Pediatric Hematology and Oncology, 29:472–478, 2012Copyright C© Informa Healthcare USA, Inc.ISSN: 0888-0018 print / 1521-0669 onlineDOI: 10.3109/08880018.2012.705230

THALASSEMIA

Indices Used in Differentiation of Thalassemia Traitfrom Iron Deficiency Anemia in PediatricPopulation: Are They Reliable?

Burcin Nalbantoglu, MD,1 Savas Guzel,2 Volkan Buyukyalcın, MD,2

M. Metin Donma, MD,1 Eda Celik Guzel, MD,3 Aysin Nalbantoglu, MD,4

Erkut Karasu, MD,1 and Burcu Ozdilek, MD1

1Department of Pediatrics, Faculty of Medicine, Namik Kemal University, Tekirdag, Turkey;2Department of Biochemistry, Faculty of Medicine, Namik Kemal University, Tekirdag,Turkey; 3Department of Family Medicine, Faculty of Medicine, Namik Kemal University,Tekirdag, Turkey; 4Department of Pediatrics, Muratlı State Hospital, Tekirdag, Turkey

Background: Iron deficiency (IDA) and beta thalassemia trait (TT) are the most common causesof hypochromia and microcytosis. Many indices have been defined to quickly discriminate thesesimilar entities via parameters obtained from automated blood cell analyzers. However, studiesin the pediatric age group are scarce and their results are controversial. Methods: We calculatedeight discrimination indices [Mentzer Index (MI), England and Fraser Index (E&F), Srivastava In-dex (S), Green and King Index (G&K), Shine and Lal Index (S&L), red blood cell (RBC) count, RBCdistribution width, and red blood cell distribution width Index (RDWI)] in 100 patients. We calcu-lated sensitivity (SENS), specificity (SPEC), positive and negative predictive value (PPV and NPV),and Youden’s Index (YI) of each discrimination index. Results: None of the discrimination indicesshowed a SENS and SPEC of 100%. The highest SENS was obtained with S&L (87.1%), while thehighest SPEC was obtained with E&F formula (100%). The highest YI value was obtained with E&Fformula (58.1%). Conclusion: In our study, none of the formulas appears reliable in discriminatingbetween TT and IDA patients. The evaluation of iron status and measurement of hemoglobin A2(HbA2) remain the most reliable investigations to differentiate between TT and IDA patients.

Keywords anemia, diagnosis, differential, iron deficiency, thalassemia

INTRODUCTION

Microcytic anemias are among the most common types of anemia encountered bypediatricians. The two most common causes of microcytic anemia are iron deficiency(IDA) and thalassemia trait (TT) [1]. To date, many discrimination indices have beenreported using red blood cell (RBC) indices obtained by automated blood count.Measurement of serum iron level (<30 mg/dL), serum iron binding capacity (SIBC)(>350), serum ferritin level (<10 ng/mL), and quantitation of hemoglobin A2 (HbA2)(>3.5%) are next steps for correct diagnosis [2–4].

Received 29 March 2012; accepted 18 June 2012.Address correspondence to Burcin Nalbantoglu, MD, Department of Pediatrics, Faculty ofMedicine, Namik Kemal University, Ugur Mumcu Street, Tekirdag 59100, Turkey. E-mail:bnalbantoglu@nku.edu.tr

Indices for Thalassemia and Iron Deficiency

However, these measurements are expensive and time-consuming for publichealth economy, particularly in countries with high prevalence of microcytosis andhypochromia, as in our country [4, 5]. Also in our country, general practitioners haveno chance to do these measurements except automated blood cell count. Various in-dices and formulas incorporating red cell indices derived from automated red cellanalyzers, such as RBC, hemoglobin (Hb), mean cell volume (MCV), and red cell dis-tribution width (RDW) in various combinations have been introduced to easily dis-criminate between IDA and TT patients [5–14]. However, none of these indices has asensitivity (SENS) and specificity (SPEC) of 100% in discrimination of IDA and TT. Thediagnostic reliability of these methods is controversial and differs from one country toother [4, 5]. Also, there are only a few studies conducted in pediatric group comparingreliability of these methods.

The aim of this study is to evaluate retrospectively the diagnostic reliability of someRBC indices and formulas in the differentiation of TT and IDA in a pediatric age grouppatients coming from the Thrace region, the European side of Turkey.

MATERIALS AND METHODS

This study is done retrospectively, from January 2011 to November 2011, with a cohortof 100 children of both sexes (48 males, 52 females) aged between 2 and 14 years (meanage 6.3 ± 1.6 years) with mild hypochromic microcytic anemia. Among these, 38 wereshown to be affected by IDA and 62 by TT. Diagnosis was performed in all patientsbased on the RBC count, Hb levels, iron assessment (serum iron (<30 mg/dL), SIBC(>350), serum ferritin (<10 ng/mL), transferin saturation <12% for IDA, and quanti-tative identification of HbA2 (>3.5%) for TT by high performance liquid chromatogra-phy. The patients with Hb value >9 g/dL were accepted as mild anemic according toWHO criteria, and the patients with Hb value <9 g/dL were excluded because moresevere anemia is not to be confused with TT [6]. Also, the patients having diagnosisof both IDA and TT at the same time were excluded. None of the children had acuteinfection based on C-reactive protein (CRP). None of the patients had any underlyingchronic diseases except IDA or TT. The children with evidence of inflammatory, ma-lignancy, recent surgery, or blood transfusion were excluded from the study. All of thedata were collected at the time of diagnosis. The reliability of following discriminationindices and formulas were calculated by using RBC indices as defined below:

1. Mentzer Index (MI): MCV/RBC;2. Shine and Lal Index (S&L): MCV2 × MCH × 0.01;3. England and Fraser (E&F): MCV − RBC − (5× Hb) − k (k is calculated to be 5.19 in

our counter as described in original report);4. Srivastava Index (S): MCH/RBC;5. Green and King Index (G&K): MCV2 × RDW/100 × Hb;6. RDW Index (RDWI): MCV × RDW/RBC;7. RDW;8. RBC count.

The differential values for each discrimination index were applied as defined inthe original published reports [7–15]. Complete blood counts were obtained by theCoulter Counter Max M System (London, UK). Serum iron and SIBC were deter-mined calorimetrically, and ferritin was measured turbidimetrically by using Hitachi-911 Automatic Analyzer (Boehringer Manheim, Tokyo, Japan) and commerciallyavailable kits. The values of HbA2 were determined by high performance liquidchromatography.

Copyright C© Informa Healthcare USA, Inc.

B. Nalbantoglu et al.

SENS, SPEC, positive and negative predictive values (PPV and NPV), efficiency(EFF), and Youden Index (YI) were evaluated for each index and formula as follow-ing:

1. SENS: TP/(TP + FN)2. SPEC: TN/(TN + FP)3. PPV: TP/(TP + FP)4. NPV: TN/(TN + FN)5. EFF: (TP + TN)/(TP + TN + FP + FN)6. YI: (SENS + SPEC) − 100.

Where TP means true positive, TN means true negative, FN means false negative, andFP means false positive.

Informed consent was obtained from parents. This study was approved by our in-stitute’s ethical committee.

RESULTS

The study included 100 children of both sexes (52 girls) aged between 2 and 14 years.The values of RBC, Hb, hematocrit, MCV, and RDW in patients with IDA and TT areshown in Table 1. Significant differences are observed in the RBC and MCV values.

The discriminant indices as well as SENS, SPEC, PPV, NPV, and YI are shown in Ta-ble 2. Our data show that S&L presents the highest SENS (87.1%), followed by RDWI(80.6%), RBC (77%), MI (67.8%), S (61.3%), E&F (58.1%), G&K (41.9%), and RDW(16.1%).

Furthermore, the E&F formula shows the highest SPEC (100%), followed by S(84.2%), G&K (84.2%), RDW (79.2%), MI (73.6%), RDWI (47.3%), S&L (31.5%), and RBC(10.5%).

The highest PPV is obtained with E&F formula (100%), followed by S (86.4%), G&K(81.3%), MI (80.8%), RDWI (71.4%), S&L (67.5%), RBC (65.8%), and RDW (55.5%). Thelowest NPV is obtained with RBC (22.2%), followed by RDW (36.6%), G&K (47.1%),S (57.1%), MI (58.3%), E&F (59.3%), G&K (60%), and RDWI (60.1%). Most reliableYI (58.1%) is obtained with E&F formula, followed by S (45.5%), MI (41.4%), RDWI(27.9%), G&K (26.1%), S&L (18.6%), RDW (−5%), and RBC (−12.5%).

TABLE 1 Hematological Data of the Groups

IDA (n: 38) BTT (n: 62)

Age (year) 6.31 ( ± 4.92) 5.8 ( ± 3.67)RBC (×1012/L) 4.82 ( ± 0.41)∗∗ 5.46 (0.74)∗∗

Hb (g/dL) 9.75 (1.32) 10.12 (0.95)MCV (fL) 70.02 (7.89)∗∗ 63.46 (9.62)∗∗

MCH (pg) 22.45 (3.82) 20.79 (3.39)RDW (%) 15.77 (2.68) 16.59 (2.78)Fe (µg/dL) 21.21 (42.06)∗∗ 74.16 (35.75)∗∗

UIBC (mg/dL) 398.15 (93.4)∗∗∗ 266.02 (70.91)∗∗∗

Ferritin (ng/mL) 6.09 (19.46)∗∗∗ 54.74 (52.35)∗∗∗

∗∗P < .01 (Mann–Whitney U-test).∗∗∗P < .001 (Mann–Whitney U-test).

Pediatric Hematology and Oncology

Indices for Thalassemia and Iron Deficiency

TABLE 2 Sensitivity (SENS), Specificity (SPEC), Positive Predictive Value (PPV), NegativePredictive Value (NPV), and Youden’s Index (YI) of Each Discrimination Index in DiscriminatingIDA from TT

Index Cut-off

BTT(male: 28,

female: 34)

IDA(male: 20,

female: 18) Sensitivity Specificity PPV NPVYouden’s

index

MI <13 42 10 67.8 73.6 80.8 58.3 41.4>13 20 28

S&L <1530 54 26 87.1 31.5 67.5 60 18.6>1530 8 12

E&F <0 36 0 58.1 100 100 59.3 58.1>0 26 38

S <3.8 38 6 61.3 84.2 86.4 57.1 45.5>3.8 24 32

G&K <65 26 6 41.9 84.2 81.3 47.1 26.1>65 36 32

RDWI <220 50 20 80.6 47.3 71.4 60 27.9>220 12 18<4.4 60 38 96.8 0 61.2 0 −3.2

RDW <14 10 8 16.1 79.2 55.5 36.6 −5>14 52 30

RBC >5 48 34 77 10.5 65.8 22.2 −12.5<5 14 4

Note. MI: Mentzer Index; S&L: Shine and Lal Index; S: Srivastava Index; E&F: England and FraserIndex; G&K: Green and King Index; RDWI: RDW Index.

DISCUSSION

The most frequently encountered diseases with mild microcytic anemia are TT andIDA [16, 17]. IDA represents the most common nutritional deficiency with the high-est prevalence among young children. Furthermore, in young children, IDA with ane-mia and also without overt anemia can result in severe delayed cognitive developmentthat may respond to early iron therapy [18]. However, differentiating TT from IDA iswarranted because the thalassemia heterozygote should not be given iron in a vain at-tempt to normalize MCV [19, 20]. Serum ferritin, serum iron, transferin saturation, andHbA2 level should be obtained for accurate diagnosis of IDA and TT, but these meth-ods are time-consuming and require laboratory staff. Until now, many investigatorsdefined many mathematical indices to discriminate IDA and TT with only completeblood count. However, none of these indices aimed to discriminate IDA and TT, whichhave 100% SENS and SPEC in any study.

Data from our study shows that both TT and IDA subjects have microcytosis, andfrequently hypochromia with significant differences in RBC and MCV values. The ma-jority of subjects with TT and IDA enrolled in the study presented with mild anemia(Hb 10.75 ± 1.62 g/dL and 11.12 ± 0.95 g /dL, respectively). The RDW, an index ofanisocytosis, was increased in both TT and IDA subjects, but without significant dif-ferences in relation to the mild degree of anemia.

In our study, none of the evaluated formulas or indices appears reliable in discrimi-nating between TT and IDA subjects, although in original reports, the authors reportedthe SENS of those indices as approximately 100% in detection of TT or IDA. However,later studies have shown that these indices correctly identify only 61–91% of the pa-tients with microcytic anemia due to TT or IDA.

Also, the results of our study were different from the outcomes of other studies ondifferent populations. The G&K was valid according to the results of Shen et al. and thestudies done by Sirdah et al. and Ntaios et al [4, 21, 22]. For the S&L, Rathod et al. [23]

Copyright C© Informa Healthcare USA, Inc.

B. Nalbantoglu et al.

showed its high validity in discriminating IDA and TT. However, Sirdah et al. [4] and AlFadhli et al. [24] proved that it had low validity. Demir et al. [5] and Shen et al. [21] alsofound it to be an invalid index in pediatric patients, which confirmed our findings.

The YI takes into account both SENS and SPEC, and gives an appropriate measureof validity of a particular question or technique. The comparison of YI with other in-dices was first made in a study by Demir et al. [5]. They found the YI of RBC countand RDWI were the highest, and they were the most reliable discrimination indicesin differentiation between TT from IDA. In our study, YI of E&F formula (58.1%) wasthe highest, but all the other formulas showed low values for YI. Also in our study,the highest PPV was obtained with E&F formula (100%), and the lowest NPV was ob-tained with RBC count (22.2%). Data from our study are in agreement with other re-cent reports [25]. But, despite the fact that we conducted our study in the same countrywith Demir et al., we found different results. It is possible that the differences we haveshown with Demir et al.’s report and with some other previous reports could be relatedto different molecular spectrum of β-thalassemia disorder in various countries, evenin different regions of the same country. Ferrara et al. [25, 26] have reported the distri-bution of different β-chain defects that they have found in 310 β-thalassemia carriers,with variable occurrence of the different β-chain defects, with a higher prevalence ofβ0 alterations. It is possible that the severity of β-chain molecular mutations as wellas the degree of anemia of IDA subjects may influence RBC parameters. Rund et al.[27] reported that different MCV values are significantly correlated with different β-thalassemia mutations. Controversy of previous studies on the accuracy of these in-dices may be explained by the different sample size, which was much larger in someof them [28]. Another reason may be the fact that some investigators excluded IDA pa-tients with lower Hb levels as our study [22]. However, in a study by Al Fadhli et al. [24],the authors concluded that E&F is highly reliable in distinguishing TT from IDA. Onan attempt to explain the different results compared with Demir’s study, the authorssuggested that these might be caused by the different sample size (63 in Demir’s studycompared with 103 in Al Fadhli’s study) as well as by the different mean age (<16 and>16 years, respectively). However, if we compare the Al Fadhli’s study with two otherwell-conducted studies, these differences do not exist, as all patients were >16 years ofage and the sample size was 103, 111, and 493, respectively. This means that there maybe a different explanation for the results in Al Fadhli’s study, other than the sample sizeand age of the patients [22, 28] (Table 3).

Besides the indices evaluated in our study, there are some recent indices reported.Eivazi-Ziaei et al. [29] found that the red cell flags (RCF) could be used for the

TABLE 3 Summary of the Previous Studies that Investigated the Reliability of the Indices Used inDifferentiation of IDA from TT

RBC RDW Mentzer Green and King RDWI E&F S&L S

Demir et al. [5]YI 82 30 48 67 80 56 0 37Ntaios et al. [22]YI 62.67 −7.32 58.08 70.86 56.3 63.23 – –Sirdah et al. [4]YI∗ – – 83.2 84.3 84.18 82.55 73.81 80.65Ferrara et al. [25]YI 63 −7.11 56.9 64 56.2 64.2 – –Beyan et al. [28]YI 73.7 2.6 64.6 65.5 63.4 57.6 11.1 50.2Al Fadhli et al. [24]YI 78.7 4.5 60.1 47.6 49.8 98.2 0.5 54.9Shen et al. [21]YI 65.9 −0.6 55.1 79.8 80.9 31.7 −0.8 36.0YI∗ 69.5 61.7 63.5 85.4 81.4 80.3 −0.8 38.0

Note. RBC: red blood cell count; RDW: red cell distribution width; RDWI: red cell distributionwidth index; E&F: England and Fraser Index; S&L: Shine and Lal Index; S: Srivastava Index; YI:Youden Index.YI∗: Youden Index was calculated according to the revised cut-off values for each index.

Pediatric Hematology and Oncology

Indices for Thalassemia and Iron Deficiency

discrimination of IDA and TT. Ceylan et al. investigated the role of reticulocytic indicesin the differential diagnosis of IDA and TT. They evaluated several reticulocytic indices,such as the absolute reticulocyte count, the percentage of corrected reticulocytes, themean Hb concentration of reticulocytes, the mean volume of the reticulocytes, and themean cell Hb concentration of the reticulocytes, only to find that absolute reticulocytecount and percentage of corrected reticulocytes were significantly different betweenIDA and TT [30]. Ntaios and Chatzinikolaou [31] speculated that the combination oferythrocytic and reticulocytic parameters might add on their efficiency to differentiateIDA and TT, and the incorporation of them in a common formula with possibly morediscriminative power for these two types of anemia. But, to our opinion more studiesare needed to confirm the validity and reliability of both of the indices.

The SENS and SPEC of calculated formulas differ from one country to another, evenin the different regions of a same country. So, every population must develop theirunique formula for a reliable discrimination between IDA anemia and β-TT. Up todate, none of the indices seems reliable enough to discriminate IDA from TT. Al-though, to establish serum iron status and HbA2 are expensive and time-consuming,these tests must be done on every patient with microcytosis for differential diagnosisfor an accurate diagnosis if possible.

CONCLUSION

In our study, none of the formulas appears reliable in discriminating between TT andIDA patients. The evaluation of iron status and measurement of HbA2 remain the mostreliable investigations to differentiate between TT and IDA patients.

Declaration of InterestThe authors report no conflicts of interest. The authors alone are responsible for thecontent and writing of the paper.

REFERENCES

[1] Kiss T, Ali M, Levine M, Lafferty J.D. An algorithm to aid in the investigation of thalassemia trait inmulticultural populations. Arch Pathol Lab Med. 2000;124(9):1320–1323.

[2] Chrobak L. Microcytic and hypochromic anemias. Vintr Lek. 2001;47:166–174.[3] Ntaios G, Chatzinikolaou A, Saouli Z, et al. Discrimination indices as screening tests for β-

thalassemic trait. Ann Hematol. 2008;87:329–330.[4] Sirdah M, Tarazi E, Al Najjar E, Al Haddad R. Evaluation of the diagnostic reliability of different RBC

indices and formulas in the differentiation of the β-thalassemia minor from iron deficiency in Pales-tinian population. Int J Lab Hematol. 2008;30:324–330.

[5] Demir A, Yarali N, Fisgin T, et al. Most reliable indices in differentiation between thalassemia traitand iron deficiency anemia. Pediatr Int. 2002;44:612–616.

[6] Demaeyer E. Preventing and controlling iron deficiency anaemia through primary health care. AGuide for Health Administrators and Programme Managers. Geneva: World Health Organization;1989.

[7] Mentzer WC. Differentiation of iron deficiency from thalassemia trait. Lancet. 1973;1:882.[8] Green R, King R. A new red blood cell discriminant incorporating volume dispersion for differentia-

tion iron deficiency anemia from thalassemia minor. Blood Cells. 1989;15:481–495.[9] Jayabose S, Giavanelli J, Levendoglu Tugal O, et al. Differentiating iron deficiency anemia from tha-

lassemia minor by using an RDW-based index. J Pediatr Hematol. 1999;21:314.[10] England JM, Fraser PM. Differentiation of iron deficiency from thalassemia trait by routine blood

count. Lancet. 1973;1:449–452.[11] Bassman JD, Feinstein DI. Quantitative anisocytosis as a discriminant between iron deficiency and

thalassemia minor. Blood. 1979;53:288–293.[12] England JM, Fraser P. Discrimination between iron deficiency and heterozygous-thalassemia syn-

dromes in differential diagnosis of microcytosis. Lancet. 1979;1:145–148.

Copyright C© Informa Healthcare USA, Inc.

B. Nalbantoglu et al.

[13] Perutelli P. Red blood cell distribution width in microcytosis. Haemotologica. 1989;74:221–222.[14] Shine I, Lal S. A strategy to detect beta-thalassemia minor. Lancet. 1977;1:692–694.[15] Srivastava PC, Bevington JM. Iron deficiency and-or thalassemia trait. Lancet. 1973;1:832.[16] Wharton BA. Iron deficiency in children: detection and prevention. Br. J Haematol.

1999;106:270–280.[17] Oski FA. Iron deficiency in infancy and childhood. N Engl J Med. 1993;329:190–193.[18] Umbreit J. Iron deficiency: a concise review. Am J Hematol. 2005;78:225–231.[19] Olivieri NF. The beta-thalassemias. N Engl J Med. 1999;341:99–109.[20] Lukens JN. The thalassemias and related disorder: an overview. In: Lee GR, et al., eds. Wintrobe’s

Clinical Hematology. 10th ed. Giza: Mass Publishing; 1999:405–433.[21] Shen C, Jiang YM, Shi H, et al. Evaluation of indices in differentiation between iron deficiency anemia

and β thalassemia trait for Chinese children. J Pediatr Hematol Oncol. 2010;32(6):218–222.[22] Ntaios G, Chatzinikolaou A, Saouli Z, et al. Discrimination indices as screening tests for β-

thalassemic trait. Ann Hematol. 2007;86:487–491.[23] Rathod DA, Kaur A, Patel V, et al. Usefulness of cell counter-based parameters and formulas in de-

tection of β-thalassemia trait in areas of high prevalence. Am J Clin Pathol. 2007;128:585–589.[24] Al Fadhli SM, Al-Awadhi AM, AlKhaldi D. Validity assessment of nine discriminant functions used

for the differentiation between iron deficiency anemias and thalassemia minor. J Trop Pediatr.2007;53:93–97.

[25] Ferrara M, Capozzi L, Russo R, et al. Reliability of red blood cell indices and formulas to discriminatebetween β thalassemia trait and iron deficiency in children. Hematology. 2010;15(2):112–115.

[26] Ferrara M, Matarese SM, Francese M, et al. Hematological and molecular analysis of beta-thalassemia and Hb Lepore in Campania, Italy. Hemoglobin. 2001;25(1):29–34.

[27] Rund D, Filon D, Strauss N, et al. Mean corpuscular volume of heterozygotes for beta-thalassemiacorrelates with the severity of mutations. Blood. 1992;79:238–243.

[28] Beyan C, Kaptan K, Ifran A. Predictive value of discrimination indices in differential diagnosis of irondeficiency anemia and beta-thalassemia trait. Eur J Hematol. 2007;78:524–526.

[29] Eivazi-Ziaei J, Dastgiri S, Pourebrahim S, et al. Usefulness of red blood cell flags in diagnosing anddifferentiating thalassemia trait from iron deficiency anemia. Hematology. 2008;13:253–256.

[30] Ceylan C, Miskioglu M, Colak H, et al. Evaluation of reticulocyte parameters in iron deficiency, vita-min B (12) deficiency and beta-thalassemia minor patients. Int J Lab Hematol. 2007;29:327–334.

[31] Ntaios G, Chatzinikolaou A. Incorporation of erythrocytic and reticulocytic parameters in a commonformula: will it provide a more accurate differentiation between iron deficiency anemia and beta-thalassemic trait?. Int J Lab Hematol. 2009;31:579–580.

Pediatric Hematology and Oncology