A ferritin result of 18 nanograms per millilitre sits comfortably inside the reference range printed by most laboratories, which commonly stops at 15. Under a guideline the American Society of Hematology published in Blood Advances in September, that same number should be read as iron deficiency in an adult, and as iron deficiency in a menstruating or pregnant person too.
That gap between what a report says and what the haematology society says is the substance of the document. Its five recommendations move the diagnostic threshold upward for every population it covers, and its background sections argue that the familiar lower limit of normal is itself partly responsible for iron deficiency being missed.
What the guideline recommends, population by population
For infants and children aged 9 months to 4 years, the panel suggests a serum ferritin threshold of 20 ng/mL or below in place of 12 ng/mL. For adults who are not menstruating or pregnant, it suggests 30 ng/mL or below in place of 15. For menstruating individuals, it suggests 30 ng/mL and explicitly suggests against 15. For pregnant individuals, it suggests 30 ng/mL, and recommends against a 15 ng/mL threshold as a strong recommendation.
Two of the recommendations carry a second, higher number. Adults with symptoms of iron deficiency or ongoing risk factors may appropriately be diagnosed at 50 ng/mL or below; so may menstruating people with heavy or otherwise abnormal uterine bleeding, symptoms, additional risk factors, upcoming surgery, or plans to become pregnant. Pregnancy-specific risks such as multiple gestation, hyperemesis gravidarum, or an interval under 18 months between deliveries carry the same 50 ng/mL figure.
The symptoms the panel lists are broader than the ones most people associate with anaemia: fatigue, reduced quality of life, reduced cognitive function, restless legs syndrome, pica or ice craving, and reduced exercise tolerance. On examination it lists spoon nails, cracks at the corners of the mouth, tongue depapillation, and diffuse scalp hair loss.
Diagnosis, the guideline stresses, does not depend on anaemia being present. Iron deficiency can exist with a normal haemoglobin, which is the population the thresholds are chiefly aimed at.
Why the bottom of the range sits where it does
The panel’s explanation is unusually direct about the flaw it is trying to correct. The reference intervals that laboratories use were derived from studies of “apparently healthy individuals”, without accounting for the fact that a portion of those people were themselves iron deficient. A lower limit built that way will classify deficiency as normal.
Most laboratories, the guideline notes, report against that population-derived lower limit rather than a clinical decision limit, the value at which a result should be flagged as abnormal regardless of the reference range. The panel suggests laboratories switch to a decision limit at the recommended thresholds and attach an explanatory comment to the report.
The same sample can read 7 or 14
Precision is the awkward part, and the guideline does not hide it. A serum ferritin of 10 ng/mL may be reported anywhere from 7 to 14 ng/mL across laboratories; a value of 30 may be reported between 21 and 42. The panel cites proficiency-testing data from 97 Canadian laboratories showing differences of roughly 30% to 40% at the low end, plus comparisons between assays in which agreement was poor, with bias ranging from −11.5 to 44 ng/mL, and one platform reading 36% to 44% higher than another in people with chronic kidney disease.
That variability is one reason the recommendations are framed as approximations. It also cuts both ways: a result of 33 does not establish that someone is iron replete, and the guideline treats the 25 and 30 ng/mL thresholds for women of reproductive age as effectively interchangeable.
Where inflammation changes the arithmetic
Ferritin rises with inflammation, so a normal-looking result in a sick patient is weak evidence. For adults with anaemia of inflammation, the panel suggests measuring transferrin saturation alongside ferritin rather than ferritin alone, and says iron deficiency may be considered when transferrin saturation is under 20% or ferritin is under 100 ng/mL. Transferrin saturation should be drawn fasting, five to nine hours after eating and with multivitamin or iron supplements held.
The certainty here is the weakest in the document, and the panel says so: very low certainty, with too few prospective trials to fix a specific threshold. It also declined to make any recommendation for children with inflammation, citing an absence of data.
The screening question the panel left open
Recommendations on who should be tested were out of scope. The document nonetheless closes with a position: the panel believes it is reasonable to check serum ferritin in all pregnant individuals and in other high-risk groups, with or without anaemia. It notes that the American College of Obstetricians and Gynecologists does not currently recommend routine screening in pregnancy, while a 2026 joint clinical report from the American Academy of Pediatrics and the American Society of Pediatric Hematology/Oncology recommends universal screening for anaemia and iron deficiency in young children and in menstruating individuals after menarche.
Other bodies have already moved in the same direction: the National Bleeding Disorder Foundation’s 2026 guidance uses 50 ng/mL for people with inherited bleeding disorders, and ACOG uses 30 ng/mL in pregnancy. The World Health Organization’s older figures, 12 ng/mL for children and 15 ng/mL for healthy adults, are the ones this guideline is explicitly departing from; the panel describes them as highly specific but insensitive, and says it found little evidence supporting their use in any population.
What the guideline does not establish
Almost every recommendation is conditional, resting on low certainty in the evidence, and the panel’s own limitations section attributes that to the state of the literature rather than to the thresholds themselves.
There is no trial showing that diagnosing at 30 rather than 15 ng/mL improves outcomes. The thresholds were derived by comparing candidate cut-offs against bone marrow iron staining, a reference standard the panel itself calls subject to sampling bias and subjective grading and now rarely justified on its own. Research priorities listed in the document include hepcidin assays, better data on iron measurement in inflammation, and outcome studies of screening in young children and in pregnancy.
The scale of what is being missed is not in dispute. WHO estimates that 40% of children aged 6 to 59 months, 37% of pregnant women and 30% of women aged 15 to 49 have anaemia, with dietary iron deficiency among the largest causes, and that half a billion women aged 15 to 49 and 269 million children aged 6 to 59 months are affected worldwide.
For readers, the practical advice has not changed. The Office on Women’s Health lists fatigue, weakness, brittle nails and unusual cravings among the signs of iron-deficiency anaemia, notes that up to 5% of women of childbearing age develop it from heavy menstrual bleeding, and is explicit that iron pills should not be started without talking to a doctor or nurse; the pills commonly cause stomach upset, constipation or diarrhoea. The recommended daily intake for women aged 19 to 50 is 18 mg, rising to 27 mg in pregnancy. Our earlier look at the global burden of iron deficiency sets out where the condition falls hardest, and the nutrient-gap problem in plant-based diets explains why vegetarian and vegan eating appears on the guideline’s risk list.
A result that sits just under a laboratory’s lower limit is a reason to ask a clinician what it means in context, not a reason to treat it yourself. As the medical disclaimer on this site notes, nothing here is medical advice, and the diagnosis of iron deficiency depends on symptoms, history, other blood indices and what is causing the loss in the first place.