International Journal of Laboratory Hematology

The Official journal of the International Society for Laboratory Hematology

ORIGINAL ARTICLE INTERNAT IONAL JOURNAL OF LABORATO RY HEMATO LOGY

Empty iron stores in children and young adults—the diagnostic

accuracy of MCV, MCH, and MCHC
A. E. 
ASBERG*, G. MIKKELSEN † , M. W. AUNE ‡ , A. 
ASBERG †

*Department of Pediatrics,
Trondheim University Hospital,
Trondheim, Norway
†
Department of Clinical
Chemistry, Trondheim
University Hospital, Trondheim,
Norway
‡
Department of Immunology
and Transfusion Medicine,
Trondheim University Hospital,
Trondheim, Norway
Correspondence:
Ann Elisabeth  Asberg, Depart-
ment of pediatrics, St. Olavs
Hospital, 7006 Trondheim,
Norway.
Tel.: +47 72574846;
Fax: +47 72575501;
E-mail: ann.asberg@stolav.no
doi:10.1111/ijlh.12132
Received 5 March 2013;
accepted for publication 9 July
2013
Keywords
Erythrocyte indexes,
hypochromic anemia, iron
deficiency anemia, mean cell
volume, mean corpuscular
hemoglobin, mean corpuscular
hemoglobin concentration
S U M M A RY
Introduction: Erythrocyte mean cell volume (MCV) is used clinically
to classify anemia, and normal values may be used to exclude iron
deficiency. We have studied the diagnostic accuracy of MCV and
the related measures mean cell hemoglobin (MCH) and mean cell
hemoglobin concentration (MCHC) in diagnosing empty iron stores
in children and young adults.
Methods: Diagnostic accuracy of MCV, MCH, and MCHC was studied
by ROC curve analysis in 6443 ambulant patients aged 0.5–
25 years, of which 476 were anemic. In all patients, blood hemoglo-
bin, MCV, MCH, and serum ferritin were measured in specimens
sampled at the same time. MCHC was calculated as MCH divided by
MCV. The gold standard of empty iron stores was s-ferritin <10, 15,
or 20 lg/L. The cutoff limit of MCV giving 90% sensitivity in diag-
nosing serum ferritin <15 lg/L was constructed using quantile
regression.
Results: Generally, MCH was slightly more accurate than MCV and
MCHC. In the whole study population, the area under the ROC
curve was 0.68–0.93 for MCV, 0.73–0.96 for MCH, and 0.68–0.87
for MCHC; and 0.70–0.86, 0.71–0.89, and 0.68–0.88, respectively,
in the anemic subpopulation. At the cutoff limits of MCV giving a
sensitivity of 90% at all ages in anemic patients, the specificity was
about 50%.
Conclusion: Mean cell hemoglobin, MCH, and MCHC are only mod-
erately accurate in diagnosing empty iron stores in children and
young adults, and normal values of these tests do not exclude
empty iron stores in anemic patients.

of iron-deficient children have iron deficiency anemia,

INTRODUCTION
which is supposed to be microcytic [2]. Reading diag-
Iron deficiency is common in children in Norway [1] as nostic guidelines, one gets the impression that iron
well as in other parts of the world [2]. About one-third deficiency anemia is always microcytic [3,4], although

98 © 2013 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2014, 36, 98–104
 A. E. 
ASBERG ET AL. | DIAGNOSTIC ACCURACY OF ERYTHROCYTE INDEXES
we have known for decades that it may very well be
normocytic [5]. In fact, the diagnostic accuracy of
mean cell volume (MCV) in diagnosing iron deficiency
is fairly moderate in adults with anemia [6] and very
little investigated in children. In this study, we evalu-
ated the diagnostic accuracy of MCV in diagnosing
empty iron stores in Norwegian children and young
adults with and without anemia, using s-ferritin as the
gold standard. We compared the diagnostic accuracy of
MCV and the related measures mean cell hemoglobin
(MCH) and mean cell hemoglobin concentration
(MCHC).
METHODS
Population
We used laboratory data from the Clinics of Laboratory
Medicine at Trondheim University Hospital. These labo-
ratories serve nearly all inpatient and outpatient activi-
ties in the hospital and receive specimens from primary
care physicians and other healthcare facilities for outpa-
tients in the County of Sør-Trøndelag. The hospital is a
tertiary care centre for about 6 88 000 inhabitants and
a secondary care centre for about 2 98 000 inhabitants.
Specimens sent to the hospital for analysis are mostly
from primary care centers.
We collected data between October 10, 2005, and
March 19, 2012, from ambulant patients that were
<25 years old at the time of specimen sampling and
kept only records for patients with complete data sets
on serum ferritin (s-ferritin), blood hemoglobin
(b-hemoglobin), MCV, and MCH analyzed in speci-
mens sampled at the same time. Then all duplicate
records were deleted, so that the data file only con-
tained one (the oldest) record for each individual.
Finally, to secure that as many as possible of the
records represented a new clinical event and not a
control situation, we deleted all records from 2005
and 2006. As a result, the record used for all patients
was the first record to satisfy the criteria since October
2005, so even for patients registered with a record
from January 2007, at least 14 months had gone.
Laboratory analysis
B-hemoglobin, MCV, and MCH were mostly analyzed
on Sysmex XE-2100 (Sysmex, Kobe, Japan) and few
© 2013 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2014, 36, 98–104
99
specimens on either Abbott CELL-DYN 4000, Siemens
Bayer ADVIA 120 or ABX Micros 60, with reagents
from the manufacturers. Sysmex XE-2100 was the
main instrument; however, in certain cases, when
the main instrument did not yield valid results for
some parameters, another instrument would be used.
The laboratory secured that b-hemoglobin, MCV, and
MCH from the different instruments were directly
comparable, using fresh blood specimens analyzed on
Sysmex XE-2100 to calibrate the other instruments.
Specimens from three patients were assayed on each
instrument every weekday to check agreement
between the instruments. On the Sysmex XE-2100,
b-hemoglobin was measured using the sodium lauryl
sulfate (SLS)-hemoglobin method with a calibrator
traceable to the reference method recommended by
the International Council for Standardization in Hae-
matology (ICSH). B-hematocrit was measured using
the erythrocyte aperture–impedance pulse height
detection method, MCV was calculated as b-hemato-
crit divided by blood erythrocyte particle concentra-
tion, MCH as b-hemoglobin divided by blood
erythrocyte particle concentration, and MCHC as MCH
divided by MCV. MCV was calibrated with a calibrator
traceable to reference methods that follow the recom-
mendations of ICSH and The Clinical and Laboratory
Standards Institute (CLSI) standards for MCV. Using
Sysmex XE-2100, we have previously found MCV to
be sufficiently stable in specimens kept at room tem-
perature (20 °C) for 14 h. If pre-analytical storage time
exceeded 14 h, MCV was analyzed using the instru-
ment ABX Micros 60. In this way, MCV could be reli-
ably analyzed in specimens kept at room temperature
for up to 48 h. Between-day coefficients of variation
for b-hemoglobin, MCV, and MCH were 0.8% at
12.3 g/dL, 0.8% at 82 fL, and 0.9% at 28 pg/L, respec-
tively. In classifying patients as anemic, we used the
following local b-hemoglobin lower reference limits
(in g/dL) in healthy persons: 0–1 days, 14.5; 1–2 days,
14.0; 2–7 days, 14.3; 1–2 weeks, 13.5; 2–4 weeks,
10.8; 1–2 months, 9.0; 2–12 months, 10.0; 1–4 years,
10.5; 4–8 years, 10.8; 8–14 years, 11.1; girls >14 years,
11.7; boys >14 years, 13.4. For MCV, MCH, and
MCHC, we used the following local lower reference
limits in healthy persons: MCV (in fL): 0–5 days, 95;
5 days-2 months, 87; 2–12 months, 72; 1–7 years, 75;
7–14 years, 77; >14 years 82. MCH (in pg): 0–7 days,
31.0; 1–4 weeks, 28.0; 1–2 months, 29.0; 2–4 months,
100 A. E. 
ASBERG ET AL. | DIAGNOSTIC ACCURACY OF ERYTHROCYTE INDEXES

27.0; 4–12 months, 24.0; 1–4 years, 23.9; 4–12 years,

(32–35)

(32–35)

(32–35)

(32–36)

(32–36)
MCHC
24.1; 12–18 years, 25.0; >18 years, 27.1. MCHC (in g/
dL): 0–2 months, 25.8; 2 months-4 years, 27.1;

33

34

34

34

34
Table 1. Population characteristics for age groups of 5-year intervals. Median values are given for s-ferritin (lg/L), b-hemoglobin (b-Hb) (g/dL),
4–9 years, 30.6; 9–12 years, 31.2; >12 years, 31.7.

(23–29)

(25–30)

(25–31)

(26–32)

(27–32)
All biochemical analyses were carried out on a

MCH
Roche Modular P system (Roche Diagnostics GmbH,

27

28

28

30

30
Mannheim, Germany), with reagents from the manu-
facturer (s-ferritin, s-iron and s-transferrin and s-CRP)

(71–85)

(76–88)

(76–90)

(78–93)

(82–94)
and Diagnostic Systems GmbH, Holzheim, Germany

MCV
(s-CRP). S-ferritin was measured using a ‘sandwich’

79

82

84

87

88
immunological method with electrochemilumines-
cence detection. The calibrator was traceable to WHO

(10.6–13.4)

(11.5–14.3)

(11.9–15.1)

(13.0–16.6)

(13.7–16.9)
Ferritin 80/602 First International Standard. Between-
day coefficients of variation were 6.9% at 5.0 lg/L

B-Hb
and 4.1% at 50 lg/L.

12.1

12.8

13.5

14.8

15.2
All analyses were monitored using appropriate
internal and external quality control systems.

(20–191)

(35–318)
Males (n = 2450)

(12–94)

(14–84)
S-ferritin

(9–85)
MCV (fL), MCH (pg), and MCHC (g/dL). The number in parenthesis is 5 and 95 percentiles
Statistical analysis

120
26

34

37

67
The diagnostic accuracy of the various tests was stud-

189

352

508

667

734
ied using receiver operating characteristic (ROC) curve

n
analysis [7]. We did separate ROC curve analysis for

(31–35)

(32–35)

(32–35)

(31–35)

(32–35)
females and males, using different cutoff values of
MCHC

s-ferritin, 10, 15, and 20 lg/L, as gold standards for 33

34

34

33

33
empty iron stores. These analyses were repeated in
the subgroup of anemic patients. We made 12 statisti-
(21–29)

(25–30)

(25–31)

(24–32)

(25–32)
cal comparisons between the diagnostic accuracy of
MCH

different tests, so we considered P-values <0.05/

27

28

29

29

30
12 = 0.0042 as statistically significant. To investigate
the influence of age and gender on certain percentiles
(69–85)

(75–89)

(78–92)

(77–94)

(78–95)
of MCV in anemic patients, we used quantile regres-
MCV

sion with multivariable fractional polynomials [8] to

80

83

85

88

89

find the simplest nonlinear transformation (if any) of

continuous variables. Medians were compared using
(10.2–13.5)

(11.4–14.0)

(11.3–14.6)

(11.0–14.7)

(11.2–14.8)

the Wilcoxon rank sum (Mann–Whitney) test. For all

statistical analyses, we used the Stata software (http://
B-Hb

12.1

12.9

13.3

13.3

13.3

www.stata.com/), version 12.

The study was approved by The Regional Commit-
Females (n = 3993)

tee for Medical and Health Research Ethics (case no

(6–109)

(7–129)
(11–99)
S-ferritin

(8–82)

(8–94)

2012/160/REK midt).
26

33

32

33

38

R E S U LT S
107

289

601

1273

1723

Characteristics of the study population of 6 443

n

patients are given in Tables 1 and 2. S-CRP was mea-

10–14

15–19

20–24

sured in 1 460 patients and was <10 mg/L in 1321

Age

0–4

5–9

(90.5%). Median s-ferritin was 42 lg/L in the group

© 2013 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2014, 36, 98–104
 A. E. 
ASBERG ET AL. | DIAGNOSTIC ACCURACY OF ERYTHROCYTE INDEXES 101

Table 2. The number of patients with anemia and s-ferritin below 10, 15, and 20 lg/L is given for age groups of 5-
year intervals. The number in parenthesis is percentages of total for each interval

Females (n = 3993) Males (n = 2450)

S-ferritin below S-ferritin below

Age Total Anemia 10 lg/L 15 lg/L 20 lg/L Total Anemia 10 lg/L 15 lg/L 20 lg/L

0–4 107 9 (8.4) 9 (8.4) 21 (19.6) 35 (32.7) 189 7 (3.7) 10 (5.3) 24 (12.7) 46 (24.3)
5–9 289 9 (3.1) 8 (2.8) 25 (8.7) 49 (17.0) 352 5 (1.4) 7 (2.0) 26 (7.4) 53 (15.1)
10–14 601 31 (5.2) 34 (5.7) 75 (12.5) 124 (20.6) 508 52 (10.2) 12 (2.4) 26 (5.1) 59 (11.6)
15–19 1273 127 (10.0) 134 (10.5) 215 (16.9) 334 (26.2) 667 56 (8.4) 12 (1.8) 23 (3.5) 32 (4.8)
20–24 1723 153 (8.9) 146 (8.5) 240 (13.9) 361 (21.0) 734 27 (3.7) 6 (0.8) 14 (1.9) 14 (1.9)

Table 3. Diagnostic accuracy of MCV, MCH, and MCHC in diagnosing empty iron stores in all 6443 patients

Area under the ROC curve (95% confidence

Number of patients interval)
Limit of s-ferritin
defining empty With empty Without empty
iron stores iron stores iron stores Total MCV MCH MCHC

Females
<10 lg/L 331 3662 3993 0.810 0.857* 0.819
(0.782–0.838) (0.834–0.881) (0.794–0.845)
<15 lg/L 576 3417 0.741 0.787* 0.729
(0.718–0.765) (0.766–0.808) (0.705–0.754)
<20 lg/L 903 3090 0.680 0.725* 0.682
(0.660–0.702) (0.706–0.745) (0.661–0.703)
Males
<10 lg/L 47 2403 2450 0.929 0.959* 0.865
(0.902–0.956) (0.943–0.975) (0.804–0.926)
<15 lg/L 113 2337 0.805 0.847* 0.744
(0.760–0.851) (0.811–0.884) (0.697–0.791)
<20 lg/L 204 2246 0.804 0.834* 0.685
(0.774–0.834) (0.809–0.859) (0.645–0.724)

*The area under the ROC curve of MCH is statistically significantly larger than the area under the ROC curve of the
second most accurate test (P < 0.002).

of 1 321 patients with s-CRP <10 mg/L compared
with 65 lg/L in the 139 patients with higher s-CRP
(P < 0.0001). Median s-ferritin was 40 lg/L in those
4983 patients, where s-CRP was not measured, not
significantly different from the group with s-CRP
<10 mg/L (P = 0.19). S-creatinine was measured in
4565 patients and found to be normal in 4539
(99.4%).
Table 3 gives the area under the ROC curve of
MCV, MCH, and MCHC for three cutoff values of the
gold-standard s-ferritin in the whole study population.
For all definitions of empty iron stores and in both
females and males, MCH had a statistically sign-
ificantly larger area under the ROC curve than the
second most accurate test (P < 0.002 for all compari-
sons). All tests showed a better diagnostic accuracy in
males than in females. Table 4 gives corresponding
results for the anemic subpopulation of 476 patients.
In this subpopulation, MCH showed better diagnostic
accuracy than the second most accurate test in all
groups, but the differences did not reach statistical
significance.
In the anemic subpopulation of 476 patients, using
the lower, age-specific reference limits as cutoff limits,

© 2013 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2014, 36, 98–104
102 A. E. 
ASBERG ET AL. | DIAGNOSTIC ACCURACY OF ERYTHROCYTE INDEXES

Table 4. Diagnostic accuracy of MCV and MCH in diagnosing empty iron stores in the subgroup of 476 anemic
patients

Area under the ROC curve

Number of patients (95% confidence interval)
Limit of s-ferritin
defining empty With empty Without empty
iron stores iron stores iron stores Total MCV MCH MCHC

Females
<10 lg/L 178 151 329 0.757 0.774 0.726
(0.702–0.811) (0.721–0.826) (0.671–0.780)
<15 lg/L 208 121 0.719 0.737 0.700
(0.659–0.780) (0.678–0.796) (0.643–0.758)
<20 lg/L 230 99 0.700 0.714 0.680
(0.633–0.768) (0.649–0.780) (0.618–0.742)
Males
<10 lg/L 29 118 147 0.856 0.894 0.881
(0.790–0.923) (0.843–0.946) (0.814–0.947)
<15 lg/L 40 107 0.790 0.823 0.805
(0.709–0.872) (0.745–0.902) (0.726–0.884)
<20 lg/L 49 98 0.766 0.792 0.756
(0.685–0.847) (0.714–0.870) (0.674–0.837)

the sensitivity and specificity (95% confidence

interval) of MCV were 71.8% (66.2–77.4%) and
66.7% (60.5–72.8%), respectively, when empty iron
stores were defined as s-ferritin <15 lg/L. The corre-
sponding figures for MCH were 71.4% (65.7–77.0%)
and 74.1% (68.4–79.8%), and for MCHC 51.6%
(45.4–57.8%) and 82.9% (78.0–87.8%), respectively.
Figure 1 shows the 90 percentile of MCV of anemic
patients with empty iron stores defined as s-ferritin
<15 lg/L. Choosing the 90 percentile as cutoff limit
for diagnosing empty iron stores implies choosing a
sensitivity of 90%. The 90 percentile increases from
78 fL in children at 1 year of age to 90 fL in adults at
25 years of age. The lower reference limits are below
the 90 percentiles at all ages, indicating a lower than
90% sensitivity if the lower reference limits are used
as cutoff limits. Also shown in Figure 1 is the 50 per-
centile of MCV in anemic patients without empty iron
stores, which is close to the 90 percentile of MCV in
anemic patients with empty iron stores. Choosing a
sensitivity of 90% therefore implies choosing a speci-
ficity of about 50%. These percentiles were not statis-
tically significantly associated with neither gender nor
an interaction factor between gender and age, and
there was no statistically significant nonlinear associa-
tion with age.
DISCUSSION
Eighty years after its introduction [9], MCV is still
used to classify anemia into micro-, normo- and mac-
rocytic forms, while MCH and MCHC are seldom
mentioned in this context. We show that MCH gener-
ally is more accurate than MCV in diagnosing empty
iron stores, while MCHC is less accurate. Interestingly,
the tests are less accurate in the subpopulation of ane-
mic patients where they are supposed be most useful
(Table 4). In fact, MCV, MCH, and MCHC are only
moderately accurate in diagnosing empty iron stores
in the anemic subpopulation, with an area under the
ROC curve of 0.7–0.8 when empty iron stores are
defined as s-ferritin <15 lg/L. For MCV, our findings
are close to the results found by Guyatt in a meta-
analysis of 436 anemic patients above 18 year of age
where iron deficiency was determined by bone mar-
row examination [6]. They estimated the area under
the ROC curve for MCV to be 0.76.
When using MCV to classify anemia, we ought not
to falsely classify iron-deficient anemic patients as iron
replete. Normal MCV does not exclude iron defi-
ciency, as shown by Beutler several decades ago [5]
and verified also in our study where values from
many patients with empty iron stores are higher than

© 2013 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2014, 36, 98–104
 A. E. 
ASBERG ET AL. | DIAGNOSTIC ACCURACY OF ERYTHROCYTE INDEXES
Figure 1. MCV in 248 anemic patients with empty
iron stores (green dots) and 227 anemic patients
without empty iron stores (red dots) plotted against
age. Empty iron stores were defined as s-ferritin
<15 lg/L. The upper green line is the 90 percentile
of MCV in patients with empty iron stores, while the
red line is the 50 percentile of MCV in patients
without empty iron stores. The stepped black line
indicates lower reference limits of MCV in healthy
persons above 6 months. Choosing the 90 percentile
(upper green line) as cutoff limit for empty iron
stores means choosing a sensitivity of 90%. The 90
percentile in patients with empty iron stores is close
to the 50 percentile in patients without empty iron
stores, indicating that a cutoff limit with 90%
sensitivity has close to 50% specificity. The results
from one patient without empty iron stores and
MCV of 124 fL were omitted from the figure.
the lower reference limit (Figure 1). To exclude empty
iron stores, ideally the sensitivity should be 100% and
definitely not <90%. The decision limit giving 90%
sensitivity, that is, the 90 percentile in the population
with empty iron stores is above the normal lower ref-
erence limits (Figure 1). However, higher sensitivity
comes at the cost of lower specificity, which is about
50% because the 90 percentile of MCV in anemic
patients with empty iron stores is quite close to the 50
percentile of anemic patients without empty iron
stores (Figure 1). In adults, Seward et al.[10] found
that MCV at a cutoff limit of 88 fL had a sensitivity of
88% and specificity of 39% in diagnosing s-ferritin
<15 lg/L. Their findings are similar to ours, and they
concluded that MCV was not sufficiently accurate to
be a useful criterion for the selection of more
© 2013 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2014, 36, 98–104
103
definitive tests. Our study was not designed to evalu-
ate whether MCV is a useful criterion for deciding on
further tests.
Why MCH performed better than MCV in our
study is not clear and a bit surprising. MCH may be a
more robust measure, depending on b-hemoglobin
and b-erythrocytes, which both are accurate and
precise, while MCV may increase during prolonged
specimen storage before analysis [11]. However, this
problem should be minimized by the standard proce-
dures in the laboratory (Materials and Methods). Any-
way, the difference between the diagnostic accuracy
of MCV and MCH was small, which is to be expected
given the close correlation between the two measures
(spearman correlation coefficient of 0.872 in this
population).
Likewise, we have no explanation why MCV, MCH,
and MCHC were more accurate in males than females,
a phenomenon not mentioned in the meta-analysis of
Guyatt et al.[6] This phenomenon seems to be age
dependent, as we found no differences between boys
and girls below 10 years of age (data not shown).
We used s-ferritin as a gold standard to determine
empty iron stores and tried different cutoff values
between 10 and 20 lg/L. In this range of cutoff values,
s-ferritin is a rather specific and less sensitive test [6]. It
may be argued that s-ferritin is not a valid gold stan-
dard, because it increases in several inflammatory con-
ditions. We did find a higher median s-ferritin in the
139 patients with elevated s-CRP. However, the
patients in this study were ambulant, and s-CRP was
<10 mg/L in 90% of those examined, so most of them
probably did not have any inflammation. Furthermore,
median s-ferritin in the majority with no s-CRP result
was not statistically significantly different from those
with normal s-CRP. Also s-ferritin may have less accu-
racy in patients with kidney failure, but more than
99% of the patients had normal s-creatinine. At the
moment, s-ferritin is the single best test for diagnosing
iron deficiency when iron content in bone marrow is
used as a gold standard [6]. Even bone marrow iron
may not always reflect the storage iron [12,13]. Any-
way, s-ferritin was the only gold standard that could be
applied in this study. For MCV, MCH, and MCHC the
diagnostic accuracy increased as the gold-standard limit
decreased, probably because a more stringent (lower)
limit isolated a more severely iron-deficient population.
We used a s-ferritin limit of 15 lg/L when estimating
104 A. E. 
ASBERG ET AL. | DIAGNOSTIC ACCURACY OF ERYTHROCYTE INDEXES

the percentile functions of MCV in anemic patients.
This choice was somewhat of a compromise, as we
thought 10 lg/L might be too low (low sensitivity) and
20 lg/L might be too high (low specificity). Using bone
marrow iron as the gold standard, Hallberg et al.[14]
found that s-ferritin ≤15 ug/L was the optimal limit for
an assay calibrated against WHO Ferritin 80/602 First
International Standard.
The findings in this clinical population may not
be transferable to other populations. First, the popu-
lation is mostly northwest European, with a very
low prevalence of thalassemia, so iron deficiency is
by far the most prevalent cause of microcytic
anemia. In a population with an appreciable num-
ber of thalassemia, MCV, MCH, and MCHC may
show even lower accuracy. Secondly, we used the
local lower b-hemoglobin reference limit of healthy
persons as cutoff limits to define anemia. These lim-
its may not be applicable in other populations and
health care systems.
In conclusion, MCV, MCH, and MCHC are only
moderately accurate in diagnosing empty iron stores
in children and young adults. These tests are no more
accurate in anemic than in nonanemic patients. It
should once again be pointed out that normal values
of MCV, MCH, and MCHC do not exclude empty iron
stores in anemic patients.
AC K N OW L E D G E M E N T S
We thank Frode Width Gran for valuable assistance
with data extraction.

REFERENCES diagnosis of iron-deficiency anemia: an the hematology analyser used: a stability

1. Hay G, Sandstad B, Whitelaw A, Borch-
Iohnsen B. Iron status in a group of
Norwegian children aged 6-24 months.
Acta Paediatr 2004;93:592–8.
2. Mahoney DH. Iron deficiency in infants
and young children: screening, prevention,
clinical manifestations, and diagnosis. In:
UpToDate. Basow DS (ed). Waltham: UpTo-
Date, 2012.
3. Janus J, Moerschel SK. Evaluation of ane-
mia in children. Am Fam Physician
2010;81:1462–71.
4. Irwin JJ, Kirchner JT. Anemia in chil-
dren. Am Fam Physician 2001;64:1379–
86.
5. Beutler E. The red cell indices in the diag-
nosis of iron-deficiency anemia. Ann Intern
Med 1959;50:313–22.
6. Guyatt GH, Oxman AD, Ali M, Willan A,
McIlroy W, Patterson C. Laboratory
overview. J Gen Intern Med 1992;7:145–
53.
7. Metz CE. Basic principles of ROC analysis.
Semin Nucl Med 1978;8:283–98.
8. Royston P, Sauerbrei W. Building multivar-
iable regression models with continuous
covariates in clinical epidemiology–with an
emphasis on fractional polynomials. Meth-
ods Inf Med 2005;44:561–71.
9. Wintrobe MM. Classification of the Anemias
on the Basis of Differences in the Size and
Hemoglobin Content of the Red Corpuscles.
Proc Soc Exp Biol Med 1930;27:1071–3.
10. Seward SJ, Safran C, Marton KI, Robinson
SH. Does the mean corpuscular volume
help physicians evaluate hospitalized
patients with anemia? J Gen Intern Med
1990;5:187–91.
11. Imeri F, Herklotz R, Risch L, Arbetsleitner
C, Zerlauth M, Risch GM, Huber AR. Sta-
bility of hematological analytes depends on
study with Bayer Advia 120, Beckman
Coulter LH 750 and Sysmex XE 2100. Clin
Chim Acta 2008;397:68–71.
12. Ganti AK, Moazzam N, Laroia S, Tendulkar
K, Potti A, Mehdi SA. Predictive value of
absent bone marrow iron stores in the clin-
ical diagnosis of iron deficiency anemia. In
Vivo 2003;17:389–92.
13. Thomason RW, Almiski MS. Evidence that
stainable bone marrow iron following par-
enteral iron therapy does not correlate with
serum iron studies and may not represent
readily available storage iron. Am J Clin
Pathol 2009;131:580–5.
14. Hallberg L, Bengtsson C, Lapidus L, Lind-
stedt G, Lundberg PA, Hulten L. Screening
for iron deficiency: an analysis based on
bone-marrow examinations and serum fer-
ritin determinations in a population sample
of women. Br J Haematol 1993;85:787–98.

© 2013 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2014, 36, 98–104
Copyright of International Journal of Laboratory Hematology is the property of Wiley-
Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv
without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.