Quantitative Measurement of Thromboelastography as a
Function of Platelet Count
Kyoko Oshita, MD*, Toshiharu Az-ma, MD, PhD*†, Yasuhiro Osawa, MD*, and
Osafumi Yuge, MD, PhD*
*Department of Anesthesiology and Critical Care Medicine, Hiroshima University School of Medicine, Minami-ku; and
†Department of Anesthesia, Hiroshima Red Cross and Atomic Bomb Survivors Hospital, Naka-ku, Hiroshima, Japan
to guide platelet transfusion has thus been advocated
by several investigators (2–6). However, it is not clear
whether TEG variables, assessed from the shape of its
tracing, are regulated by platelets quantitatively, be-
cause these variables are concurrently under the influ-
ence of coagulation factors. We examined the modu-
lation of TEG variables by the amount of platelet,
which was controlled by the dilution of platelets into
the patient’s plasma without affecting other factors.
We also attempted to evaluate the critical platelet
counts in coagulation by using this technique.
hromboelastography (TEG) analyzes the status
of blood coagulation, including abnormalities as-
sociated with low platelet count (1). Using TEG
After approval by the local institutional review board,
six volunteers (26–38 yr old; two male, four female)
scheduled for elective surgery gave their informed,
written consent to participate in our study. No
patients received anticoagulant and/or antiplatelet
medications. No patient had any abnormality in coag-
ulation, either clinically or by measurements of pro-
thrombin time (PT), activated partial thromboplastin
time (aPTT), fibrinogen, bleeding time, and platelet
count. They received 0.5 mg of atropine and 50 mg of
hydroxyzine IM as premedication approximately 1 h
before anesthesia. Blood samples were obtained just be-
fore the induction of general anesthetic from a 20-gauge
plastic cannula inserted in a forearm vein using a two-
syringe technique. After the first 6–10 mL of blood was
drawn and discarded, the second blood sample was
drawn into a polypropylene tube containing 1.2 mL of
3.8% sodium citrate, resulting in 12 mL of citrated blood.
Platelet-rich plasma (PRP) was obtained by the centrifu-
gation of citrated blood at 240g for 10 min. A one-half
volume of PRP was further centrifuged at 3000g for
0:5 vol/vol) to give a series of plasma containing various
amounts of platelet from each patient. The platelet count
in each sample was measured by an automated blood
cell/platelet counter (MEK-6108; Nihon Kohden, Tokyo,
Japan) after mixing PRP and PPP. PT, aPTT, and the
concentration of fibrinogen of these samples were meas-
ured by the Coagulation Laboratory (ACL 2000; Instru-
mentation Laboratory, Milano, Italy). There was no sig-
nificant difference in these values among the samples in
each series, which confirmed that the process of mixing
PRP and PPP at various ratios did not influence coagu-
TEG of citrated plasma was performed by adding
Ca2?. In brief, 250 ?L of the PRP/PPP mixture was
pipetted into the prewarmed (37°C) cuvette of a
thromboelastograph (Hellige, Freiburg, Germany).
Coagulation was achieved by adding 50 ?L of 0.4%
CaCl2to the plasma. CaCl2was uniformly mixed by
lowering and raising the pin of the thromboelasto-
graph three to four times before layering liquid par-
affin on the surface of plasma. TEG variables (reaction
time [r], coagulation time [k], and maximal amplitude
[MA]) were measured from each tracing (Fig. 1). All
samples were measured within 3 h after the blood
Statistical analysis included linear regression and
analysis of variance, followed by Bonferroni’s test for
multiple comparison, as indicated in Results. The sta-
tistical significance of the linear regression was con-
firmed by analysis of variance. A value of P ? 0.05
was considered significant. Data are expressed as
mean ? sd.
This study was supported in part by Grant-in-Aid for Scientific
Research 09771160 from the Ministry of Education, Science and
Culture of Japan.
Accepted for publication April 16, 1999.
Address correspondence and reprint requests to Toshiharu Az-ma,
MD, PhD, Department of Anesthesiology and Critical Care Medicine,
Hiroshima University School of Medicine, Minami-ku, Hiroshima 734-
8551, Japan. Address e-mail to firstname.lastname@example.org.
©1999 by the International Anesthesia Research Society
Anesth Analg 1999;89:296–90003-2999/99
We studied the TEG tracings of 36 samples obtained
from six individual series of citrated plasma contain-
ing various amounts of platelet. Nine samples were
omitted because the platelet count was under the de-
tection limit of the particle counter. TEG variables
were plotted against the logarithm of platelet count
(log10[platelet (/?L)]). A potent linear relationship
between MA and log10[platelet (/?L)] was observed
(R2? 0.739, P ? 0.0001) (Fig. 2A-1). An inverse linear
regression of log10[platelet (/?L)] with k was also
seen, although the statistical significance was less po-
tent than with MA (R2? 0.356, P ? 0.001) (Fig. 2B-1).
We found from each individual plot that log10[platelet
(/?L)] was significantly related to MA in every series,
whereas two of the six series showed no significant
relationship of log10[platelet (/?L)] with k (Fig. 2).
In contrast,no significant
log10[platelet (/?L)] and r was observed in every se-
ries (Table 1).
To determine the critical platelet counts in coagula-
tion, we defined the normal limit of TEG as the 2sd
limit of each TEG variable from samples containing
normal platelet counts (?150 ? 103/?L) (Table 2).
From each individual plot of MA and k against
log10[platelet (/?L)], we calculated the platelet counts
at which a linear regression line surpasses the lower
and longer limits of normal ranges for MA and k,
respectively. The platelet count showing the lower
limit of normal MA range (46.7 mm) was 58 ? 29 ?
103/?L, and the longer limit of k (8.5 min) was 145 ?
62 ? 103/?L. The values of individual critical platelet
counts are shown in Table 1.
The samples were also grouped into four levels
according to the logarithm of platelet count. A sig-
nificant decrease in MA and a prolongation in k at
platelet counts ?66 ? 103/?L, compared with the
corresponding values at normal platelet counts
(?150 ? 103/?L), were observed (Table 2).
Platelet aggregation to form a plug at injured vascular
endothelium is the initial step of hemostasis. The fol-
lowing coagulation cascade occurs on the surface of
aggregated platelets because phospholipids in the
Figure 1. The thromboelastogram measurement was commenced
by adding Ca2?in the citrated plasma. Reaction time (r) was defined
as the time from CaCl2was added in the cuvette until the amplitude
of the thromboelastogram tracing reached 2 mm. Coagulation time
(k) was defined as the interval when the trace amplitude reached
20 mm after r. Maximal amplitude (MA) was defined as the greatest
amplitude on thromboelastogram tracing.
Figure 2. A-1, The relationship of the maximal amplitude (MA) in
thromboelastography (TEG) with the logarithm of platelet count
(log10[platelet (/?L)]). Six individual series of citrated plasma con-
taining various amounts of platelet were obtained by mixing
platelet-rich plasma and platelet-poor plasma from each patient at
various ratios (n ? 27). The measurement of TEG was commenced
by adding Ca2?in the citrated plasma. A-2 and A-3, Two represen-
tative MA-log10[platelet (/?L)] plots obtained from individual se-
ries (Patients 1 and 3, respectively). A similar significant linear
relationship of log10[platelet (/?L)] with MA was observed in every
series. B-1, The relationship of the coagulation time (k) with
log10[platelet (/?L)] obtained from the same patients shown in A-1
(n ? 27). B-2 and B-3, Two representative k-log10[platelet (/?L)]
plots obtained from the corresponding patients shown in A-2 and
A-3. A significant linear relationships of log10[platelet (/?L)] with k
was observed from four individual series in total six patients.
THROMBOELASTOGRAPHY AND PLATELET COUNT
OSHITA ET AL.
platelet membrane are required for the activation of
coagulation factor X and prothrombin (7), which sug-
gests that the amount of platelet influences clot for-
mation. The quantitative abnormality of platelets is
thus an important factor in deciding whether the in-
dication of therapeutics (e.g., platelet transfusion, re-
gional anesthesia) is proper during perioperative
management. TEG has been advocated as a useful
guide of blood transfusion practice in cardiac surgery
(2), liver transplantation (3–5), and intensive care med-
icine (6). The diagnostic use of TEG for patients at risk
of coagulopathy, in whom epidural anesthesia is indi-
cated, has also been attempted by several investigators
(8,9). It has been reported that several TEG variables
are associated with platelet dysfunction (8–10),
whereas the resultant changes in TEG are also influ-
enced by coagulation factors, causing the evaluation
We measured TEG of citrated plasma after recalcifica-
tion. Our method made it possible to analyze the mod-
ulation of TEG merely by the amount of platelet because
we could obtain a series of plasma with various platelet
counts from one patient at a time without affecting the
concentration of coagulation factors. Using this method,
a potent linear relationship of log10[platelet (/?L)] with
MA was demonstrated. We then attempted to determine
the critical level of platelet count in coagulation by two
approaches. First, from an individual linear regression
line of MA against log10[platelet (/?L)], we determined
the critical platelet count at which MA value is less than
normal platelet counts (?150 ? 103/?L). Because MA
reflects clot strengthening (1), it has been suggested that
the abnormality in hemostasis occurs under the calcu-
lated value (58 ? 29 ? 103/?L). However, the wide-
spread variance of these individual critical platelet
counts indicated that the significance of platelet count
regarding the strength of clot varied among individuals,
possibly due to the complex interaction of platelet func-
tion with other factors relating to blood coagulation. We
also studied the individual plot of k as a function of
platelet count. The k time is thought to be affected by the
amount of platelet, as well as the concentration of co-
agulation factors, because it represents the velocity to
form a clot as a result of fibrin cross-linkage at the
surface of platelets (1). However, a linear-relationship of
log10[platelet (/?L)] with k was observed only in two
thirds of the six individual series of platelet-containing
every series. This indicates that the amount of platelet is
not the principal factor of the clot-forming rate and sug-
gests that the calculated critical platelet count for k
Table 1. Critical Platelet Count Determined from the 2sd Limit of TEG Variables at Normal Platelet
Counts (?150 ? 103/?L)
69 ? 4658 ? 29
n ? samples from each subject.
TEG ? thromboelastogram.
* P ? 0.05.
aThe critical platelet count was calculated individually from the linear-regression line.
Table 2. Changes in TEG Variables
13.6 ? 3.8
11.0 ? 2.4
12.8 ? 2.9
12.9 ? 2.5
P ? 0.5082
10.1 ? 4.8*
7.2 ? 1.9
6.0 ? 1.4
4.7 ? 1.9†
P ? 0.0238
11 35.5 ? 10.6*
48.8 ? 4.4†
54.0 ? 6.5†
58.9 ? 6.1†
P ? 0.0001
Data represent mean ? sd.
n ? samples in each group.
TEG ? thromboelastogram, ANOVA ? analysis of variance, MA ? maximal amplitude.
* Significant difference from normal platelet counts (?150 ? 103/?L).
† Significant difference from ?66 ? 103/?L platelet counts.
THROMBOELASTOGRAPHY AND PLATELET COUNT
OSHITA ET AL.ANESTH ANALG
(69 ? 46 ? 103/?L) has further limiting value, compared Download full-text
with that for MA.
The second approach to determining the critical
platelet count was to evaluate the change in MA
among the groups classified by the logarithm of plate-
let count. We found that the MA of the group at
platelet counts ?66 ? 103/?L was significantly
smaller than that of the group at normal platelet
counts (?150 ? 103/?L). The k time of the group at
?66 ? 103/?L platelet counts was also significantly
prolonged compared with that of the group at normal
platelet counts, the same as shown for the change in
In summary, two variables of TEG, MA and k, were
linearly related with log10[platelet (/?L)]. We propose
that the platelet count ?66 ? 103/?L is implicated in
the risk of dysfunction to form a clot, whereas the
critical platelet count evaluated by using TEG varied
1. Mallett SV, Cox DJA. Thrombelastography. Br J Anaesth 1992;
2. Tuman KJ, Spiess BD, McCarthy RJ, et al. Comparison of vis-
coelastic measures of coagulation after cardiopulmonary by-
pass. Anesth Analg 1989;69:69–75.
3. Kang YG, Martin DJ, Marquez J, et al. Intraoperative changes in
blood coagulation and thromboelastographic monitoring in
liver transplantation. Anesth Analg 1985;64:888–96.
4. McNicol PL, Liu G, Harley ID, et al. Patterns of coagulopathy
during liver transplantation: experience with the first 75 cases
using thromboelastography. Anaesth Intensive Care 1994;22:
5. Hardings SA, Mallett SV, Peachey TD, et al. Use of heparinase
modified thromboelastography in liver transplantation. Br J
6. Clayton DG, Miro AM, Kramer DJ, et al. Quantification of
thromboelastographic changes after blood component transfu-
sion in patients with liver disease in the intensive care unit.
Anesth Analg 1995;81:272–8.
7. Wintrobe MM, Lee GR, Boggs DR, et al. Blood coagulation. In:
Wintrobe MM, ed. Clinical hematology. Philadelphia: Lea &
8. Wong CA, Liu S, Glassenberg R. Comparison of thromboelas-
tography with common coagulation tests in preeclamptic and
healthy parturients. Reg Anesth 1995;20:521–7.
9. Orlikowski CEP, Rocke DA, Murray WB, et al. Thrombelastog-
raphy changes in pre-eclampsia and eclampsia. Br J Anaesth
10. Zuckerman L, Cohen E, Vagher JP, et al. Comparison of throm-
boelastography with common coagulation tests.
THROMBOELASTOGRAPHY AND PLATELET COUNT
OSHITA ET AL.