The night-sweats interval is wide enough to rule out almost nothing.

The trials were too dissimilar to combine, so the review's only pooled estimate is of red clover, a different plant. Four trials of extracts delivering more than 30 mg of genistein a day did consistently reduce flushes, so the negative verdict is not uniform across preparations.

The largest of those changes is 7 mg/dL, under 5 percent of a typical level. Reductions were bigger in people who started with high cholesterol, so anyone starting from normal should expect less. Only 11 trials, none of them published after 2006.

The rise in thyroid stimulating hormone only just cleared significance, its interval reaching down to 0.001. The pooled trials disagreed sharply with one another. Everyone studied was thyroid-healthy and presumed iodine-replete, so the worry that soy matters more when iodine is scarce goes untested.

Most of these trials enrolled men by age rather than by a low testosterone reading, so many participants were not deficient. A gain in lean mass is not by itself a gain in anything a man would notice. In the 16-trial pooling, whose men averaged 77 years old, no musculoskeletal difference from placebo was left by 12 months, which could mean the effect fades or only that fewer trials reported that far out.

The relative risk of 8.14 runs across an interval from 1.87 to 35.40, which reflects how few men had the event rather than a precise estimate. It rests on four trials, and those trials needed to run only twelve weeks, short against how long testosterone is actually taken.

About 65 men were screened for every one enrolled, so these are not the men who walk into a clinic. One year of treatment is short against how long testosterone is actually taken. In the pooled analysis the sexual-desire interval ran from 0.01 to 0.34, close enough to zero to be worth naming.

The trial was powered for artery thickening rather than for sexual outcomes, so the null on sex is a secondary result rather than a definitive one. Its authors say plainly that it should not be read as establishing cardiovascular safety. Haematocrit and prostate-specific antigen both rose more on testosterone.

Men had to have a partner willing to attempt intercourse at least twice a month, which selects a particular and comparatively motivated group. Twelve outcomes were tested, so some of the ten significant ones may be the arithmetic of testing many things. And the missing threshold cuts both ways: it undercuts any single diagnostic cutoff, and it also means no reading identifies in advance who will benefit.

This is a snapshot, so it can say which symptoms travel with low testosterone and not whether raising it would relieve them. That fatigue and low mood failed to cluster is a statement about how symptoms group, not proof they have nothing to do with testosterone. Testosterone here was measured by mass spectrometry on morning samples, a stricter standard than much routine clinical testing, so the thresholds may not transfer to an ordinary lab result.

Thirty-four people finished the trial, but the treatment effects were computed on 26 — the ones who had actually reacted to the task at baseline, picked out after the fact. An industry grant funded the study, and the paper says the funder had no hand in the analysis. The other ingredients were phosphatidylserine, alpha-glycerylphosphorylcholine and chamomile, each in a small dose. Why the two effects arrived two hours apart is unexplained. Performance on the task itself never moved.

Thirty-six people enrolled and 34 finished, but the effects were computed on 26, chosen after testing and before unblinding for having reacted to the task at baseline. Four people gave too little saliva and were left out of the cortisol analysis, and the paper does not say how many remained. The drink was not theanine alone: it also carried 25 mg of alpha-glycerylphosphorylcholine, 1 mg of phosphatidylserine and 10 mg of chamomile. Before dosing and at one hour the task did not reliably raise cortisol, so the rise the drink blunted appeared at one session out of three. A grant from a company, Neurobrands, paid for the trial, and the paper says the funder had no input into its design, conduct or interpretation.

This was whole matcha, theanine and catechins and caffeine together, so no result here belongs to theanine specifically. The two results that moved were secondary outcomes, which carry less weight than the ones the trial was built to test. And everyone in it already had cognitive decline, which is not a healthy adult drinking green tea.

Every outcome here is what people reported about their own sleep; nothing was recorded by a wrist monitor or in a sleep lab. Few of the pooled trials tested theanine on its own rather than mixed with other ingredients, so its individual share stays blurred. And the authors say the best dose and duration are still undetermined.

Twenty-six of the 34 who finished were analysed, and they were the ones who had reacted to the task at baseline. The drink was not theanine alone: phosphatidylserine, alpha-glycerylphosphorylcholine and chamomile came with it, in small doses. An industry grant funded the trial, and the paper says the funder had no hand in its design or analysis. A brain-wave finding in the same trial turned up only in the more anxious participants and did not track the stress or cortisol results, so it stands unexplained.

The outcome is relapse, and in the underlying trials relapse was not reliably told apart from withdrawal. Slow means longer than four weeks, not the months-long schedules now advised. Venlafaxine was not separated from other antidepressants. The therapy comparison rests on two studies.

Only five comparisons split the symptoms this way, out of ten trials in the pooling. Severity was measured inconsistently, in some trials by counting days spent indoors, which mixes how ill someone felt with whether they could take time off work.

A guideline grades other people's evidence rather than producing any, and most of its positions are worded as suggestions, which is how low certainty gets expressed. It sets aside anyone with an established medical reason to be tested or treated, so it speaks only about people without one. Two corrections to it were published in 2025 and its paediatric advice has drawn published rebuttals.

Each of these trials recruited a different population and measured a different outcome, so what they share is a pattern rather than one replicated result. Three of the four used large intermittent doses, a schedule some analyses argue is the least effective. In one of them, the group whose levels reached the recommended target fell more often than the group whose levels did not.

The outcome in both studies is a blood level; neither measured whether anyone felt or functioned better. The 87 percent figure comes from 33 people on a single-blind protocol at a weekly dose, which makes it a statement about that regimen rather than about a daily D3 tablet against a daily D2 one. The pooled comparison searched the literature only to July 2011.

Participants were picked for being unusually healthy for their age, with no major health event in five years and good mobility, so a frailer group might respond differently. Testing six outcomes at once, the authors demanded a p value below 0.01, a strict bar that would hide a modest real effect. Three years is short for both fractures and cognition.

Nobody was recruited for having low vitamin D, low bone mass or osteoporosis, which is exactly the group the supplement is normally prescribed to. Fractures were reported by participants once a year and then checked against medical records. Hip fractures were few enough that their estimate spans a 30 percent reduction to a 47 percent increase.

Both trials gave very large intermittent doses, so nothing here suggests a daily tablet carries the same risk. The falls trial enrolled 200 people and had no placebo group, comparing one dose against another rather than against nothing. The annual-dose trial's fracture interval starts at exactly no effect, and its explanation for the timing, a spike in blood levels just after each dose, is a hypothesis it raises rather than one it tests.

Neither trial screened people in for low vitamin D, so both test adding more to people who mostly had enough rather than correcting a shortfall. The Australian participants averaged 77 nanomoles per litre before they started, in a country with a great deal of sun. Five years may be short for cancers that take longer than that to surface.

The over-75 recommendation rests on a very small reduction in deaths, graded as high certainty but never translated into a number of lives. Most of the guideline's positions are worded as suggestions rather than recommendations, the wording used when certainty is low. The trials behind it used widely varying doses and often let participants keep taking their own vitamin D, so the right dose is left unresolved.

Baseline protein intake was not consistently controlled across the trials, so some of these people may already have been eating enough for a supplement to add nothing. The handgrip analysis was highly inconsistent between studies. Appendicular muscle mass did improve, but on two trials that disagreed sharply with each other. Older adults only, so this says nothing about young trained people.

The height gain is a standardised effect of 0.12, which is real across a population and invisible in any one child. The diarrhoea figure came from 19,468 children while the mortality analyses carried over 130,000 each, which is part of why the nulls in this review are the more certain results. The trials ran mostly in low- and middle-income countries where zinc shortage is common.

The children were mostly under five, in low- and middle-income countries, on 10 to 15 mg of zinc a day for a median of 26 weeks, so this is not evidence about supplementing a well-fed child anywhere else. The review's own summary describes zinc as probably reducing deaths from chest infection and from malaria, but both of those ranges cross into no effect and the authors add that they could not rule out an increased risk of death. A null on mortality is not a null on everything the same trials measured.

Nine trials and 1,449 people, graded low certainty. The review pooled lozenges, syrups, tablets and nasal gels as one treatment, with the gluconate lozenge doses alone spanning 45 to 276 mg a day. It also excluded any product mixing zinc with other vitamins or herbs, which is most of what a pharmacy actually sells.

The figure comes from roughly a week of high-dose lozenges taken during a cold, not from a daily supplement taken for months. It counts non-serious events only, and no treatment study recorded a serious one. Sixteen trials contributed, most at unclear or high risk of bias in at least one area.

Most of these trials ran in Asian countries where zinc shortage is common, and the review's authors say plainly that the evidence does not support giving zinc to well-nourished children or in low-risk settings. Vomiting rose in both age groups, by about 57% in the over-sixes, which is a real cost in a child already losing fluid. Death and hospitalisation, the outcomes that matter most, sit at very low certainty and are not settled here.

Neither finding comes from a trial designed to test the pairing: one is a subgroup pattern and the other a set of marker outcomes graded very low to moderate certainty. The usual explanation, that two similar metals compete for the same route into the body, is not tested by either source. All of it concerns children given substantial doses of both minerals in settings where deficiency is common, not the small amounts in a multivitamin.

This was an open treatment series with partly historical comparisons, because withholding an antidote to build a control group was not available. It dates from 1988 and the regimens have changed since, most notably to a shorter two-bag drip. The time bands rest on when people said they took the tablets, which is unreliable after an overdose and missing altogether when doses were spread across days. What survives all of that is the steepness of the gradient, which later guidelines reproduce.

One specialist transplant unit between 1992 and 2008, so these are the most severe cases and the mortality figures are not overdose mortality generally. Ingestion times and patterns were reconstructed from records after the fact. The repeat-dose group did die more often, 37.3% against 27.8%, but that gap did not hold up once age and alcohol misuse were accounted for, so it travels with those rather than clearly causing the worse outcome. Treatment of both patterns has changed since.

One urban public hospital clinic where 78% of parents had limited health literacy, so these are not general-population error rates. The benefit was concentrated in that group and was not detectable among parents with adequate literacy. It was a supervised measurement task rather than several days of dosing at home, and no child was harmed in it. The product tested was the older concentrated infant drops with a dropper, a format the United States has since changed.

Neither number marks a point where injury begins, and reading either as one is the most common mistake made with this drug. This is one product from one manufacturer in one country: other acetaminophen products, Tylenol's own included, print different strengths and different ceilings, and labels outside the United States differ again. The two figures also count differently, one per product and one across everything taken, which is exactly what makes them easy to swap.

None of these three figures is the dose at which injury begins, and the 10,000 mg line is set deliberately below where harm is expected so that clinicians act early. That threshold is also 200 mg per kilogram of body weight wherever that comes to less, so it is lower for a smaller person. It is Australian and New Zealand guidance, and the United Kingdom and United States set their own numbers. The 24,000 mg median describes people who had already reached liver failure, not a dose at which anyone should expect to.

All of them were older outpatients with diabetes, many on metformin, so this is not a general older-adult sample. Forty-four percent already had questionable dementia at the start, which may put some of the decline beyond reach. Borderline-low is not outright deficiency.

This is an association measured at a single point in time, so it cannot show that the fortified food did the lifting. The thinned stomach lining was inferred from a blood ratio rather than confirmed by biopsy, so the group is drawn approximately.

The survey figures were reviewed in 2009 and predate current fortification and supplement habits. The 148 pmol/L line between deficient and marginal is itself contested, and moving it moves both percentages. The review was written to argue for fortifying flour, a stated policy position.

This measures what was in the blood, not whether anyone took a supplement, so a powder started after the draw would be invisible. The intervals are wide enough to include modest risk either way, leucine running from 0.57 to 1.33. Estimates below 1 are not evidence of protection.

Only five of twelve studies could be pooled, and the main analysis compares before with after inside the supplemented group rather than against a control, so it cannot separate the supplement from everything else that changed. Cirrhosis is a malnourished state. None of this transfers to a well-fed person.

Ten trials is a small base, and the review reports no pooled effect size at all. Soreness is self-reported. The dose ceiling is where the evidence stops, not where a harm begins, and the review is explicit that it cannot speak to anything above it.

A food-composition database, not a study: it says how much is in what, and nothing about what these amino acids do. Values are averages across sampled products and shift with cut, brand and cooking loss. The figures are per 100 g, not per portion.

Ten men, one four-hour window, and all of them already trained. The comparator was placebo, so this shows branched-chain amino acids beat nothing rather than that they match protein. The authors put the rise at roughly half what intact protein produces in similar designs, which this study never measured directly.

An overview of reviews inherits every weakness of the trials underneath it, and the same small primary studies are counted inside several of the eleven. The authors grade those reviews and report the recurring flaws: no justification for excluding studies, and no funding disclosure for the primary trials.

Fifteen men per group, healthy and aged around 71, at rest with no exercise. The doses are not matched: 6 g of branched-chain amino acids against 30 g of milk protein is the choice a shopper faces, not a gram-for-gram test of which does more. Nothing here was measured beyond five hours.

The direction rule is settled; the amount of biotin needed to trigger it is not. That threshold differs by manufacturer, by test and by how recent the reagent is, and cannot be predicted from one analyser to another. The panel study behind it added biotin to pooled serum in a laboratory rather than measuring people who had swallowed any.

This is a reference chapter, and it reports no incidence figure. The high doses used here treat a specific inherited fault in recycling biotin, and say nothing about biotin in people who do not have it. Reading across from this condition to general supplementation is precisely the error its existence invites. This reflects the chapter as it stood in December 2025.

A clinical reference chapter rather than primary research: it synthesises rather than measures, and puts no number on how rare rare is. The at-risk list is qualitative, with no estimate of how much any of those exposures raises the risk. These chapters are revised continuously, and this reflects the text as it stood in September 2025.

The dose figures are what the agency was aware of in 2019, not a survey of what is on shelves now. The 0.03mg intake figure is itself extrapolated from the biotin content of human milk rather than measured against a requirement, and no Recommended Dietary Allowance was ever set, for lack of evidence.

The table names devices the agency judged had not addressed the risk, and puts no number on the size of the error on any of them. Several entries are point-of-care strip tests rather than the laboratory analysers most hospitals run, and some clearances date to the 1990s. The page carries no version marker, so the count of fourteen is a snapshot of one day in September 2026.

These are case reports, the weakest design there is: uncontrolled, unblinded and uniformly positive, which is the pattern publication bias produces. No randomised trial appeared in the search at all. The conditions involved are individually rare, so the group in which biotin did help is very small.

Three studies is a very thin base, and the review is limited by that rather than by its method: PubMed only, no registration, no pooled estimate. What it found is an absence of demonstrated benefit in people who are not deficient, which is not the same as demonstrated absence of benefit. The authors are practising dermatologists writing about a supplement their patients ask for.

23 samples, and the exposure is multiple sclerosis treatment at doses far above any hair supplement, so the blood levels here are an upper bound rather than what a shopper reaches. Two authors worked for the company developing that high-dose product, which had a commercial interest in characterising this interference. Roche assays only, and the removal step is not perfectly neutral: it shifts parathyroid hormone by about 8%.

Absence of reported harm is not demonstrated safety; it is what an unset upper limit actually means, and the panel that could not set one said so. That panel was writing in 2012 about intravenous nutrition products, before the interference literature existed, and is silent on blood tests.

A single case, paired with a review of other single cases: this shows that something can happen, never how often. The interference was inferred from how fast the results normalised rather than by measuring biotin in the sample. Assay formulations from 2016 have since been revised for biotin tolerance, and this patient's own dose is not a market-typical exposure.

Nobody in this study was misdiagnosed; what was measured was the assay, not a patient outcome. Only three assays were tested, so it cannot say what any particular hospital runs. Whether interference of this size changes a decision depends on where a person's true troponin sits relative to the cut-off, which the design does not test.

Nobody in this experiment had swallowed any biotin; it was added to plasma in a laboratory, and added biotin may not behave quite like biotin absorbed alongside its own breakdown products. The two highest concentrations tested sit far above what ordinary supplement users have been shown to carry. What was measured was the size of the error, not a missed diagnosis.

Something else in cola, or about the people who drink it, may carry the risk, and the authors could not say what. Caffeine intake was not flat either: women in the middle fifth had 12 to 13% higher risk than the lowest. Both intake and diagnosis were reported by the women themselves.

With 17 men on coffee, the trial was sized to catch a difference of 220 ml of urine or more, so a smaller loss could have gone unseen. Larger doses were not tested, and the researchers note a measurable effect on urine is likely once a single dose passes about 300 mg of caffeine. The men were fully hydrated and had fasted overnight. The litre went down in half an hour, not as cups spread across a day.

Twenty people, run by the maker of the cannabidiol formulation, and built to measure blood levels rather than outcomes — it establishes the exposure change, not its consequences. The dose is a prescription epilepsy one; whether the same happens at consumer amounts was not tested.

The prescription figures are set per kilogram of body weight; the milligram equivalents here use the label's own 75kg reference, and many of the patients in these trials are children. The consumer average has a standard deviation of 40.7mg, nearly as wide as the mean itself.

Single doses in a residential unit, about six people per formulation, so this is not a test of daily use. Cannabidiol did not convert to THC by any route tried here, but only the routes and formulations tried are covered.

This is the agency's stated position under US federal law, not a court ruling and not a description of other countries or of individual states. The page was last updated in July 2024 and the law in this area moves.

Urine, not blood, so these concentrations cannot be set beside the plasma figures on the prescription label. About six people took each formulation, and each took a single dose in a residential unit rather than using the product over time.

A pilot of thirty people cannot rule out a small effect. One objective measure did move — sleep efficiency favoured cannabidiol at week two — while every measure the sleepers reported themselves did not. And 150mg nightly is about three times what people typically take.

Each study sets its own tolerance — a tenth either way here, a stricter band for the drinks — so the percentages describe different tests and should not be pooled. The samples are snapshots too: one online channel in 2016, one US state later, with single-figure counts inside some categories.

These rates come from patients with severe epilepsy on cannabidiol at 10 to 25mg/kg/day, far above what consumers take. NIH's LiverTox reference gives 21% for the valproate-and-clobazam group where the label gives 30%, attributing its figure to extension studies rather than to these trials.

Both are dated regional snapshots — products bought online in 2016 and in one US state in 2020 — and neither describes what is on shelves now. The synthetic cannabinoids turned up in vaping products specifically, so nothing here says the same of oils or edibles.

Both tests soaked teeth in the drink for a day or a week at a stretch, with no saliva rinsing between sips. In one of them the damage did not track a drink's measured acidity at all, so phosphoric acid's own share never separates from the rest of cola. The seven-day test used bovine enamel, a stand-in for human enamel rather than the thing itself. Twelve of the nineteen studies in the review are lab work too.

The joint trial tested one branded product, and two of its authors work for the manufacturer and for the contract firm that ran it, so an independent repeat is still missing. Several outcomes in the largest synthesis, skin roughness among them, came out flat.

Finding the fragment in blood shows it survives digestion; it does not show it did anything afterwards. The signalling step rests on animal and cell-dish work rather than on the human trials the reviews cover. That headline pools 113 trials through 16 earlier reviews, so its precision is theirs rather than its own, and several outcomes within it, including skin roughness and two joint sub-scores, came out flat. The peptide work came from a research institute owned by a collagen manufacturer.

These are absorption curves for a few marker fragments, not a clearance time for the ingredient as a whole. Collagen from different animal sources may behave differently. And finding the tough fragment in blood shows it lasts, not that it does anything while it is there.

The pharmacology paper is a review and reports no human half-life figure. Its interest is in engineering cordycepin into something usable, with an enzyme inhibitor, a redesigned molecule or a new delivery system, none of which exists in a capsule. And it never asks the separate question of whether the amounts in a supplement would matter even if they did survive.

Two commercial supplements were tested, so this describes two products rather than a market. The digestion was simulated in a laboratory model, which estimates what could be released into the gut, not what reaches the bloodstream or any tissue. No health outcome was measured anywhere in the study.

Ten samples in total, five per group, which is small for a statement about a market. Two of the authors work for a company selling authentication testing, which is a commercial interest in the standard markers failing. The paper set out to propose better markers, so that failure is its starting premise rather than a measured rate.

Both are method papers: they show substitution happens without reporting how often it happens across a defined slice of the market. The products were bought in Asia, so the rate may not carry over to Western shelves. And DNA says which species is present, not how much of the product it accounts for.

The review appraised each study for bias but never pooled them, so there is no combined estimate here, only five separate results. It covers Cordyceps militaris alone, leaving out both the wild caterpillar fungus and the fermented mycelium products that fill much of the retail shelf.

This is a review of the Chinese industry rather than Western retail, and it describes those names rather than counting how often they appear. One author works for a cordyceps manufacturer, though the paper argues against loose labelling rather than for it. The authors say two of the fungi they discuss need further study before their place is settled.

Forty people took the dose, and each took it once, so this shows what one sharp rise does rather than what months of raised cortisol do. The loss was confined to recalling facts and handling space, which leaves much of what memory does unaffected.

The group was small and elderly, and the published summary never says how many people were in it. In humans this is a correlation; the evidence that cortisol itself damages the structure comes from the rat work the same paper cites. Ageing or illness could equally shrink that structure and raise cortisol at once, which would draw this same pattern with no link between them.

One hospital, 250 people. Facing major surgery can itself distort cortisol patterns, so this may not reflect ordinary life. A flatter drop may also mark someone who is already sicker, rather than adding risk of its own.

This is a narrative review drawing human and animal work together, not new data. And the arrow could run either way: a stress-disrupted cortisol system might raise autoimmune risk, or years of autoimmune illness might be what disrupts the system. Nothing here separates the two.

The authors describe the effects as small. Everyone measured was young and healthy, at an average age of 20, so this says nothing about older adults or about anyone with a sleep or stress disorder. And these are the same people watched over time rather than an experiment, so something else in their days could be driving both ends.

These were tiny experiments, 13 men in one of them, run over one or two nights. All the participants were young men, so nothing here covers women or older adults. And a night or two of lost sleep is not years of it, so how far the flattening goes over a lifetime is unmeasured.

This is what one US label states, not a measurement of harm. Nothing here counts how often people do stack a second acetaminophen product on top, only that the label names it as the thing to avoid.

The two milligram figures are worked out from published molecular weights rather than quoted from a label, so they are close but should not be treated as an exact conversion factor. Both label versions were read on 6 September 2026, and printed strengths do change.

The source is a clinical reference chapter summarising other work rather than a study. It names these effects without giving a rate for any of them, so this says what the drug does at ordinary doses, not how likely each is in a given person.

A panel of experts weighs evidence and then votes, so strong records a committee's confidence rather than an effect size. The list carries its own exception: treating a severe allergic reaction may still be appropriate. The harms bundled into one line run from dry mouth to contested dementia evidence.

The instruction sits on the allergy liquid. The dedicated sleep aids handle the question differently, by excluding anyone under 12 outright, so the sentence belongs to this label, not to diphenhydramine labelling in general. A label requirement is not evidence about how often harm follows.

Twenty-one people, about seven per group, with standard deviations of 4.2, 2.5 and 1.8 hours around those means. The dose was 1.25 mg per kilogram, roughly 86 mg in the adults, well above the 25 to 50 mg a tablet gives. A separate trial at 25 mg found no age difference at all, most likely because of that gap.

A label count is not a product count and not a measure of how many people take these. The same physical product carries several records across repackagers and store brands, so duplicates sit inside that figure. Counts drift as labels are added and withdrawn, which is why the date belongs with the number.

A label records what a regulator permits a company to print, not what the evidence shows. The dependence claim is narrowly defensible: diphenhydramine is not a controlled substance, and the one trial that looked found no rebound insomnia on stopping. This is one product from one manufacturer in one country.

These are calls to a poison centre, so mild poisonings nobody phoned about are missing and the mix looks worse than all ingestions would. The amount swallowed is what someone reported afterwards, unreliable in both directions. Above 1.5 g, coma and seizures grew more common while delirium did not.

The search deliberately began at 2003, so three is a count inside that window rather than of the whole literature. One database was searched. The second author worked for a company selling over-the-counter sleep products, melatonin among them, though its own diphenhydramine line comes off badly too.

Fifteen healthy men, no women, nobody with a sleep problem. The dosing was twice daily and the outcome was daytime sleepiness, so a nightly bedtime habit is an inference rather than something this trial watched. Four days shows tolerance arriving and says nothing about weeks.

The 300 mg figure comes from a clinical reference chapter for the allergy indication, not from an over-the-counter label, so quoting it about sleep would misdescribe what those products allow. The 50 mg figure is read off one US sleep-aid label, and other markets word their limits differently.

The label tells users not to drive while taking it and says nothing about the following morning, which is where the driving evidence sits. One product from one manufacturer in one country.

This was read off US label records on 6 September 2026, for one brand family. Brand compositions change, and a product carrying the same name in another country may contain something else again.

A position stand is expert consensus with no stated search and no risk-of-bias grading, and panels in this field have included researchers with sports-drink funding. It was written before the 2015 consensus on exercise-related low blood sodium and before the field studies that found sodium supplements did not prevent it. Advising people to individualise because sweat varies is not evidence that individualising improves any outcome, which is the step the personalised-hydration industry takes.

These are general-population reference values, not sport nutrition, and they describe what a healthy adult is thought to need over a day rather than what anyone should drink during exercise. The salt equivalent is arithmetic from the sodium figure, not a separately measured value. The report's own opening observation is that what you eat does not usually move your blood electrolyte levels, because the kidneys and hormones hold them steady.

One real exception sits inside the largest of these studies: among American footballers alone, sweat measurements did carry a weak signal, and in the other ten sports none did. Beyond that, cramping was self-reported throughout, and serum electrolytes were sampled before and after a race rather than inside the muscle at the moment a cramp struck, which is where the theory actually locates the problem. Forty-three crampers is enough to rule out a dominant effect, not a small one.

Twenty crampers is a small group, and the finish-line enzyme gap is borderline at p = 0.034 with spreads wide enough to swallow it. The blood was drawn after the cramp, so muscle damage could as easily follow a cramp as cause one. Crampers and non-crampers only diverged in pace from 25 kilometres onward, which points back at effort. The paper's own abstract calls 20 of 88 a rate of 24 percent; it is 22.7 percent.

Five desert stage races, which is what extreme heat does across six days, not what a hot afternoon does. Only 174 of the 266 runners enrolled gave a post-race sample. The link between hot races and low blood sodium is not statistically reliable, its interval running from 0.9 to 25.9; the link with high blood sodium is much firmer. And the 4.7 kilogram average carries a standard deviation twice its own size.

Both are narrative reviews with no stated search and no risk-of-bias grading, so which literature they cite is at the authors' discretion. The lead author of one chaired the consensus conferences on this condition, which makes the summary authoritative and not independent of the position it summarises. 'Conclusively identified' is their own confidence wording, not a graded rating.

One race, one temperate day in 2002, and a blood sample at the finish line rather than an experiment. Only 488 of the 766 runners who signed up gave a usable sample, so the 13 percent could sit either side of the truth. What people drank was reconstructed from a questionnaire they filled in after four hours of running.

These are the cases somebody wrote up. Severe and unusual presentations get published and mild ones do not, so this describes the case record rather than everyone it happened to. The search was a manual sweep of two databases, not a registered protocol. Women appear more often as a share of participants, not in greater numbers, a distinction the review itself insists on.

This concerns water, not electrolyte products, which were not tested here, so it undercuts the theory the hangover sachets rest on rather than testing the sachets themselves. It is a narrative review with no protocol or stated search, by a group that has argued this position for years. Much of the underlying evidence is people recalling their own drinking and their own hangovers. The authors do report that dehydration after alcohol is real, just mild and short-lived.

Fully hydrated, fasted men, all 72 of them male, drinking a litre in half an hour. Retention was measured as urine volume over four hours, not as anything a person would notice. And milk brings protein, fat and lactose along with its sodium and potassium, so what did the holding is not separated out here.

This is a reference-intake document for the general population, not exercise guidance, and it says nothing about what to drink during a race. The 2,300 milligram figure is a chronic-disease risk-reduction value introduced in the same 2019 report, not a toxicity threshold, and the child and adolescent versions were extrapolated from the adult one rather than measured. Anyone whose kidneys or medication affect how they handle potassium sits outside all of it.

This is transport physiology rather than a trial, and the two-to-one ratio is standard textbook mechanism restated by the review, not a new finding of it. The review was written to argue for developing drugs that improve on rehydration therapy, which frames how generously that therapy is described. Its own authors note the treatment reverses dehydration without shortening the illness or reducing stool output.

A survey of what runners believe; nothing in it tests whether the capsules work. Participants opted into a longitudinal tracking study, so they are more engaged with the sport than average, which makes the belief's persistence more striking rather than less. One country dominates the sample, one sport, 2018 data. The authors' verdict that the evidence is settled rests partly on their own earlier studies.

A chemical identity record, not a study: it says what these compounds weigh and nothing about what they do in a body. Both percentages were calculated from the retrieved molecular weights rather than read off the record, so they carry that rounding.

The men were fully hydrated and fasted when they drank, which is nobody's state when they reach for one of these, and retention behaves differently in someone already short of fluid. All 72 were male. The outcome is urine volume over four hours, a stand-in for retention rather than for feeling or performing better. The drinks were not matched for sodium, so this ranks these particular products, not electrolytes as an idea.

The analysis came out of a sports-drink manufacturer's own sweat-testing database, which cuts against the simplest commercial reading but travels with the finding either way. The people tested had chosen to come for sweat testing, so they skew athletic. Two of the three models used forearm patches, an indirect estimate; the whole-body method covered 48 tests. Unexplained variance includes measurement error, so this shows these variables do not predict sweat sodium, not that nothing could.

Twenty-five percent of twelve people is three people, so the familiar claim that a quarter to a third of Westerners are producers is consistent with this study rather than established by it. Its lasting contribution is the test, not the prevalence, and the work was done at a company that sells soy.

The clinical cases that established this used folic acid above 5mg a day in the 1940s and 1950s, roughly ten times a modern supplement and far above what fortified flour delivers, so it does not transfer straight to a 400 microgram tablet. The authors of the review say themselves that the modern evidence is correlative or from uncontrolled observation, and that their proposed mechanism has not been tested. They are also the group that proposed it.

The high-dose figure is deliberately absent here. The Canadian guideline states its tiers qualitatively rather than in milligrams, so no number can honestly be quoted from it, and an old trial dose is not a current recommendation. Both documents are graded syntheses of other people's work, resting partly on expert opinion where trials do not exist, and both address populations whose flour has been fortified since the 1990s.

None of this was randomised. These are before-and-after comparisons, so anything else that changed over the same years counts too: prenatal screening improved, supplement campaigns ran alongside the fortified flour, and registries differed in what they counted. What this design cannot do is separate the flour from the campaign that arrived with it.

A risk factor is not a diagnosis, and nothing here shows that carriers need a different supplement or a different dose. Pooled case-control genetic studies are prone to small-study effects and to population mismatch between cases and controls, so a significant pooled result is much weaker than the same result from trials. No odds ratio is reported, so the size of the association cannot be stated. The association did not hold in mixed populations, and nobody has explained why.

The 121 million figure is a model, not a count, built from country surveys of uneven quality and published with an 80% uncertainty range of 112.8 to 131.5 million. It also says nothing about who was or was not taking folic acid. The week-four timing comes from a clinical reference chapter that gives no study for it.

Milder knees showed a trend toward benefit with no significance test attached to it, so it is not a result and should not be reported as one. The trial used glucosamine hydrochloride rather than the sulfate salt, which is one of the two ways it differs from the three-year trials reporting a benefit.

This is one manufacturer's label, so its choices are evidence about this product rather than about US labelling rules in general. The document also disagrees with itself: the index title says tablet where the Drug Facts panel and carton say capsule, and three different effective dates appear in it.

Three in ten had knee pain without X-ray-confirmed osteoarthritis, so this is not purely a trial in established disease. Twenty-four weeks is short for a structural endpoint when the trials reporting benefit ran three years. It used a hydrochloride drink, which sulfate proponents reject.

Fifteen people and a single dose cannot rule out rare reactions, and the authors limit their conclusion to the manufacturers' products they tested rather than to glucosamine in general. Nothing here tests a poorly purified product or daily use. Chondroitin comes from cattle or shark cartilage.

Thirty-two people finished that trial, and the gap between glutathione and placebo ran close to the significance line, so the honest reading is no proven benefit rather than proven uselessness. The August recall reports came from the pharmacies themselves rather than independent checking, and neither notice names a death or a hospital admission.

This is a second analysis of the same 31 participants, not an independent replication, so it inherits that trial's size and design unchanged. A null result across the markers measured narrows the candidates rather than showing that no mechanism exists. Muscle fibre size specifically was unaffected, which sits awkwardly next to the whole-muscle scan difference the earlier report described, and the authors do not reconcile the two.

This is a reference monograph summarising labelling and formulary convention, not a study of what either ceiling does to anyone. The figures are United States labelling and differ by country. Monograph entries are revised continuously, so a later version may read differently.

Eleven trials and 312 people in total, and the authors name the small number of studies as their main limitation. The search closed in August 2013 and nothing comparable has replaced it. No pooled dose is reported, so these results should not be read across to the intakes used in polycystic ovary syndrome supplements.

A registry entry records that a trial was declared, not what it found, so none of these counts is evidence about how inositol works. Trials often publish in journals without ever posting back to the registry, so a raw non-posting count overstates true non-reporting. This registry is US-centred and under-covers the Italian and Indian trials that dominate this field.

The subgroup finding that carries this claim is reported without a formal test of interaction, so it is suggestive rather than demonstrated. Six of the eighteen studies had no control group, and their effect sizes run two to three times larger, which is what an uncontrolled design predicts. Only 204 of the 1,408 participants were children or adolescents.

These are modelled ranges built from intake data and isotope studies, not measurements taken in one population, which is why they are given as spans rather than single numbers. The review is a narrative one and lists the nature of the meat factor, and the roles of vitamin A and carotenoids, as still unresolved.

Poison-centre and poisons-service data are voluntary referrals rather than a census, so a count of zero deaths describes the cases that reached them. The UK cohort counts hospital referrals, which would miss a child who died before or too soon after arriving. And the US figures stop in 1998, straddling the 1997 packaging rule without separating out its effect.

This is a narrative review, so which studies went into it was the authors' judgement, and they work at the laboratory that produced much of the underlying isotope work. The two-week comparison rests on a single cited study. And single-meal effects are real, so this deflates the advice rather than cancelling it.

Every study underneath these means was observational. The 70 micrograms per litre comparator rests on three studies and 151 people. The inflammation correction is a statistical adjustment to survey data rather than a check against bone marrow, and the two proteins it needs are not measured alongside an ordinary ferritin test.

This is one manufacturer's label. Other iron salts — fumarate, gluconate, bisglycinate — deliver different amounts of elemental iron per tablet, so the ratio here is not a conversion factor for every product. The label itself carries three disagreeing revision dates, so treat it as a record of what a shopper is told rather than of when they were told it.

The review mixes randomised trials with observational studies. Its haemoglobin and ferritin figures came out with heterogeneity of 91% and 87%, which makes those pooled numbers too unstable to quote as point estimates. What is left is a benefit measured in blood tests and a diagnosis avoided, not in how the pregnancy or the baby turned out.

The authors read the older peak as intentional rather than accidental, which is an interpretation of the age and sex pattern rather than something the referrals recorded. These are hospital referrals to a poisons service, not all iron overdoses. And how much iron someone said they had taken predicted neither their measured blood level nor their symptoms, so the exposure data here are unreliable.

This concerns routine supplementation in women who are not anaemic, not the treatment of a diagnosed deficiency, where the case is different. The reviewers name the practical consequence themselves: international guidelines on this point disagree with each other. The same review mixes randomised trials with observational studies.

A median is not a person: half of women eat more than 12 milligrams and half less, and falling below a recommendation is not the same as being deficient. These intake figures were compiled for a 2001 report and have not been revised since. The absorption assumptions behind the recommendations predate what is now known about how the body regulates uptake.

People with hereditary haemochromatosis were deliberately left out of the derivation as exceptionally sensitive, so this figure does not protect them. The same 45 milligrams was applied to pregnancy only because data on lower doses in pregnant women did not exist. The chapter was published in 2001 and has not been revised since.

A single blood draw cannot say who felt unwell or who would have benefited from treatment. The headline figure moves with the cutoff chosen, and below 30 is one of several in use. No inflammation adjustment was applied, which by other work would push the estimate higher. The authors say what functional deficiency means in an otherwise well adult is still unresolved.

The abstract never says how many men took part, which is the first thing you would want from a dose-response study. The adverse signal is a blood chemistry change, not a symptom anyone reported. Healthy young men, fed briefly, so nothing here reaches long-term intake.

Nine trials is a small pool and the searches ended in December 2013, so a decade of later work is missing. The two outcomes are measured on different clocks: synthesis takes hours, lean mass takes weeks to months, so part of the contrast is a contrast of study durations. Older adults only.

Eight men per arm, one acute measurement, nothing past four and a half hours and nothing on muscle size. Part-funded and co-authored with a commercial protein-beverage research centre, a direct interest in a result about beverage formulation. A fifth arm added isoleucine and valine, and its value was never published.

Sixty-eight people were recruited and 41 finished, against the researchers' own target of 60. The company that makes the capsule funded the trial and pays five of the authors. Three of the four people who quit over side effects were taking lion's mane.

The mycelium here grew in stirred liquid, not on grain, so this settles which tissue makes erinacines and not how much of a grain-grown powder is mushroom. Only four erinacines were measured, and the assay's detection floor is not stated.

This paper set out to test safety, not to measure composition, so the oat description is an incidental disclosure rather than a finding. No independent laboratory has ever assayed a commercial lion's mane product and published the result.

This panel carries no new data of its own and was answering a narrow question about one disease, so its refusal to recommend lion's mane there is not a verdict on memory claims in general. What does travel is its flat report that the memory finding stands unrepeated.

People who eat mushrooms differ from people who do not in diet quality, income and how much they move, and any of those could be doing the work. The observational half covers all mushroom species, so none of that signal belongs to lion's mane in particular.

The review's verdict that lion's mane is the most promising of the four is a ranking inside eleven papers, partly earned by having the most of them and by being safer than fly agaric. Several of those five come from one research group, so they are less independent than the count suggests.

These are not four points on one scale. Fruiting-body powder and mycelium are different tissue carrying different compounds, so a gram of one is not a gram of the other, and the range above is what has been tried rather than what has been optimised.

The same review sorts those eighteen ingredients by evidence and puts lion's mane in its best-supported group, but that sorting is one team's judgement with no stated inclusion rules and no bias assessment. Its own closing paragraph says the category lacks a solid scientific foundation.

Nobody swallowed anything: these were extracts applied to platelets in a tube, at concentrations matching no serving or capsule. The same paper's introduction cites an earlier report that a lion's mane extract did interfere with one platelet receptor, so even the tube evidence is split.

2000mg per kilogram was the highest dose tried, not a limit anyone established, and a rat no-harm level is not a human intake limit. No human safety measure was taken. The company selling the powder ran and wrote it. An expert panel review records one anaphylactic reaction to fresh lion's mane.

The 28-day stress result that gets quoted came in at p=0.051, just outside the usual cutoff for calling a result real. The authors call this a pilot too small to conclude from. What was in the capsule was described only as a proprietary mushroom complex.

Everyone in these trials already had one of those conditions, so a healthy adult has no reason to expect the same shift. The trials ran between one and six months, which says nothing about a year on it. And 543 people across 11 trials is a modest base for a figure this precise.

The drip is given in a clinic under supervision, and nothing about that result transfers to a capsule taken at home. Some of the pooled studies assigned participants poorly, by the authors' own account. And each half of this rests on fewer than a thousand people.

Blood is not tissue. The one trial that biopsied muscle to check delivery found NAD-related metabolites raised there, including the products of clearing the compound away, while the muscle's mitochondria performed no differently. In mice, oral doses of both compounds are taken apart into plain nicotinamide in the liver; the same compounds injected arrive intact.

The notification list records that the FDA responded to each submission on a given date. It does not say whether the response objected, and the agency does not approve dietary ingredients in any case. The submission and response dates come straight from the FDA's own published list; the date of the reversal itself comes from legal and trade reporting, because the agency's letter could not be retrieved.

Open-label, fourteen days, healthy adults, so this establishes a route rather than an outcome. The bacterial step was shown in fermentation vessels seeded with human gut microbes rather than watched in the participants themselves. Nine of the twelve authors work for Nestle, whose health science arm sells NAD-related products.

A narrative appraisal rather than a systematic review, so its selection of evidence cannot be reproduced by anyone else. Its authors include the group running the pharmaceutical NMN trials, which makes this the field appraising itself. Its phrase about raising NAD in blood and some tissues is deliberately narrower than the marketing and should not be read as delivery throughout the body.

Fourteen days and open-label, so the comparison is of absorption, not of benefit, and neither arm measured anything a person would feel. Nine of the twelve authors work for Nestle, whose health science arm sells NAD-related products, which is worth noting mainly because the result cuts against both branded products rather than favouring either.

This was one result among many in a study built to map how fast NAD is made and consumed across tissues, not a comparison it was powered to make. It tracks where the molecule goes, not whether anything follows from its arriving. No equivalent measurement has been made in people.

Thirty-six people is small for a vascular endpoint, so this cannot rule out a modest real effect. The paper's own closing line says NMN showed potential in alleviating arterial stiffness, which is a firmer reading than its numbers support. The work was done in the laboratories of a Japanese supplement company, though the authors declare no competing interests.