HEMATOPATHOLOGY
Original Article
Clinical Significance of Immature Reticulocyte Fraction
Determined by Automated Reticulocyte Counting
CHUNG-CHE CHANG, MD, PhD, AND LAWRENCE KASS, MD
The Sysmex R-3000 (TOA Medical Electronics, Kobe, Japan) eval-
uates maturation of reticulocytes by quantitating the fraction of
reticulocytes within low-, middle-, and high-fluorescence inten-
sity regions. We defined the immature reticulocyte fraction (IRF)
as the sum of the fraction of high-fluorescence intensity regions
plus the fraction of middle-fluorescence intensity regions. Then,
we studied the clinical significance of IRF in the evaluation of
anemia by comparing the IRF with the absolute reticulocyte
count (ARC) and with the reticulocyte production index (RPI)
and by r e v i e w i n g p e r t i n e n t clinical i n f o r m a t i o n a b o u t t h e
patients. In the study, 132 specimens from 102 patients undergo-
ing evaluation of anemia were analyzed. By using simple regres-
sion analysis, our results showed that the IRF has a weak but sig-
nificantly positive correlation with ARC and with RPI, indicating
that IRF is an additional useful parameter to evaluate the erythro-
poietic activity in anemia. Interpretation by integrating IRF and
reticulocyte enumeration (ARC and RPI) provided useful infor-
mation for further subclassification of anemia. Increased IRF
Besides the actual reticulocyte count, assessment of
reticulocyte maturation is important for evaluating the
degree of effective erythropoiesis, for understanding
the pathophysiologic changes of anemia, and for the
differential diagnosis of anemia. 1-13 Historically, reticu-
locytes have been classified into different maturational
stages by morphology. 1-4 Various reticulocyte indexes
have been used to estimate these maturational stages,
but they are of limited clinical applicability because of
variability in manual microscopic methods. 1 " 4,14
Automated flow cytometric reticulocyte coun-
ters have increased the precision of reticulocyte
From the Department of Pathology, MetroHealth Medical Center and
Case Western Reserve University School of Medicine, Cleveland, Ohio.
M a n u s c r i p t received A u g u s t 16, 1996; revision accepted
December 3,1996.
Address reprint requests to Dr Kass: Department of Pathology,
MetroHealth Medical Center, Case Western Reserve University
School of Medicine, 2500 MetroHealth Drive, Cleveland, OH
44109-1998.
69
(IRF >0.23) and increased ARC generally indicated an adequate
Downloaded from https://academic.oup.com/ajcp/article/108/1/69/1757187 by guest on 29 October 2020
erythroid response to anemia. All but three specimens with an
IRF less than 0.23 showed an RPI of 2 or less. These specimens
were from patients with underlying diseases known to lead to
decreased erythropoietic activity, predominantly chronic renal
insufficiency. S p e c i m e n s w i t h a s u b n o r m a l or n o r m a l ARC
(with a corresponding RPI <2) but with an IRF of more than
0.23 were from patients with various u n d e r l y i n g conditions,
i n c l u d i n g acute i n f e c t i o n , iron deficiency a n e m i a , h u m a n
immunodeficiency virus infection, sickle disease with crisis,
pregnancy, and myelodysplastic syndrome. Our results indicate
that an IRF of 0.23 or less in patients with anemia reflects bone
m a r r o w t h a t is n o n r e s p o n s i v e or u n d e r r e s p o n s i v e to t h e
anemia. Patients with an increased IRF (IRF >0.23) may require
further examination to clarify the cause of the anemia. (Key
w o r d s : I m m a t u r e reticulocyte fraction; Absolute reticulocyte
c o u n t ; R e t i c u l o c y t e p r o d u c t i o n i n d e x ) Am J C l i n P a t h o l
1997;108:69-73.
enumeration. 1 5 - 1 8 Furthermore, some counters, like
the Sysmex R-3000 (TOA Medical Electronics, Kobe,
Japan), can evaluate the maturation of reticulocytes
by quantitating the fraction of reticulocytes within
low (LFR)- , middle (MFR)-, and high-fluorescence
(HFR) intensity regions. The HFR represents the
most immature reticulocytes; MFR represents imma-
ture reticulocytes, and LFR represents mature reticu-
locytes. However, to date, the clinical relevance of
these distinctions is not completely clear.
In the p r e s e n t study, we o u t l i n e d t w o goals.
Initially, we defined the immature reticulocyte frac-
tion (IRF) as the sum of the fraction of HFR plus the
fraction of MFR, as proposed by Davis. 19 First, we
asked whether the IRF was clinically informative
about the bone marrow response to anemia, by com-
paring the IRF to the absolute reticulocyte count
(ARC) and to the reticulocyte production index (RPI).
The RPI is a mathematical expression of the erythro-
poietic capacity of bone marrow. 2 ' 3 According to
Hillman and Finch, 3 this numerical index can be
obtained as follows:
70 HEMATOPATHOLOGY
Original Article

Patient hematocrit Patient reticulocyte %
Ideal hematocrit Reticulocyte maturation (in days)
With the use of the reticulocyte maturation time, a
compensation can be made for premature release of
reticulocytes (so-called shift reticulocytes) from the
bone marrow under the influence of erythropoietin in
patients with anemia. Clinically, an RPI less than 2 has
been widely accepted as an indicator of inadequate
bone marrow response to anemia. 2,3 Second, we asked
by Davis19 who stated that the normal range for the IRF
value was 0.05 to 0.22 in the American population. In this
group of healthy persons without anemia, the normal
reference value for ARC was 30 to 82 x 10 9 /L. This value
agrees with published normal values for ARC.20,21
By using simple regression analysis, our results
showed that ARC had a weak but significant positive
correlation with IRF for samples from healthy persons
without anemia (Fig 1, A, R2 = 0.27, P=.018). A similar
positive correlation also existed for samples from

Downloaded from https://academic.oup.com/ajcp/article/108/1/69/1757187 by guest on 29 October 2020

whether the IRF representing immature reticulocytes
used along with the ARC or with the RPI was helpful
in the differential diagnosis of anemia.

MATERIALS AND METHODS

Between December 1995, and March 1996, reticulo-

cyte counts were performed on 132 specimens of
peripheral venous blood anticoagulated with EDTA
from 102 patients undergoing evaluation of anemia.
There were 71 females and 31 males, ages 12 to 92
(mean, 52.6 years). As normal controls, specimens
from 20 healthy persons (10 women and 10 men, ages
18 to 81 years; mean, 43.6 years) who had normal val-
ues for hemoglobin, hematocrit, mean corpuscular
volume, mean corpuscular hemoglobin, mean corpus-
cular hemoglobin concentration, leukocyte count,
platelet count, and leukocyte differential count were
analyzed in the same way as the patient specimens.
Specimens were prepared according to manufac-
turer's guidelines 20 and were analyzed within 4 hours
of specimen collection. The reticulocyte percentage, 1.4
B
ARC, and fractions of HFR, MFR, and LFR in the retic- B

ulocyte population were determined by the Sysmex R- y = - 0.36463 + 2.5220X R2 = 0.25

3000, and hematocrits were determined by the Coulter
STKS (Coulter, Hialeah, Ha). Reticulocytes were iden- B

tified in the Sysmex R-3000 by fluorescence based on B

binding of auramine O to RNA. Depending on the flu- B

orescence intensity, the reticulocytes were further dif- a. / ^ B B
DC
ferentiated into HFR, MFR, and LFR. 20 With these ^ ^
data, the IRF and RPI were calculated. 0.6
B " V ^ "
B
B
Finally, completed charts were available for 90 of the
B
102 patients. We reviewed these charts and recorded ^ B

pertinent clinical information. Simple regression analy-

sis and the x2 were applied for statistical analysis.
—, a , r-
0.0 0.1
RESULTS IRF

FIG 1. A, Correlation between the absolute reticulocyte count

The mean value and SD for healthy persons without
anemia in this study for IRF were 0.136 and 0.046, respec-
tively (normal reference value is 0.044-0.228). These
results are consistent with the results reported recently
(ARC) and the immature reticulocyte fraction (IRF) for samples
from healthy persons without anemia. B, Correlation between the
reticulocyte production index (RPI) and the IRF for samples from
healthy persons without anemia.

AJCP • July 1997

 CHANG AND KASS 71
Significance of Immature Reticulocyte Fraction

patients (Fig 2, A, R2 = 0.18, P=.0001). In healthy persons

B El
without anemia and patients with anemia, the correla-
tion between the RPI and IRF was similar to but slightly
y = - 17.959 + 515.29x R2 = 0.18
weaker than that between the ARC and IRF (Fig 1, B, R2
= 0.25, P=.02 for healthy persons without anemia; Fig 2, 500-
B, R2 = 0.15, P-0001 for patients with anemia).
As shown in sections a and b of Figure 2, A, 45 of the
132 specimens had an increased ARC (>82 x 10 9 /L). As o
depicted (see Fig 2, A, section a), 26 (58%) of these 45 a.
<
specimens had an IRF of 0.23 or more (the upper limit of
normal range for IRF). Nine of the 27 specimens showed

Downloaded from https://academic.oup.com/ajcp/article/108/1/69/1757187 by guest on 29 October 2020

an RPI greater than 2. The majority of these specimens
were from patients with sickle disease with crises. Of the
45 specimens, 19 (42%) had an IRF of less than 0.23 (see
Fig 2, A, section b). All but three of the 19 specimens
showed an RPI of 2 or less. These specimens were
mainly from patients with hypothyroidism, sarcoidosis,
eating disorder with malnutrition, and neurosyphilis.
The remaining 87 specimens had a normal or sub-
normal ARC (<82 x 10 9 /L). All of these specimens
showed an RPI of 2 or less. Of the 87 specimens, 59
y = -9.8161e-3 + 4.0089x R2 = 0.15 B
(68%) showed an IRF of less than 0.23 (see Fig 2, A, sec- 6.0-

tions c through e). Thus, the specimens with a normal B

or subnormal ARC were more likely to have an IRF of B

5.0 - a
less than 0.23 compared with specimens with an
increased ARC (68% vs 42%; P=.008, %2)- This trend is 4.0 -
B

consistent with the results of simple regression analysis. Q. B

However, there were 28 other specimens that also had B
3.0-
B
a normal or subnormal ARC but had an IRF of 0.23 or B
B
higher (see Fig 2, A, sections f and g). A review of the B
2.0- B E) B
patient charts revealed that specimens with a normal or B
B
B °G> B ^_^,- - " ' ^ B
subnormal ARC were from two different groups of #3 B B | ^
1.0-
S B
patients corresponding with the value of IRF. Specimens B a
with a normal or subnormal ARC and an IRF of less than B e
0.0 -
0.23 were from patients with anemia and underlying dis-
eases known to lead to decreased erythropoietic activity, IRF
such as chronic renal insufficiency and eating disorder,
predominantly chronic renal insufficiency (see Fig 2, A,
sections c through e). Conversely, specimens with a nor- FlG 2. A, Correlation between the absolute reticulocyte count (ARC)
and the immature reticulocyte fraction (IRF) for samples from
mal or subnormal ARC but an IRF of 0.23 or higher were patients. Section a, sickle disease with crisis (n = 8), physiologic ane-
from patients with anemia associated with various under- mia secondary to respiratory insufficiency (n = 1)*, pregnancy (n =
lying conditions, including acute infection, iron deficiency 2), acute infection (n = 3), lead poisoning (n = 1); section b, sickle dis-
anemia, human immunodeficiency virus infection, sickle ease with crisis (n = 1), physiologic anemia secondary to respiratory
insufficiency (n= 1)*, hypothyroidism (n = 4), sarcoidosis (n = 2),
disease with crisis, pregnancy, and myelodysplastic syn-
neurosyphilis (n = 1), eating disorder with malnutrition (n= 2); sec-
drome (see Fig 2, A, sections f and g). tion c, chronic renal failure (n = 30), eating disorder with malnutri-
tion (n = 3); section d, chronic renal failure (n = 1); section e, chronic
renal failure (n = 8); section f, iron deficiency anemia (n = 3), human
DISCUSSION immunodeficiency virus (HIV) infection (n = 3), acute infection (n =
2); section g, iron deficiency anemia (n = 2), HIV infection (n = 2),
Automated reticulocyte counting has helped labora- pregnancy (n = 1), acute infection (n = 5), sickle disease with crisis (n
tories not only to improve precision in reticulocyte = 4), myelodysplastic syndrome (n = 1). 'Different specimens from
enumeration, but also to stimulate interest in the clini- the same patient. B, Correlation between the reticulocyte production
index (RPI) and the IRF for samples from patients.
cal utility of quantitating the immaturity of reticulocyte

Vol. 108 • No. 1

72 HEMATOPATHOLOGY
Original Article
populations. Limited pilot studies indicated that the
level of immaturity of reticulocyte populations could
have clinical utility in the assessment of erythropoietic
activity in cases of bone marrow transplantation 4-8 and
anemia. 10 - 11 However, the clinical significance of the
level of immaturity of reticulocyte populations is still
unclear. Furthermore, to date, no international consen-
sus on the definitions of terms and the methods for
measuring reticulocyte immaturity exists. 4 ' 9 ' 10 ' 16,22-25
In our study, the term immature reticulocyte fraction rep-
resents the immature reticulocyte population. The advan-
tages of our definition are as follows. First, HFR has been
shown to have a much higher coefficient of variation than
MFR.16 Thus, the use of IRF should significantly decrease
the coefficient of variation compared with using the frac-
tion of HFR alone, a value that has been used by some
investigators as the sole indicator of reticulocyte immatu-
rity.23'24 Second, IRF gave a pathophysiologic meaning for
representing total immature reticulocytes and showed a
significant positive correlation with the ARC and with the
RPI in our study. The associations between the fraction of
HFR and the ARC and between the fraction of HFR and
the RPI (R2 = 0.064, for ARC; R2 = 0.042 for RPI) were
much less significant than those between the IRF and
ARC and between the DRF and RPI. Recently, Davis19 sug-
gested that the determination of IRF may be the best way
to detect changes in erythropoietic activity. Our studies
support his suggestion.
As shown by regression analysis in the present
study, a weak but statistically significant positive cor-
relation between the ARC and IRF and between the
RPI and IRF, agreed with the findings by Davis et al. 26
By using a different flow cytometry-based method
with thiazole orange, their results also showed weak
significant positive correlation between the IRF and
ARC. Our study further confirmed that these correla-
tions are independent of the fluorochrome and the
instrumentation. In their study, a correlation between
the RPI and IRF was not specifically addressed.
The weak but significant correlations between the
ARC and ERF and between the RPI and IRF indicated that
the IRF is also a useful parameter to evaluate erythropoi-
etic activity in anemia. Further analysis by interpreting
the integration of IRF and reticulocyte enumeration (ARC
and RPI) is necessary and may help to further understand
erythropoiesis in anemias and to subclassify them.
In our study, the majority of specimens with an
increased ARC (>82 x 10 9 /L) showed an IRF of 0.23 or
higher. This finding indicated that an increased level
of erythropoietic bone marrow activity could result in
an increased IRF. The increased IRF suggested that an
erythropoietic bone marrow response could manifest
AJCP • July 1997
itself through the enhanced (increased or accelerated)
release of immature reticulocytes from bone marrow.
This effect has been suggested to be erythropoietin-
related, at least partially. 26 ' 27
It is noteworthy that all but three specimens with an
IRF of less than 0.23 showed an RPI of 2 or less (see Fig
2, A, sections b through e). The RPI was defined before
the advent of the automated reticulocyte counters and
has been widely accepted as an indicator of bone mar-
row response to anemia. 2 ' 3 Clinically, it has been widely
Downloaded from https://academic.oup.com/ajcp/article/108/1/69/1757187 by guest on 29 October 2020
accepted that an RPI of 2 or less indicates a bone mar-
row that is nonresponsive or hyporesponsive to anemia.
Accordingly, an IRF less than 0.23 is an indicator,
corresponding with an RPI of 2 or less, for patients
with anemia and bone marrow that is nonresponsive
or underresponsive to anemia. This finding is further
supported by the observation that these specimens
were from patients with chronic diseases known to
lead to decreased erythropoietic activity, such as
chronic renal failure, eating disorder with malnutri-
tion, hypothyroidism, sarcoidosis, neurosyphilis, and,
particularly, chronic renal failure. Our results are con-
sistent with those of Davis et al. 10 ' 26 By using a differ-
ent automated reticulocyte counter, they suggested
that only the patients with anemia with low reticulo-
cyte counts and low mean fluorescence intensity of the
reticulocyte population have true hypoproliferation.
The significance of the three specimens among those
with an IRF of less than 0.23 with an RPI greater than 2 is
uncertain. It has been suggested that in cases of reduced
erythropoiesis, application of an RPI correction may con-
ceal a failure of the bone marrow response, ie, a falsely
elevated RPI, because the shift does not occur fully or at
all. These three specimens could reflect this situation.
Thus, in cases of an RPI greater than 2 but a low IRF, this
possibility must be considered in the differential diagno-
sis of anemia. Our findings support the concept that the
ERF may be better than, or at least as good as, the RPI for
evaluating hyporesponsive marrow. Furthermore, the ERF
had the advantage of being directly obtained from the
automated reticulocyte counter over an off-line or labora-
tory information system-generated RPI calculation.
Another set of parameters, a normal or subnormal
ARC but an ERF of 0.23 or higher, defines an additional
group of patients with various underlying conditions (see
Fig 2, A, sections f and g). This finding may be clinically
important because the majority of the patients need fur-
ther immediate examination for proper diagnosis and
treatment. This finding further highlighted the value of
the ERF in the differential diagnosis of anemia. Without
the information provided by the ERF, these patients would
be believed to have the same pathophysiologic causes as
 CHANG AND KASS
Significance oflmmm
those with a normal or subnormal ARC (with a corre-
sponding RPI <2) and an IRF of less than 0.23.
However, the pathophysiologic changes of this set of
parameters remains unclear. Tatsumi and Izumi 13 and
Watanabe et al 12 suggested that the increased IRF found
in patients with myelodysplastic syndrome, leukemias,
or megaloblastic anemia may result from a qualitative
abnormality of erythropoiesis. Wells et al11 and, more
recently, Davis et al 26 reported that the reticulocyte fluo-
rescent intensity was significantly elevated in patients
with iron deficiency anemia because of an increase of
transferrin receptor messenger RNA content. Mercelina-
Roumans et al 28 stated that a significant increase of more
immature reticulocytes occurred during pregnancy due
to uncertain mechanisms. Further studies with bone
marrow examination and measurement of the erythro-
poietin level in patients with this set of parameters are
critical to fully evaluate the pathophysiologic causes.
Our results indicate that the IRF combined with the
ARC or the RPI can help to determine the adequacy of
marrow response to anemia and the underlying cause
of anemia. Our findings suggest that the IRF parameter
provides a basis for the further differential diagnosis of
anemias. An IRF of 0.23 or less in patients with anemia
suggests that the anemia is accompanied by a bone
marrow that is truly nonresponsive or underresponsive
to the anemia. These patients are most likely to have
underlying diseases leading to decreased erythropoietic
activity, predominantly chronic renal insufficiency.
Patients with an IRF of 0.23 or more may require further
examination to clarify the various causes of the anemia.
Acknowledgment: We t h a n k Dr S. A m i n i , D e p a r t m e n t of
Biostatistics, Case Western Reserve University, for his critical
review of the statistical data.
REFERENCES
1. Houwen B. Reticulocyte maturation. Blood Cells. 1992;18:167-186.
2. Hillman RS. Characteristics of marrow production and reticulo-
cyte maturation in normal man in response to anemia. / Clin
Invest. 1969;48:443-453.
3. Hillman RS, Finch CA. Erythropoiesis: normal and abnormal.
Semin Hematol. 1967;4:327-336.
4. Spanish Multicentric Study Group for Hematopoietic Recovery.
Flow cytometric reticulocyte quantification in the evaluation
of hematologic recovery. Eur J Haematol. 1995;53:293-297.
5. Davis BH, Bigelow N, Ball ED, Mills L, Cornwell GG III. Utility
of flow cytometric reticulocyte quantification as a predictor
of engraftment in autologous bone marrow transplantation.
Am ) Hematol. 1989;32:81-87.
6. Davies SV, Cavill I, Bentley N, et al. Evaluation of erythropoiesis
after bone marrow transplantation: quantitative reticulocyte
counting. Br J Haematol. 1992;81:12-17.
7. Lazarus HM, Chahine A, Lacerna K, et al. Kinetics of erythroge-
nesis after bone m a r r o w t r a n s p l a n t . Am ] Clin Pathol.
1992;97:574-583.
Vol. 108
73
Reticulocyte Fraction
8. Batjer JD, Riddell K, Fritsma GA. Predicting bone marrow trans-
plant engraftment by automated flow cytometric reticulocyte
analysis. Lab Med. 1994;25:22-26.
9. Kuse R. The appearance of reticulocytes with medium or high
RNA content is a sensitive indicator of beginning granulo-
cyte recovery after aplasiogenic cytostatic drug therapy in
patients with AML. Ann Hematol. 1993;66:213-214.
10. Davis BH, Bigelow NC. Row cytometric reticulocyte quantifica-
tion using thiazole orange provides clinically useful reticulo-
cyte maturity index. Arch Pathol Lab Med. 1989;113:684-689.
11. Wells DA, Daigneault-Creech CA, Simrell CR. Effect of iron sta-
tus on reticulocyte mean channel fluorescence. Am ] Clin
Pathol. 1992;97:130-134.
12. Watanabe K, Kawai Y, Takeuchi K, et al. Reticulocyte maturity
Downloaded from https://academic.oup.com/ajcp/article/108/1/69/1757187 by guest on 29 October 2020
as an indicator for estimating qualitative abnormality of ery-
thropoiesis. / Clin Pathol. 1994;47:736-739.
13. Tatsumi N, Izumi T. Reticulocyte maturation index as a useful
diagnostic parameter. Sysmex J Int. 1991;1:23-28.
14. Savage RA, Skoog DP, Rabinovitch A. Analytic inaccuracy and
imprecision in reticulocyte counting: a preliminary report
from the College of American Pathologists Reticulocyte
Project. Blood Cells. 1985;11:97-112.
15. Davis BH, Bigelow NC, Koepke JA, et al. Row cytometric retic-
ulocyte analysis: multiinstitutional interlaboratory correla-
tion study. Am J Clin Pathol. 1994;102:468^77.
16. Buttarello M, Bulian P, Venudo A, Rizzotti P. Laboratory evalua-
tion of the Miles H-3 automated reticulocyte counter: com-
parative study with manual reference method and Sysmex
R-1000. Arch Pathol Lab Med. 1995;119:1141-1148.
17. Davis BH, DiCorato M, Bigelow NC, Langweiler MH. Proposal
for standardization of flow cytometric reticulocyte maturity
index (RMI) measurements. Cytometry. 1993;14:318-326.
18. Tichelli A. Gratwohl A, Driessen A, et al. Evaluation of the
Sysmex R-1000: an automated reticulocyte analyzer. Am f
Clin Pathol. 1990;93;70-78.
19. Davis, BH. Immature reticulocyte fraction (IRF): by any name, a
useful clinical p a r a m e t e r of erythropoietic activity. Lab
Hematol. 1996;2:2-8.
20. Operator's Manual: Automated Reticulocyte Analyzer R-3000®.
Kobe, Japan: TOA Medical Electronics Co Ltd; 1992.
21. Tietz NW, ed. Clinical Guide to Laboratory Tests. 3rd ed.
Philadelphia, Pa: WB Saunders Co; 1995.
22. Tanke HJ, Rothbarth PH, Vossen JM, Koper GJ, Ploem JS. Flow
cytometry of reticulocytes applied to clinical hematology.
B/ood.l983;61:1091-1097.
23. Lesesve JF, Lacombe F, Marit G, et al. High fluorescence reticulo-
cytes are an i n d i c a t o r of b o n e m a r r o w recovery after
chemotherapy. Eur J Haematol. 1995;54:61-63.
24. Noguerol P, Perez J. High fluorescence reticulocytes increase in
peripheral blood is the first sign of circulating stem cells
after a p l a s i a - i n d u c e d c h e m o t h e r a p y . Exp Hematol.
1993;21:1138.
25. Serke S, H u h n D. Improved specificity of determination of
immature erythrocytes (reticulocytes) by multiparametre
flow cytometry and thiazole orange using combined staining
with monoclonal antibody (anti-glycophorin-A). Clin Lab
Haematol. 1993;15:33^4.
26. Davis BH, Ornvold K, Bigelow NC. Row cytometric reticulocyte
maturity index: a useful laboratory parameter of erythropoi-
etic activity in anemia. Cytometry. 1995;22:35-39.
27. Kendall RG, Jeffries R, Cavill I, Norfolk DR. Relationship
between e n d o g e n o u s erythropoietin levels, reticulocyte
count, and reticulocyte RNA distribution: a study of anemic
patients with and without renal failure. Ann N Y Acad Sci.
1994;718:353-355.
28. Mercelina-Roumans PEAM, Ubachs JMH, Wersch JWJ. The
reticulocyte count and its subtractions in smoking and non-
smoking pregnant women. Eur J Clin Chem Clin Biochem.
1995;33:263-265.