Journal of Feline Medicine and Surgery (2011) 13, 953e958

doi:10.1016/j.jfms.2011.07.014

Comparison of platelet clumping and complete

blood count results with Sysmex XT-2000iV in
feline blood sampled on EDTA or EDTA plus CTAD
(citrate, theophylline, adenosine and dipyridamole)
Fanny Granat DVM, Anne Geffré DVM, Jean-Pierre Braun,
Catherine Trumel DVM, PhD, Dipl ECVCP*

Département des Sciences
Cliniques, Ecole Nationale
Vétérinaire, 23 chemin des Capelles,
31 076 Toulouse Cedex 3, France
Date accepted: 25 July 2011
False thrombocytopenia may result from platelet aggregation, especially in feline
ethylenediamine tetra-acetic acid (EDTA) blood specimens. Citrate, theophylline,
adenosine and dipyridamole (CTAD) was added to 46 feline EDTA specimens to
test its anti-aggregation action. Platelet aggregation was estimated from blood
films and a complete blood count was performed with a Sysmex XT-2000iV
analyser. Platelet aggregation score was >2 in 11/46 EDTA tubes and only in one
EDTA þ CTAD specimen. The platelet count was higher in all
CTAD-supplemented tubes except one, medians measured by cytometry being
225.5  109/l and 249.0  109/l in EDTA and EDTA þ CTAD, respectively
(P ¼ 0.007). Adding CTAD had statistically and analytically significant but
moderate effects on other blood variables, the most intense variations being
observed for reticulocytes (about 3% higher in EDTA specimens) and
reticulocyte indexes. Addition of CTAD to EDTA when sampling feline blood is
a useful option to reduce platelet clumping.
Ó 2011 Published by Elsevier Ltd on behalf of ISFM and AAFP.

T
he measurement of platelet concentration is es-
sential for many diagnostic and therapeutic
purposes. Its accuracy can be markedly af-
fected by platelet aggregation which generally de-
creases the count and can thus result in apparent
thrombocytopenia, regardless of laboratory method
or species.1,2 Platelet clumping is often observed in fe-
line blood specimens, even in normal cats.3 Further-
more, feline platelet and red blood cell (RBC) sizes
often overlap because of the large size of platelets
and small size of RBCs in this species.3,4 As a conse-
quence, (i) the platelet count for feline specimens
with aggregates is often inaccurate regardless of the
laboratory method used; (ii) the mean platelet volume
(MPV) and mean corpuscular volume (MCV), white
blood cell count (WBC) and differential can also be in-
accurate in feline whole blood specimens when
impedance-based analysers are used.1,5,6
Although size-overlapping problems can be resolved
by flow cytometry technology, the occurrence of platelet
aggregates needs to be reduced at the pre-analytical
*Corresponding author. Tel: þ33-561-193-831; Fax: þ33-561-193-
978. E-mail: c.trumel@envt.fr
step. The anticoagulant citrate, theophylline, adenosine
and dipyridamole (CTAD) has been reported to reduce
feline platelet aggregation,7 and in one study in humans,
a combination of ethylenediamine tetra-acetic acid
(EDTA) and CTAD was more efficient than CTAD alone
at 4 C.8,9 Other reagents have also been used to obtain
resting platelets, such as kanamycine10,11 and prosta-
glandin E1 (PGE1)12e14 but are less easy to handle,
whereas CTAD tubes are commercially available. The
purpose of this study was thus to compare the degree
of aggregation in feline blood sampled on EDTA or
EDTA plus CTAD (EDCT), and to see if the addition of
CTAD to EDTA altered the results of other blood
variables.
Materials and methods
Forty-six blood specimens were selected among speci-
mens collected from cats admitted to different services
of the Hospital of the Veterinary School of Toulouse, as
part of routine diagnoses or disease monitoring. The
only inclusion criteria were that only fresh specimens
(<1 h) were used and that the volume of specimen col-
lected was sufficient for the study. Blood tubes were

1098-612X/11/120953+06 $36.00/0 Ó 2011 Published by Elsevier Ltd on behalf of ISFM and AAFP.
954 F Granat et al

collected from the jugular vein into a 5 ml K3-EDTA vac-
uum tube (Venoject; Terumo, Europe NV, Leuven,
Belgium) with a 0.8  40 mm needle (Venosafe Multi-
sample; Terumo, Europe NV, Leuven, Belgium). The
tubes were gently homogenised by 10 inversions imme-
diately after collection. Incorrectly filled tubes, or tubes
with macroscopic clots, were excluded. Three millilitres
of each specimen were then pipetted into a tube contain-
ing CTAD (Vacuette; Greiner Bio-one GmbH, Krems-
münster, Austria), which was homogenised by 10
inversions (EDCT tubes) and placed for approximately
20 min on a mechanical mixer (Speci Mix, CT06478,
Oxford, USA) before analysis.
A blood film was prepared from each specimen, air-
dried and stained (modified May-Grünwald-Giemsa,
Aerospray, Wescor), fixed, coverslip-mounted and
stored in darkness. To determine the platelet count
and degree of platelet aggregation, smears were exam-
ined by the same trained veterinarian (FG) under light
microscopy (BX60 microscope WH10X/22, Olympus,
Japan). The film platelet count (PLT-F) was estimated
by calculating the mean number of platelets per oil-
immersion field (10 randomly selected fields in the
monolayer area) multiplied by a pre-established factor
of 15 (109 platelets/l).15 The degree of aggregation was
estimated using the following scoring system7: 0 ¼ no
aggregate, 1e5 ¼ aggregates of 2e4 platelets, 5e9 plate-
lets, 10e19 platelets, 20e29 platelets and  30 platelets,
respectively. The mean of the 10 values observed on
2  5 oil-immersion fields randomly selected on the
edge and tail of the film was calculated. Arbitrarily
EDTA specimens with a score <2 were considered as
not or little aggregated (NLA) and specimens with
a score 2 were considered as highly aggregated (HA).
A complete automated blood count was performed
on each specimen using the Sysmex XT-2000iV
(Sysmex, Kobe, Japan). This included WBC, RBC,
PLT, haemoglobin (HGB), haematocrit (HCT), MCV,
mean corpuscular haemoglobin (MCH), mean
corpuscular haemoglobin concentration (MCHC),
red blood cell distribution width as coefficient of var-
iation (RDW-CV), red blood cell distribution width as
percentage (RDW-CV), reticulocyte percentage (RET
%), and count (RET), reticulocyte indexes [low fluores-
cence ratio (LFR), medium fluorescence ratio (MFR),
high fluorescence ratio (HFR), immature reticulocyte
fraction (IRF)] and the differential leukocyte count.
The PLT and RBC counts were estimated by two dif-
ferent methods: flow cytometry and impedance cell
counting (PLT-O, RBC-O and PLT-I, RBC-I) (Table 1).
All counts obtained for EDCT tubes were multiplied
by a factor (1.12) to compensate for the dilution.
The analyser was subjected to a quality control using
the manufacturer’s control solutions at three levels
(Sysmex e-check L1, L2 and L3 levels); imprecision of
measurements was evaluated according to CLSI16 rec-
ommendations and is reported in Table 2.
The results obtained with the two anticoagulants
were compared by Student’s paired t-test when the var-
iances were homogeneous (Fisher’s test), and otherwise
by Wilcoxon’s test. The number of cases when the differ-
ences were higher than could be expected from within-
run imprecision, and the number of cases when the
differences could account for a different classification
of results, according to recently published feline refer-
ence intervals,17 were also counted. The three different
PLT counts (PLT-O, PLT-I and PLT-F) were compared
by analysis of variance (ANOVA) for paired measure-
ments and Tukey’s test. Calculations were performed
with an Excel spreadsheet and Analyse-It (Leeds, UK).
As most distributions were significantly different from
Gaussian, the results are reported as median and mini-
mumemaximum range between brackets.
Results
All 46 paired feline specimens were analysed. Some
results were not given by the analyser, however, due

Table 1. Comparison of platelet count by impedance (PLT-I) and optical (PLT-O) methods with Sysmex
XT-2000iV and by evaluation from blood film (PLT-F) according to Norman’s platelet aggregation score in
feline blood specimens collected in EDTA or EDTA þ CTAD (EDCT). (HA and NLA subgroups based on
EDTA score of aggregation 2 and <2, respectively; comparison between the EDTA and EDCT results
by Student’s t-paired test or Wilcoxon’s test according to homogeneity or heterogeneity of variances.)
Analyte Group n EDTA EDCT P
Median Minemax Median Minemax
9
PLT-I (10 /l) All 45 122.0 1.0e345.0 166.4 3.4e338.4 0.057
HA 11 24.0 1.0e111.0 156.3 24.6e338.4 0.002
NLA 34 167.0 4.0e345.0 178.7 3.4e328.3 0.004
PLT-O (109/l) All 46 225.5 8.0e518.0 249.0 27.9e513.7 0.007
HA 11 80.0 24.0e242.0 240.1 134.0e424.3 <0.001
NLA 35 262.0 8.0e518.0 255.7 27.9e513.7 0.209
PLT-F (109/l) All 46 345.0 25.5e850.0 383.9 8.4e1080.4 0.022
HA 11 160.0 47.5e477.5 390.8 78.2e764.9 0.026
NLA 35 407.5 12.5e850.0 376.9 8.4e1080.4 0.419
Table 2. Comparison of blood variable results obtained in feline blood collected in EDTA and EDCT (EDTA þ CTAD) tubes (comparison by
Student’s t-paired or Wilcoxon’s test according to homogeneity of variances); number of cases for which the differences were higher than analytical
variability and of cases which would have been classified differently according to the limits of the reference interval in Moritz et al.17

Comparison of platelet clumping and complete blood count results

Variable EDTA EDCT P PassingeBablok equation Repeatability n cases > n cases >
EDCT $
EDCT ¼a EDTA þ b CV (%) analytical differently
vs EDTA variability classified
Median Minemax Median Minemax r a b according
to RI
RBC-I (1012/l) 7.87 1.74e11.48 7.97 1.76e11.51 <0.001 1.01[1.00; 1.01] 0.04[0.02; 0.09] 0.5 14/46 0/46
RBC-O (1012/l) 7.29 1.55e10.53 7.35 1.56e10.81 0.035 1.02[1.00; 1.04] 0.06[0.21; 0.05] 0.5 11/46 0/46
HGB (g/l) 112.0 30.0e162.0 112.8 30.2e161.9 0.161 1.00[0.99; 1.01] 0.50[0.28; 2.04] 0.4 5/46 0/46
HCT (l/l) 0.361 0.123e0.502 0.377 0.125e0.523 <0.001 1.05[1.03; 1.06] 0.04[0.56; 0.50] 0.5 44/46 1/46
MCV (fl) 45.0 35.1e78.2 46.8 36.1e78.5 <0.001 1.04[1.01; 1.06] 0.16[0.81; 1.07] 0.2 43/46 3/46
MCH (pg) 13.8 10.8e18.4 13.9 10.4e17.7 <0.001 1.00[1.00; 1.05] 0.10[0.75; 0.10] 0.8 2/46 ND
MCHC (g/l) 306.0 235.0e331.0 290.0 226.0e314.0 <0.001 0.94[0.89; 1.00] 3.37[13.50; 19.27] 0.8 40/46 ND
RDW-SD (fl) 30.8 27.0e71.8 31.1 27.0e70.2 0.054 1.08[1.00; 1.17] 2.19[4.90; 0.40] 0.5 23/45 ND
RDW-CV (%) 21.8 18.2e30.2 20.9 17.1e28.6 <0.001 0.94[0.89; 1.00] 0.20[1.10; 1.44] 0.4 44/46 ND
Reticulocytes (109/l) 36.35 1.80e269.20 34.95 1.12e239.41 0.010 0.86[0.76; 0.97] 11.82[24.24; 54.73] 3.0 34/46 ND
Reticulocytes (%) 0.53 0.08e15.47 0.48 0.05e13.57 0.028 0.86[0.75; 0.97] 0.02[0.03; 0.07] 3.2 30/46 ND
LFR (%) 93.3 68.9e100.0 93.7 61.7e100.0 0.192 0.87[0.56; 1.25] 12.89[23.01; 41.55] 2.2 6/46 ND
MFR (%) 5.4 0.0e20.8 4.7 0.0e25.4 0.383 0.91[0.61; 1.33] 0.00[2.22; 1.60] 8.4 29/46 ND
HFR (%) 1.5 0.0e10.3 1.0 0.0e12.9 0.115 1.07[1.08/3.23] 0.68[4.53/3.17]* 17.4 33/46 ND
IRF (%) 6.7 0.0e31.1 6.4 0.0e38.3 0.192 0.87[0.57; 1.26] 0.26[2.01; 1.98] 6.8 25/46 ND
WBC (109/l) 10.28 3.82e79.69 10.09 3.73e77.26 <0.001 0.97[0.96; 0.99] 0.02[0.20; 0.12] 1.3 12/46 0/46
Neutrophils (109/l) 6.40 1.16e38.03 5.78 0.85e38.46 <0.001 0.99[0.96; 1.01] 0.07[0.27; 0.06] 1.6 12/43 0/46
Lymphocytes (109/l) 1.93 0.31e7.35 1.56 0.26e7.47 0.138 0.98[0.93; 1.04] 0.06[0.16; 0.04] 5.7 10/43 0/46
Monocytes (109/l) 0.36 0.05e6.99 0.37 0.04e6.06 0.333 0.99[0.94; 1.03] 0.00[0.01; 0.02] 13.2 8/46 0/46
Eosinophils (109/l) 0.33 0.03e2.37 0.32 0.01e2.38 0.013 0.98[0.92; 1.03] 0.01[0.03; 0.00] 5.9 6/46 0/46
r ¼ Spearman’s correlation coefficient; a and b are slope and intercept respectively, with 95% confidence interval between brackets. ND ¼ not determined because
reference limits are not indicated. CVs were determined with manufacturer’s control solution at ‘normal’ level according to CLSI; LFR, MFR, HFR.
*Deming’s equation because PassingeBablok’s was not computable.

955
956 F Granat et al
to modifications induced by the cat’s illness. For ex-
ample, some differential counts were incomplete and
one PLT-I was excluded because the cat presented mi-
crocytosis which had induced an absurd platelet
count (w5.1012/l) with the impedance measurement.
The numbers of comparisons used for each variable
are reported in Table 1.
The overall degree of aggregation was lower in
EDCT than in EDTA specimens, with medians of 0.75
and 0.30, respectively (P  0.0002) (Fig 1). Moreover,
the distributions of the aggregation scores for the
EDTA tubes were bimodal: 11 specimens had a score
higher than 2 and 35 a score lower than 2, with me-
dians of 3.17 and 0.58, respectively (P < 0.0001). In
the EDCT specimens, the aggregation scores for the
NLA and HA subgroups based on aggregation in
EDTA, were not significantly different (P ¼ 0.1428).
The aggregation scores were lower in EDCT tubes
than in EDTA tubes in the HA and NLA subgroups
(P < 0.001 and P ¼ 0.021, respectively).
The PLT counts were lower in EDTA than EDCT
tubes with all three methods (Table 1). In the HA sub-
group, the platelet count was consistently lower in
EDTA specimens than in EDCT specimens, and the
platelet counts in EDTA were at least two or three times
lower than in EDCT specimens. Eight out of 11 cases in
the HA subgroup had much lower counts in EDTA
than in EDCT, whichever PLT count method was
used. Moreover, the PLT counts differed significantly
according to method, irrespective of the aggregation
score or anticoagulant used, (ANOVA for repeated se-
ries, P < 0.0001), PLT-F being significantly higher than
PLT-O, itself higher than PLT-I (Tukey’s test, P < 0.05).
No statistically significant difference was observed
between results obtained in EDTA and EDCT for
HGB, RDW-SD, reticulocyte indexes, and lymphocyte
and monocyte counts (Table 2). All other variables
Fig 1. Comparison of platelet aggregation scores in feline
specimens according to anticoagulant (EDTA or
EDCT ¼ EDTA þ CTAD). Boxes represent median and quar-
tiles for the 46 feline blood specimens (All) and in HA
(n ¼ 11) and NLA (n ¼ 35) subgroups based on platelet ag-
gregation score 2 and <2, respectively, in EDTA specimens.
showed significant differences, many of which were
larger than could be due to analytical variability. How-
ever, none of the latter differences had any effect on clin-
ical classification of the results according to the reference
limits, except for one and three cases out of 46 for HCT
and MCV, respectively (Table 2). RBC-I, RBC-O, HCT,
MCV and MCH were higher in EDCT specimens
whereas MCHC, RDW-CV, RET%, RET, WBC and neu-
trophil count were lower in EDCT than in EDTA tubes.
Moreover, most correlations were good (r  0.94) and
most PassingeBablok’s equations had slope and inter-
cept values for which the confidence interval contained
1 or 0, respectively (Table 2). However, the correlation of
reticulocyte count was weaker (r ¼ 0.91) and correlation
of reticulocyte indexes was low.
Discussion
The decision to use CTAD as a platelet clumping inhib-
itor was based on the commercial availability of this an-
ticoagulant which can easily be used by any practitioner.
In this study, we chose to add CTAD to EDTA: (i) as this
combination had been shown to be more efficient in pre-
venting human plateleteleukocyte aggregation and
platelet activation than CTAD alone at 4 C9; (ii) because
this study was performed on specimens submitted for
CBC as EDTA tubes (it was thus not possible to test
the effects of CTAD alone) and, (iii) even though the ad-
dition of CTAD produced an additional dilution factor
requiring correction and possibly altering cell morphol-
ogy. To our knowledge, addition of CTAD to EDTA had
not been previously tested in feline haematology and
possible effects of addition of CTAD to EDTA on routine
blood results could not be inferred from previously re-
ported of CTAD on blood variables.7
In this study, addition of CTAD to EDTA decreased
platelet aggregation in feline specimens. However, this
result was obtained in a limited number of cases, as
only 11/46 specimens (24%) showed significant platelet
clumping in EDTA (classified in the HA subgroup).
This low proportion of aggregated specimens could
be explained by the fact that the blood specimens
were collected by experienced clinicians from a large
vessel (the jugular vein) with a vacuum system. These
conditions have been reported to limit spontaneous
platelet clumping18,19 but no original study could be
found. In one case, addition of CTAD failed to prevent
platelet aggregation. The tube was from a diseased cat,
which had to be re-sampled because macroscopic clots
were visible in the first specimen.
The decreased platelet clumping produced by CTAD
resulted in higher platelet counts, whatever the tech-
nique. However, the results regarding platelet indexes
are lacking because MPV, PDW and PCT are not re-
ported by the Sysmex XT-2000iV in feline blood. This
is due to the manufacturer’s choices of methodology:
this analyser calculates the platelet indexes from imped-
ance curves, which are considered not accurate enough
in feline blood to validly calculate these variables. With
an impedance analyser, the feline plateletcrit has been
 Comparison of platelet clumping and complete blood count results
shown to be higher and MPV lower in the presence of
CTAD alone.7 These results were not confirmed in hu-
man specimens but analyses were performed by cytom-
etry, with an ADVIA 120.9 This discrepancy could be
explained; (i) by the different haematology methods
used; (ii) the inability of impedance measurements to
give reliable results for platelets in cats, because the
platelet size distribution curve never reaches baseline
within the PLTwindow in this species. This could justify
the manufacturer’s choice to not provide platelet in-
dexes obtained by impedance, even if this is a limit of
the analyser.
Finally, after having confirmed platelet clumping in-
hibition by CTAD, it was necessary to assess the possible
effects of CTAD addition on other blood variables,
which could not be inferred from effects reported on
CTAD alone.7 This study showed that the complete
blood counts obtained with EDTA or EDTA plus
CTAD were very similar, even though they were often
statistically different. Correlation between the two sets
of values was weaker for reticulocyte count than for
the other variables for which it was 0.94. Correlations
were low for reticulocyte indexes. To our knowledge,
there is no information on the relevance and use of these
indexes to diagnose regenerative anaemia in cats. The
high rate of ‘analytically significant’ differences largely
results from the very good repeatability of this analyser,
which makes small differences larger than analytical
variability for specimens analysed within the same
series.
These results contrast with previous studies which
reported that RBC and WBC were lower with
CTAD,7 and that MCV was lower in CTAD and higher
in CTAD plus EDTA than in EDTA alone.7,9 These dis-
crepancies could be explained by differences between
methods. To our knowledge there is no publication
with which our results for reticulocytes and indexes
can be compared.
Even though most results for EDTA alone and EDTA
plus CTAD were very similar, specific reference inter-
vals should be determined or transferred when avail-
able according to international recommendations.20e22
To conclude, adding CTAD to EDTA tubes to deter-
mine complete blood counts in felines can be a valid
option until specific tubes become available.
Conflict of interest
The Sysmex XT-2000iV analyser used in this study
was provided as a free loan by the manufacturer
who was not involved in any step or funding of this
study.
References
1. Norman EJ, Barron RC, Nash AS, Clampitt RB. Preva-
lence of low automated platelet counts in cats: compar-
ison with prevalence of thrombocytopenia based on
blood smear estimation. Vet Clin Pathol 2001; 30:
137e40.
957
2. Koplitz SL, Scott MA, Cohn LA. Effects of platelet
clumping on platelet concentrations measured by use
of impedance or buffy coat analysis in dogs. J Am Vet
Med Assoc 2001; 219: 1552e6.
3. Zelmanovic D, Hetherington EJ. Automated analysis of
feline platelets in whole blood, including platelet count,
mean platelet volume, and activation state. Vet Clin
Pathol 1998; 27: 2e9.
4. Weiser MG, Kociba GJ. Platelet concentration and plate-
let volume distribution in healthy cats. Am J Vet Res 1984;
45: 518e22.
5. Dawson H, Hoff B, Grift E, Tvedten H, Shoukri M. Val-
idation of the Coulter AcT Diff hematology analyzer
for analysis of blood of common domestic animals.
Vet Clin Pathol 2000; 29: 132e6.
6. Dewhurst EC, Crawford E, Cue S, Dodkin S, German AJ,
Papasouliotis K. Analysis of canine and feline haemo-
grams using the VetScan HMT analyser. J Small Anim
Pract 2003; 44: 443e8.
7. Norman EJ, Barron RC, Nash AS, Clampitt RB. Evalua-
tion of a citrate-based anticoagulant with platelet inhib-
itory activity for feline blood cell Counts. Vet Clin
Pathol 2001; 30: 124e32.
8. Mody M, Lazarus AH, Semple JW, Freedman J. Preanalyt-
ical requirements for flow cytometric evaluation of plate-
let activation: choice of anticoagulant. Transfus Med 1999;
9: 147e54.
9. Macey M, Azam U, McCarthy D, et al. Evaluation of the
anticoagulants EDTA and citrate, theophylline, adeno-
sine, and dipyridamole (CTAD) for assessing platelet ac-
tivation on the ADVIA 120 hematology system.
Clin Chem 2002; 48: 891e9.
10. Ahn HL, Jo YI, Choi YS, et al. EDTA-dependent pseudo-
thrombocytopenia confirmed by supplementation of
kanamycin; a case report. Korean J Intern Med 2002; 17:
65e8.
11. Sakurai S, Shiojima I, Tanigawa T, Nakahara K. Amino-
glycosides prevent and dissociate the aggregation of
platelets in patients with EDTA-dependent pseudo-
thrombocytopenia. Br J Haematol 1997; 99: 817e23.
12. Gryglewski RJ, Korbut R, Ocetkiewicz A. De-aggrega-
tory action of prostacyclin in vivo and its enhancement
by theophylline. Prostaglandins 1978; 15: 637e44.
13. Welles EG, Bourne C, Tyler JW, Boudreaux MK. Detec-
tion of activated feline platelets in platelet-rich plasma
by use of fluorescein-labeled antibodies and flow cytom-
etry. Vet Pathol 1994; 31: 553e60.
14. Tvedten H, Johansson P. Feline platelet counting with
prostaglandin E1 on the Sysmex XT-2000iV. Vet Clin
Pathol 2010; 39: 190e2.
15. Georges JW. Ocular field width and platelet estimates
(letter). Vet Clin Pathol 1999; 28: 126.
16. NCCLS.. Document EP5-A2 evaluation of precision
performance of quantitative measurement methods;
approved guideline. 2nd edn. Wayne, PA: NCCLS,
2004.
17. Moritz A, Fickenscher Y, Meyer K, Failing K, Weiss DJ.
Canine and feline hematology reference values for the
ADVIA 120 hematology system. Vet Clin Pathol 2004;
33: 32e8.
18. Mackin A. Platelet disorders. In: August JR, ed. Consul-
tations in feline internal medicine, vol 5: Saint Louis, MI:
Elsevier Saunders, 2006; 575e84.
19. Meinkoth JH, Allison RW. Sample collection and han-
dling: getting accurate results. Vet Clin North Am Small
Anim Pract 2007; 37: 203e19.
958 F Granat et al

20. Geffre A, Friedrichs K, Harr K, Concordet D, Trumel C,
Braun JP. Reference values: a review. Vet Clin Pathol 2009;
38: 288e98.
21. Geffré A, Concordet D, Trumel C, Braun JP. Validation of
pre-existing reference intervals: can the procedure be
applied to canine haemostasis? J Vet Diagn Invest 2011;
21: 803e14.
22. CLSI. Defining, establishing, and verifying reference in-
tervals in the clinical laboratory; approved guideline.
3rd edn. Wayne, PA: CLSI, 2008.

Available online at www.sciencedirect.com