International Journal of Laboratory Hematology

The Official journal of the International Society for Laboratory Hematology

REVIEW ARTICLE INTERNAT IONAL JOURNAL OF LABORATO RY HEMATO LOGY

Verification and quality control of routine hematology analyzers

J. Y. VIS, A. HUISMAN

Department of Clinical
Chemistry and Hematology,
University Medical Center
Utrecht, Utrecht, The
Netherlands
Correspondence:
Albert Huisman, Department
of Clinical Chemistry and
Hematology, University Medical
Center Utrecht, G.03.550,
Heidelberglaan 100, 3584 CX
Utrecht, The Netherlands.
Tel.: +31 887558104;
Fax: +31 88 7555418;
E-mail: a.huisman@umcutrecht.nl
doi:10.1111/ijlh.12503
accepted for publication 23 Feb-
ruary 2016
Keywords
Hematology Analyzer, Verifica-
tion, Complete Blood Count
S U M M A RY
Verification of hematology analyzers (automated blood cell coun-
ters) is mandatory before new hematology analyzers may be used
in routine clinical care. The verification process consists of several
items which comprise among others: precision, accuracy, compara-
bility, carryover, background and linearity throughout the expected
range of results. Yet, which standard should be met or which veri-
fication limit be used is at the discretion of the laboratory specialist.
This paper offers practical guidance on verification and quality con-
trol of automated hematology analyzers and provides an expert
opinion on the performance standard that should be met by the
contemporary generation of hematology analyzers. Therefore
(i) the state-of-the-art performance of hematology analyzers for
complete blood count parameters is summarized, (ii) considera-
tions, challenges, and pitfalls concerning the development of a veri-
fication plan are discussed, (iii) guidance is given regarding the
establishment of reference intervals, and (iv) different methods on
quality control of hematology analyzers are reviewed.

(surgery, transfusion, drugs, etcetera) [4, 5]. The ana-

INTRODUCTION
Hematology analyzers (HA) or automated blood cell
counters are analytically and technically highly com-
plex automated analyzers [1–3]. These analyzers are
used to measure the complete blood count (CBC, also
known as full blood count). The CBC consists of sev-
eral parameters that are measured simultaneously and
all concern the number, concentration (in case of
hemoglobin), and some additional (descriptive) param-
eters of the different blood cells (Table 1). The CBC is
measured daily in virtually all medical laboratories
worldwide to screen for disease or abnormalities and
in the follow-up of all kinds of medical therapies
lytical results generated by HA are therefore the basis
of numerous medical interventions, and it is of para-
mount importance that these results are analytically
valid. Therefore, during the implementing process of a
new HA in a clinical laboratory, it is mandatory that a
verification process is performed to assure that the
analytical performance is up to standard. Which stan-
dard should be met or which criteria should be used is
at the discretion of the laboratory specialist. This article
provides guidance in choosing the verification limits
for the most common blood count parameters of HA.
Manufacturers validate diagnostic analyzers before
they enter the market in concurrence with different

100 © 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109
 J. Y. VIS AND A. HUISMAN | VERIFICATION OF HEMATOLOGY ANALYZERS 101

Table 1. CBC parameters [9]

Parameter (recommended
reporting unit) Description Critical values

Hemoglobin (Hb) concentration Hb concentration in blood <70 g/L (<7.0 g/dL)

(g/dL, g/L or mmol/L)
Hematocrit (Ht) (% or L/L) Fraction of Whole blood that
consists of RBC, sometimes also referred
as Packed Cell Volume (PCV).
Red Blood Cell (RBC) count Number of RBC (or Erythrocytes)
(91012/L) per volume blood
Mean Cell Volume (MCV) (fL) Average volume of RBC
Mean Cell Hemoglobin (MCH) Average amount of
(pg or fmol) hemoglobin per RBC
Red Cell Distribution Width The variation of cellular volume
(RDW) (%CV or fL) of the RBC population
Reticulocyte count (9109/L) Number of reticulocytes per
volume blood
Nucleated Red Blood Cell Number of Nucleated red Blood Cells
(NRBC) (9109/L) (also called erytroblast or normoblast)
per volume blood
White Blood Cell (WBC) Number of WBC (or Leucocytes) <2.0 9 109/L
count (9109/L) per volume blood and >40 9 109/L
White Blood Cell differential White Blood Cell differential count: Neutrophil Granulocytes
count (9109/L) Usually 5 part differential: < 0.5 9 109/L
Neutrophil Granulocytes
Eosinophil Granulocytes
Basophil Granulocytes
Lymphocytes
Monocytes
Platelet count (9109/L) Number of platelets (or thrombocytes) <50 9 109/L
per volume blood
Mean Platelet Volume Average volume of a platelet
(MPV) (fL)

There are numerous other parameters also reported by some HA, examples of these parameters are mean cell hemoglo-
bin concentration (MCHC), microcytic RBC and macrocytic RBC, hyperchromic RBC, and hypochromic RBC, frag-
mented RBCs, immature reticulocytes, reticulocyte hemoglobin content, immature neutrophil granulocytes (IG’s),
reticulated (or immature) platelets [5, 10, 11]. The nomenclature and definition may vary depending on the type of
HA. The suggested medical decision limits (critical values) are based on experience of the authors and the current liter-
ature [4, 12, 13].
fL, femtoliter (910
15 L); pg, picogram (910
12 g); fmol, femtomole (910
15 mole).

(inter)national requirements, leading for example to
FDA approval (USA) or a CE mark (European Union).
This validation is often performed under ideal circum-
stances. In clinical laboratory practice, the daily ana-
lytic performance may differ. Several international
guidelines and regulations, such as ICSH guidelines
[6] and CLSI guidelines [7] and the ISO 15189 stan-
dard [8], dictate that the laboratory verifies the
performance of the HA. Among other items, these
guidelines describe several verification items, which
comprise among other precision, accuracy, compara-
bility and linearity throughout the expected range of
results. In contrast, detailed specifications of the mini-
mal analytic performance on different levels are gen-
erally not stated, leaving the laboratory specialist to
answer these questions.

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109
102 J. Y. VIS AND A. HUISMAN | VERIFICATION OF HEMATOLOGY ANALYZERS
Besides complying with these guidelines, it is
important to have information on the reliability of
laboratory results, especially around critical cutoff val-
ues for diagnosis and treatment decisions. If the ana-
lytical performance around these crucial values is
uncertain or deviates too much, the new method or
analyzer should not be implemented. This paper offers
practical guidance on verification and quality control
of automated HA and provides an expert opinion on
the performance standard that should be met.
V E R I F I C AT I O N O R VA L I DAT I O N ?
According to ISO 15189, the laboratory should only
implement HA that are validated for the intended
purpose [8]. Validation, for example, according to
CLSI guidelines [7], should in principle be performed
by the manufacturer for the intended scope, and veri-
fication by the clinical laboratory of the manufacturers
claim is sufficient. Caution with different body fluids
(e.g., synovial, peritoneal, pleural, pericardial, cere-
brospinal fluid (CSF), or wound drainages) and differ-
ent tubes (plain, heparin, or citrate vs. EDTA) is
warranted, as not all HA have yet been validated for
these purposes. A full validation may also be required
for adjusted methods and measurements outside of
the analytical measurement range. There are several
challenges for the manufacturers during the validation
process:
First, certain parameters must be compared to a
reference method. Most of the current reference
methods are outdated such as hematocrit, red blood
cell count, white blood cell count, and white blood
cell differential [9]. In such cases, improved perfor-
mance of the new method is compared or calibrated
to a relatively poor standard, which leads to a poorer
performance than analytical and technically feasible.
Unfortunately, in some countries, the registration
authorities demand a comparison with these reference
methods and/or with a current generation analyzer.
This requirement could thus lead to poorer analytical
performance than necessary and also hampers innova-
tion.
Second, it is challenging to compare the ‘flagging’
of a new instrument with an existing method. Sam-
ples are for instance ‘flagged’ to identify pathological
samples that need rerun in a different mode, revision,
or additional diagnostic work-up (i.e., microscopic
© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109
review or flow cytometry). The underlying algorithms
that generate these flags may be totally different,
making a comparison between two types of HA chal-
lenging. Usually manufacturers try to make sure that
all pathological samples are recognized, for example,
the samples with malignant WBCs (blasts). To avoid
missing these pathological samples, the flagging algo-
rithm generates flags with a low positive predictive
value (PPV) [and usually a high negative predictive
value (NPV)], that is, a lot of normal samples get a
suspicious flagging and need (microscopic) review,
thereby unnecessary increasing the workload of the
laboratory.
Lastly, establishing reference intervals by manufac-
turers can be arduously, especially in the pediatric
population. Obtaining samples from healthy children
and neonates, has shown to be difficult. Reference
intervals in the pediatric population are often based
on publications from decades ago, using technology
that is no longer widely available, making it question-
able whether these reference ranges can be applied in
the present time; or there are new parameters or
parameters using a new analytical technique for
which there are no published reference ranges avail-
able [14]. Therefore, manufacturers should make con-
siderable efforts to generate reference intervals in
these groups for their new HA.
VERIFICATION
The ISO 15189 states that the laboratory shall verify
upon installation and before use that the equipment
is capable of achieving the necessary performance and
that it complies with requirements relevant to any
examinations concerned [8]. Both the ICSH and CSLI
have developed guidelines for the verification of HA
[6, 7, 9]. According to these guidelines, verification
studies should be performed on precision (repro-
ducibility), accuracy (method comparison), analytical
sensitivity (limit of detection), analytical specificity
including interfering substances, reference ranges,
patient correlation studies, linearity, limits of detec-
tion, carryover, and the analytical measuring range.
However, the ICSH and CLSI guidelines do not specify
the limits of acceptability of these studies throughout
the different parameters [9]. Nor do these guidelines
suggest a minimal range in which the reliability of the
parameters must be established. The extensiveness of
 J. Y. VIS AND A. HUISMAN | VERIFICATION OF HEMATOLOGY ANALYZERS
the verification study depends upon the availability of
independent evaluation data (preferably published in
the peer-reviewed literature), the extent of CBC
parameters reported by the laboratory, and the range
of samples available for the evaluation of the equip-
ment. As hematology parameters include cell differen-
tiations and cell indices, some (pre)analytical
characteristics differ from more regular laboratory
tests expressed per concentration plasma. Therefore,
some in-depth knowledge about hematology and
hematology parameters is required to interpret the
verification analyses.
Before starting a verification process, we should
establish the medically allowable error. The combina-
tion of different aspects of test performance forms the
total analytical error. The total analytical error reflects
the analytic reliability of a test result. If a laboratory
wants to verify that it produces reliable results in a
clinical perspective, there are several considerations.
Firstly, it is advised to establish which levels are clini-
cal relevant. Table 1 includes some suggested medical
decision limits. As some cutoff points distinguish rele-
vant patient groups, the reliability around these cutoff
points is important. For instance, if platelet transfu-
sions are given below a platelet count of 10 9 109/L,
it is sensible to verify the test performance around this
concentration. Secondly, the total analytical error
around this cutoff point should be acceptable for clini-
cal decision making. This medically allowable error
may depend on the expectations of the clinicians, the
biological variation, and the current state-of-the-art
performance.
Samples
In the evaluation study, fresh human whole blood
samples, anticoagulated with K2EDTA (or K3EDTA),
should be used and processed within 4–8 h after blood
sampling if stored at room temperature [9, 15]. If
samples are refrigerated, hematological parameters are
stable for a longer time period [16–18]. Special atten-
tion is warranted to guarantee that all tube types that
may enter the laboratory will be included in the veri-
fication study. If micro tubes (e.g., for pediatric
patients and capillary sampling) are used, these should
be evaluated, as the relatively low sample volume in
these containers may influence sample handling by
the HA (e.g., insufficient mixing or problems related
© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109
103
to the dead volume of the HA). It must also be guar-
anteed that samples with a very broad range of differ-
ent types of underlying pathology will be included in
the study, with results that encompass the entire ana-
lytical range, such as samples with low platelet and/or
WBC count, samples with high cell counts, samples
with abnormal cells (blasts) and samples with known
interference such as cryoglobulins, high bilirubin, and
high triglycerides.
Precision
Precision is the closeness of agreement between
repeated measurements of a sample (Figure 1) [9, 15].
Within-run precision (also known as reproducibility
or repeatability) usually consists of a single run of 20
measurements and is reported as a coefficient of varia-
tion [CV (%)]. Between-batch precision [reported as
CV (%)] is based on single measurements that are
repeated every day for a 20-day period and is affected
by random error and drift. Usually, stabilized quality
control samples are used to establish the between-
batch precision. The manufacturers’ claims about both
within-run and between-batch precision must be veri-
fied. In addition, we should always attempt to reach
at least the current state-of-the-art CV (%) for repro-
ducibility, so we can provide physicians with the most
accurate information and thereby provide
Probability
Accuracy
Reference value
x Value
Precision
Figure 1. Accuracy is the proximity of a measurement
result to the true value and mainly dependent on
systematic error (the term ‘bias’ should be avoided);
precision is the reproducibility of the measurements
and mainly dependent on random error.
104 J. Y. VIS AND A. HUISMAN | VERIFICATION OF HEMATOLOGY ANALYZERS

opportunities to improve medical decision making and
quality of care. The current state-of-the-art CV (%)
for the most relevant CBC parameters can be found in
Table 2. We encourage manufacturers to develop and
innovate HA in such a way that the current state of
the art is achievable for all new HA that enter the
market.
Accuracy and comparability
Accuracy (also known as trueness) is used to describe
the closeness of a measurement to the true value
(Figure 1) [9, 15]. The use of a ‘true value’ for the
CBC is hard to apply in daily laboratory practice, as
reference methods only exists for a few selected
parameters and these reference methods tend to be
rather impractical [9]. For the leukocyte differential
count, the microscopic evaluation of a slide is cur-
rently regarded as the reference method (400-cell
manual differential) [33]. However, this method suf-
fers from several disadvantages such as statistical
error, slide distribution error, and morphological inter-
pretation error. As a surrogate, usually in a verifica-
tion process, the HA under evaluation is compared

Table 2. Limits of acceptable imprecision: current state of the art

State-of-the-art State-of-the-art ‘Ricos’ criteria

[CV (%)] reproducibility [CV (%)] between-batch [CV (%)]

Hb 0.9 1.0 1.43

Ht 1.2 1.4 1.35
RBC 1.1 1.1 1.60
MCV 0.6 0.8 0.70
MCH 1.1 1.5 0.70
RDW 2.0 2.0 1.80
Reticulocytes 10 10 5.5*
WBC
High level (>10 9 109/L) 1.5 1.5 –
Normal level (1–10 9 109/L) 2.5 2.5 5.73
Low level (<1.0 9 109/L) 6.0 6.0 –
Neutrophils (abs)
Normal level 0.5–8.0 9 109/L 2.5 2.5 8.55
Low level (<0.5 9 109/L) 10 10 –
Eosinophils (abs) 10 10 10.5
Basophils (abs) 20 20 14.0
Lymphocytes (abs) 3.5 3.5 5.10
Monocytes (abs) 8.5 8.5 8.9
Platelets
Normal range 3.0 3.0 4.6
Low (~50 9 109/L) 4.5 4.5 –
Very low (10–20 9 109/L) 5.0 5.0 –
MPV 2.5 2.5 2.15

In a verification report, the results for imprecision (repeatability and between batch) should be reported including the
mean and CV (%). The state-of-the-art CV (%) for imprecision (reproducibility and between-batch) is based on the
current literature [17, 19–29] and the experience of the authors. The fourth column represents the ‘Ricos criteria’ that
illustrate biological variation [30, 31], and these performance criteria are based on biological variability. There is con-
troversy regarding the use of these criteria for method verification [32]; therefore, we suggest to use the current state-
of-the-art CV (%) (columns 2 and 3) as criteria for acceptability of the HA under verification.
*For reticulocytes, the ‘Ricos criteria’ depend on the method used, for a fluorescent method the % varies between 0.8
and 6.5. In clinical laboratories or settings where the HA are not used to guide platelet transfusion therapy or, for
example, therapy with antibiotics in case of neutropenia a higher %CV can be considered for platelet count and neu-
trophil count in the lower range with a maximum of 10% CV for both.

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109
 J. Y. VIS AND A. HUISMAN | VERIFICATION OF HEMATOLOGY ANALYZERS 105

with the instrument in routine. In the comparability
study (patient correlation studies), as many samples as
possible (generally 400 samples or more for each
parameter) should be compared using both HA. Fur-
thermore, it is of importance to include normal and
abnormal samples in approximate equal proportions.
The difference between the HA under evaluation and
the current analyzer or reference method should be as
low as possible (Table 3a). The parameter from the
HA under evaluation can be compared with the
current HA using linear regression. Perfect concor-
dance results in a correlation coefficient (r value) of 1.
Yet, if the new HA is much more accurate than the
old HA, this will negatively affect the correlation coef-
ficient, and thus, the correlation coefficient should be
interpreted with caution. In addition, some parame-
ters are known for their poor correlation, such as the
basophil count.
Sensitivity and specificity

Table 3. (a) Analytical accuracy, (b) Flagging Clinical sensitivity and specificity of flagging by HA
efficiency should be determined by comparing whether flagged
or unflagged samples accurately distinguish samples
State of the art (%)
with or without morphological abnormalities, respec-
(a)* tively, based on microscopic review and the micro-
Hb 1.3 scopic 400-cell differential count reference [33]. Yet,
Ht 1.8 not all morphological abnormalities are of equal clini-
RBC 3.2
cal relevance. We suggest that the accuracy of flagging
MCV 2.0
WBC 4.4 is at least evaluated for blast cells, promyelocytes,
Neutrophils (abs) 3.2 myelocytes, promonocytes, abnormal/atypical (sus-
Eosinophils (abs) 13 pected malignant) lymphocytes, fragmentocytes, plate-
Basophils (abs) 32 let clumps, red blood cell aggregation, and (depending
Lymphocytes (abs) 5.0 on the patient population) malaria-infected red blood
Monocytes (abs) 15
Platelets 6.4
cells. The specificity (% unflagged samples among all
morphological normal samples) and sensitivity (%
(b)† flagged samples among all morphological abnormal
Specificity >70
Sensitivity >90
samples) should be calculated. The current state of the
NPV >95 art is listed in Table 3b, although not all laboratories
PPV >60 are yet able to meet this current standard. Whereas
Overal efficiency >75 sensitivity is important to identify all pathology and
thus to provide accurate results, specificity is mainly
State-of-the-art Blast Variant Platelet important to limit the amount of unnecessary micro-
flagging (%)‡ cells lymphocytes clumps
scopic differentiations, thus to increase efficiently. To
Sensitivity >95 >80 >80 evaluate the predictive values and the overall effi-
Specificity >95 >95 >98 ciency of flagging in clinical laboratory practice, we
*The state of the art for accuracy is based on the current should select random samples to represent the normal
literature [4, 23] and the experience of the authors. Abs, workflow, as these values are influenced by the inci-
absolute count. dence of morphological abnormalities. The NPV, PPV,
†
Based on the current state of the art for aggregate of and overall efficiency can be calculated from the
various pathological findings (e.g., blasts, variant lym-
number of true positives (TP – abnormal morphology
phocytes, left shift (immature granulocytes), platelet
clumps) [20, 24, 26, 28, 34, 35] and the experience of with flag), true negatives (TN – normal morphology
the authors. It must be kept in mind that the outcome of without flag), false positives (FP – normal morphology
these calculations is dependent on the percentage and with flag), and false negatives (FN – abnormal mor-
the nature of the pathological samples. phology without flag) [15]. It is not feasible to calcu-
‡
Based on the current state of the art for various patho- late a reliable sensitivity and specificity for all types of
logical findings [20, 24, 26, 28, 34, 35] and the experi-
flagging in among the broad variety of samples we
ence of the authors.
suggest including the most useful flags.

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109
106 J. Y. VIS AND A. HUISMAN | VERIFICATION OF HEMATOLOGY ANALYZERS
TN
Specificity ¼  100:
(TN + FP)
TP
Sensitivity ¼  100:
(TP+FN)
TN
NPV ¼  100:
(TN+FN)
TP
PPV ¼  100:
(TP+FP)
TP+TN
Overall efficiency ¼  100:
(TP + FP+TN+FN)
Carryover, background, and linearity
Carryover is defined as the amount of analyte carried
by the HA from one sample measurement into the
subsequent measurement [6, 7, 9]. This may erro-
neously affect (usually increase) the reported concen-
tration in the subsequent sample. It is mainly of
importance for carryover from high to low concentra-
tions of Hb, RBC, WBC, and platelets. The percentage
carryover is assessed by analyzing a sample with a
high concentration three times (H1, H2, H3) followed
by analyzing sample with a low concentration three
times (L1, L2, L3). Percentage carryover is calculated
as follows:
ðL1
L3Þ
 100:
ðH3
L3Þ
Bear in mind that even a carryover percentage as
small as 2% may influence the results profoundly. For
instance, at a carryover of 2%, a sample with a true
platelet count of 1000 9 109/L (H3) may falsely
increase the platelet count of a subsequent sample of
5 9 109/L (L3, true value) to 25 9 109/L (L1), an
false increase of 500% that can be prevented by dupli-
cated measurements in such circumstances. If the car-
ryover is only 0.5% in the same situation, the second
sample will be measured at 10 9 109/L instead of the
true value of 5 9 109/L.
Background (limit of blank) refers to any signal
(noise) that is measured but does not originate from
the parameter of interest. Background may be caused
by an interfering substance, for example, signals from
blood free reagents or electronic noise caused by the
HA [36]. Optimally, the background counts should
be zero. A negligible background count is most
important in body fluids with very low cell counts
like CSF. The lower limit of detection refers to the
© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109
Table 4. State of the art for carryover, background,
and linearity
Carryover (%) Background Linearity (r)
Hb <0.4 <0.07 g/dL >0.99
RBC <0.5 <0.01 9 1012/L >0.99
WBC <0.2 <0.03 9 109/L >0.99
Plt <0.5 <0.5 9 109/L >0.99
The state of the art for carryover, background, and lin-
earity are based on the current literature [22, 23, 37]
and the experience of the authors.
lowest analyte concentration that produces a signal
that can be reliable distinguished form background
noise. It is based on the background and repro-
ducibility of a low concentration sample and usefully
lies below the reported range. Based on the CLSI
guideline [7], the lower limit of quantitation is the
level on which the imprecision for WBCs is 15%CV
and for platelets 25% CV. Ideally, the lower limit of
quantitation should be <0.1 9 109/L for WBCs and
<1.0 9 109/L for platelets. The limit of quantitation
should be used to define the lower limit of the
reported range (Table 4).
Linearity (analytical measuring range) is the ability
of the HA to provide a result that is proportional to
the analyte that is measured over a defined concen-
tration range [6, 7, 9]. There should be a linear rela-
tion over a large concentration range at various
dilutions for the parameter that is determined. Obvi-
ously, this does only apply for parameters expressed
per plasma volume and not for cell indices. To evalu-
ate linearity, a sample with a high (or artificially ele-
vated) concentration of a certain parameter is diluted
with diluent or AB plasma and measured in duplicate.
In the optimal situation, linear regression will show a
regression line with an intercept of zero and a slope
of 1. The correlation coefficient (r value) should be
close to 1. As linearity is usually not a big issue for
HA, the biggest challenge in linearity evaluation may
be to carry out a perfect duplicate dilution series.
R E F E R E N C E I N T E RVA L S
Several characteristics such as age, sex and ethnic
group may influence CBC parameters. Therefore, an
accurate and careful verification process to verify the
 J. Y. VIS AND A. HUISMAN | VERIFICATION OF HEMATOLOGY ANALYZERS
reference interval for each reportable parameter is
necessary. ICSH recommends to include at least 120
healthy persons (60 male/60 female) to establish
whether the reference interval is the same as the cur-
rent HA [6]. If appropriate also pediatric reference
intervals should be verified, although, in practice,
usually published reference intervals or those pro-
vided by the manufacturer of the HA are used. Yet, if
there is good concordance between the old and new
HA as is demonstrated by a comparison study, one
may argue whether reference values should be
re-established for each new HA.
Q UA L I T Y C O N T R O L
For intra-instrument quality control (internal quality
control), the measurement of quality control samples
on a daily basis (or more often) is obligatory [7]. Usu-
ally, these quality control samples are obtained from
the manufacturer and they consist of two or three
levels (low, medium, high). There are several pitfalls.
First, these samples are usually manipulated to
lengthen the shelf life; therefore, they may behave
differently than ordinary patient material. Second, the
manufacturer target’s limits are often very broad, and
subtle changes in analyzer behavior may thus be
missed. Therefore, it is recommended to adjust the
target range after a run-in period of several measure-
ments, for example, to the mean  2 standard devia-
tions. A benefit in using these quality control samples
is that they may be used to judge the instrument pre-
cision over time (i.e., drift) using a Levey–Jennings
graph; however, it must be kept in mind that at the
end of the shelf life the quality of the control samples
may deteriorate [38, 39].
An attractive alternative approach is the use of the
so-called moving average. This statistical method uses
the fact that the analytical parameters of the CBC in a
large population are stable over time, changes in the
average value thus possibly represent an analytical
problem. This method has proven to be very robust
and cheap, however it cannot fully replace the use of
quality control samples [38, 39].
If a laboratory has multiple HA and/or different
analytical techniques that are used for the same
parameter, interinstrument quality control should be
performed. All methods should lead to the same result
in each sample. If different methods lead to
© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109
107
unexpected discrepancies between reported results,
this may lead to misinterpretation and possibly unnec-
essary clinical intervention [40, 41]. To assure harmo-
nization of reported results between different
analyzers and methods, an interinstrument quality
control comparison is needed. Usually this can be
achieved by measuring patient samples (thereby
avoiding so-called matrix effect of stabilized quality
control samples) on these multiple systems and com-
paring the results. It is recommended that multiple
HA are compared at least on a weekly basis, using at
least three samples or more [38, 39].
The participation in an External Quality Assess-
ment (EQA) Scheme (also known as proficiency test-
ing scheme) is mandatory [38, 39]. The EQA samples
should be handled in the same way as routine patient
samples to acquire a fair comparison. EQA usually
uses manipulated samples which are composed to
resemble pathological samples (e.g., extremely high or
low cell counts). The samples are manipulated to
lengthen the shelf life and reduce the sensitivity of
the sample to transport related problems (temperature
fluctuations and vibration). Therefore, direct compar-
ison with patient results is not possible. Laboratories
must use the EQA results to compare their results
with (inter)national consensus or reference results
and use these to improve their quality and to harmo-
nize their HA with a consensus group. It must be kept
in mind that, in the absence of a reference method,
the results should be compared with a group of HA
that use the same analytical technique, as consensus
values may be systematically biased by inaccuracy of
most dominant group method [42]. Moreover, EQA
material should not be used for calibration purposes
(because a EQA consensus value and a true value
(calibrator) are not necessarily similar). In some coun-
tries, regulatory agencies abuse the EQA results to
evaluate the quality of the participating laboratory.
This has a negative influence on the self-improvement
of the laboratories, and this practice should be
discouraged [38, 39].
CONCLUSION
To ensure the best quality of results reported by HA, a
thorough verification progress is advised. This article
provides practical guidance for verification of HA.
Limits of acceptability for HA measurement
108 J. Y. VIS AND A. HUISMAN | VERIFICATION OF HEMATOLOGY ANALYZERS

characteristics based on the current state of the art

have been summarized. We have described pitfalls
and challenges within the verification process of HA
and suggest that clinical laboratory verification instruc-
tions take these considerations into account. In addi-
tion, different approaches for quality control have been
given to ensure optimal quality of parameters reported
by HA.
AU T H O R S H I P C O N T R I B U T I O N S A N D
C O N F L I C T O F I N T E R E S T S TAT E M E N T
JV and AH wrote, revised, and approved the manu-
script. JV and AH: The department of clinical chemistry
and Hematology (UMC Utrecht, Netherlands) has
received funding for contract research from Abbott
Diagnostics (Santa Clara, CA, USA).

REFERENCES
1. Davis BH, Barnes PW. Automated cell analy-
sis: principles. In: Laboratory Hematology
Practice. Kottke-Marchant K (ed). Oxford,
UK: John Wiley and Sons Ltd; 2012: 26–32.
2. Bene MC, Lacombe F. Differential leukocyte
analysis. In: Laboratory Hematology Prac-
tice. Kottke-Marchant K (ed). Oxford, UK:
John Wiley and Sons Ltd; 2012: 33–47.
3. Green R, Wachsmann-Hogiu S. Develop-
ment, history, and future of automated cell
counters. Clin Lab Med 2015;35:1–10.
4. Buttarello M. Quality specification in
haematology: the automated blood cell
count. Clin Chim Acta 2004;346:45–54.
5. Buttarello M, Plebani M. Automated blood
cell counts: state of the art. Am J Clin
Pathol 2008;130:104–16.
6. Briggs C, Culp N, Davis B, d’Onofrio G, Zini
G, Machin SJ. ICSH guidelines for the eval-
uation of blood cell analysers including
those used for differential leucocyte and
reticulocyte counting. Int J Lab Hematol
2014;36:613–27.
7. Rabinovitch A, Barnes P, Curcio KM, Dor-
man J, Huisman A, Nguyen L, O’Neil P. Val-
idation, Verification, and Quality Assurance
of Automated Hematology Analyzers; H26-
A2. Wayne, PA: Clinical and Laboratory
Standards Institute; 2010. ISSN 0273-3099,
ISBN 1-56238-728-6.
8. ISO. Medical Laboratories – Particular
Requirements for Quality and Competence
ISO 15189. Geneva: ISO; 2012.
9. Verbrugge SE, Huisman A. Verification and
standardization of blood cell counters for
routine clinical laboratory tests. Clin Lab
Med 2015;35:183–96.
10. Briggs C. Quality counts: new parameters
in blood cell counting. Int J Lab Hematol
2009;31:277–97.
11. Lecompte TP, Bernimoulin MP. Novel
parameters in blood cell counters. Clin Lab
Med 2015;35:209–24.
12. McFarlane A, Aslan B, Raby A, Bourner G,
Padmore R. Critical values in hematology.
Int J Lab Hematol 2015;37:36–43.
13. Schapkaitz E, Levy B. Critical limits for
urgent clinician notification at South Afri-
can intensive care units. Int J Lab Hematol
2015;37:620–5.
14. Wong EC. Hematology analyzers: special
considerations for pediatric patients. Clin
Lab Med 2015;35:165–81.
15. Briggs C. Evaluation of hematology analyz-
ers. In: Laboratory Hematology Prac-
tice. Kottke-Marchant K (ed.). Oxford,
UK: John Wiley and Sons Ltd; 2012: 96–
102.
16. Ashenden M, Sharpe K, Plowman J, Allbon
G, Lobigs L, Baron A, Gore CJ. Stability of
athlete blood passport parameters during
air freight. Int J Lab Hematol 2014;36:505–
13.
17. Longair I, Briggs C, Machin SJ. Perfor-
mance evaluation of the Celltac F haema-
tology analyser. Int J Lab Hematol 2011;
33:357–68.
18. Joshi A, McVicker W, Segalla R, Favaloro
E, Luu V, Vanniasinkam T. Determining
the stability of complete blood count
parameters in stored blood samples using
the SYSMEX XE-5000 automated haema-
tology analyser. Int J Lab Hematol
2015;37:705–14.
19. Xiang D, Yue J, Lan Y, Sha C, Ren S, Li Y,
Li M, Wang C. Evaluation of Mindray BC-
5000 hematology analyzer: a new
miniature 5-part WBC differential instru-
ment. Int J Lab Hemato 2015;37:597–605.
20. Bruegel M, Nagel D, Funk M, Fuhrmann P,
Zander J, Teupser D. Comparison of five
automated hematology analyzers in a uni-
versity hospital setting: Abbott Cell-Dyn
Sapphire, Beckman Coulter DxH 800, Sie-
mens Advia 2120i, Sysmex XE-5000, and
Sysmex XN-2000. Clin Chem Lab Med
2015;53:1057–71.
21. Seo JY, Lee ST, Kim SH. Performance eval-
uation of the new hematology analyzer
Sysmex XN-series. Int J Lab Hematol
2015;37:155–64.
22. Grillone R, Grimaldi E, Scopacasa F, Dente
B. Evaluation of the fully automated hema-
tological analyzer Mindray BC 6800:
comparison with Horiba ABX Pentra
DX120. Int J Lab Hematol 2014;36:e55–8.
23. Briggs C, Longair I, Kumar P, Singh D,
Machin SJ. Performance evaluation of the
Sysmex haematology XN modular system. J
Clin Pathol 2012;65:1024–30.
24. Hedley BD, Keeney M, Chin-Yee I, Brown
W. Initial performance evaluation of the Uni-
Cel(R) DxH 800 Coulter(R) cellular analysis
system. Int J Lab Hematol 2011;33:45–56.
25. Hotton J, Broothaers J, Swaelens C, Can-
tinieaux B. Performance and abnormal cell
flagging comparisons of three automated
blood cell counters: Cell-Dyn Sapphire,
DxH-800, and XN-2000. Am J Clin Pathol
2013;140:845–52.
26. Jean A, Boutet C, Lenormand B, Callat
MP, Buchonnet G, Barbay V, Basuyau JP,
Vasse M. The new haematology analyzer
DxH 800: an evaluation of the analytical
performances and leucocyte flags, compar-
ison with the LH 755. Int J Lab Hematol
2011;33:138–45.
27. Kang SH, Kim HK, Ham CK, Lee DS, Cho
HI. Comparison of four hematology analyz-
ers, CELL-DYN Sapphire, ADVIA 120,
Coulter LH 750, and Sysmex XE-2100, in
terms of clinical usefulness. Int J Lab
Hematol 2008;30:480–6.
28. Meintker L, Ringwald J, Rauh M, Krause
SW. Comparison of automated differential
blood cell counts from Abbott Sapphire,
Siemens Advia 120, Beckman Coulter DxH
800, and Sysmex XE-2100 in normal and
pathologic samples. Am J Clin Pathol
2013;139:641–50.
29. Leers MP, Goertz H, Feller A, Hoffmann JJ.
Performance evaluation of the Abbott
CELL-DYN Ruby and the Sysmex XT-2000i
haematology analysers. Int J Lab Hematol
2011;33:19–29.
30. Perich C, Minchinela J, Ricos C, Fernandez-
Calle P, Alvarez V, Domenech MV, Simon
M, Biosca C, Boned B, Garcia-Lario JV,
Cava F, Fernandez-Fernandez P, Fraser CG.
Biological variation database: structure and
criteria used for generation and update. Clin
Chem Lab Med 2015;53:299–305.

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109

31. Ricos C, Alvarez V, Perich C, Fernandez-
Calle P, Minchinela J, Cava F, Biosca C,
Boned B, Domenech M, Garcia-Lario JV,
Simon M, Fernandez PF, Diaz-Garzon J,
Gonzalez-Lao E. Rationale for using data
on biological variation. Clin Chem Lab Med
2015;53:863–70.
32. Topic E, Nikolac N, Panteghini M, Theodor-
sson E, Salvagno GL, Miler M, et al. How to
assess the quality of your analytical method?
Clin Chem Lab Med 2015;53:1707–18.
33. Koepke AA, vanAssendelft OW, Brindza LJ,
Davis BH, Fernandes BJ, Gewirtz AS, Rabi-
novitch A. Reference Leukocyte (WBC) Dif-
ferential Count (Proportional) and
Evaluation of Instrumental Methods. H20-
A2. Wayne, PA: CLSI; 2007.
34. Tan BT, Nava AJ, George TI. Evaluation of
the Beckman Coulter UniCel DxH 800, Beck-
man Coulter LH 780, and Abbott Diagnostics
© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2016, 38 (Suppl. 1), 100–109
J. Y. VIS AND A. HUISMAN | VERIFICATION OF HEMATOLOGY ANALYZERS
Cell-Dyn Sapphire hematology analyzers on
adult specimens in a tertiary care hospital.
Am J Clin Pathol 2011;135:939–51.
35. Tan BT, Nava AJ, George TI. Evaluation of
the Beckman Coulter UniCel DxH 800 and
Abbott Diagnostics Cell-Dyn Sapphire
hematology analyzers on pediatric and
neonatal specimens in a tertiary care hospi-
tal. Am J Clin Pathol 2011;135:929–38.
36. Armbruster DA, Pry T. Limit of blank, limit
of detection and limit of quantitation. Clin
Biochem Rev 2008;29(Suppl. 1):S49–52.
37. Grimaldi E, Scopacasa F. Evaluation of
the Abbott CELL-DYN 4000 hematology
analyzer. Am J Clin Pathol 2000;113:497–
505.
38. Cembrowski GS. Hematology quality prac-
tices. In: Laboratory Hematology Practice.
Kottke-Marchant K (ed). Oxford, UK: John
Wiley and Sons Ltd; 2012: 686–706.
109
39. Cembrowski GS, Clarke G. Quality control
of automated cell counters. Clin Lab Med
2015;35:59–71.
40. Huisman A. Discrepancy in hemoglobin
results between automated analyzers caused
by differences in hemoglobin standardiza-
tion [abstract]. Int J Lab Hematol 2014;36
(Suppl. 1):128.
41. de Vooght KM, Groenendaal F, Bierings
MB, van Solinge WW, Huisman A. Falsely
elevated point-of-care hematocrit and calcu-
lated hemoglobin concentration due to
extreme leukocytosis. Ann Hematol
2014;93:1949–50.
42. Boonen KJ, Curvers J, Timmerman AA,
Steurs D, van de Kerkhof D. Trueness in
the measurement of haemoglobin: consen-
sus or reference method? Clin Chem Lab
Med 2012;50:511–4.