From www.bloodjournal.org by guest on October 26, 2016. For personal use only.

Bleeding Time and Bleeding: An Analysis of the Relationship of the Bleeding

Time Test With Parameters of Surgical Bleeding
By Raffaele De Caterina, Maria Lanza, Giampiero Manca, Giuliana Buti Strata,
Stefan0 Maffei, and Leonard0 Salvatore
The bleeding time is currently the only clinically available
comprehensivetest t o explore primary hemostasis. It is currently performed mostly as a screening procedure before
surgery, t o detect otherwise unknown defects in plateletvessel wall interactions, but its use in this specific setting
has been seriously questioned by recent reanalyses of previously published literature. We studied the relationship of
the bleeding time from a standardized cutaneous incision
with other parameters ofbleeding derived from the analysis
of the bleeding time curve and prospectively investigated
possible correlations of these alternative parameters, as well
as of the bleeding time, with a number of indices of actual
bleeding during or after coronary bypass surgery. Four parameters (bleedingtime, total bleeding, peakbleeding rate,
and time t o peak bleeding) were derived from the analysis
of bleeding time curves measuringblood losses from a standardized cutaneous incision at 30-second intervals in 118
subjects. Parameters
from the
bleedingtime curve were subsequently obtained in duplicate as a preoperative assessment in 40 patients with a negative bleeding history and no
recent intake of non-steroidal anti-inflammatory drugs who
were undergoing elective pure coronary bypass surgery
performed by thesame operator. These parameters were correlated in simple linear regression analysiswith estimates of

surgical bleeding (chest tube drainage, transfusion requireof platelet

ments,percentageofhematocrit,percentage
level decrease, and times t o hematocrit and platelet level
nadir) and then,in multiple regression analysis, with indices
of operation complexity (number ofbypasses, total duration
of the operation, and duration of the extracorporeal circulation). Bleeding time was significantly correlated, among parameters derived from the bleeding time curve, with total
bleeding and peak bleeding rate, but not withtime to peak
bleeding. Bleeding time, total bleeding, and peak bleeding
rate were similarly affected by acute interventions with intravenous aspirin (500 mg) and sublingual nitroglycerin (0.3
mg). None of these parameters,
which were obtained in duplicate in each patient preoperatively, was significantly related t o actual indices of bleeding at surgery. Thus, in patients with a negative history of bleeding and no recent
intake of non-steroidal anti-inflammatory drugs, higher values for bleeding time and bleeding time-related parameters
are not associated with higher indices of perioperative and
postoperative bleeding at coronary bypass surgery. Therefore, we do not recommendthe use of the test in thissetting
to predict perioperative or postoperative bleeding.
0 1994 by The American Societyof Hematology.

so far there is no proof of it. There is still space, therefore,

for further development of the field, including further advancements in method development and better selection of
the clinical settings in which to administer the test. More
recent encouraging reports have indeed, at least partially,
refueled the d i s c ~ s s i o n . ~ ~
We hypothesized that the main reason for the lack of a
relationship between the cutaneous bleeding time and surgical bleeding lies in the multiple determinants of surgical
bleeding, which might obscure, by their preponderant
weight, the possible predictive capacity of the bleeding time
test. However, an additional reason may lie, on the other
hand, in the intrinsic variability of the test itself, which is
therefore amenable to possible further improvements.
To verify these hypotheses, we first determined the relationship between the bleeding time and estimates of bleeding
from the cutaneous incision in an attempt to exploit addi-

HE BLEEDING TIME is a widely used and popular

test to explore primary hemostasis. The most common
use of the bleeding time is as preoperative screening of
potentially dangerous platelet disorders. Because surgery entails a major challenge to hemostasis, which may be fatal
in case of hemostatic defects, it makes sense to search for
preoperative screening of potentially dangerous bleeding. Indeed, the risk of bleeding from platelet disorders is not reflectedby alterations of commonly used coagulation tests
such as the prothrombin time (PT) and the partial thromboplastin time (m).
Because the bleeding time is prolonged
in congenital or acquired platelet defects, it has been common reasoning for years that the test could provide a screening for hemorrhagic tendencies from otherwise occult platelet disorders. However, doubts about the predictive value of
the bleeding time with respect to surgical bleeding have
appeared more and more often in the recent
In 1990,
an exhaustive review of published reports on the test concluded, among other points, that no evidence exists that the
bleeding time is indeed a predictor of surgical losses during
or after surgical operations and that no evidence exists that
efforts in standardizations have resulted inany significant
enhancement in the utility of the test. Although most other
reviews appearing at the same time or after this publication
have substantially agreed with these con~lusions,~~-~
they
have not necessarily shared an overall skepticism on the
possible clinical usefulness of the t e ~ t . Also,
~ , ~ , it~ has been
pointed out that the lack of data assessing predictive value
of the bleeding time test with respect to individual patients
does not in and of itself negate the possibility that the bleeding time is a predictor of bleeding in certain conditions.8The
bleeding time, it has been said, might predict bleeding, but
Blood, Vol 84, No 10 (November 15). 1994 pp 3363-3370

From the CNR lnstitute of Clinical Physiology, and the Division

of Cardiac Surgery, Spedali Riuniti di Santa Chiara, Pisa, Italy.
Submitted February 23, 1994; accepted July 12, 1994.
Supported in part by grants from National Research Council and
Minister0 dell Universith e della Ricerca Scientijica, Italy.
Address reprint requests to Raffaele De Caterina, PhD, Vascular
Medicine and Atherosclerosis Unit, Cardiovascular Division, Brighamand Womens Hospital, Harvard Medical School, 221 Longwood Ave, LMRC 3, Boston, MA 021 15.
The publication costs of this article were defrayed in part by page
charge payment. This article must therefore be hereby marked
advertisement in accordance with 18
U.S.C. section 1734 solely to
indicate this fact.
0 1994 by The American Society of Hematology.
0006-4971/94/8410-0004$3.00/0
3363

From www.bloodjournal.org by guest on October 26, 2016. For personal use only.

3364

DE CATERINA ET AL

tional information available from the current performance

of the bleeding time testand to search for additive/alternative
parameters that might reduce test variability and possibly
offer more direct and accurate predictors of surgical bleeding. We then analyzed the relationship of the bleeding time
and of related parameters with several estimates of surgical
bleeding in the course of one single type of operation (elective, pure coronary bypass surgery) performed by the same
surgeon to test the impact of these further efforts at standardization on the prediction of surgical bleeding.
MATERIALS AND METHODS
Patients. Duplicate measurements of the bleeding time were performed in a cohort of 118 patients (subset 1: 97 men and 21 women;
age [mean 2 SDI, 58.1 2 9.8) admitted to a cardiological clinic for
the evaluation of a chest pain syndrome or as a preoperative assessment in 40 further patients with angiographically proven coronary
artery disease scheduled for elective aorto-coronary bypass surgery
(subset 2: 33 men and 7 women; age, 62.3 -C 7.2). The only inclusion
criteria for subset 1 patients was the absence of any history ofintake
of non-steroidal anti-inflammatory agents for 3 days and of aspirincontaining medications for 10 days before the test. These patients
were allocated to the assessment of performance of the test, ie, the
intraoperator and interoperator variability of the bleeding time and
related parameters (n = 66 and n = 21, respectively; see below);
the correlations of two alternative methods to estimate total bleeding
from the cutaneous incision (n = 37); and the acute effect of drugs
(n = 20). Patients in subset 2 were selected for the part of the study
aimed at assessing correlations of the bleeding time andrelated
parameters with parameters of surgical bleeding if they fulfilled
the following criteria: patients were scheduled for elective coronary
bypass surgery by the same cardiac surgeon in the absence of any
need for other concomitant surgical correction (ie, valvular heart
disease); patients had an absence of any history of bleeding; patients
within the normal range
had values of PT and activated F'IT (m)
(international normalized ratio [INRI c1.2 and aPTT <32 seconds,
respectively); and patients had an absence of any history of intake
of non-steroidal anti-inflammatory agents for 3 days and of aspirincontaining medications in the 10 days before the test.
Reasons for this selection weretheneed for an assessment of
baseline bleeding time performance in the absence of pharmacologic
interventions and for the study of an otherwise hemostatically normal
patient population to assess the correlation of test performances with
actual surgical bleeding. Patients finally selected for the prospective
study of surgical bleeding prediction represented 40 of 112 patients
screened. Reasons for the 72 exclusions in this series were a concurrent extra-coronary (valvular) surgery in 10 cases and a positive
history for non-steroidal anti-inflammatory drug intake in 65 cases.
All patients, specifically those allocated to drug interventions,
gave informed consent to the study.
Test procedure. Bleeding times were performed by the technique described by Mielke et all4 in which a horizontal incision
(parallel to the elbow crest) on the lateral aspect of the volar surface
of the forearm is made after the inflation of a sphygmomanometric
cuff at 40 mmHg. Care was taken throughout all measurements to
the following technical details: tests were performed on patients
lying inbedin a confortable 45" position, between 9 and I 1 AM,
after a light breakfast taken at 7 AM; skin was cleansed with ethyl
ether and shaved, when necessary, more than 4 minutes before the
incision; the incision was performed within I minute after the application of the cuff; and clamps were applied to the cuff tubings to
maintain the pressure in the cuff constant throughout the duration
of thetest. Tests were performed in duplicate in each patient on
both arms in random order. All tests were performed by two opera-

tors (among the authors of the study) whose performances proved

to be similar (see below). For the part of the study aimed atassessing
interoperator variability, the two operators performed the tests at a
minimum interval of 15 minutes on different arms in random order
of operator (right arm first); for all other studies, the same operator
performed the duplicate measurements on the same patient on different arms, again at a minimum interval of15 minutes (right arm
first). Incisions were performed by a disposable device (Surgicutt:
Ortho Diagnostics, Raritan, NJ); the operators were aware of the
possible alterations of results caused by the application of an inconsistent pressure ofthe device on the skinandwere instructed to
apply moderate pressure on the forearm, ie, sufficient to ensure a
complete contact of the inferior side ofthe device withthe skin.
Whatman No. 1 filter paper (Whatman, Hillsboro, OR) was used to
blot the blood oozing from the incision, and care was taken not to
touch the wound directly.
Analysis of test results. Squares of Whatman No. I filter paper
of approximately 1.5 cm X 1.5 cm were cut and preweighed individually on a Mettler precision balance (sensitive to variations of 0. I
mg) and arranged in a rack ready for the blotting of each incision
at 30-second intervals. Blotting was performed withtheaid of a
forceps. Reweighing of the filter paper after blotting allowed the
determination (by subtraction of the preblotting value) of the weight
of the estimated blood loss during that time interval. This estimate
was correlated to the actual amount ofblood loss, as estimated
alternatively by blood collection in preweighed borosilicate glass
microhematocrit capillary tubes (BRI; Moduloohm, Henlev, Denmark),15withthe difference being exactly accounted for by the
amount ofwater in plasma, which quickly evaporated during the
drying of the blotting filter paper (the correction factor was, for all
practical purposes, the plasmacrit value). The estimate of blood loss
allowed, for each bleeding time determination, the construction of
a curve of blood accumulation over time (Fig IA). This proved in
all cases to be asymmetrical, with an early peak and a long tail. The
following four parameters were defined in each curve and used for
all subsequent analysis (Fig 1B): (1) thebleeding time, defined
conventionally as the time up to the first blotting not giving rise to
a detectable blood staining of the filter paper; (2) the total bleeding,
definedasthe area underthe curve obtained by thesum ofthe
estimates at each time point; (3) the peak bleeding rate, defined as
the maximum weight of dried blood in the 30-second intervals; and
(4) the time to peak bleeding, defined as the time up to the peak
bleeding rate.
Pharmacologic interventions. Among subset 1 patients, 10 (7
men and 3 women; age, 5 1.7 2 4.0 years) were studied before and
5 minutes after an intravenous injection of 500 mg of acetyl-salicylic
acid (in the form of lysine salt; Aspkgic; Lirca Synthelabo, Milan,
Italy), a dose consistently producing in our hands a complete abolition of platelet aggregatory responses to 1 mmoliL arachidonic acid
and greater than 98% inhibition of thromboxane generation in serum.
Ten similar patients (8 men and 2 women; age, 55.4 % 2.5 years)
were studied before and 5 minutes after a sublingual administration
of 0.3 mg of nitroglycerin as a tablet (Trinitrina; Carlo Erba. Milan.
Italy).
Coronary bypasssurgery.
Coronary bypass surgery wasperformed by the same experienced car&ac surgeon using left internal
mammary artery for the left anterior descending coronary artery and
saphenous vein with or without the right mammary artery for other
bypasses, the number of which was determined according to conventional criteria derived from coronary angiography. The time of entire
duration of surgery and the time of the extracorporeal circulation
were carefully recorded. Cardio-pulmonary bypass was performed
using standard techniques with a membrane oxygenator, the performances of which, in terms of platelet activation (as determined by
the increase of platelet factor 4 and P-thromboglobulin), had been

From www.bloodjournal.org by guest on October 26, 2016. For personal use only.

BLEEDING TIME AND BLEEDING

3365

corporeal circulation. Multiple regression analysis was also performed to assess the possible independent effect of bleeding time
parameters and parameters related to the length of operation on one
side and estimates of surgical bleeding on the other. Comparison
of the bleeding time curves before and after drug interventions was
performed by the Kolgomorov-Smirnov statistiscs. The level of significance was set at P < .05.

RESULTS
0,5 1 1 . 5 2 2.5 3 3 . 5 4 4 . 5 5 5 . 5 6 6 , 5 7 7.5

TIME,min

BLEEDING TIME

PEAK BLEEDING
RATE

TIMETOPEAKBLEEDING

TIME,
min

Fig 1. (A) Anexample of the curveofbloodaccumulationover

time (the bleeding time curve) after a standardized cutaneous incision, as obtained experimentally by weighing the dried blood after
blotting of the wound at 30-second intervals. (B) The same curve,
with interpolation of the values, showing the following parameters
assessed: the bleeding time, total bleeding, peak bleeding rate, and
time to peak bleeding. (See text for further details.)

Baseline values of bleeding time and related parameters

in 46 patients of subset l receiving double measurements
and allocated to the study of intraobservervariability are
reported in Table 1. Bleeding time was not related in our
study to baseline Hct values ( R = .05) in the narrow range
of Hct values for our patients (33% to 47%).
Correlations of bleeding time with bleeding parameters
derived from the analysis of the bleeding time curve.
The
bleeding time proved to be correlated significantly with the
total bleeding rate and the peak bleeding rate, butno relationship was found with time to peak (Fig 2). The trend to a
clear relationshipbetween the bleeding time, the total bleeding rate, and the peak bleeding rate was also evident from
analysis of the curves before and after acute pharmacologic
interventionswith drugs reported to prolong the bleeding
time, such as aspirin and nitroglycerin (Table 2 and Fig 3).
These interventions prolonged bleeding time and increased
total bleeding and peak bleeding rate, but did not affect the
time to peak. Interestingly, the global analysis of bleeding
timecurvesbefore
andafterpharmacologicinterventions
proved to be statistically more powerful in detecting differences(Kolgomorov-Smirnov
statistcson
completedata
shown in Fig 3).
Assessment and comparisons of the variability of the various parameters derived ,from the analysis of the bleeding
time curve. The coefficients of variation between duplicate
measurents for the bleeding time and related parameters are
shown in Fig 4. Despite the theoretical attractivenessof alternative bleeding parameters, they
proved to be affected by
greater variability than the bleeding time itself, both when
duplicate measurements were performedby the same operator (intraoperator variability) and when these were performed
by different operators (interoperator variability).
Correlations of the bleeding time and related parameters
with estimates qf surgical losses. There was norelationship
between any of the bleeding time parameters and any of the
parameters of surgical bleeding at simple linear regression
analysis (Table 3). The bleeding time in patients with lower
and higher than median values for all indices of bleeding

previously assessed16 with heparin anticoagulation and subsequent

administration of protamine sulfate. A cold potassium blood cardioplegic solution was used to induce and maintain ischemic arrest.
Iced saline topical cooling and systemic hypothermia at 28C were
used. The mediastinum was drained with one tube.
Assessment ofsurgical bleeding. The amount of perioperative and
postoperative blood losses were determined by computation ofthe
following parameters: ( I ) the amount of blood collected through the
chest tube drainage; (2) the amount of blood transfusions administered
(these were either whole blood or plasma and were administered
according to an integrated estimate by the operator of perioperative
and early postoperative losses); (3) the percentage of the maximum
postoperative hematocrit (Hct) decrease; (4)the time to Hct nadir; (5)
the percentage of the maximum postoperative platelet level decrease;
and (6) the time to platelet level nadir. For computation of parameters
3 through 6, Hct and platelet count were performed three times on
the operation day and then daily for at least 1 week.
Staristical analysis. Estimates of intraoperative and interoperative variability of the bleeding time and related parameters were
Table 1. Baseline Values of the Bleeding Time and Related
obtained by calculating the percentage of the coefficient of variation
Parameters in Averaged Duplicate Measurements in 66
(CV = [SD/mean] X 100) of the duplicate measurements. These
Coronary Artery Disease Patients
estimates were compared between each other by analysis of variance
Peak BleedingTimeto
Peak
and the protected Fisher least significant difference test. Linear reBleeding
Time
Total
Bleeding
Rate
Bleeding
gression analysis was used to correlate the means of the two determi(mid
(mg)
(mg/30 S)
(rnin)
nations for all the bleeding time parameters used among each other
13.7 Mean
5.7
3.2
1.o
and with the various parameters of surgical bleeding. These were
1.8
0.4
SD
1.8
10.3
also in turn correlated with the number of bypasses applied in each
32
75
56
40
% CV
patient, the duration of the operation, and the duration of the extra-

From www.bloodjournal.org by guest on October 26, 2016. For personal use only.

3366

DE CATERINA ET AL

y =

112.92

R"2

44.595X
D

0.441

= 0.66

R
El

n = 6 9

10

12

BT (min)

y = 3.0692

R"2 = 0.249

5.1627X

loo]

R = 0.50

.1 2 ; . t i

y = 0.54772

10

BT (min)

6.635Oe-2x

R"2

= 0.063
R

parameters and parameters related to the complexity of the

surgical procedure (duration of the operation, duration of
the extracorporeal circulation, and the number of bypasses
[dependent variables]). The bleeding time or related parameters did not explain part of the residual variability after taking
into account parameters of surgical complexity. For example,
in multiple regression analysis of chest tube drainage (independent variable) versus the number of coronary artery bypasses, the addition of the bleeding time as a second dependent variable did not improve the R* value significantly (.205
before and .208 after) and did not decrease the standard error
of the estimate (45.8 before and 46.5 after), indicating a lack
of improvement in the prediction of the estimate." This was
also shown by the absence of any significant trend towards
a relationship between bleeding time parameters and parameters of actual bleeding at surgery taking into account relatively homogeneous groups of operations (Fig 5). Also, correction of the bleeding time values for the packed red blood
cell volume, as recently suggested by one group,"'did not
improve the relationship between the bleeding time and indices of operative bleeding. Indeed, the coefficient of correlation ( R ) between postoperative bleeding, as assessed by chest
tube drainage, cumulative transfusion units, Hct decrease.
time to Hct nadir, percentage of platelet level decrease, or
time toplatelet level nadir, versus the meanof the twobleeding time determinations corrected for the preoperative Hct
still remained clearly nonsignificant ( R = .17, .OO, .OO, .14,
.08, and .l5 for each of these correlations, respectively).

DISCUSSION

0.25

First described in 1901 by Milian," the bleeding time has

been practiced for several decades in the way described in
Table 2. Values of the Bleeding Time and Related Parameters
in Single Measurements Before and After Treatment
With IntravenousAspirin (500 mg, n = 10) or Sublingual
Nitroglycerin (0.3 mg, n = 10)
~~

!
2

I
4

#
6

12 8

,
10

Bleeding
Time
(min)

BT (min)
Fig 2. Simple linearregressionanalysis
between the bleeding
time (BT; independent variable, x, on the abscissas) and related parameters (dependent variables, y, on the ordinates)derived from the
analysis ofthe bleedingtime curve: total bleeding (A), peakbleeding
rate (B), and time to peak bleeding (C). The equations from linear
regression analysisand the relative coefficients ofcorrelation ( R I are
shown. Rvalues are significant forthe first two correlations, but not
for the third one.

analyzed wasnot significantly different. This finding was

also true for anyof the bleeding time-related parameters
investigated (data not shown). Because surgical bleeding appeared to be influenced, to some extent, by the duration and
complexity of the surgical procedure ( R = .45, P < .05, at
correlation of number of bypasses v chest tube drainage),
multiple regression analysis was also performed testing the
existence of a significant relationship between parameters
of surgical losses (independent variable) and bleeding time

Before ASA
Mean
SD
% CV70
After ASA
Mean
21.5
SD
% CV
P (before v after)
Before NTG
Mean
SD
% CV
After NTG
34.7
Mean
SD
% CV
P (before v after)

Total
Bleeding
(rng)

Peak
Bleeding
Rate
(mgI30 S)

Time to
Peak
Bleeding
(min)

6.2
1.9
31

20.3
14.2

4.5
1.9
42

1.2
0.3
25

9.5
2.2
23
1.05

45.0
48

6.1
2.2
36

<.05

<.05

1.2
0.4
33
NS

7.4
1.o
14

23.9
11.4
48

4.5
1.9
42

1.2
0.4
33

8.6
1.3
15
1.05

13.0
31

5.5
2.5
45
NS

1.9
0.6
32
NS

1.05

Abbreviations: ASA, aspirin; NTG, nitroglycerin; NS, not significant.

From www.bloodjournal.org by guest on October 26, 2016. For personal use only.

3367

BLEEDING TIME AND BLEEDING

6
1

5m

pre-ASA

post-ASA

t-

+I

n = 10

md,
E

10

12

TIME (min)

TT

IY
T

hT

tive history of bleeding in situations in which the hemostatic

system is seriously challenged, such as during major surgery.
As clearly formulated, the focus of the dispute is whether
the bleeding time is useful to predict clinically significant
bleeding sufficiently in advance of its occurrence to allow
helpfulinterventionandtoprovide
informationnotalready available by means of other tests that would in any
event have been performed. The evidence gathered so far
is largely negative in this regard.
In particular, it has been
stressed that at present there is no evidence that the utility
of the bleeding time has been enhanced by recent advances
in standardization and that there isno evidence thatthe
bleeding time is a predictorof the risk of h e m ~ r r h a g eThese
.~
two statements are someway interconnected. It may be that
a lack of clinical usefulness to predict bleeding is caused by
variabilityintrinsic to the test as performed so far. If so,
attempts at further improving the methodology of the test
are fully justified. This is even more so because no practical
alternatives to the bleeding time asascreening procedure
for defects of primary hemostasis are on the horizon.

pre-NTG

post-NTG

INTRA-OBSERVER VARIABILITY

n = 10
70

n
U)

h
0

+I

z
10

12

TIME (min)

5I

6ol T
50

Fig 3. Modifications of the bleeding time curves after the intrave- >
0
nous administration of aspirin
(500 mg, A) or the sublingual administration of nitroglycerin (B). Values at each time point represent the
mean ? SEM of determinations in 10 different subject for each drug
studied. The curves before and aftereach pharmacological interventions differ more ( P < .01 at Kolgomorov-Smirnov statistics) than
the bleeding time( P< .05 at Students t-test for paired data). Note
the clear increase in the size of the curves after each pharmacologic
intervention(increase in total bleeding), with anincrease in the
bleeding time and the
peak bleedingrate, but n o t in t h e t i m e peak.
to

BT

TOTAL

PEAK
BLEEDING
RATE

TIME
TO PEAK
BLEEDING

INTER-OBSERVER
VARIABILITY

n
1910 by Duke.19 Subsequentimportant modifications attempted at standardizing the procedure were introduced by
Ivy et aI2and Mielke et a1.I4The test is currently performed
by means of disposabledevices producing an incision of
relatively consistent depth and lengthand measuringthe time
intercurring from the incision to the arrest of the hemorrhage
afterperiodicblotting
of thewound. Practicallyunquestioned as to its clinical usefulness until the mid-l980s, despite the lack of appropriately controlled studies addressing
the issue of its predictive ability with regard
to the risk of
actual bleeding, the validity of the bleeding time as an
useful
clinical test has been seriously questioned by a number of
inve~tigator?~and, particularly, by the accurate and complete reanalysis of the literature up to 1986 by Rodgers and
L e ~ i nUnder
. ~ discussion is not so much the value of the test
as an epidemiologic or pharmacologic tool to explore primary hemostasis, but rather its practical usefulness
in predicting bleeding episodes inpatients with a previously nega-

U)

+I

5a
8
>

40

30
20

0
10
A
Y

B T PEAK TOTAL
BLEEDING
BLEEDING
RATE

TIME
TO PEAK
BLEEDING

Fig 4. The variability in the practical performance of the bleeding

time test, as related
to the measurement of bleeding time and related
parameters. The percent coefficient of variation for
the four parameters derived from the
analysis of the bleeding time curves
(CV, as an
indices of the variability of the measurements)
in t h e t w omeasurements obtained in each patient by the same operator (intraobserver
variability) or by t w o different operators (interobserver variability)
are shown.

From www.bloodjournal.org by guest on October 26, 2016. For personal use only.

3368

DE CATERINAET

Table 3. Linear Regression Analysisof Bleeding Time and Related

ParametersVersus Indices of Perioperativeand Postoperative
Bleeding During Bypass Surgery

Bleeding time
Bleeding time
Bleeding time
Bleeding time
Bleeding time

Linear Regression
Equation

Chest tube drainage

Transfusion units
% Hct decrease
Time to Hct nadir
% Platelet level
decrease
Bleeding time
Time
to
platelet
level nadir

y = 729.33

Total
Total
Total
Total
Total

y =

bleeding
bleeding
bleeding
bleeding
bleeding

Total bleeding

Peak bleeding
Peak bleeding
Peak bleeding
Peak bleeding
Peak bleeding
Peak bleeding

Chest tube drainage

Transfusion units
% Hct decrease
Timeto Hct nadir
% Platelet level
decrease
Time to platelet
level nadir
Chest tube drainage
Transfusion units
% decrease
Hct
Time to Hct nadir
% Platelet level
decrease
Time to platelet
level nadir

+ 32.60~

y = 41.87 - 0.46~

Coefficient of
Correlation
(R)

.l4

.00

y = 38.36 + 0.24~

.05

y = 46.89 - 0.67~

.l7
.03

y = 2.23 - 0.03~

.03

y = 3.01 - 0.14~

+ 3.83~
+ 0.06~

856.21

y = 40.37

y = 37.12 + 0.18~
y =
y =

2.39 - 0.01~
39.64 + 0.02~

.l 1

.oo
.27
.09
.00

y = 2.14 - 0.01~

.l4

y = 808.45 + 24.54~
y = 3.69 + 0.12~

.l4
.07
.27

y = 36.46 + 0.98~
y = 2.31 - 3.74~

AL

bleeding are both continuous variables. We tested the hypothesis that the hemostatic competence of a patient is related to the results of the bleeding time test.
We first determined the relationship among the various
parameters derived from the bleeding time procedure. The
significant correlation found between bleeding time on one
side and total bleeding and peak bleeding rate on the other
allows us to reach the conclusion that the bleeding time is
indeed a measure of the total amount of blood loss from the
standardized cutaneous incision. However, the relatively low
correlation coefficients between the bleeding time and some
of the alternative parameters derived from the analysis of the
bleeding time curve (see Fig 2) also suggested the exitence of
bleeding determinants during the test procedure not assessed
by the bleeding time. Therefore, determination of parameters
of actual bleeding during the bleeding time procedure might
offer better correlates with blood losses at surgery. Pharmacologic interventions increasing the bleeding time (aspirin
and nitroglycerin in our case) also increase the total amount

.05

y = 39.13 + 0.19~

.03

y = 2.14 - 0.04~

.08

We first sought to determine whether the bleeding time is

actually a correlate of bleeding from the cutaneous incision.
This was investigated by carefully measuring the amount of
blood loss in each 30-second interval of the present stateof-the-art technique, which allowed us to construct, for each
bleeding time determination, a curve of blood loss over time.
A similar attempt had been previously described with similar
patterns and comparably similar baseline re~ults,''~ but
before the introduction of the template to standardize the
cutaneous incision. This approach had never been applied,
to our knowledge, to prospectively evaluate the clinical relationship with surgical bleeding. This empirical attempt is
justified by the following considerations. (l) Assessment of
alternative parameters is less affected by the relatively subjective determination of the moment of arrest of the cutaneous hemorrhage. (2) alternative parameters derived from the
analysis of the bleeding time curve might be better predictors
of bleeding in other districts of the circulation; indeed total
bleeding is expressed in the same units (weight) of the parameters to whichprediction is targeted (blood loss). (3) Due
to some variability intrinsic in any test (for the bleeding
time, see Rogers and Levin"), duplicate measurements might
be better than single determinations; these had never been
performed in previous prediction studies.
We limited our study to patients without medications obviously able to prolong the bleeding time in order to have a
baseline assessment of the normal distribution of bleeding
time and bleeding values in the absence of foreign interventions. We also decided to perform linear regression analysis
rather than receiver-operator characteristic (ROC) analysis
(assessing specificity, sensitivity, and positive and negative
predictive values) in the absence of any a priori criteria to
set cutoff values, on the assumption that bleeding time and

BP23
y = 3.5393

0.16057X

R"2 = 0.005
R=0.07

El
B

D
B

--

B
Q

'B

--.

QQ

10

bleeding time ( m i d

BP23
2000

y = 546.86

77.424~ R"2 = 0.082

R=0.2Y
B

1000
B

0 Qm

El

O !
2

10

bleeding time ( m i d
Fig 5. Linear regression analysis
of the bleedingtime (BT; x, on the
abscissas)and indices of postoperativebleeding, y, on the ordinates
(chest tube drainage in [AI and transfusion units in [B]) in bypass
surgery patients homogeneous by operation complex'm (three or
more bypasses [BPI applied; operation time, 2.3 to 4.5 hours; time
of extracorporeal circulation, 65 to 135 minutes). Both correlation
coefficients ( R I are nonsignificant. Transfusionunits equal the number of units of whole blood plus the number of units of plasma.

From www.bloodjournal.org by guest on October 26, 2016. For personal use only.

BLEEDING TIME AND BLEEDING

3369

Indeed, recent preoperative aspirin intake was associated

of blood loss and the peak bleeding rate during the procewith excessive bleeding in one of these studies." Although
dure. However, comparisons of the intraoperator and interopthe aspirin-induced prolongation of the bleeding time has
erator variabilities for the various parameters investigated
been well known for a long time:' a demonstration of the
indeed proved that the bleeding time is affected by a lower
aspirin-induced increase in postoperative blood losses has
variability than total bleeding and peak bleeding rate. The
awaited the conclusion of a large-scale multicenter ~ t u d y . ~ '
coefficient of variation for intraindividual variability of the
However, we feel that confirmation by other groups of the
bleeding time reported by us (13%,Fig 4) is comparable to
findings of the only two studies reporting a predictive value
(and actually slightly better than) the 24% and 17% preof the bleeding time on perioperative bleedingI0.l3is needed
viously reported with the Simplate I and Simplate I1 devices,
to recommend the use of the bleeding time in the preoperarespectively.28,29 However, even in this case, the coefficient
tive screening of cardiac surgery patients. Our negative conof variation is greater than lo%, which is larger than that
clusion pertains to patients with no history of bleeding or of
found for the vast majority of laboratory tests, despite being
recent intake of non-steroidal anti-inflammatory drugs. In
obtained in our case by two trained operators who are likely
this setting, duplicate performance of the bleeding time and
to be better than the average physicians or technicians usuthe assessment of parameters related to it (otherwise useful,
ally administering the test in a clinical setting.
in our hands, to document the effects of drugs affecting
We subsequently compared all parameters derived from
primary hemostasis) does allow us to establish not even a
the analysis of the bleeding time curve with a variety of
trend towards significant correlations with actual bleeding.
indices of actual bleeding during surgery. To minimize some
Based on our results, we suggest not using the bleeding
surgical determinants of variable bleeding values, we elected
time test in the preoperative screening of patients undergoing
to study only one type of surgical operations (elective isocoronary surgery without history of bleeding and of recent
lated coronary bypass surgery), all performed by the same
intake of non-steroidal anti-inflammatory drugs.
surgeon. In consideration of the possibly preponderant value
that length and complexity of the operation mayplay in
determining surgical bleeding, we performed multiple linear
ACKNOWLEDGMENT
regression analysis, including variables such as duration of
The authors thank Dr M. Benedetti for his help in the surgical
the operation, duration of the extracorporeal circulation, and
patient selection and all the medical and paramedical personnel in
the number of bypasses applied. Even accounting for all
the Division of Cardiac Surgery, Spedali Riuniti di Santa Chiara,
these factors, no significant relationship was disclosed beand the C.N.R.Institute of Clinical Physiology, Pisa, for their collabtween any value derived from the bleeding time test and any
oration in making the study possible. Thanks also to Dr Stuart E.
index of actual bleeding.
Lind (Brigham and Women's Hospital, Boston, MA) for his critical
This negative conclusion is in line with a number of previreview of the manuscript.
ous attempts at correlating surgical losses with the results
of the tests?' Neither the selection and careful definition
REFERENCES
of alternative bleeding time parameters nor the attempts at
1.
White
GI,
Marder
V,
Colman R, Hirsh J, Salzman E: Hemostastandardizing some of the surgical variables and in choosing
sis and Thrombosis. Basic Principles and Clinical Practice. Philadelmultiple indicators of surgical bleeding were able to revert
phia, PA, Lippincott, 1987, p 1048
previously expressed skeptical 0pinions.4.~This conclusion
2. Lind S: Prolonged bleeding time. Am J Med 77:305, 1984
is also in agreement with that of a recently reported study
3. Bums E, Lawrence C: Bleeding time: A guide to its diagnostic
aimed at relating preoperative and postoperative bleeding
and clinical utility. Arch Pathol Lab Med 113:1219, 1989
time with surgical blood losses.12 In this report, postopera4. Rodgers R, Levin J: A critical reappraisal of the bleeding time.
tive, but not preoperative, bleeding time was related to postSemin Thromb Hemost 16:1, 1990
5 . Triplett D: The bleeding time: Neither pariah or panacea. Arch
operative blood losses after cardiopulmonary bypass. Our
Pathol Lab Med 113:1207, 1989
conclusion is at variance, on the other hand, from that of two
6. Editorial. The bleeding time. Lancet 337:1447, 1991
recently reported studies. In one of these, 159 consecutive
7. Lind S: The bleeding time does not predict surgical bleeding.
patients undergoing coronary bypass surgery were studied
Blood 17~2547,1991
in a multivariate analysis to determine clinical and laboratory
8. Bums E Bleeding time: A guide to its diagnostic and clinical
predictors of excessive postoperative packed red blood cell
utility. Arch Path01 Lab Med 114 1187, 1990 (letter)
transfusion.loIn this study, preoperative bleeding time ap9. Gimple L, Gold H, Leinbach R, Coller B, Werner W, Yasuda
peared to be an useful predictor of excessive postoperative
T, Johns J, Ziskind A, Finkelstein D, Collen D: Correlation between
blood transfusion, which is the only indicator of bleeding
template bleeding times and spontaneous bleeding during treatment
with recombinant tissue-type plasminogen activator. Circulation
reported. Another recent report by the same group has re80:581, 1989
cently reemphasized the same finding using the ratio of preIO. Ferraris V, Gildengorin V: Predictors of excessive blood use
operative bleeding time to preoperative packed red blood
after coronary bypass grafting. J Thorac Cardiovasc Surg 98:492,
cell volume to predict the need of postoperative blood trans1989
fusion.13 We have attempted to use the same index in our
11. Gluszko P, Maring J, Edmunds L: Bleeding time test. J Thorac
patients, without improving the ralationship of the bleeding
Cardiovasc Surg 101:173, 1991
time versus indices of operative bleeding. A possible reason
12. Khuri S, Wolfe A, Josa M, Axford T, Szymaski I, Assousa
for discrepancy is the inclusion of patients with drug-induced
S, Ragno G, Pate1 M, Silverman A, Park M, Valeri R: Hematologic
prolongation of the bleeding time in the quoted s t u d i e ~ . ' ~ . ' ~ changes during and after cardiopulmonary bypass and their relation-

From www.bloodjournal.org by guest on October 26, 2016. For personal use only.

DE CATERINA ET AL

3370

ship to the bleeding time and nonsurgical blood loss. J Thorac

Cardiovasc Surg 104:94, 1992
13. Ferraris V, Berry W, Klingman R: Comparison of blood reinfusion techniques used during coronary artery bypass grafting. Ann
Thorac Surg 56:433, 1993
14. Mielke C, Kaneshiro M, Maher I, Weiner J, Rapaport S:
The standardized normal Ivy bleeding time and its prolongation by
aspirin. Blood 34:204, 1969
15. De Nicola P: Critical evaluation of methods for the study
of blood coagulation, fibrinolysis and hemostasis. Thromb Diath
Haemomh 7:325, 1962 (suppl 1)
16. Benedetti M, De Caterina R, Bionda A, Gardinali M, Cicardi
M, Maffei S, Gazzetti P, Pistolesi P, Vernazza F, Michelassi C,
Giordani R, Salvatore L: Blood-artificial surface interactions during
cardiopulmonary bypass. A comparative study of four oxygenators.
Int J Art Org 13:488, 1990
17. Draper N, Smith H: Applied Regression Analysis. New York,
NY,Wiley, 1980
18. Milian M: Contribution a Ibtude de la coagulation du sang.
C R Soc Biol (Paris) 53556, 1901
19. Duke W: The relation of blood platelets to hemorrhagic disease. Description of a method of determining the bleeding time and
coagulation time and report of three cases of hemorrhagic disease
relieved by transfusion. JAMA 55:1185, 1910
20. Ivy A, Nelson D, Bucher G: The standardization of certain
factors in the cutaneous venostasis bleeding time technique. J
Lab Clin Med 26:1812, 1941
2 I . Rodgers R, Lewin J: Bleeding time: A guide to its diagnostic
and clinical utility. Arch Pathol Lab Med 114:1187, 1990 (letter)
22. Adelson E, Crosby W: A new method to measure bleeding

time: The immersion method. Acta Haematol (Basel) 18:281,

I957
23. Willoughby M, Allington M: The rate of blood loss from skin
punctures during the Ivy bleeding time test. J Clin Pathol 14:381, 1961
24. Sutor A, Bowie EJW, Thompson JJ, Didisheim P, Mertens
B, Owen CJ: Bleeding from standardized skin punctures: Automatic
technic for recording time, intensity, and pattern of bleeding. Am J
Clin Pathol 5.5541, 1971
25. Bowie EJW: The bleeding time. Prog Hemost Thromb 2:249,
1974
26. Sutor A, Bowie EJW, Owen CJ: Quantitative bleeding time
(hemorrhagometry). A review. Mayo Clin Proc 52:238, 1977
27. Rodgers R, Levin J: Bleeding time tables. A tabular summary
of pertinent literature. Semin Thromb Hemost 16:21, 1990
28. Sheridan B, Fraser E, Tay W, Hayne 0: The Simplate bleeding time: A retrospective analysis. Blood 62:268a, 1983 (abstr, suppl
1)

29. Babson S, Babson A: Development and evaluation of a disposable device for performing simultaneous duplicate bleeding time
determinations. Am J Clin Pathol 70:406, 1978
30. Beaumont J-L, Willie A, Lentgre J: Influence de Iacide acetilsalicylique sur Ihbmostase. Bull Mem Soc Med Hop Paris 7 I : 1077.
1955
3 I . Goldman S, Copeland J , Moritz M, Henderson W, Zadina K,
Ovitt T, Doherty J, Read R, Chesler E, Sako Y, Lancaster L, Emery
R, Sharma G, Josa M, Pacold I, Montoye A, Parich D, Sethi G, Holt
J , Kirklin J, Shabetai R, Moores W, Aldridge J, Masud Z, DeMots
H, Floten S, Haakerson C, Harker L: Improvement in early saphenous vein graft patency after coronary artery bypass surgery with
antiplatelet therapy: Results of a Veterans Administration Cooperative Study. Circulation 77: 1324, 1988

From www.bloodjournal.org by guest on October 26, 2016. For personal use only.

1994 84: 3363-3370

Bleeding time and bleeding: an analysis of the relationship of the

bleeding time test with parameters of surgical bleeding
R De Caterina, M Lanza, G Manca, GB Strata, S Maffei and L Salvatore

Updated information and services can be found at:

http://www.bloodjournal.org/content/84/10/3363.full.html
Articles on similar topics can be found in the following Blood collections
Information about reproducing this article in parts or in its entirety may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#repub_requests
Information about ordering reprints may be found online at:
http://www.bloodjournal.org/site/misc/rights.xhtml#reprints
Information about subscriptions and ASH membership may be found online at:
http://www.bloodjournal.org/site/subscriptions/index.xhtml

Blood (print ISSN 0006-4971, online ISSN 1528-0020), is published weekly by the American
Society of Hematology, 2021 L St, NW, Suite 900, Washington DC 20036.
Copyright 2011 by The American Society of Hematology; all rights reserved.