Iron Markers

Iron

Serum iron

The amount of iron circulating in the blood at the moment the sample was taken. On its own it is the least useful of the iron markers, and the reasons are worth setting out properly, because most explanations of why get it wrong.

It is not mainly a time-of-day problem. Serum iron does have a daily rhythm and it is a real one, driven from inside and not by meals. What it does not have is a rhythm you can predict in an individual. In the study that sampled healthy people through the day, morning levels were higher than afternoon levels in only half of them. The variation between one day and the next was about the same size as the variation within a single day. Its conclusion was that restricting iron collection to a particular time of day does not improve the reliability of the result. Separately, a very large set of routine results found iron running high across a long window from mid-morning into the afternoon, with the peak falling in different places in men, women, children and teenagers. So the familiar advice to test in the morning does not fix this marker, and anyone relying on it is solving the smaller half of the problem.

Sampled weekly in healthy people, serum iron varies by about 28% within the same person. That is among the highest figures for any routine blood test. Put through the arithmetic, 2 serum iron results from the same man have to differ by roughly 44% downwards or 77% upwards before the difference is more likely real than noise, and in premenopausal women the figures are wider still. On the best current estimate the spread within one person is as large as the spread between people, which is the formal way of saying that the population reference range is a weak comparator for a serum iron. An older compilation of published studies puts that ratio lower, so the exact figure is contested and the direction is not.

An iron supplement is measured as the supplement. An oral iron dose lifts serum iron within hours by an amount that can exceed the entire reference range. The pharmacokinetic literature uses a rise of 20 micromoles per litre as its worked example of what a dose does. Whatever the exact figure, it is larger than most of the differences people are trying to detect, which is why every guideline that asks for iron studies asks for them off supplements.

Food matters, and the bodies disagree about how long to fast. The measurement side is reasonably clear. Iron takes roughly 5 hours after a meal to come back to baseline, so a sample taken shortly after eating is reading the meal. Past about 10 hours without food it starts climbing again, and at 12 hours or more it reads above what it otherwise would, so on those data the cleanest window is 5 to 9 hours and not overnight.

The guidance side does not agree with that. The 2026 American haematology guideline says transferrin saturation should be drawn fasting, including fasting from supplements. The 2025 sleep medicine guideline asks for a morning sample with no iron-containing supplements or foods for 24 hours beforehand. Against them, the screening programme that tested over 100,000 people for genetic iron overload found no advantage to fasting samples for that purpose. UK haematology guidance says transferrin saturation need not be measured fasting provided a borderline result is repeated or checked fasting afterwards. So a fast is asked for by the newest guidance, the measurement data say a very long one is counterproductive, and nobody has reconciled the 2.

What everything agrees on is the supplement. No iron-containing supplement for 24 hours before the test is the instruction with the most behind it, and it is the one most often left out. Ferritin is unaffected by either food or supplements, so none of this applies to it.

The sample itself changes the number too. Damaged red cells release iron, so a haemolysed sample reads high. On the most widely used platform a 10% shift in serum iron appears at a degree of haemolysis below the level the manufacturer declares as interfering. So a sample can be reported without a haemolysis flag and still carry a measurable iron error. The direction is always upwards.

Where the numbers come from

No professional body publishes a threshold for serum iron, and the reason is everything above. NHS guidance says so directly, that it varies with time of day and with recent meals and does not reflect iron stores. The population reference range is assay-dependent.

Serum iron earns its place as the numerator of transferrin saturation, which is the number that carries the guideline thresholds. What the marrow does when iron runs short, and why the red cells change before the blood count does, is under Blood cells, in the Body Guide. Read on its own it is closer to noise than to information. There is one genuine exception, which is that in advanced kidney disease the amount of carrier protein falls, so a fixed transferrin saturation corresponds to a progressively lower absolute iron, and at least one group argues the absolute figure is the better guide in that specific setting. That is an argument about the numerator and not a case for reading serum iron in a healthy person.

Units

The UK reports serum iron in micromoles per litre and the US in micrograms per decilitre. Divide the American figure by 5.585. The same factor applies to binding capacity, which is why transferrin saturation comes out the same percentage either way.

Total Iron Binding Capacity

TIBC

How much iron the blood could carry if every binding site were occupied. It is a measure of how much transferrin is present, arrived at indirectly.

The body makes more transferrin when iron is short, so this number tends to rise when stores are low and fall when they are high. It also falls with inflammation, with liver disease and with poor nutrition, because transferrin is made in the liver and behaves partly as a nutritional marker. That matters, because it means a low binding capacity has 2 very different explanations and they point opposite ways.

The relationship to transferrin itself is arithmetic. Binding capacity in micromoles per litre is about 25 times transferrin in grams per litre, from the fact that each transferrin molecule carries 2 iron atoms.

Where the numbers come from

There is no published threshold for this marker. The population reference range is assay-dependent.

It is an inferior way of measuring transferrin, and that should be said. Measuring transferrin directly is more precise, has an international reference material behind it and binding capacity does not. Binding capacity measured directly also reads a little higher than the capacity worked out from transferrin, so it overstates the real carrying capacity slightly. Where a lab reports transferrin instead, that is the better number of the 2.

There is an honest counterweight, and 2 parts to it. The first is that binding capacity is by some distance the steadiest number in this group. Sampled repeatedly in the same person it varies far less than serum iron does, which makes it the one member of the profile whose movement is more likely to mean something. The second is that a low binding capacity has been reported to predict death in people on dialysis, after allowing for iron, ferritin and inflammation. Transferrin itself has been put forward as predicting outcomes in heart failure, but only in preliminary work. In both cases it is working as a marker of nutrition and inflammation and not of iron, which is a different job from the one it is on the panel to do.

Its use here is as the denominator of transferrin saturation, which is where the guideline figures in this group actually sit.

Unsaturated Iron Binding Capacity

UIBC

The portion of that carrying capacity currently sitting empty.

The arithmetic runs the opposite way to the names, and most descriptions have it backwards. In many labs the unsaturated capacity is the number that is actually measured, and the total capacity is then serum iron added to it. Where that is the route, the total is the derived figure and this one is the direct one, which is the opposite of what the names suggest.

Either way, given any 2 of serum iron, total capacity and unsaturated capacity, the third is arithmetic.

Where the numbers come from

There is no published threshold for this marker. The population reference range is assay-dependent.

So this is a reported line and not an independent piece of information. It is on your report because it is part of the profile, not because it adds a fifth thing to know. Of the 5 numbers in this group, 3 are measured and 2 are worked out from the others.

Transferrin saturation

TSAT

The percentage of the blood's iron-carrying capacity that is actually occupied. It is more informative than serum iron alone, because it corrects for how much carrier protein is present, and it is the number that carries the published thresholds at both ends.

The arithmetic is not standardised, and that is a bigger problem than it sounds. Most labs calculate it as serum iron divided by total binding capacity. Others calculate it from transferrin directly, multiplying transferrin by a factor to get a capacity first. With transferrin in grams per litre the stoichiometric factor is about 25.1, from a molecular weight of roughly 79,600 and 2 iron atoms per molecule. Surveys of labs have found factors in use either side of it. Whichever number your lab used is not printed on the report. Run the same serum through those different routes and the transferrin saturation moves. Near the top of the range that matters, because the figures that prompt investigation for genetic iron overload sit only 5 percentage points apart, and a method difference can be the whole of that gap. This is the reason 2 transferrin saturations from different labs are not comparable in the way 2 haemoglobins are.

On top of that it inherits serum iron's problems in full. It carries the daily rhythm, the response to a supplement, the response to a meal and the effect of a long fast, and it carries the variability of 2 assay-dependent numbers instead of one.

Here is what that looks like in practice. In a group of amateur runners taken through a 100km race, average transferrin saturation was 32% a week before, 13% immediately after finishing and 43% a day later. The last 2 of those fall within a single day, they are more than 3 times apart, and only the first of the 3 describes iron status. One caveat on that study, which cuts both ways. It calculated transferrin saturation by one of the non-standard routes described above, so the absolute figures carry that as well as the biology. The direction and the size of the swing do not depend on it.

Where the numbers come from, at the low end

The American Society of Hematology's 2026 guideline on diagnosing iron deficiency uses a transferrin saturation below 20%, and it uses it in one specific place, which is in someone who has anaemia of inflammation. There the recommendation is to measure both ferritin and transferrin saturation instead of ferritin alone, and to call it iron deficiency on either a transferrin saturation below 20% or a ferritin below 100. It sets no transferrin saturation threshold at all for people without inflammation. The kidney guidance uses 20% in the same way, alongside a ferritin below 30, to define profound iron deficiency in someone with kidney disease and no anaemia.

The figure of 16% that circulates is usually attributed to the World Health Organization, and its 2020 ferritin guideline does not contain it. That guideline publishes no transferrin saturation threshold at all. The 16% appears in a much older joint document on iron deficiency anaemia from 2001, and even there it describes what is usually seen rather than setting a cut-off. That is a different thing from current guidance and should be described as such.

Where the numbers come from, at the high end, and the bodies do not agree on the sexes

European liver guidance is the most specific. Take someone already known to carry 2 copies of the common haemochromatosis variant. A transferrin saturation above 45% with a ferritin above 200 in women, or above 50% with a ferritin above 300 in men, is enough to diagnose haemochromatosis without imaging or a biopsy. In anyone with a different genotype, the same blood picture is not enough and the diagnosis needs iron shown in the liver.

UK haematology practice splits at 40% in women and 50% in men. American liver guidance uses 45% for everybody. So all 3 are in the same territory, and which side of a line a woman at 43% falls on depends on which country wrote the guidance. Publishing all 3 is more useful than picking one.

What the figure is actually for is deciding who gets a genetic test, and the yield is known. A man with a ferritin above 300 and a transferrin saturation above 50% turns out to carry 2 copies of the variant about 1 time in 5. A woman with a ferritin above 200 and a transferrin saturation above 40% about 1 time in 6. The guidance is clear that most of the rest still have iron overload of some kind, so the genetic test coming back negative is not the same as the picture being a false alarm.

What a high ferritin with a normal transferrin saturation means

British Society for Haematology guidance on investigating a raised ferritin makes transferrin saturation the discriminator. Its position is that iron overload is an uncommon cause of a raised ferritin and that a normal transferrin saturation points away from it. A high ferritin on its own means very little. A high ferritin with a normal transferrin saturation points away from iron overload rather than towards it.

The same guidance adds the caveat, and it matters. An acute infection, menstrual bleeding or a recent blood donation can pull a transferrin saturation temporarily into the normal range in somebody who does have iron overload. So a normal result points away from it without completely excluding it, and a single reading in the wrong week is not the end of the question.

Kidney disease and heart failure use different figures, and one of them has outrun its own guideline

Kidney guidance was rewritten in 2026, the first update in 14 years. On haemodialysis it suggests starting iron if the ferritin is at or below 500 and the transferrin saturation at or below 30%. For people not on dialysis, and for those on peritoneal dialysis, it is tighter and comes in 2 branches, a ferritin below 100 with a transferrin saturation below 40%, or a ferritin between 100 and 300 with a transferrin saturation below 25%. It also gives figures for holding iron back, as a practice point and not a graded recommendation, at a ferritin above 700 or a transferrin saturation at or above 40%. Those numbers exist because iron is being given by infusion in that setting, and they are not general-population figures.

Heart failure is the interesting one. Its guideline definition of iron deficiency leans on a ferritin below 100. Since that definition was written, a large study following over 2,000 people with heart failure found that ferritin was not associated with outcomes at all, while a low transferrin saturation and a low serum iron both were. In the same study, iron deficiency defined by the current criteria did not predict death. The guideline has not caught up with that, and until it does, a heart failure iron assessment built on ferritin alone is being run on the weaker of the 2 numbers.

Ferritin

The protein iron is stored in. Serum ferritin reflects how much iron is banked and not how much is circulating, which makes it the most useful single marker of iron stores, and it is usually the first to move when stores are falling, well before a blood count changes.

One important caveat: ferritin is also an acute phase reactant. It rises with inflammation, infection and liver inflammation, which can mask low stores that are real. That is why hs-CRP is often measured alongside it, and Inflammation, in the Body Guide sets out what an acute phase reactant is and why several markers move together. How the liver decides how much iron to release, and why storage and supply can move in opposite directions, is in Iron, in the Body Guide.

Where the numbers come from, and this changed in September 2026

Three bodies publish a figure for iron deficiency and they no longer sit close together.

The American Society of Hematology published its first guideline on diagnosing iron deficiency in August 2026 and announced it on 16 September. For adults who are neither menstruating nor pregnant it suggests a ferritin at or below 30 micrograms per litre, and it words the recommendation as using 30 instead of 15. For menstruating people it suggests 30 and suggests against 15. Where there are symptoms, risk factors, heavy periods, surgery coming up or a pregnancy being planned, it says 50 is appropriate. In children aged 9 months to 4 years it suggests 20 instead of 12. Every one of those is a conditional recommendation on low certainty evidence, which the guideline says plainly. Its recommendation against 15 in pregnancy is the only strong one it makes, and its recommendation about inflammation rests on very low certainty.

The World Health Organization's 2020 guideline takes ferritin below 15 micrograms per litre in adults and below 12 in children under 5, rising to below 70 in adults where infection or inflammation is present. For risk of iron overload it gives above 200 micrograms per litre in men and non-menstruating women, above 150 in menstruating women, and above 500 in anyone unwell.

British Society of Gastroenterology guidance is the document UK practice actually runs on, and it is more granular. Below 15 is highly specific for iron deficiency. Below 30 generally indicates low stores. Where there is inflammation, 45 has been suggested as the best trade-off between missing deficiency and overcalling it. And above 150, absolute iron deficiency is unlikely even with inflammation present.

So for an adult, the American figure is now exactly double the World Health Organization's, and the American figure for inflammation, 100, sits between the British 45 and the British 150. There is no reconciling these. They are 3 bodies reading the same evidence and landing in different places, and a ferritin of 25 is iron deficiency in America, low stores in Britain and normal to the World Health Organization.

Why the numbers moved, and where the World Health Organization's came from

This is the part that explains the gap. The World Health Organization's 2020 review concluded the evidence was not sufficient to justify changing the existing cut-offs, and those cut-offs trace back to expert opinion, based on the finding of absent iron stores in bone marrow. They are not wrong. They were simply never derived from population data.

The work that replaced that approach asked a different question, which is at what ferritin iron-restricted red cell production actually begins, measured by haemoglobin starting to fall and the markers of iron-starved marrow starting to rise. Pooled across 12 countries it puts the threshold at about 24.8 micrograms per litre in women and 22.1 in children. A 2026 study in blood donors took it further and split it by age and sex, at about 25 in women aged 18 to 49, 31 in women aged 50 to 75 and 33 in men. Those figures are not what the American guideline adopted, since it settled on the rounder 30 and 20. They are what moved it. One honest limit on them is that the 12-country pooling drew on surveys in Africa, Asia, Europe and Central America and not on a Western clinical population.

The consequence of the cut-off is enormous and it has been counted. Across a quarter of a million Swiss primary care patients, moving the threshold from 15 to 30 to 45 took iron deficiency diagnoses from 10.9 to 29.9 to 48.3 per 1,000 patient-years. For iron deficiency without anaemia specifically it went from 4.1 to 14.6 to 25.8, a 6-fold change from the number alone with no change in the people.

One line from the new guideline is worth having in front of you. Its own good practice statement is that a full blood count alone is not adequate to identify iron deficiency, whether or not anaemia is present. That is a haematology body saying that the test most people are given instead of a ferritin does not do the job.

On UK guidance specifically, there is no NICE guideline on iron deficiency anaemia. There is a NICE Clinical Knowledge Summary, which takes ferritin below 30 as iron deficiency and 30 to 100 as possible iron deficiency where symptoms or other markers support it. A summary is not a guideline, and the 2 do not carry the same weight.

The inflammation problem, as numbers instead of a warning

Every body handles it differently and none of them handles it well. The World Health Organization raises its own deficiency threshold from 15 to 70 when infection or inflammation is present. British gastroenterology guidance uses 45 and sets a ceiling of 150 above which absolute deficiency is unlikely. The American guideline abandons the single number and says to measure transferrin saturation as well, calling it deficiency on either a transferrin saturation below 20% or a ferritin below 100.

What does not exist is a published correction you can apply to one person's ferritin from their CRP. Correction factors exist for adjusting whole populations in nutrition surveys. Nothing validated exists for the individual, which means reading a ferritin alongside a raised hs-CRP is a judgement and not a calculation. That is the honest state of it.

The assay problem, and it runs the opposite way to what people assume

Ferritin immunoassays are poorly standardised between manufacturers, so the same sample returns meaningfully different numbers on different platforms. The scale is the surprise. In a study that ran the same dialysis samples through 10 ferritin assays and 5 transferrin saturation assays, the ferritin results spanned a range of about 63% between the highest and lowest method, and the transferrin saturation results only about 10%.

So the marker treated as the dependable one is the less comparable of the 2 between labs, and the marker treated as noisy is the more comparable. Comparing a ferritin taken here with one taken elsewhere is comparing 2 assays as much as 2 time points, and where a number is being tracked over years it needs to be tracked on one platform.

What a change between 2 results has to be before it means anything

Roughly 40 to 50%. That is combining the day-to-day biological swing with a good analyser, and it is the figure to hold against any comparison of 2 ferritins from the same lab. Day-to-day variation in women runs close to double that in men, so for a woman the figure may be nearer 70 to 75%.

In practice, a move from 30 to 40 is inside the noise. A move from 30 to 50 is probably real in a man and marginal in a woman. Anyone reading a 10-point change as progress or deterioration is reading the assay.

There is a related point about the reference range. Estimates of how the spread between people compares with the spread within one person vary a great deal for ferritin, and none of them makes a population reference range a strong tool for an individual, so the person's own previous result is the better comparator. It is the same conclusion the HbA1c entry reaches by a different route.

Where the optimal ferritin figures come from, one by one

Numbers like 50, 70 and 100 circulate as optimal. Some of them are real and most of them are not, and they are worth separating.

Restless legs is the real one. American sleep medicine guidance from 2025 sets out that in restless legs syndrome, iron should be started as oral or intravenous iron if the ferritin is at or below 75 or the transferrin saturation below 20%, as intravenous iron only if the ferritin sits between 75 and 100, and in children if the ferritin is below 50. That is a named body with a published figure, which is more than any other claim in this area has. It comes with 2 caveats the guidance states itself. The consensus figures have not been empirically tested. And they are explicitly different from the figures used for the general population.

The underlying reasoning is about iron in the brain and not iron in the body, and the sleep medicine figures stop at 100. The separate international restless legs group goes further and puts intravenous ferric carboxymaltose on the table at a ferritin up to 300, which is frankly normal, and that is a different body making a wider claim. Consistent with that, a study of female blood donors found no association between restless legs and ferritin at all. So the threshold is real, it is specific to one condition, and it is not evidence that 75 is a good general target. On what actually works, intravenous ferric carboxymaltose is the only iron formulation to earn a strong recommendation, oral ferrous sulphate is conditional, gabapentin and pregabalin outrank oral iron, dopamine agonists are recommended against for routine use and cabergoline is strongly recommended against.

Fatigue is the one everybody quotes, and the honest version is more interesting than the marketing. Pooled across trials of iron in people who are iron deficient without being anaemic, self-reported fatigue improves by a small to moderate amount, consistently and with little disagreement between trials. Objective physical capacity does not improve at all. Across 9 trials the effect on maximal oxygen uptake is a clean null. So iron reliably changes how tired people say they feel and reliably does not change what they can do.

Where the effect sits matters more than whether it exists. In one trial the benefit was confined to women with a ferritin at or below 50. In another, recruited at 50 or below, the trial was null overall and the entire effect sat in the women who started below 15. In a third, the cognitive benefit was confined to those below 15.

And then the trial that anyone quoting a threshold of 50 has to answer. Over 400 non-anaemic blood donors with a ferritin at or below 50, given intravenous iron or placebo, properly double-blinded. Ferritin rose by over 100 micrograms per litre and haemoglobin rose measurably. Fatigue did not move at all, and it did not move in any of the lower bands the investigators looked at either, which went down to below 25, below 20, below 15 and below 10. That last part matters, because it is the evidence that argues hardest against the reading we give at the end. Fatigue is also about as placebo-sensitive an endpoint as exists, and the reviews of this literature single out the difficulty of blinding oral iron, which discolours stools and reliably upsets stomachs.

Our own reading, stated as ours, and it is a reading and not a conclusion. The oral iron trials put the effect below 50 and concentrate it below 15 to 20. The one properly blinded intravenous trial found nothing anywhere, including in those low bands. So the honest position is that the effect is probably real, probably small, probably confined to low ferritin, and not demonstrated by the strongest single trial available. Treating 50 as a symptomatic threshold is a defensible clinical hedge and is not a finding.

Hair loss has no threshold and no trial. No body publishes a ferritin figure for hair loss. No randomised trial has tested iron against placebo for hair loss of any kind. Searching the trial registries turns up observational studies and one trial of a multi-ingredient supplement, which is a different thing and is sometimes presented as if it were iron. What exists is association, inconsistent between the different types of hair loss. The figures quoted in this space, usually 40 or 70, are inferences from those associations and not thresholds anybody has set or tested.

Cognition and endurance performance are both inferred. There are a few trials in the cognitive space, no threshold and no body. For endurance performance, sports medicine does use working ferritin figures, but the pooled trial evidence is the same maximal oxygen uptake null described above, so the performance claim is the weaker of the 2 and not the stronger.

The upper end, where most raised ferritins are not iron

UK haematology guidance states it as a strong recommendation. Reactive causes of a raised ferritin, meaning malignancy, inflammatory disorders, kidney failure, liver disease and metabolic syndrome, should always be considered, because all of them are considerably more common than true iron overload.

A raised ferritin is also not rare. In primary care, about 1 in 5 white adult men have a ferritin above 300. In women above 200 it runs at about 3% between 30 and 50, 10% between 50 and 70 and 17% over 70. So the finding by itself is common and the explanation usually is not iron.

Alcohol is the most common single explanation nobody volunteers. Above 2 drinks a day, every marker of iron overload is significantly more common, and the same national survey found that drinking any amount of alcohol carries a 40% lower risk of iron deficiency anaemia, which is a real finding and not a reason to drink. UK haematology guidance adds, from other work, that beer does more of this than wine or spirits.

Fatty liver is the other big one. Roughly 1 in 4 people with metabolic fatty liver disease have a raised ferritin, pooled across studies that varied a great deal in how they defined it. It is not harmless in that setting. A raised ferritin there goes with about double the rate of liver-related events, and above about 1.5 times the upper limit of normal it has been reported to associate with advanced scarring independently of how much iron is actually in the liver. What it does not do is call for treatment aimed at the iron. Why a fatty liver raises a storage protein in the first place is set out under Liver, in the Body Guide. UK guidance states, as a strong recommendation, that there is no evidence to support removing blood to lower ferritin in fatty liver disease.

Ethnicity belongs here too. A ferritin above 1,000 is 2 to 3 times more common in Black and Asian populations than in white ones, despite an almost complete absence of the iron-loading genotypes in both. So the same very high number carries a different set of likely explanations depending on who it came from.

What guidance says to do with a high one. In someone otherwise well, with an unexplained and moderately raised ferritin below 1,000 and a normal transferrin saturation, a period of observation with a repeat in 3 to 6 months is reasonable. Persistently above 1,000 warrants referral to a liver specialist. Above 10,000 the rare inflammatory syndromes come into the picture, although even at that level kidney disease, liver disease, infection and malignancy still account for most of it. In a hospital series of ferritins above 1,000, malignancy was the most common cause at about 1 in 4 and iron overload was second at about 1 in 5.

Haemochromatosis, with the numbers that are usually left out

The common form comes from carrying 2 copies of one variant in a gene called HFE. In UK data on people of European background, that is about 1 in 156, which makes it one of the most common inherited conditions in this country.

The part that gets left out is how often it causes anything. In an Australian cohort followed for over a decade, documented iron-overload disease appeared in about 28% of men carrying 2 copies and about 1% of women, and the figure in women rests on so few events that its range runs from almost nothing to 6%. So most carriers never develop the disease, and the sex difference is not subtle.

In the men it does affect, the stakes are real. Risk of primary liver cancer by age 75 runs at about 7% in male carriers of 2 copies against about 0.6% in men carrying neither variant. Half the liver cancers that occurred in those men happened in men who had no haemochromatosis diagnosis when they were first assessed, which is the argument for looking and not the argument for worrying. In women neither liver cancer nor death reached significance.

A 2025 finding that is treatable and easily missed. Calcium deposits in the knee joint came out nearly 4 times as common in people carrying 2 copies, and higher again in the men. That second figure rests on 14 cases and its range is wide enough that only the direction is safe. Most of the affected men had knee pain and no haemochromatosis diagnosis.

Where treatment targets sit, if it is diagnosed, is a ferritin below 50 while iron is being removed and below 100 once that is maintained. And the honest gap is in the middle. Whether removing blood from a carrier whose ferritin sits between 300 and 1,000 does any good has never been established, and the trial set up to answer it was designed precisely because the evidence was lacking.

Biotin does affect this one

Unlike the blood sugar markers, ferritin is measured on an immunoassay of the kind biotin interferes with. The manufacturer of the most widely used platform sets an interference limit and states that samples should not be taken from anyone on high-dose biotin until at least 8 hours after a dose. That matters more here than almost anywhere else on the menu, because biotin is the hair, skin and nails supplement, and the people taking it are disproportionately the people buying an iron panel to investigate hair loss or fatigue.

Exercise makes this number lie, and for longer than people expect

Ferritin behaves as an inflammatory marker during and after hard endurance work, and the size of it is not trivial. After a 100km race it read about 38% above baseline immediately and was still about 37% above it 24 hours later. The hs-CRP at 24 hours was roughly 44 times baseline. So a panel taken the day after a hard event reads as both inflamed and iron-replete at once, and the transferrin saturation alongside it reads high, having read very low the day before.

How long it lasts is the part that surprises people. After a 56km race ferritin took 6 days to return to baseline, and with daily training continuing it can stay raised for 6 to 14 days. In runners who were tested at rest, 14% had a subnormal ferritin against 2% of controls, so this is a population where the marker matters most and is hardest to read.

Sports science practice for testing an athlete is a morning sample after at least a day of rest, with only low to moderate activity for 2 or 3 days beforehand, and no illness or infection in the preceding week. After a race or a very hard block, the 6-day figure argues for waiting a week. One honest limit on all of that. The quantitative work is from ultra-endurance events in small groups, and there is no published figure for what an hour in the gym does. Do not assume it scales.

In the post

Ferritin travels well and serum iron does not, and that split is what decides whether a posted sample can be read.

A UK study across several NHS trusts found ferritin acceptable for up to 24 hours in unspun whole blood at room temperature. Serum iron in the same study was stable for at least 12 hours, and its recommendation was that blood should ideally be spun within 12 hours for iron. Its authors also said that for iron specifically a delay out to 24 hours was likely to have little clinical impact. It was a small study, in 10 volunteers. A 2026 study built specifically to test home collection found iron consistently outside acceptable limits in unspun whole blood out to 72 hours, under more than one storage temperature.

Putting those together, the variable that decides it is whether the tube was spun before the journey and how warm the journey got. A separated serum iron holds for days. An unspun one is fine for a day and not for 3, and because transferrin saturation is calculated from serum iron, transferrin saturation is no more dependable than its numerator.

One thing to note about hand hygiene before a finger-prick sample. Washing hands thoroughly and taking off hand cream is standard practice for any capillary sample and it is sound. The specific claim that iron-oxide pigments in cosmetics contaminate an iron result is one we could not trace to a primary source, so we are not putting a direction or a size on it.

What to do with a low ferritin, and the dosing argument

A dose of 60mg of iron or more raises hepcidin, the hormone that shuts down iron absorption, for about 24 hours, and it subsides by 48. Measured with tracer studies, the fraction absorbed from a dose given every other day is 40 to 50% higher than from the same dose given daily. To deliver the same total iron you give twice the daily dose on alternate days. The same work found the daily rise in hepcidin is made worse by an afternoon or evening dose on top of a morning one. Splitting a dose across the day therefore raises hepcidin more, although in the study that tested it, splitting did not actually reduce how much iron was absorbed.

That is the mechanism and it is solid. The outcomes are more modest, and that should be said plainly. In a properly blinded 6-month trial at equal total dose, alternate-day and daily dosing produced identical ferritin. What alternate-day did deliver was about a third fewer days of gut side effects, counted on the days iron was actually taken. It also left fewer people iron deficient at 6 months, by a margin that was only just statistically significant. A 2025 pooling of 11 trials found the 2 comparably effective on haemoglobin and concluded that alternate-day dosing was better tolerated. Its own pooled figure for side effects showed no difference, and it rated the certainty very low throughout.

No guideline has adopted it. UK gastroenterology guidance still recommends one tablet a day as a strong recommendation, while acknowledging the absorption data in its own discussion. So the defensible position is that alternate-day dosing is better tolerated with equivalent stores at the end, which is a real reason to prefer it, and it is not faster or more effective, and no body recommends it.

On vitamin C, the mechanism and the outcome disagree. Ascorbate clearly increases absorption of non-haem iron in tracer studies. The one adequately sized trial, in over 400 people with iron deficiency anaemia, gave iron with 200mg of vitamin C 3 times a day or iron alone, and found them equivalent on both haemoglobin and ferritin. UK guidance recommends against routine co-administration on the strength of it. Both of those are true and the trial was open-label with a generous equivalence margin, so publishing the disagreement is the honest option.

What sits beyond these 5, including things we do not sell

Three markers get proposed as better than ferritin and none of them is ready to replace it. Hepcidin is the hormone at the centre of the system, so measuring it directly is the obvious move. The assays are not standardised and no body recommends it. Soluble transferrin receptor rises when the marrow is short of iron and is much less disturbed by inflammation than ferritin, which is exactly the property ferritin lacks, though less disturbed is not the same as unaffected. It has the same standardisation problem. Reticulocyte haemoglobin content reads the iron actually going into new red cells, and on some blood count analysers it comes free. It is not part of a standard UK panel. We do not sell any of the 3. If the field solves the inflammation problem it will probably be with one of them.

Ferritin and testosterone

This is not a guideline position and is presented as what the published work shows.

Testosterone suppresses hepcidin, the hormone controlling how much iron is released into circulation, and stimulates erythropoietin. Both push red cell production up, and the rise in haematocrit is the most common adverse effect of testosterone therapy.

What that does to ferritin is less consistent than the internet suggests. Studies have found hepcidin falling by around a third, with ferritin falling alongside it. Others have found ferritin and hepcidin both falling early, with a rebound back towards baseline by 6 months while testosterone was still being given. At least one found hepcidin falling and haematocrit rising while ferritin did not change significantly, with transferrin saturation and serum iron falling instead. The consistent findings across all of them are hepcidin suppression, a rise in haematocrit, and markers of increased iron use. Ferritin often falls but does not always.

The practical reading. A falling ferritin in someone whose haematocrit is climbing is a recognised pattern rather than a puzzle, and a stable ferritin does not mean iron is not being consumed.

Units

Ferritin is reported in micrograms per litre in the UK and nanograms per millilitre in the US. Those are the same number, so no conversion is needed and a ferritin figure transfers directly between American and British sources. That is unusual, and it makes ferritin the one marker in this group where an American article about your result can be read straight off. It does not extend to the rest of the group. Serum iron and binding capacity are reported in different units on each side of the Atlantic and do need converting, as set out in their own entries above.

The upper limit of normal is a lab matter and not a guideline one. UK labs differ in where they put it, with figures in the region of 300 to 400 micrograms per litre for adult men and 150 to 200 for adult women in common use. Given how poorly the assays agree with each other, the figure on your own report is the one that applies to your own result.