Francois Lauzier

Donald M. Arnold
Christian Rabbat
Diane Heels-Ansdell
Ryan Zarychanski
Peter Dodek
Betty Jean Ashley
Martin Albert
Kosar Khwaja
Marlies Ostermann
Yoanna Skrobik
Robert Fowler
Lauralyn McIntyre
Joseph L. Nates
Tim Karachi
Renato D. Lopes
Nicole Zytaruk
Simon Finfer
Mark Crowther
Deborah Cook
Risk factors and impact of major bleeding
in critically ill patients receiving heparin
thromboprophylaxis
Received: 5 December 2012
Accepted: 22 July 2013
Published online: 14 August 2013
Springer-Verlag Berlin Heidelberg and
ESICM 2013
For the PROTECT Investigators, the
Canadian Critical Care Trials Group, and
the Australian and New Zealand Intensive
Care Society Clinical Trials Group.
Electronic supplementary material
The online version of this article (doi:10.1007/
s00134-013-3044-3) contains supplementary
material, whichis available toauthorizedusers.
F. Lauzier (
)
)
Division of Critical Care, Departments of
Medicine and of Anesthesiology, Centre de
recherche du CHU de Quebec, Axe Sante
des populations et pratiques optimales en
sante, Universite Laval, 1401, 18e Rue,
Quebec, QC G1J 1Z4, Canada
e-mail: francois.lauzier@med.ulaval.ca
Tel.: ?1-418-649-0252
Fax: ?1-418-649-5733
D. M. Arnold C. Rabbat
T. Karachi M. Crowther D. Cook
Department of Medicine, McMaster
University, Hamilton, Canada
D. M. Arnold D. Heels-Ansdell
N. Zytaruk M. Crowther D. Cook
Department of Clinical Epidemiology and
Biostatistics, McMaster University, Hamilton,
Canada
D. M. Arnold
Canadian Blood Services, Hamilton, ON,
Canada
R. Zarychanski
Department of Internal Medicine,
University of Manitoba, Winnipeg, MB,
Canada
R. Zarychanski
Department of Medical Oncology and
Hematology, CancerCare Manitoba,
Winnipeg, MB, Canada
P. Dodek B. J. Ashley
Division of Critical Care Medicine, Center
for Health Evaluation and Outcome
Sciences, St. Pauls Hospital, University of
British Columbia, Vancouver, BC, Canada
M. Albert
Departments of Medicine and of Critical
Care, Centre de recherche de lHopital du
Sacre-Coeur de Montreal, Universite de
Montreal, Quebec, Canada
K. Khwaja
Departments of Surgery and of Critical
Care, McGill University Health Centre,
Montreal, QC, Canada
M. Ostermann
Department of Critical Care, Kings College
London, Guys and St Thomas Foundation
Hospital, London, UK
Y. Skrobik
Lise and Jean Saine Critical Care Chair,
Department of Medicine, Universite de
Montreal; Regroupement de Soins Critiques
respiratoires, Reseau de Sante Respiratoire,
Quebec, Canada
R. Fowler
Departments of Critical Care Medicine and of
Medicine, Sunnybrook Health Sciences Centre,
University of Toronto, Toronto, ON, Canada
L. McIntyre
Division of Critical Care, Department of
Medicine, University of Ottawa, Ottawa,
ON, Canada
J. L. Nates
Department of Critical Care, The University
of Texas MD Anderson Cancer Center,
Houston, TX, USA
R. D. Lopes
Division of Cardiovascular Medicine, Duke
Clinical Research Institute, Durham, NC, USA
S. Finfer
Department of Intensive Care, Royal North
Shore Hospital and the George Institute for
Global Health, University of Sydney,
Sydney, NSW, Australia
Intensive Care Med (2013) 39:21352143
DOI 10.1007/s00134-013-3044-3
ORIGINAL
Abstract Purpose: Bleeding fre-
quently complicates critical illness
and may have serious consequences.
Our objectives are to describe the
predictors of major bleeding and the
association between bleeding and
mortality in medicalsurgical criti-
cally ill patients receiving heparin
thromboprophylaxis. Methods: We
prospectively studied patients from
67 intensive care units and six coun-
tries enrolled in a
thromboprophylaxis trial
(NCT00182143) comparing daltepa-
rin with unfractionated heparin.
Patients with trauma, orthopedic sur-
gery or neurosurgery were excluded.
Trained research coordinators used a
validated tool to document bleeding,
which underwent duplicate indepen-
dent blinded adjudication. Major
bleeding was dened as hypovolemic
shock, bleeding into critical sites,
requiring an invasive intervention or
transfusion of at least two units of red
blood cells, or associated with hypo-
tension or tachycardia in the absence
of other causes. Adjusted Cox pro-
portional hazard regression analysis
was used to identify major bleeding
predictors and the association
between bleeding and mortality.
Results: Among 3,746 patients,
bleeding occurred in 208 [5.6 %,
95 % condence interval (CI)
4.96.3 %]. Time-dependent predic-
tors were prolonged activated partial
thromboplastin time [hazard ratio
(HR) 1.10, 1.051.14 per 10 s
increase], lower platelet count (HR
1.16, 1.091.24 per 50 9 10
9
/L
decrease), therapeutic heparin (HR
3.26, 1.726.17), antiplatelet agents
(HR 1.38, 1.021.88), renal replace-
ment therapy (HR 1.75, 1.202.56),
and recent surgery (HR 1.64,
1.012.65). Type of pharmacologic
thromboprophylaxis was not associ-
ated with bleeding. Patients with
bleeding had a higher risk of in-hos-
pital death (HR 2.09, 1.692.57).
Conclusions: As major bleeding has
modiable risk factors and is associ-
ated with in-hospital mortality,
strategies to mitigate these factors
should be evaluated in critically ill
patients.
Keywords Hemorrhage
Critical care Mortality Embolism
Thrombosis Platelet aggregation
inhibitors Heparin
Introduction
Major bleeding is associated with increased morbidity and
mortality in acutely ill patients who have acute coronary
syndrome [1], in patients undergoing peripheral vascular
surgery [2], in recipients of bone marrow transplant [3],
and in heterogeneous population of patients hospitalized
for medical conditions or undergoing orthopedic or
abdominal surgery [4]. However, such association has not
been demonstrated in critically ill patients. Studies to date
estimating bleeding rates, analyzing predisposing factors
and bleeding consequences during critical illness have
been small and underpowered.
Critically ill patients may be predisposed to major
bleeding because of clinical conditions such as throm-
bocytopenia, platelet dysfunction, coagulation factor
deciencies, use of antiplatelet or anticoagulant agents,
and invasive procedures such as insertion of central
venous catheters. A better understanding of the incidence,
the risk factors, and the consequences of major bleeding
may help to anticipate such events during critical illness,
prevent their occurrence, potentially improve clinically
relevant outcomes, and decrease associated healthcare
costs.
The three objectives of this study are to: (1) describe
the incidence, timing, and anatomical locations of major
bleeding in critically ill patients; (2) evaluate the risk
factors for major bleeding in these patients; and (3)
investigate the association between major bleeding and
blood product transfusion, duration of mechanical venti-
lation, intensive care unit (ICU) and hospital length of
stay, and mortality. We hypothesized that major bleeding
risk in critically ill patients is related to the use of med-
ication affecting coagulation such as anticoagulant and
antiplatelet agents, to abnormal coagulation parameters
such as low platelet counts, high international normalized
ratio (INR), and prolonged activated partial thrombo-
plastin time (aPTT), and to renal failure requiring
replacement therapy. We also hypothesized that major
bleeding would be associated with prolonged length of
stay, increased blood product transfusion, and increased
risk of ICU and in-hospital mortality. Part of this study
was previously published in abstract form [5].
Patients and methods
Study design
We analyzed a cohort of patients participating in an
international, multicenter, blinded thromboprophylaxis
trial that compared subcutaneous unfractionated heparin
5,000 units twice daily versus subcutaneous dalteparin
5,000 units daily for thromboprophylaxis in critically ill
medicalsurgical patients (NCT00182143) [6, 7]. Patients
C18 years of age, weighing C45 kg, and expected to
remain in the ICU C72 h were eligible. Exclusion criteria
were admissions for trauma, neurosurgery or orthopedic
surgery, therapeutic anticoagulation at baseline, prophy-
lactic or therapeutic heparin for at least 72 h prior to ICU
admission, contraindication to heparin or blood products,
pregnancy, limitation of life support, platelet count
B75 9 10
9
/L or abnormal coagulation testing (INR or
2136
aPTT greater than twice the upper normal limit), or
enrollment in a related trial. Research coordinators col-
lected daily data during the ICU stay related to life-
sustaining therapy, laboratory tests, indwelling catheters
or devices, drugs affecting thrombotic or bleeding risk,
and details of bleeding events using a validated bleeding
assessment tool developed for critically ill patients [8].
All management decisions, including thresholds for
transfusions, withholding of thromboprophylaxis, thera-
peutic drugs, use of invasive devices, and surgical
interventions were at the discretion of the treating
physicians.
Bleeding adjudication
We adjudicated bleeding events independently and in
duplicate by using a standardized pretested measurement
tool (Electronic Supplementary Material) [9]. Adjudica-
tors were blinded to center and treatment allocation.
Bleeding was dened as major if, in the absence of
another cause, the patient fullled one of the following
criteria: (1) hypovolemic shock, (2) bleeding into critical
sites, (3) invasive intervention requirement, or (4) need
for transfusion of C2 units of red blood cells, or was
associated with a decrease of C20 mmHg in systolic
blood pressure or an increase of at least 20 beats/min in
heart rate without another explanation. Bleeding that did
not satisfy these criteria was dened as minor.
Statistical analysis
We present categorical data as counts and percentages,
and continuous data as mean (standard deviation) or
median (25th percentile75th percentile) if data were
skewed. The incidence of bleeding is expressed as a
proportion.
We performed multivariable Cox proportional hazards
regression analysis to identify independent risk factors for
major bleeding. Based on a priori plans, we included the
following baseline characteristics: age, Acute Physiology
and Chronic Health Evaluation (APACHE) II score [10],
medical versus surgical admission, and pre-ICU end-stage
dialysis-dependent renal failure. We also included time-
dependent covariates: advanced life support (mechanical
ventilation, vasopressor/inotrope infusion, and renal
replacement therapy), surgery (requiring an incision
either at the bedside or in the operating room), coagula-
tion parameters (highest INR, longest aPTT, and lowest
platelet count), medications [stress ulcer prophylaxis,
therapeutic anticoagulation, activated protein C, prophy-
lactic dalteparin, prophylactic unfractionated heparin, and
antiplatelet agents (acetylsalicylic acid or clopidogrel)].
All time-dependent variables were considered in the
3 days prior to major bleeding, or 7 days prior to bleeding
in the case of antiplatelet agents [11]. The nal model
included all covariates regardless of their p value on
univariable analyses. We evaluated rst-order interactions
and tested the assumption of nonproportional hazards by
evaluating the interaction between each baseline variable
with time. To assess the association between major
bleeding and mortality, we performed multivariable Cox
proportional hazards regression analysis with adjustment
for age, APACHE II score [10], and use of inotropes,
vasopressors, and mechanical ventilation at study enroll-
ment. We used similar models to assess the time to rst
transfusion of each blood product. Twenty-four patients
who were missing any of the covariates were excluded
from multivariable analyses. To compare the duration of
mechanical ventilation and length of stay between
patients with or without major bleeding, we used a two-
sided Wilcoxon test. All results are presented with 95 %
CI; p values \0.05 were considered statistically signi-
cant. Statistical analyses were performed using SAS
version 9.2 (SAS Institute, Cary, NC).
Results
Among 6,034 patients fullling all inclusion criteria and
no exclusion criteria, 3,764 (62.4 %) were randomized.
The consent rate was 82.3 % (3,764 for 4,574 patients
approached for consent). Eighteen patients withdrew
consent. The study cohort therefore included 3,746
patients enrolled between May 2006 and June 2010.
Baseline characteristics, including admission type and
diagnosis on ICU admission, are presented in Table 1
according to the occurrence of major bleeding. Most
patients were male, admitted for respiratory failure or
sepsis, and were mechanically ventilated. Bleeding
occurred in 491 patients (13.1 %, 95 % CI 12.114.2 %),
and was classied as major in 208 patients (5.6 %, 95 %
CI 4.96.3 %) and minor in 283 patients (7.6 %, 95 % CI
6.88.4 %). The most common sites of major bleeding
were the gastrointestinal tract, a surgical site, the respira-
tory tract, and the retroperitoneal space (Table 2). Among
the entire cohort, seven patients (0.2 %, 95 % CI
0.10.4 %) developed intracranial hemorrhage during
their ICU stay and 31 patients (0.8 %, 95 % CI 0.61.2 %)
had two or more major bleeding episodes. Among the 208
patients with major bleeding, 36 (17.3 %) had hypovole-
mic shock, 37 (17.8 %) had bleeding at a critical site, 73
(35.1 %) required an invasive intervention, and 183
(88.0 %) had clinically important bleeding as described in
the Patients and methods section (criterion 4). Major
bleeding criteria were satised for a median of 2 days
(13 days). The median time from ICU admission to rst
major bleeding was 9.5 days (615.5 days). Invasive
procedures and therapeutic interventions potentially
inuencing the risk of bleeding are outlined in Table 3.
2137
Factors independently associated with major bleeding
No baseline patient characteristic was independently
associated with the risk of major bleeding. We found six
signicant time-dependent predictors of major bleeding
(Table 4): prolonged aPTT, low platelet count, use of
therapeutic heparin, use of antiplatelet agents, use of renal
replacement therapy, and any surgical procedure. Similar
results were obtained when backwards-stepwise elimina-
tion analyses were performed (data not shown). When we
examined all two-way interactions of predictors in the
model after backwards-stepwise elimination, we found
only one signicant interaction. The association between
lower platelet count and major bleeding was stronger in
the absence of concomitant acetylsalicylic acid or clopi-
dogrel (HR 1.25, 1.151.36) than with either drug (HR
1.06, 0.961.17) (p value for interaction = 0.0095).
Platelet counts were lower at the time of bleeding in
patients who did not receive antiplatelet agents compared
with those who did [median 149 9 10
9
/L (104223 9
10
9
/L) versus 212 9 10
9
/L (128274 9 10
9
/L), p =
0.005]. There was no interaction between use of thera-
peutic heparin and aPTT (p = 0.41), between use of
therapeutic heparin and renal replacement therapy
(p = 0.54), or between renal replacement therapy and
use of antiplatelet agents (p = 0.38). Neither dalteparin
nor unfractionated heparin thromboprophylaxis was
associated with major bleeding. When we tested the
Table 1 Baseline
characteristics
Patient characteristics Patients with major
bleeding, n = 208
Patients without major
bleeding, n = 3,538
p-Value
Age, mean (SD), years 63.4 (16.3) 61.3 (16.5) 0.08
Female sex, n (%) 89 (42.8) 1,525 (43.3)
a
0.89
APACHE II score, mean (SD) 23.5 (7.6) 21.4 (7.8)
b
\0.001
Body mass index, mean (SD), kg/m
2
28.0 (7.3)
c
28.3 (7.7)
d
0.57
Medical admission, n (%) 155 (74.5) 2,676 (75.6) 0.74
Personal history of VTE, n (%) 9 (4.3) 111 (3.2)
a
0.31
Admitting diagnosis, n (%)
Respiratory 71 (34.1) 1,630 (46.3) \0.001
Sepsis 51 (24.5) 498 (14.2)
Gastrointestinal 36 (17.3) 484 (13.8)
Cardiovascular 17 (8.2) 319 (9.1)
a
Neurologic 8 (3.8) 221 (6.3)
Renal 6 (2.9) 59 (1.7)
Metabolic 4 (1.9) 140 (4.0)
Othersurgical 9 (4.3) 109 (3.1)
Othermedical 6 (2.9) 59 (1.7)
History of malignancy, n (%) 8 (3.8) 142 (4.0)
a
1.0
Baseline life support, n (%)
Mechanical ventilation 195 (93.8) 3,163 (90.0)
e
0.09
Vasopressors 121 (58.2) 1,556 (44.3)
e
\0.001
Chronic dialysis 8 (3.8) 110 (3.1)
a
0.54
This table presents the characteristics of enrolled patients according to major bleeding status. Due to
some missing data, denominators vary for some summary measures (
a
n = 3,519;
b
n = 3,517;
c
n = 197;
d
n = 3,400;
e
n = 3,516)
SD standard deviation, APACHE Acute Physiology and Chronic Health Evaluation, VTE venous
thromboembolism
Table 2 Sites of major
bleeding
Site of bleeding Number of
patients, n (%)
% of patients with
major bleeding n = 208
(95 % CI)
% of overall cohort
n = 3,746 (95 % CI)
Gastrointestinal 108 51.9 (45.258.6) 2.9 (2.43.5)
Surgical site
a
63 30.3 (24.436.8) 1.7 (1.32.1)
Respiratory 33 15.9 (11.521.4) 0.9 (0.61.2)
Retroperitoneal 17 8.2 (5.212.7) 0.5 (0.30.7)
Intracranial 7 3.4 (1.66.8) 0.2 (0.10.4)
Pericardial 3 1.4 (0.54.2) 0.1 (0.030.2)
Other 20 9.6 (6.314.4) 0.5 (0.30.8)
This table presents the sites of major bleeding among 208 patients. Sites of bleeding are not mutually
exclusive
a
15 patients had major bleeding at central venous catheter insertion site
2138
Table 3 Incidence of potential
risk factors for bleeding
All patients,
n = 3,724
a
Patients with major
bleeding, n = 208
Use of antiplatelet agents 1,244 (33.4) 80 (38.5)
Acetylsalicylic acid 1,204 (32.3) 78 (37.5)
Clopidogrel 194 (5.2) 10 (4.8)
Concomitant therapy 154 (4.1) 8 (3.8)
Use of therapeutic heparin 144 (3.9) 13 (6.3)
Renal replacement therapy 502 (13.5) 56 (26.9)
Modality
Intermittent 334 (9.0) 31 (14.9)
Continuous 258 (6.9) 26 (12.5)
Sustained low-efciency dialysis 36 (1.0) 4 (1.9)
Circuit anticoagulation
Unfractionated heparin 289 (7.8) 22 (10.6)
Saline ushes 251 (6.7) 21 (10.1)
Citrate 160 (4.3) 19 (9.1)
Danaparoid 7 (0.2) 1 (0.5)
Surgery 678 (18.2) 19 (9.1)
Head and neck 366 (9.8) 10 (4.8)
Thoracic 74 (2.0) 2 (1.0)
Gastrointestinal 221 (5.9) 7 (3.4)
Plastic 51 (1.4) 1 (0.5)
Cardiac 17 (0.5) 0 (0.0)
Vascular 14 (0.4) 0 (0.0)
Orthopedic 15 (0.4) 0 (0.0)
This table presents therapeutic interventions and all surgical procedures in all patients in the second
column. In the third column, among patients who had major bleeding, we present these interventions up
to 3 days prior to major bleeding (or 7 days for use of antiplatelet agents)
a
Twenty-two patients had missing daily data. Therefore, 3,724 patients contributed to the risk factor
analysis
Table 4 Risk factors for major
bleeding
Hazard ratio
(95 % CI)
Adjusted
p value
Baseline characteristics
Age (10 year increase) 0.99 (0.891.09) 0.77
APACHE II (10 point increase) 1.04 (0.851.27) 0.70
Medical admission 1.31 (0.881.93) 0.18
End-stage renal disease 0.88 (0.411.88) 0.74
Time-dependent factors
Life support
Invasive mechanical ventilation 1.04 (0.651.67) 0.87
Vasopressor infusion 1.27 (0.921.74) 0.14
Any renal replacement therapy 1.75 (1.202.56) 0.004
Laboratory results
Highest INR (0.5 unit increase) 1.03 (0.931.14) 0.58
Highest aPTT (10 s increase) 1.10 (1.051.14) \0.001
Lowest platelet count (50 9 10
9
/L decrease) 1.16 (1.091.24) \0.001
Other interventions
Surgical procedure 1.64 (1.012.65) 0.045
Stress ulcer prophylaxis 1.09 (0.661.82) 0.73
Acetylsalicylic acid or clopidogrel 1.38 (1.021.88) 0.04
Therapeutic heparin 3.26 (1.726.17) \0.001
Prophylactic low-molecular-weight heparin 1.16 (0.781.73) 0.46
Prophylactic unfractionated heparin 0.93 (0.621.39) 0.72
Activated protein C 1.26 (0.503.17) 0.62
This table presents the independent risk factors for major bleeding using multivariable time-to-event
Cox regression analysis. Time-dependent risk factors were considered in the preceding 3 days, or
7 days in the case of acetylsalicylic acid or clopidogrel. All covariates were included in the model
APACHE Acute Physiology and Chronic Health Evaluation, INR international normalized ratio, aPTT
activated partial thromboplastin time
2139
assumption of nonproportional hazards, we did not nd
any signicant interactions between time and any of the
variables.
Clinical outcomes related to major bleeding
When compared with patients who did not have major
bleeding, patients who had major bleeding had longer
median duration of mechanical ventilation, and longer
ICU and hospital length of stay, and received signicantly
more red blood cell and frozen plasma transfusions
(Table 5). Most transfused patients received leukode-
pleted blood products, as universal leukodepletion was
implemented in Canada and the UK since 1999 and in
Australia since 2008 and 86.6 % of patients enrolled were
from those countries. Among patients with major bleed-
ing who died, the median time from the rst day of
bleeding to death was 11 (525) days. After adjustment
for baseline characteristics, patients who had major
bleeding were signicantly more likely to be transfused
and had signicantly higher ICU mortality (HR 1.64,
1.272.10) and hospital mortality (HR 2.09, 1.692.57)
(Table 6).
Discussion
In medicalsurgical critically ill patients, we observed an
incidence of major bleeding of approximately 6 %. We
also found that prolonged aPTT, lower platelet count,
treatment with therapeutic heparin, antiplatelet agents,
renal replacement therapy, and surgery were indepen-
dently associated with major bleeding. Neither baseline
characteristics nor type of heparin thromboprophylaxis
predicted major bleeding. When adjusting for other con-
founders, major bleeding was associated with greater
transfusion of blood products, and a twofold increased
risk of ICU and hospital mortality.
Our ndings are consistent with the 5.2 % (95 % CI
3.67.6 %) major bleeding rate found in a single-center
observational study of consecutive unselected medical
surgical ICU patients, all of whom received prophylactic
unfractionated heparin [8], and similar to a 7.2 % (95 %
CI 4.012.8 %) major bleeding rate found in a multicenter
observational study in which all medicalsurgical ICU
patients received prophylactic low-molecular-weight
heparin [12]. Our results differ from those of a multi-
center observational study of non-critically ill medical
patients [13], which found a 1.2 % major bleeding
Table 5 Blood products transfused, duration of mechanical ventilation, and ICU and hospital length of stay
Patients with major
bleeding, n = 208
Patients without major
bleeding, n = 3,538
a
Unadjusted
p value
Units of red blood cells transfused (ever),
b
median (25th75th) 8 (415), n = 193 2 (14), n = 1,034 \0.001
Units of platelets transfused (ever),
b
median (25th75th) 5 (2.511.5), n = 52 5 (26), n = 61 0.15
Units of frozen plasma transfused (ever),
b
median (25th75th) 6 (211), n = 91 3 (25), n = 190 \0.001
Units of cryoprecipitate transfused (ever),
b
median (25th75th) 7 (310), n = 20 5 (510), n = 3 0.89
Duration of invasive mechanical ventilation (days), median (25th75th) 16.5 (1030) 5 (210) \0.001
Duration of ICU stay (days), median (25th75th) 23 (1539) 9 (615) \0.001
Duration of hospital stay (days), median (25th75th) 36 (2166) 21 (1238) \0.001
This table presents unadjusted comparison of the amount of blood
products transfused per patient and duration of mechanical venti-
lation, and ICU and hospital stay in patients who had major
bleeding and patients without major bleeding
a
n = 3,516 who have daily data for transfusion
b
Per patient who had transfusions
Table 6 Multivariable analysis of transfusion of blood products and mortality
Patients with major
bleeding, n = 208
Patients without major
bleeding, n = 3,538
a
Adjusted
b
hazard
ratio (95 % CI)
Adjusted
b
p value
Red blood cell transfusion (ever), n (%) 193 (92.8) 1,034 (29.4) 8.54 (6.7010.89) \0.001
Platelet transfusion (ever), n (%) 52 (25.0) 61 (1.7) 18.61 (11.7129.57) \0.001
Frozen plasma transfusion (ever), n (%) 91 (43.8) 190 (5.4) 15.81 (11.2722.18) \0.001
Cryoprecipitate transfusion (ever), n (%) 20 (9.6) 3 (0.1) 97.20 (31.59299.05) \0.001
ICU mortality, n (%) 82 (39.4) 506 (14.3) 1.64 (1.272.10) \0.001
Hospital mortality, n (%) 106 (51.0) 767 (21.7) 2.09 (1.692.57) \0.001
This table presents the outcomes associated with major bleeding
ICU intensive care unit
a
n = 3,516 who have daily data for transfusion and ventilation
information
b
Adjusted for age, APACHE II score, baseline use of inotropes/
vasopressors, and baseline mechanical ventilation (invasive or
noninvasive). All covariates were included in the models
2140
incidence during the rst 14 days of hospitalization and
did not evaluate time-dependent bleeding risk factors.
In our analysis of baseline and time-dependent risk
factors for major bleeding, we identied that renal
replacement therapy conferred an increased risk for major
bleeding, which may reect the platelet dysfunction
associated with renal failure [14] and the associated
increased bleeding risk [1517]. Most renal replacement
therapy used in this study was intermittent and associated
with saline ushes rather than heparin to maintain circuit
patency. Nevertheless, there was no interaction between
renal replacement therapy, therapeutic heparin, and aPTT.
We also found that prolonged aPTT was an independent
predictor of major bleeding, as previously described in
patients who have acute coronary syndromes in whom
every 10 s increase in aPTT was found to be associated
with a 7 % increase in major bleeding [18]. The inde-
pendent association between therapeutic unfractionated
heparin and aPTT, as highlighted by the absence of sig-
nicant interaction between those variables, may be
explained by the fact that the therapeutic efcacy of
heparin correlates more with the dose administered rather
than the aPTT [19, 20]; moreover, maximal aPTT does
not necessarily reect the time spent in the therapeutic
range. In contrast, we found no association between
abnormal INR and bleeding. This observation is congru-
ent with previous studies demonstrating that prolonged
INR does not predict bleeding in this population [21] and
supports the observation that INR is not an adequate
surrogate marker of decits in coagulation factors among
critically ill patients who are not receiving vitamin K
antagonists [22]. Our nding that lower platelet count
predicts major bleeding is consistent with a previous
single-center study (HR 1.7 for every 50 9 10
9
/L
decrease in platelet count) [8] and is in contrast with other
studies in this population using univariate analyses [23,
24]. We found that treatment with antiplatelet agents was
an independent predictor of bleeding, which was also
identied in observational studies that enrolled patients
who received prophylactic unfractionated heparin [8] or
low-molecular-weight heparin [25]. We identied an
increased risk of major bleeding after invasive proce-
dures, mainly thoracic surgeries such as pneumonectomy,
abdominal surgeries such as bowel resection, and tra-
cheotomies. This association between bleeding and
surgery was independent of admitting diagnosis, disease
severity, and type of thromboprophylaxis used.
We identied interactions that suggest platelet count
was a stronger predictor of bleeding in the absence of
concomitant acetylsalicylic acid or clopidogrel. This
likely reects the prescribing practice of physicians who
avoided anti-platelet agents in patients with lower platelet
count. These patients may have been perceived to be at a
higher risk of bleeding.
We also observed an association between major
bleeding and mortality. This nding parallels previous
studies in acutely ill patients which showed that major
bleeding is associated with a signicant risk of death in
patients with acute coronary syndromes [1, 26] and in
bone marrow transplant recipients [3]. The underlying
pathophysiology leading to death remains unclear, but is
not likely directly related to bleeding as suggested by the
median time between major bleeding and death [11 days
(525 days)]. Mortality risk could to some extent be
related to an increased risk of cardiovascular or other life-
threatening complications occurring over a more pro-
longed ICU stay. Increased mortality could to some extent
be related to the use of blood products, which may be
associated with an increased risk of infection [27] and
acute lung injury [28]. It may also be related to with-
holding of benecial therapies, such as antiplatelet agents
in patients with acute coronary syndromes due to a per-
ceived risk of recurrent bleeding, or due to residual
uncontrolled confounding.
The 5.6 % rate of major bleeding that we observed
might be lower than in other cohorts of critically ill
patients because we excluded patients from this throm-
boprophylaxis trial if, upon ICU admission, they had
major hemorrhage, trauma, platelet counts \75 9 10
9
/L
or abnormal coagulation testing, or were already receiv-
ing therapeutic anticoagulation. Our ndings do not
reect predictors or consequences of bleeding in trauma
patients, in patients with major bleeding on admission to
the ICU, or in patients who underwent orthopedic surgery
or neurosurgery. As with all regression analyses of
observational studies, these results are hypothesis-gener-
ating but do not prove causality. In examining the risk
factors for major bleeding and risk factors for death, we
did not analyze all possible confounders such as acute
kidney injury that did not require renal replacement
therapy and the specic variable of ICU-acquired sepsis
[29]; however, we included vasopressor/inotrope depen-
dency as a marker of hemodynamic instability common in
sepsis. The association between bleeding and surgical
procedures could have been underestimated since we
considered surgery as a potential risk factor only if it was
performed within 3 days of bleeding. Finally, identifying
risk factors for major bleeding depends on the prevalence
of these characteristics in the population studied; For
example, only eight (3.8 %) patients who had major
bleeding received concomitant acetylsalicylic acid and
clopidogrel therapy, which limits the ability to detect
whether this drug combination was safe.
Strengths of this study include the comprehensive
daily screening for bleeding by trained blinded research
coordinators. We used a clinically useful, ICU-specic,
validated bleeding tool with high interrater reliability [8]
and a rigorous blinded duplicate adjudication process for
all major bleeds after an initial calibration exercise that
demonstrated high agreement [9]. We examined inde-
pendent predictors of major bleeding using the Cox
model, which assesses the association between each risk
2141
factor and the hazard rate of bleeding over time, adjusts
for other factors in the model, and allows for censoring
due to death or discharge. We also used Cox regression
models to assess the association between major bleeding
and mortality to adjust for baseline characteristics and to
account for immortal time bias [30]. This is the largest
study to describe the independent association between
platelet count and bleeding in the ICU, with prior study
sample sizes ranging from 145 to 362 patients [31].
Enrollment of a heterogeneous group of patients from 67
centers in six countries enhances the generalizability of
our ndings.
Conclusions
In this multicenter study, we found that approximately
6 % of medicalsurgical critically ill patients receiving
heparin thromboprophylaxis developed major bleeding.
Independent predictors of major bleeding were renal
replacement therapy, surgical interventions, therapeutic
anticoagulation, prolonged aPTT, or lower platelet count
all in the preceding 3 days, and antiplatelet agents in the
preceding 7 days. There was no association between type
of heparin for thromboprophylaxis (unfractionated or low
molecular weight) and increased risk of bleeding. Patients
who had major bleeding compared with those who did not
had a signicantly increased rate of blood product trans-
fusion, increased duration of mechanical ventilation,
longer ICU and hospital lengths of stay, and increased
ICU and hospital mortality. Strategies to mitigate bleed-
ing risk factors such as use of narrow therapeutic
windows of anticoagulation, more careful assessment of
short-term riskbenet proles of anticoagulant and
antiplatelet drugs, and deferral of nonurgent surgical
interventions may be warranted.
Acknowledgments The authors are very grateful to the research
coordinators for their help with this trial. The trial was designed by
the PROTECT (Prophylaxis for ThromboEmbolism in Critical
Care Trial) Steering Committee, PROTECT investigators, and the
Canadian Critical Care Trials Group. PROTECT was supported by
the Canadian Critical Care Trials Group and the Australian and
New Zealand Intensive Care Society Clinical Trials Group. The list
of PROTECT collaborators is provided in the Electronic Supple-
mentary Material.
Funding sources F Lauzier is a recipient of a research career
award from the Fonds de la recherche du Quebec-Sante. DM
Arnold holds a New Investigator Award from the Canadian Insti-
tutes of Health Research in partnership with Hoffmann-LaRoche.
R Zarychanski receives salary support from CIHR as part of an
RCT Mentoring Award. R Fowler is a Clinician Scientist of the
Heart and Stroke Foundation of Ontario. M Crowther holds a
Career Investigator Award from the Heart and Stroke Foundation of
Ontario and the Leo Pharma Chair in Thromboembolism Research
at McMaster University and St Josephs Healthcare, Hamilton.
L McIntyre holds a New Investigator Award from the Canadian
Institute of Health Research/Canadian Blood Services. D Cook is a
Research Chair of the Canadian Institutes of Health Research. The
list of PROTECT collaborators is available in the Supplement
Digital Content. This study was funded by the Canadian Institute
Health Research (#MCT78568), Heart & Stroke Foundation of
Canada (#T6157, #T6950, #NA6186) and the Australian and New
Zealand College of Anesthetists Research Foundation (#07/23).
Conicts of interest RD Lopes received research grant from Bristol-
Myers Squibb and sat on advisory boards for Bristol-Myers Squibb, BI,
and Bayer. M Crowther sat on advisory boards for Leo Pharma, Pzer,
Bayer, Boehringer Ingelheim, Alexion, CSL Behring, and Artisan
Pharma. He prepared educational materials for Pzer, Octapharm, and
CSL Behring, and provided expert testimony for Bayer. M Crowthers
institution has received funding for research projects from Boehringer
Ingelheim, Octapharm, Pzer, and Leo Pharma.
References
1. Eikelboom JW, Mehta SR, Anand SS,
Xie C, Fox KAA, Yusuf S (2006)
Adverse impact of bleeding on
prognosis in patients with acute
coronary syndromes. Circulation
114:774782
2. van Hattum ES, Algra A, Lawson JA,
Eikelboom BC, Moll FL, Tangelder MJ
(2009) Bleeding increases the risk of
ischemic events in patients with
peripheral arterial disease. Circulation
120:15691576
3. Nevo S, Swan V, Enger C, Wojno KJ,
Bitton R, Shabooti M, Fuller AK, Jones
RJ, Braine HG, Vogelsang GB (1998)
Acute bleeding after bone marrow
transplantation (BMT)- incidence and
effect on survival. A quantitative
analysis in 1,402 patients. Blood
91:14691477
4. Eikelboom JW, Quinlan DJ, ODonnell
M (2009) Major bleeding, mortality,
and efcacy of fondaparinux in venous
thromboembolism prevention trials.
Circulation 120:20062011
5. Lauzier F, Arnold D, Rabbat C, Heels-
Ansdell D, Dodek P, Ashley B, Albert
M, Khwaja K, Ostermann M, Skrobik
Y, Fowler R, McIntyre L, Nates J,
Karachi T, Lopes R, Zytaruk N,
Crowther M, Cook D (2012) In
medicalsurgical ICU patients, major
bleeding is common but independent of
heparin prophylaxis. Crit Care 16:P435
6. Cook D, Meade M, Guyatt G, Walter
SD, Heels-Ansdell D, Geerts W,
Warkentin TE, Cooper DJ, Zytaruk N,
Vallance S, Berwanger O, Rocha M,
Qushmaq I, Crowther M (2011)
PROphylaxis for ThromboEmbolism in
Critical Care Trial protocol and analysis
plan. J Crit Care 26(223):e221e229
2142
7. PROTECT Investigators for the
Canadian Critical Care Trials Group
and the Australian and New Zealand
Intensive Care Society Clinical Trials
Group, Cook D D, Meade M, Guyatt G
G, Walter S, Heels-Ansdell D,
Warkentin TE, Zytaruk N, Crowther M,
Geerts W, Cooper DJ, Vallance S,
Qushmaq I, Rocha M, Berwanger O,
Vlahakis NE (2011) Dalteparin versus
unfractionated heparin in critically ill
patients. N Engl J Med 364:13051314
8. Arnold DM, Donahoe L, Clarke FJ,
Tkaczyk AJ, Heels-Ansdell D, Zytaruk
N, Cook R, Webert KE, McDonald E,
Cook DJ (2007) Bleeding during
critical illness: a prospective cohort
study using a new measurement tool.
Clin Invest Med 30:E93E102
9. Arnold D, Rabbat C, Lauzier F, Zytaruk
N, Heels-Ansdell D, Cook D (2009)
Adjudicating bleeding outcomes in a
large thrombroprophylaxis trial in
critical illness. Blood (ASH Annual
Meeting Abstracts) 114:A2471
10. Knaus WA, Draper EA, Wagner DP,
Zimmerman JE (1985) APACHE II: a
severity of disease classication system.
Crit Care Med 13:818829
11. Douketis JD, Spyropoulos AC, Spencer
FA, Mayr M, Jaffer AK, Eckman MH,
Dunn AS, Kunz R (2012) Perioperative
management of antithrombotic therapy:
antithrombotic therapy and prevention
of thrombosis, 9th ed: American
College of Chest Physicians Evidence-
Based Clinical Practice Guidelines.
Chest 141:e326Se350S
12. Douketis J, Cook D, Meade M, Guyatt
G, Geerts W, Skrobik Y, Albert M,
Granton J, Hebert P, Pagliarello G,
Marshall J, Fowler R, Freitag A, Rabbat
C, Anderson D, Zytaruk N, Heels-
Ansdell D, Crowther M, Group CCCT
(2008) Prophylaxis against deep vein
thrombosis in critically ill patients with
severe renal insufciency with the low-
molecular-weight heparin dalteparin: an
assessment of safety and
pharmacodynamics: the DIRECT study.
Arch Intern Med 168:18051812
13. Decousus H, Tapson VF, Bergmann JF,
Chong BH, Froehlich JB, Kakkar AK,
Merli GJ, Monreal M, Nakamura M,
Pavanello R, Pini M, Piovella F,
Spencer FA, Spyropoulos AC, Turpie
AG, Zotz RB, Fitzgerald G, Anderson
FA (2011) Factors at admission
associated with bleeding risk in medical
patients: ndings from the IMPROVE
investigators. Chest 139:6979
14. Di Minno G, Martinez J, McKean ML,
De La Rosa J, Burke JF, Murphy S
(1985) Platelet dysfunction in uremia.
Multifaceted defect partially corrected
by dialysis. Am J Med 79:552559
15. Fiaccadori E, Maggiore U, Clima B,
Melfa L, Rotelli C, Borghetti A (2001)
Incidence, risk factors, and prognosis of
gastrointestinal hemorrhage
complicating acute renal failure.
Kidney Int 59:15101519
16. Cook D, Heyland D, Grifth L, Cook R,
Marshall J, Pagliarello J (1999) Risk
factors for clinically important upper
gastrointestinal bleeding in patients
requiring mechanical ventilation.
Canadian Critical Care Trials Group.
Crit Care Med 27:28122817
17. Luo JC, Leu HB, Huang KW, Huang
CC, Hou MC, Lin HC, Lee FY, Lee SD
(2011) Incidence of bleeding from
gastroduodenal ulcers in patients with
end-stage renal disease receiving
hemodialysis. CMAJ 183:E1345
E1351
18. Anand SS, Yusuf S, Pogue J, Ginsberg
JS, Hirsh J, Investigators OtASfIS
(2003) Relationship of activated partial
thromboplastin time to coronary events
and bleeding in patients with acute
coronary syndromes who receive
heparin. Circulation 107:28842888
19. Anand SS, Bates S, Ginsberg JS, Levine
M, Buller H, Prins M, Haley S, Kearon
C, Hirsh J, Gent M (1999) Recurrent
venous thrombosis and heparin therapy:
an evaluation of the importance of early
activated partial thromboplastin times.
Arch Intern Med 159:20292032
20. Anand S, Ginsberg JS, Kearon C, Gent
M, Hirsh J (1996) The relation between
the activated partial thromboplastin
time response and recurrence in patients
with venous thrombosis treated with
continuous intravenous heparin. Arch
Intern Med 156:16771681
21. Ciavarella D, Reed RL, Counts RB,
Baron L, Pavlin E, Heimbach DM,
Carrico CJ (1987) Clotting factor levels
and the risk of diffuse microvascular
bleeding in the massively transfused
patient. Br J Haematol 67:365368
22. Chowdhury P, Saayman AG, Paulus U,
Findlay GP, Collins PW (2004)
Efcacy of standard dose and 30 ml/kg
fresh frozen plasma in correcting
laboratory parameters of haemostasis in
critically ill patients. Br J Haematol
125:6973
23. Strauss R, Wehler M, Mehler K,
Kreutzer D, Koebnick C, Hahn EG
(2002) Thrombocytopenia in patients in
the medical intensive care unit:
bleeding prevalence, transfusion
requirements, and outcome. Crit Care
Med 30:17651771
24. Shalansky SJ, Verma AK, Levine M,
Spinelli JJ, Dodek PM (2002) Risk
markers for thrombocytopenia in
critically ill patients: a prospective
analysis. Pharmacotherapy 22:803813
25. Cook D, Douketis J, Meade M, Guyatt
G, Zytaruk N, Granton J, Skrobik Y,
Albert M, Fowler R, Hebert P,
Pagliarello G, Friedrich J, Freitag A,
Karachi T, Rabbat C, Heels-Ansdell D,
Geerts W, Crowther M, Canadian
Critical Care Trials Group (2008)
Venous thromboembolism and bleeding
in critically ill patients with severe renal
insufciency receiving dalteparin
thromboprophylaxis: prevalence,
incidence and risk factors. Crit Care
12:R32
26. Lopes RD, Subherwal S, Holmes DN,
Thomas L, Wang TY, Rao SV, Magnus
Ohman E, Roe MT, Peterson ED,
Alexander KP (2012) The association
of in-hospital major bleeding with
short-, intermediate-, and long-term
mortality among older patients with
non-ST-segment elevation myocardial
infarction. Eur Heart J
33(16):20442053
27. Marik PE, Corwin HL (2008) Efcacy
of red blood cell transfusion in the
critically ill: a systematic review of the
literature. Crit Care Med 36:26672674
28. Gajic O, Rana R, Winters JL, Yilmaz
M, Mendez JL, Rickman OB, OByrne
MM, Evenson LK, Malinchoc M,
DeGoey SR, Afessa B, Hubmayr RD,
Moore SB (2007) Transfusion-related
acute lung injury in the critically ill:
prospective nested case-control study.
Am J Respir Crit Care Med
176:886891
29. Roman-Marchant O, Orellana-Jimenez
CE, De Backer D, Melot C, Vincent JL
(2004) Septic shock of early or late
onset: does it matter? Chest
126:173178
30. Shintani AK, Girard TD, Eden SK,
Arbogast PG, Moons KG, Ely EW
(2009) Immortal time bias in critical
care research: application of time-
varying Cox regression for
observational cohort studies. Crit Care
Med 37:29392945
31. Hui P, Cook DJ, Lim W, Fraser GA,
Arnold DM (2011) The frequency and
clinical signicance of
thrombocytopenia complicating critical
illness: a systematic review. Chest
139:271278
2143