Your lab portal loads. You scan for anything flagged in red, find nothing, and close the tab. Ferritin: 14. No flag.
You are not here because of a number. You are here because of the third hour of the afternoon, when the work stops going in and you assume it’s the sleep, or the coffee, or the month you’re having.
Three different numbers could have been applied to that 14. The clinical guideline threshold, which is 30. The level below which trials of iron for fatigue found any effect at all, which is 50. And your laboratory’s lower limit of normal, which for women often sits around 9.
Only the third one is on your report.
I went into this expecting to find that ferritin is an underused test. It isn’t — it’s ordered constantly. What I found instead is that the number coming back gets compared against a range assembled by measuring a group of people who were themselves substantially iron-deficient.
How the range got built
A reference range is a statistical statement, not a clinical one. Measure a lot of apparently healthy people, take the middle 95%, call the bottom edge the lower limit of normal.
That works when healthy people are, in fact, healthy.
Here is the problem, stated by two haematologists in an American Society of Hematology education paper: 30 to 50% of healthy women have no iron stores in their bone marrow at all. Which makes basing a ferritin cutoff on the bottom 2.5% of sampled ferritins inappropriate — the sample was full of the condition the test is meant to detect (Martens and DeLoughery, Hematology Am Soc Hematol Educ Program 2023, DOI: 10.1182/hematology.2023000494).
Feed iron-deficient women into a reference population and the “normal” range descends to meet them.
The numbers this produces are startling side by side. One university laboratory, in an analysis presented as a conference abstract rather than a peer-reviewed paper, derived its limits from presumably healthy volunteers and reported a lower limit of 7.3 ng/mL for adult women and 10.5 ng/mL for men. Across the five most-used American commercial laboratories, a conference poster put the mean female lower limit near 9 ng/mL — the weakest source in this section, and I’m citing it because it’s the only survey of American laboratories I could find.
Against that: the WHO threshold of 15 µg/L, based on historical data with cutoffs last revised in 1993 and acknowledged in the current literature to miss up to half of people with true iron deficiency. A ferritin below 30 µg/L is 92% sensitive and 98% specific and corresponds to absent marrow iron stores (Swinkels et al., Lancet Haematology 2024, 11(10):e721). A 2023 Delphi consensus settled on 30 as the clinical decision limit. The American Society of Hematology’s draft guideline suggests ≤30 ng/mL and recommends against using 15 — a strong recommendation the panel itself rates as resting on low-certainty evidence, still under public comment.
The systematic review that traced all this concluded that laboratory lower limits might not be rooted in rigorous scientific evidence and might be contributing to structural underdiagnosis of iron deficiency.
The practical version: at a cutoff of 10, one laboratory flagged 13.2% of its female patients as low. At 30, it flagged 45.7%.
And it isn’t the only threshold on that panel under question. PREFER’s authors — whose trial appears below — noted that intravenous iron raised reticulocyte counts even among women already above 120 g/L of haemoglobin, the level at which anaemia is ruled out, and concluded this raises doubts about whether that cutoff is appropriate for premenopausal menstruating women. The authors of the 2012 oral iron trial made the same point from the other direction: haemoglobin cutoffs are a surrogate, and they don’t accurately reflect everyone’s erythropoietic function. Two numbers on one blood panel, both questioned by the people who study them, both still printed as normal.
This isn’t a fringe complaint. In September 2024, LifeLabs — which handles community laboratory testing across Ontario — stopped reporting ferritin against population reference intervals and switched to clinical decision limits of under 30 µg/L for adults, aligning with updated provincial guidelines. I did not find an equivalent change at any American laboratory network, though I can’t claim to have checked them all — which means for a US reader the gap is probably still on the report.
But a low number is not a diagnosis, and here’s the trial that proves it
Before you go and get tested, the more important half of this.
PREFER randomised 290 premenopausal women — mean age 35, screened clean of thyroid, B12, folate and inflammatory abnormalities — to a single 1000 mg intravenous iron infusion or saline (Favrat et al., PLOS ONE 2014, DOI: 10.1371/journal.pone.0094217).
Fatigue improved in 65.3% on iron. And in 52.7% on saline. Odds ratio 1.68 (95% CI 1.05–2.70), P = 0.03, number needed to treat 8.
More than half the women who received salt water reported feeling less tired. The trial’s own authors attribute this to the emotional component of fatigue and to the intravenous route. Fatigue is a symptom that improves substantially when someone takes it seriously, and any article quoting the 65% without the 52.7% is telling you a different story than the trial did.
The effect ran both ways, too. The same authors noted an unusually high rate of adverse events in the placebo arm and put it down to overly cautious reporting — women receiving salt water also reported side effects. Attention moves symptoms in both directions.
The disclosures belong here rather than in a footnote, and they point toward the result rather than against it. Vifor Pharma sponsored the study and supported its design; the paper states the funders had no role in data collection, analysis or the decision to publish, but that employees of the funder were involved in study design and reviewed the manuscript as coauthors. One author is a Vifor employee. The contracted statistician received consultant fees from Vifor. Ferric carboxymaltose is Vifor’s product. All of it is printed in the paper — but this is a manufacturer-sponsored, manufacturer-co-designed, positive trial, and that is the opposite configuration from the two nulls this site has covered where the conflict pointed the other way.
Two further findings change how you read everything else.
They screened 861 women to randomise 294. Two-thirds didn’t qualify, and the main reason was transferrin saturation at or above 20% — fatigued, but not iron-deficient. The authors’ conclusion, in their own words: not every fatigued patient is automatically iron-deficient, and appropriate iron status assessment is mandatory before initiating any kind of iron therapy. That is the thesis of this article, written by a manufacturer-funded trial of an iron product.
How much ferritin rose predicted nothing. Within the treated group, the magnitude of ferritin change did not correlate with change in any endpoint. Prior trials found the same. The number you’d be chasing doesn’t track the thing you care about.
Which is why the oral trials disagree with each other
Two found a benefit. A 2003 Swiss trial randomised 144 non-anaemic women aged 18–55 with unexplained fatigue to 80 mg/day of iron or placebo for four weeks; improvement appeared, in a subgroup analysis, only in those with ferritin at or below 50 µg/L (Verdon et al., BMJ 2003). A 2012 follow-up in 198 menstruating women below that threshold ran twelve weeks and found fatigue reduced by almost half from baseline — a 19% difference against placebo (Vaucher et al., CMAJ 184(11):1247–1254). Nineteen per cent is the honest figure; “almost 50%” is what gets quoted, and it contains everything placebo did too.
One found nothing. 154 female blood donors under 50, iron-deficient without anaemia, randomised a week after donation to the same 80 mg/day or placebo for four weeks. Haemoglobin rose 5.2 g/L (P<0.01). Ferritin rose 14.8 ng/mL (P<0.01). Fatigue moved −0.15 points, 95% CI −0.9 to 0.6, P = 0.697. Nothing on aerobic capacity, mood or quality of life either.
The biology responded. The women didn’t notice.
The difference between those trials isn’t dose, duration or iron status. It’s how the participants were found. Verdon and Vaucher enrolled women who walked in complaining of tiredness. The donor trial screened people who hadn’t complained of anything.
And PREFER’s authors identify something in the two positive trials that nobody covering this topic mentions: constipation and blackened stool may have compromised patient blinding, producing higher reported response in the treatment arm and a suppressed placebo effect. Iron announces itself. In a trial with a self-reported endpoint, that matters enormously.
The cognition question, in one paragraph because that’s what it earns
PREFER measured cognition with computerised tests. Changes in the cognitive subscales were numerically larger with iron and not significantly different from placebo. The only significant cognitive finding was in the subgroup with ferritin below 15: an improvement in “power of attention” of 38.4 milliseconds (95% CI −65.8 to −11.0, P = 0.006). Thirty-eight milliseconds, in a subgroup, on a computerised battery. That is a real measurement and it is not a better Tuesday.
A 2025 meta-analysis of iron supplementation in non-anaemic people found improvements in short-term memory (d = 0.53) and cognitive intelligence (d = 0.46) — and none in attention, and none in depression — in a pool dominated by children and adolescents, with effects disappearing entirely when iron-deficient participants were excluded. If you’re here hoping iron sharpens thinking, the evidence is for fatigue, and even that is conditional.
A third of people can’t tolerate it
Ferrous sulphate raised the odds of gastrointestinal side effects against placebo with an OR of 2.32 (95% CI 1.74–3.08) across 43 trials and 6,831 adults. In absolute terms: 35% on iron reported them versus 22% on placebo. Constipation 12%, nausea 11%, diarrhoea 8%. Studies report up to 40% non-adherence (Tolkien et al., PLOS ONE 2015, MRC Human Nutrition Research, Cambridge).
That’s a meaningful cost to pay on a guess.
And iron is the one where guessing has a mechanism of harm
Roughly 1 in 200 people of northern European descent are C282Y homozygotes, the genotype behind most hereditary haemochromatosis. Elevated ferritin and transferrin saturation occur in 75–100% of male homozygotes and 40–60% of female. Clinical disease develops in about 10% of them, and men have a 24-fold higher rate of iron-overload disease than women, because menstruation is protective. Early symptoms are typically absent.
Those numbers don’t support panic. They support one claim: with iron, the test costs less than the supplement and tells you something the supplement can’t. And if you’re a man with low ferritin, that usually means blood loss from somewhere — a reason to see a doctor, not to buy capsules. No trial of iron for fatigue in men turned up in this research at all.
What to do with this
Ask for ferritin, and ask for transferrin saturation alongside it. PREFER screened out two-thirds of its fatigued volunteers on transferrin saturation; ferritin alone would have sent many of them home with a supplement they didn’t need.
When the result arrives, don’t read the flag. Read the number, and compare it against 30 rather than against whatever your laboratory prints. Under 30 is a conversation with a doctor. Between 30 and 50 with real exhaustion is the zone where two trials found something in symptomatic women and one found nothing in asymptomatic ones — a genuine ambiguity, not a hidden answer. Above 50, the fatigue is something else, and the most useful thing this article can do is save you three months and a one-in-three chance of gut side effects finding that out.
The reference range on your report isn’t wrong, exactly. It’s answering a different question than the one you’re asking: not is this person iron-deficient but is this person unusual compared to the sample we measured. When a third of that sample had empty iron stores, those two questions come apart — and the gap between them is where a lot of tired people are sitting, holding a result that says normal.
About this article
Written by Leah Elish. Leah covers nutrition claims, dose arithmetic, and the distance between a research protocol and a shopping list. Not a dietitian or clinician — traces nutritional claims back to the studies that supposedly support them.
Medical review: None. NeuriFuel does not currently have a licensed clinician on the editorial team, and this article has not been medically reviewed. We state this rather than implying an authority we do not have. See our About page for our full methodology.
Sources: Fourteen, comprising four randomised controlled trials, two meta-analyses, one systematic review of reference-interval derivation, one analysis of laboratory flagging rates, one draft clinical guideline under public comment, one haematology education paper, two population genetic cohorts, one laboratory network’s published testing change, and one WHO technical brief. Two figures come from conference abstracts and are identified as such in the text, because they were the only American laboratory data I could locate. Funding and competing interests were retrievable for two studies and are stated inline, including in full for the trial whose disclosures point toward its own positive result; for the remainder they were not available in the sources I could reach, and that absence is marked rather than assumed favourable. Values that could not be established are marked, and where sources conflict, both are given rather than one chosen quietly.
Corrections: Found an error? Write to hello@neurifuel.com with a source and we will fix it and log the correction.
Last updated: 5 August 2026
References
- Martens K, DeLoughery TG (2023). Sex, lies, and iron deficiency: a call to change ferritin reference ranges. Hematology Am Soc Hematol Educ Program 2023(1):617–621. DOI: 10.1182/hematology.2023000494.
- Swinkels DW, van Schrojenstein Lantman M, Matlung HL, Weykamp C, Thelen M (2024). The origin of ferritin reference intervals: a systematic review. The Lancet Haematology 11(10):e721.
- Ferritin flagging rates by cutoff in female patients. Journal of Applied Laboratory Medicine 6(3):765.
- American Society of Hematology. Draft recommendations for the diagnosis of iron deficiency. Public comment version; not a final guideline.
- LifeLabs / Ontario Association of Medical Laboratories (2024). Change to ferritin reporting: clinical decision limits of <30 µg/L in adults, effective 9 September 2024.
- Favrat B, Balck K, Breymann C, Hedenus M, Keller T, Mezzacasa A, Gasche C (2014). Evaluation of a Single Dose of Ferric Carboxymaltose in Fatigued, Iron-Deficient Women — PREFER: a Randomized, Placebo-Controlled Study. PLOS ONE 9(4):e94217. DOI: 10.1371/journal.pone.0094217. Sponsored by Vifor Pharma Ltd., manufacturer of ferric carboxymaltose. Full disclosures stated inline in the text.
- Verdon F, Burnand B, Stubi CL, Bonard C, Graff M et al. (2003). Iron supplementation for unexplained fatigue in non-anaemic women: double blind randomised placebo controlled trial. BMJ 326:1124. PMID: 12763985.
- Vaucher P, Druais PL, Waldvogel S, Favrat B (2012). Effect of iron supplementation on fatigue in nonanemic menstruating women with low ferritin: a randomized controlled trial. CMAJ 184(11):1247–1254.
- Waldvogel S et al. Clinical evaluation of iron treatment efficiency among non-anemic but iron-deficient female blood donors: a randomized controlled trial. ClinicalTrials.gov NCT00981877.
- Psychiatric and cognitive outcomes of iron supplementation in non-anemic children, adolescents, and menstruating adults: a meta-analysis and systematic review (2025). Neuroscience & Biobehavioral Reviews.
- Tolkien Z, Stecher L, Mander AP, Pereira DIA, Powell JJ (2015). Ferrous Sulfate Supplementation Causes Significant Gastrointestinal Side-Effects in Adults: A Systematic Review and Meta-Analysis. PLOS ONE 10(2):e0117383. DOI: 10.1371/journal.pone.0117383. MRC Human Nutrition Research, Cambridge.
- Allen KJ et al. Iron-overload-related disease in HFE hereditary hemochromatosis. New England Journal of Medicine. Melbourne Collaborative Cohort Study, n=31,192.
- Adams PC et al. Hemochromatosis and Iron-Overload Screening (HEIRS) Study, n=99,711. Prevalence of C282Y homozygosity by ancestry group.
- World Health Organization (2020). Serum ferritin concentrations for the assessment of iron status in individuals and populations: technical brief. Threshold of 15 µg/L; underlying cutoffs last revised 1993.