Expression of the Iron Hormone Hepcidin Distinguishes Different Types of Anemia in African Children

See allHide authors and affiliations

Science Translational Medicine  07 May 2014:
Vol. 6, Issue 235, pp. 235re3
DOI: 10.1126/scitranslmed.3008249


Childhood anemia is a major global health problem resulting from multiple causes. Iron supplementation addresses iron deficiency anemia but is undesirable for other types of anemia and may exacerbate infections. The peptide hormone hepcidin governs iron absorption; hepcidin transcription is mediated by iron, inflammation, and erythropoietic signals. However, the behavior of hepcidin in populations where anemia is prevalent is not well established. We show that hepcidin measurements in 1313 African children from The Gambia and Tanzania (samples taken in 2001 and 2008, respectively) could be used to identify iron deficiency anemia. A retrospective secondary analysis of published data from 25 Gambian children with either postmalarial or nonmalarial anemia demonstrated that hepcidin measurements identified individuals who incorporated >20% oral iron into their erythrocytes. Modeling showed that this sensitivity of hepcidin expression at the population level could potentially enable simple groupings of individuals with anemia into iron-responsive and non–iron-responsive subtypes and hence could guide iron supplementation for those who would most benefit.


Both host and pathogen need iron (1). Insufficient iron results in iron deficiency anemia and may impair cognitive development, especially in childhood (2, 3). Iron supplementation restores hematological indices and improves intelligence quotient in primary school age children (4). However, iron supplementation carries risks because it alters the gut microbiota (5) and promotes or exacerbates malaria and other infections (68). Furthermore, anemia in an individual may have other causes besides iron deficiency, as is the case with inherited globin disorders (which are associated with iron overload) or chronic inflammation (2). Infection may lead to chronic inflammation, sequestration of iron in macrophages, and a decrease in iron availability not only for invading pathogens but also for red blood cell formation (erythropoiesis) (9). Therefore, identification of those types of anemia that are most likely to benefit from iron supplementation is desirable.

We hypothesized that iron supplementation would be most effective, and probably safest, in individuals who are truly iron-deficient (ID). To identify this subgroup, we combined two frequently used definitions of iron deficiency: (i) a low concentration of serum ferritin (either <12 or <30 μg/liter in the presence of inflammation) and (ii) a raised index (that is, >2) of soluble transferrin receptor/log serum ferritin (sTfR-F). This definition encompasses World Health Organization (WHO) recommendations (10) and includes the sTfR-F index cutoff that identifies iron depletion in the bone marrow (11). This combined definition strictly ensured that children defined as ID had low iron stores and iron depletion in tissues.

Hepcidin is the iron regulatory hormone that controls iron absorption and distribution (12). High concentrations of hepcidin inhibit dietary iron absorption, even in the presence of iron deficiency (13, 14), whereas low concentrations of hepcidin facilitate release of iron from stores and increase its availability for erythropoiesis (12). Given the multifactorial regulation of hepcidin synthesis and its role in controlling systemic iron homeostasis, we hypothesized that plasma concentrations of hepcidin might reflect body iron status and requirements in settings where iron deficiency, iron deficiency anemia, anemia of inflammation, and high infectious burden coincide. Therefore, we evaluated whether hepcidin measurements could diagnose iron deficiency, distinguish iron deficiency anemia from anemia of inflammation, and identify those who would most likely benefit from iron supplementation. To achieve these aims, we measured hepcidin in samples from two large community cohorts of African children (aged 6 months to 6 years) from studies performed in The Gambia in 2001 and Tanzania in 2008. We also retrospectively reanalyzed data from a stable iron isotope study performed on a smaller group of anemic children from The Gambia in 2003.


ID children have greater iron incorporation in erythrocytes

To test how ID children responded to iron supplementation, we retrospectively applied our definition of iron deficiency to published data from a cohort of anemic Gambian children, who had the stable isotope 57Fe administered orally (15); cellular incorporation of the iron into erythrocytes was measured 2 weeks later (15). We found that children who met our criteria of iron deficiency used oral iron more efficiently [mean erythrocyte iron incorporation (EII), 28.9% versus 6.6% in non-ID children] (Fig. 1A).

Fig. 1. Plasma hepcidin identifies iron deficiency in African children.

(A) EII of orally administered 57Fe [Gambia-2003 cohort (15)] was increased in ID children. Iron deficiency was defined as follows: ferritin <12 μg/liter or below <30 μg/liter in the presence of inflammation, CRP >5 mg/liter or ACT >0.6 mg/liter based on WHO definitions (37, 38), and sTfR-F index >2. EII in ID children: 25.9% [95% confidence interval (CI), 18.7 to 35.8]; EII in non-ID children: 4.9% (95% CI, 2.8 to 8.5) (P < 0.0001). Red lines represent geometric mean and 95% CIs. (B) Hepcidin concentrations were significantly lower in children (Gambia-2001 and Tanzania-2008 cohorts) with iron deficiency (n = 207) (geometric mean, 1.9 ng/ml; 95% CI, 1.5 to 2.4) compared to the rest of the population (n = 1017) (geometric mean, 14.8 ng/ml; 95% CI, 13.7 to 15.9) (P < 0.0001). Red lines represent geometric mean and 95% CIs. (C) ROC curves for hepcidin to identify iron deficiency: for the Gambia-2001 cohort, the AUCROC was 0.86; for the Tanzania-2008 cohort, the AUCROC was 0.84 (not significantly different, P = 0.37); the AUCROC for the combined population was 0.85.

Hepcidin measurements identify children categorized as ID

Previously, we found that plasma hepcidin concentrations were an excellent predictor of oral iron incorporation into erythrocytes in African children (16). Because our definition of iron deficiency similarly identified children who efficiently used oral iron (Fig. 1A), we considered that plasma hepcidin might be of use to distinguish ID children within a population with mixed iron deficiency, anemia, and inflammation/infection. We measured hepcidin concentrations in plasma and hematological, inflammatory, nutritive, and iron indices in 709 children from The Gambia (2001 cohort) and 604 children from Tanzania (2008 cohort), and genotyped these children for inherited red cell disorders (table S1). In the two community cross-sectional samples combined (n = 1313), 61.2, 16.9, and 13.0% had anemia, iron deficiency, and iron deficiency anemia, respectively. The high prevalence of anemia is representative of much of sub-Saharan Africa, and the relative contribution of iron deficiency illustrates the complex etiology of anemia in such settings.

Hepcidin concentrations were significantly lower in children with iron deficiency compared to those without iron deficiency (Fig. 1B, P < 0.0001). We next generated receiver operating characteristic (ROC) curves and measured the area under the curve (AUCROC) for the ability of hepcidin to detect iron deficiency. The AUCROC for hepcidin to identify iron deficiency in Gambian children was 0.86 and in the Tanzanian group was 0.84; the difference between the two AUCROC was not significant (P = 0.37) (Fig. 1C). After combining the Gambian and Tanzanian populations, the AUCROC for hepcidin to detect iron deficiency was 0.85. Furthermore, there were no statistically significant differences in AUCROC for hepcidin to detect iron deficiency by sex (P = 0.66), age (>24/<24 months, P = 0.075), wasting (P = 0.22), anemia (distinguishing iron deficiency anemia from other causes of anemia, P = 0.37), sickle hemoglobin (HbS) carrier status (P = 0.54, The Gambia), glucose-6-phosphate dehydrogenase (G6PD) status (P = 0.92, The Gambia), or α+-thalassemia status (P = 0.47, Tanzania) (fig. S1, A to G). Asymptomatic Plasmodium falciparum parasitemia was detected in 25.3% of the children and was associated with increased plasma hepcidin compared with uninfected children (12.5 ng/ml versus 9.7 ng/ml, P = 0.009); however, the presence of parasitemia did not affect the AUCROC for hepcidin to detect iron deficiency (P = 0.24) (fig. S1H).

Hepcidin distinguishes iron deficiency anemia from anemia due to inflammation

Next, we evaluated the performance of plasma hepcidin for distinguishing anemic children (Hb <11 g/dl) with iron deficiency from those with anemia of inflammation [Hb <11 g/dl and C-reactive protein (CRP) >5 mg/liter or α1-antichymotrypsin (ACT) >0.6 mg/liter, with no iron deficiency 11.4% of the population]. Iron supplementation of individuals with anemia of inflammation is less likely to be effective and is potentially harmful (11). Hepcidin was significantly lower in children with iron deficiency anemia compared with those with anemia of inflammation (1.8 ng/ml versus 21.7 ng/ml, P < 0.0001) (Fig. 2A). The AUCROC for hepcidin to distinguish iron deficiency anemia from anemia of inflammation was 0.87 and 0.88 in the Gambian and Tanzanian groups, respectively (Fig. 2B). After combining data sets, the AUCROC for hepcidin to distinguish iron deficiency anemia from anemia of inflammation was 0.89.

Fig. 2. Hepcidin distinguishes iron deficiency anemia from anemia of inflammation/infection.

(A) Hepcidin was significantly lower in children with iron deficiency anemia (iron deficiency and Hb <11 g/dl, n = 159: geometric mean, 1.8 ng/ml; 95% CI, 1.4 to 2.3) compared with children with anemia due to inflammation/infection (defined as CRP >5 mg/liter or ACT >0.6 mg/liter and Hb <11 g/dl and no iron deficiency, n = 139: geometric mean, 21.7 ng/ml; 95% CI, 17.6 to 26.8) (P < 0.0001). Red lines represent geometric mean and 95% CIs. (B) ROC curves for the ability of hepcidin to distinguish iron deficiency anemia from anemia of inflammation/infection: for the Gambia-2001 cohort, the AUCROC was 0.87; for the Tanzania-2008 cohort, the AUCROC was 0.88 (not significantly different, P = 0.87); for the combined population of Gambia-2001 and Tanzania-2008 cohorts, the AUCROC was 0.89. (C) Youden indices [(sensitivity + specificity) − 1]: sensitivity and specificity for detection of iron deficiency at each cutoff of hepcidin. (Inset) Youden indices: sensitivity and specificity for hepcidin in the range of 2 to 12 ng/ml. The maximal Youden index occurred at a hepcidin cutoff of 5.5 ng/ml (sensitivity, 78.7%; specificity, 80.1%) and was stable between hepcidin cutoffs of 5 and 8 ng/ml (8.8% variation from maximal Youden index). (D) Youden indices: sensitivity and specificity to distinguish iron deficiency anemia from anemia of inflammation/infection at each cutoff of hepcidin. (Inset) Youden indices: sensitivity and specificity for hepcidin in the range of 2 to 12 ng/ml. The maximal Youden index was seen at a hepcidin cutoff of 5.39 ng/ml (sensitivity, 79.3%; specificity, 87.1%) and was stable between hepcidin cutoffs of 5 and 8 ng/ml (7.5% variation from maximal Youden index).

Hepcidin thresholds for defining iron deficiency and distinguishing iron deficiency anemia from anemia of inflammation are similar

We then evaluated the properties of different hepcidin cutoffs for detecting iron deficiency and for distinguishing iron deficiency anemia from anemia of inflammation in the combined population, evaluating sensitivity (correct identification of true positives), specificity (correct identification of true negatives), and Youden indices [(sensitivity/100 + specificity/100) − 1] (17) to define the cutoff of maximal combined sensitivity and specificity (table S2 and Fig. 2, C and D). Optimal Youden indices for detecting iron deficiency and distinguishing iron deficiency anemia from anemia of inflammation were seen at cutoffs of 5.5 and 5.4 ng/ml, respectively. Furthermore, for both outcomes, the Youden index was maximal and stable between hepcidin cutoffs of about 5 to 8 ng/ml (varying by less than 9% between these two hepcidin concentrations). EII at each hepcidin threshold was estimated as follows: EII = −16.7 × log10(hepcidin) + 31.9 (calculated from the 2003 Gambian cohort) (16).

Hepcidin has different diagnostic properties from ferritin and outperforms zinc protoporphyrin

Ferritin and zinc protoporphyrin (ZPP) are frequently used to assess iron status (18). Ferritin and hepcidin are both induced by iron stores and inflammation, but only hepcidin is directly suppressed by bone marrow demand for iron. In the combined Gambian 2001 and Tanzanian 2008 population, hepcidin was decreased in anemic compared to nonanemic children (8.3 ng/ml versus 14.9 ng/ml, P < 0.0001) (fig. S2A), but there was no difference in ferritin between anemic and nonanemic children (29.5 ng/ml versus 28.4 ng/ml, P = 0.53) (fig. S2B), demonstrating the nonequivalence of hepcidin and ferritin as indices of iron status and the added sensitivity of hepcidin as an indicator of erythropoietic activity. ZPP is increased by iron deficiency but is also raised in inflammatory anemias (19). Using samples from the Gambia-2001 cohort, we compared hepcidin to ZPP. The AUCROC for hepcidin was higher than that for ZPP for identifying iron deficiency (P < 0.001) and for distinguishing iron deficiency from anemia of inflammation (P = 0.001) (fig. S2, C and D).

Hepcidin predicts iron utilization

We then sought to determine whether hepcidin could identify not only iron deficiency anemia but also physiologic iron utilization as measured by cellular incorporation of iron into erythrocytes (EII). To do this, we reanalyzed data from The Gambia 2003 study of EII (15, 16). An EII >20% is considered to reflect an appropriate iron deficiency–mediated increase in iron utilization (20). Plasma hepcidin was lower in children with EII >20% compared to those with plasma hepcidin <20% (1.6 ng/ml versus 20.6 ng/ml, P < 0.0001). EII was lower in children with hepcidin >5.5 ng/ml (the hepcidin concentration at the optimal Youden index value determined above) than in children with hepcidin <5.5 ng/ml (n = 25, Fig. 3A). The AUCROC for hepcidin to predict EII >20% was 0.90 (Fig. 3B), which was not statistically different from the AUCROC for ferritin, sTfR, sTfR-F, ZPP, and Hb to predict EII >20% (0.88, 0.84, 0.86, 0.76, and 0.53, respectively; fig. S3, A to E).

Fig. 3. Plasma hepcidin identifies children who incorporate >20% of orally administered iron into erythrocytes.

(A) EII in the Gambia-2003 cohort is increased in children with hepcidin concentrations below 5.5 ng/ml. EII in children with hepcidin <5.5 ng/ml: geometric mean, 28.5% (95% CI, 21.0 to 38.7); EII in children with hepcidin exceeding 5.5 ng/ml: geometric mean, 6.2% (95% CI, 3.7 to 10.3) (P < 0.0001). Red lines represent geometric mean and corresponding 95% CIs. (B) ROC curve for hepcidin as an indicator of EII >20% in the Gambia-2003 cohort. AUCROC = 0.90 (95% CI, 0.78 to 1.00).

Modeling demonstrates potential utility of presupplementation screening for hepcidin

Finally, we modeled the potential impact of presupplementation screening on distribution of iron administration in these populations (Table 1). If a screen was based on anemia (Hb <11 g/dl), then 77% of ID children would have received iron, but iron would also have been given to 73% of children with P. falciparum parasitemia and to all children with anemia of inflammation. In contrast, if the screen was based on a hepcidin cutoff of <5.5 ng/ml, 77% of children with iron deficiency and 80% with iron deficiency anemia would have received iron, whereas only 20% of children with P. falciparum infection and 14% of children with anemia of inflammation would have received the iron supplement. Lower cutoffs of hepcidin would have reduced the proportion of individuals with infection and/or inflammation receiving iron, but also would have increased the risk that more children with iron deficiency anemia would fail to receive iron. Higher cutoffs would have ensured that more children with iron deficiency anemia received iron but could have exposed more infected children to potentially hazardous iron supplementation.

Table 1. Impact of different screening approaches on distribution of iron supplementation.
View this table:


Expression of hepcidin is regulated by a combination of various inputs derived from different physiological processes (that is, iron stores, inflammation, and erythropoiesis), which appear to act at the level of the promoter to control transcription (21). In this way, in a given individual, hepcidin generally facilitates iron availability despite variations in diet and developmental/erythropoietic requirements while retaining the ability to restrict iron under conditions of infection and inflammation (22). At the population level, this property and sensitivity of hepcidin could be exploited to identify individuals in need of iron supplementation while also distinguishing those with anemia due to iron deficiency from those with anemia due to infection or inflammation. Genetic lesions that cause increased hepcidin result in an iron deficiency anemia that is resistant to oral iron supplementation (13, 14).

Here, we show that low plasma hepcidin identifies anemic children in a Gambian and Tanzanian cohort who required iron supplementation. We determined that a hepcidin cutoff between 5 and 8 ng/ml using the Bachem enzyme-linked immunosorbent assay (ELISA) had optimal sensitivity and specificity for predicting iron deficiency and iron utilization in this cohort. The Youden index appeared to be stable across this range, implying that, in the field, small variations in assay performance or calibration were unlikely to considerably alter the clinical impact of the result. There are currently more than 20 available hepcidin assays, for which results correlate well, although absolute values vary (23, 24). Although cutoffs and reference ranges are not directly transferable between assays, interassay conversion factors are being developed (23) and could enable application of results reported here to other settings.

Smaller studies in Korean children and Australian female blood donors analyzed AUCROC for hepcidin to identify iron deficiency of 0.85 to 0.89, respectively (25, 26). Hepcidin also efficiently distinguished iron deficiency from anemia of inflammation in hospital cohorts in high-income settings of patients with AUCROC of 0.88 to 0.97 (27, 28). We have shown that hepcidin strongly correlates with EII (16) reflected by the AUCROC of 0.90 that we have reported here. Similarly, others have reported that hepcidin is the best predictor of iron absorption (29) and hematologic response to iron therapy (30), although studies in other contexts have reported that indices including ferritin may be comparable (31) or superior (32). Similarities between our data and previous findings suggest that hepcidin measurements should also be of value in screening for iron deficiency and distinguishing iron deficiency anemia from anemia of inflammation in both developed and developing countries. A limitation of our study is that noninvasive gold standard definitions for iron deficiency, against which we assessed the measurement of hepcidin, are imperfect. However, it is unlikely to be possible to measure bone marrow iron stores (a gold standard definition of iron deficiency) in a large community of nonhospitalized children such as the cohorts we studied here.

Measurement of hepcidin integrates the multiple stimuli regulating systemic iron homeostasis and compartmentalization into a single index. This provides an advantage over current approaches for measuring systemic iron status that requires combinations of separate measurements of iron stores (for example, ferritin), erythropoiesis and tissue iron requirement (for example, sTfR), and inflammation (for example, CRP). Further work should aim to define the molecular mechanisms that confer this sensitive control of hepcidin expression, and should test the utility of hepcidin for improving anemia treatment strategies in communities with complex anemia, iron deficiency, infection, inflammation, and hemoglobinopathy.


Study design

Research objectives. In children in a community rural African setting, we sought to determine the capacity of hepcidin to identify iron deficiency, to distinguish between iron deficiency anemia and anemia of inflammation/infection, and to model the effects of hepcidin screening before iron supplementation on the distribution of iron supplements.

Experimental design. Study of a diagnostic test using baseline samples from two longitudinal studies. Samples for analyses of the performance of hepcidin as an index of iron deficiency were obtained from baseline samples in two longitudinal studies (Gambia-2001 and Tanzania-2008).

Gambia-2001. Community children aged 2 to 6 years (n = 709) were recruited during the malaria season (July to August 2001) from rural villages in the West Kiang region of The Gambia, as described (33, 34). All 2- to 6-year-old children living in selected villages were eligible.

Tanzania-2008. Community children aged 6 to 60 months (n = 604) were enrolled from a highly malaria-endemic rural area—Handeni District, Tanzania; samples were collected at baseline for a randomized controlled trial evaluating effects of micronutrient supplementation (8). All children with height for age z score <−1.5 living in four selected villages were eligible.

For both studies, all samples with sufficient volume for analyses were included. Assuming a prevalence of iron deficiency of 30%, the combined sample size (n = 1313) permitted identification of sensitivities and specificities of 0.80 with a precision of ±5%.

Analyses of hepcidin as a predictor of EII were performed on samples from the following study.

Gambia-2003. Children (18 to 36 months) with postmalarial or nonmalarial anemia (Hb <11 g/dl) from local rural villages were recruited during the 2003 malaria season by Medical Research Council (MRC) Keneba, The Gambia, as described in detail previously (15, 16). Malaria was treated before administration of 57Fe. Children received 3.9 mg of 57Fe after a 2-hour fast. EII was measured after 14 days by mass spectrometry and expressed as a percentage of the dose administered (15, 16).


The Gambian studies were approved by the Gambian government and MRC Ethics Committee. The Tanzanian study was approved by the Ethical Review Committee of Wageningen University, the Netherlands, and the National Health Research Ethics Review Sub-Committee, Dar es Salaam, Tanzania. In all studies, written informed consent for all participants was provided by parents or carers.

Laboratory measurements

Gambia-2001/2003. Hb (Medonic CA 530 Hemoglobinometer), plasma ferritin (Microparticle Enzyme Immunoassay, IMx-Ferritin, Abbot Laboratories), sTfR (ELISA, R&D Systems), CRP (Dade Dimension particle enhanced turbidimetric immunoassay) or ACT (immunoturbidimetry, Cobas Mira Plus Bioanalyzer, Roche), and ZPP (Aviv Biomedical hematofluorometer) were measured. Thick and thin blood films were examined for Plasmodium parasites. Genotyping for HbS and G6PD A type deficiency was performed (34).

Tanzania-2008. P. falciparum infection was assessed by rapid diagnostic test (CareStart, G0121, Access Bio) and blood film. Hb (KX-21 Hematology Analyzer, Sysmex), CRP, ferritin, and sTfR (chemiluminescent immunoassay, UniCel DxC 880i, Beckman Coulter) were measured. Because Gambian and Tanzanian studies used different sTfR assays, values were not directly comparable; however, R&D ELISA and the Beckman Coulter assay data are linearly correlated (r = 0.96). Thus, we converted Beckman sTfR data using the following: R&D sTfR = (0.89 × Beckman sTfR) + 0.69 (35). α+-Thalassemia genotype was determined by polymerase chain reaction (36).

Hepcidin measurement

Plasma hepcidin was quantified by competitive ELISA [Hepcidin-25 (Human) EIA Kit, Bachem] by a trained scientist. Standards and samples were analyzed in duplicate or triplicate. Samples giving readings outside the standard curve linear region were repeated at appropriate dilutions. Readings with coefficient of variation >10% were repeated. The lower limit of detection (LOD) was estimated as 0.08 ng/ml on the basis of the hepcidin value corresponding to 3 SDs below the mean “no-hepcidin blank” OD450 (optical density at 450 nm); undiluted samples giving readings less than the LOD were reported as LOD/2 = 0.04 ng/ml. There are currently several available hepcidin assays, for which results generally correlate well, although absolute values reported vary. Therefore, although the cutoffs and reference ranges described in this study are not absolute values, the relationship between hepcidin and the other markers analyzed should be consistent and generally applicable.


The definitions used were as follows. Anemia: Hb <11 g/dl (10); inflammation: CRP >5 mg/liter (37) or ACT >0.6 mg/liter based on WHO definitions (38); ID: ferritin <12 μg/liter, or <30 μg/liter in the presence of inflammation, and sTfR/log10 ferritin (sTfR-F) index >2 [this definition combines (i) WHO recommendations (10) and (ii) the sTfR-F index that has been reported to identify bone marrow iron depletion in Finnish adults (11, 39)] (this was included to ensure that children with ferritin concentrations 12 to 30 μg/liter and inflammation also have evidence of tissue iron requirement, and ensures that children defined as ID were indeed likely to have low iron stores and tissue iron depletion); IDA: presence of both anemia and iron deficiency; AI: presence of anemia with concomitant inflammation, as well as no evidence of iron deficiency (ferritin ≥30 μg/liter and sTfR-F index ≤2); P falciparum–infected: presence of P falciparum by microscopy; “well” children: nonanemic, not ID, no evidence of inflammation, not infected with P. falciparum; wasting: weight for length z score <−2 SDs (contemporaneous WHO growth charts). EII (percentage of administered isotope-labeled 57Fe incorporated into erythrocytes after 14 days) >20% was considered a desirable response to iron deficiency (20).


Data were analyzed with Stata 11 (StataCorp). Log transformation was applied to normalize skewed distributions where necessary. Mean hepcidin values were compared using Student’s t tests on log-transformed data. For Gambia-2001 and Tanzania-2008 data sets, reference ranges (2.5th to 97.5th centile) were determined by estimating mean ± 1.96 SDs (log-transformed data) among total populations and specifically among well children. Nonparametric ROC curves comparing hepcidin with reference definitions were drawn, and AUCROC values were calculated. Sensitivities and specificities of selected cutoffs were evaluated, and Youden indices [which seek to identify the point where these are simultaneously optimal, that is, (sensitivity/100 + specificity/100) − 1] were calculated for specific hepcidin concentrations (17). Ninety-five percent CIs were estimated. Missing data were handled by listwise deletion.


Fig. S1. Stability of the performance of hepcidin as an index of iron deficiency in children of different sex, age, wasting, anemia, malaria, or hemoglobinopathy status.

Fig. S2. Comparison of characteristics of hepcidin with other indices of iron status.

Fig. S3. Hepcidin compared to other indices as a test for >20% of EII.

Table S1. Characteristics of the Gambia-2003, Gambia-2001, Tanzania-2008, and combined Gambia-2001/Tanzania-2008 study populations.

Table S2. Cutoffs for hepcidin to diagnose iron deficiency and distinguish iron deficiency anemia from anemia of inflammation/infection.


  1. Funding: Supported by a National Health and Medical Research Council Early Career Fellowship (Australia) and the Haematology Society of Australia and New Zealand New Investigator Award (S.-R.P.), MRC/Oxford University Clinical Academic Graduate School/Academy of Medical Sciences (S.H.A.), The Wellcome Trust (S.E.C.), INSTAPA project and European Union’s Seventh Framework Programme (FP7/2007-2013, no. 211484) (H.V.), the MRC UK (A.E.A., A.M.P., and H.D.), and The Netherlands Organisation for Scientific Research/WOTRO (grants W93-413 and WAO93-441), United Nations International Children’s Emergency Fund, Cornelis Visser Foundation and Wageningen University (Interdisciplinary Research and Education Fund), The Wellcome Trust (094780), MRC UK (MC-A760-5QX00), NIHR Oxford Biomedical Research Centre, and the Bill and Melinda Gates Foundation (“Hepcidin and Iron in Global Health”, OPP1055865). Author contributions: S.-R.P., S.H.A., A.E.A., A.M.P., and H.D. conceived and designed the study; S.H.A., J.V., S.E.C., C.P.D., A.Y.D., E.T., H.V., and A.M.P. collected the field samples; S.K., A.E.A., A.Y.D., and E.T. measured hepcidin on the samples; S.-R.P., S.H.A., L.A.E., A.E.A., A.M.P., and H.D. designed the statistical analysis; S.-R.P. and S.H.A. undertook the statistical analysis; S.-R.P., S.H.A., A.E.A., T.H., H.V., A.M.P., and H.D. drafted the manuscript; all authors edited and approved the final manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: The data for this study are available from the authors upon request.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article