This guideline provides evidence-based recommendations for the management of bleeding in major trauma patients. It has been updated from a previous version published in 2007 based on a systematic review of new evidence. Key changes include new recommendations on coagulation support, monitoring, and the use of local hemostatic measures, tourniquets, calcium, and desmopressin. The recommendations are graded based on the quality of supporting evidence. The guideline aims to improve outcomes for critically injured bleeding trauma patients through a multidisciplinary approach.
2. Rossaint et al. Critical Care 2010, 14:R52 Page 2 of 29
http://ccforum.com/content/14/2/R52
state, it may be appropriate to reassess some data from or ethical reasons. The formulation and grading of the
the past, and with time new research will doubtless lead recommendations presented here are therefore weighted
to a better understanding of the risks and benefits of to reflect both this reality and the current state-of-
different therapeutic approaches applied to this group of the-art.
patients.
In 2007, we published a European guideline for the Materials and methods
management of bleeding following major trauma that These recommendations were formulated and graded
included recommendations for specific interventions to according the Grading of Recommendations Assess-
identify and control bleeding sources using surgical, ment, Development and Evaluation (GRADE) hierarchy
physiological and pharmacological strategies [13]. The of evidence [14-16] summarised in Table 1. Comprehen-
guideline was developed by a multidisciplinary group of sive computer database literature searches were per-
European experts, including designated representatives formed using the indexed online databases MEDLINE/
from relevant professional societies, to guide the clini- PubMed and the Cochrane Library. Lists of cited litera-
cian in the early phases of treatment. Here we present ture within relevant articles were also screened. The pri-
an updated version of the guideline that incorporates a mary intention of the review was to identify prospective
renewed critical survey of the evidence published during randomised controlled trials (RCTs) and non-RCTs,
the intervening three years and a consideration of existing systematic reviews and guidelines. In the
changes in clinical practice that have taken place based absence of such evidence, case-control studies, observa-
on technologies that have become more widely available tional studies and case reports were considered.
and pharmacological agents that have entered or left the Boolean operators and Medical Subject Heading
market. Although the level of scientific evidence has (MeSH) thesaurus keywords were applied as a standar-
improved in some areas, other areas remain devoid of dised use of language to unify differences in terminology
high-level evidence, which may never exist for practical into single concepts. Appropriate MeSH headings and
Table 1 Grading of recommendations from Guyatt and colleagues [14]
Grade of Clarity of risk/benefit Quality of supporting evidence Implications
recommendation
1A
Strong Benefits clearly outweigh risk and RCTs without important limitations or Strong recommendation, can
recommendation, burdens, or vice versa overwhelming evidence from observational apply to most patients in most
high-quality evidence studies circumstances without reservation
1B
Strong Benefits clearly outweigh risk and RCTs with important limitations (inconsistent Strong recommendation, can
recommendation, burdens, or vice versa results, methodological flaws, indirect or apply to most patients in most
moderate-quality imprecise) or exceptionally strong evidence from circumstances without reservation
evidence observational studies
1C
Strong Benefits clearly outweigh risk and Observational studies or case series Strong recommendation but may
recommendation, burdens, or vice versa change when higher quality
low-quality or very evidence becomes available
low-quality evidence
2A
Weak Benefits closely balanced with risks RCTs without important limitations or Weak recommendation, best
recommendation, and burden overwhelming evidence from observational action may differ depending on
high-quality evidence studies circumstances or patient or
societal values
2B
Weak Benefits closely balanced with risks RCTs with important limitations (inconsistent Weak recommendation, best
recommendation, and burden results, methodological flaws, indirect or action may differ depending on
moderate-quality imprecise) or exceptionally strong evidence from circumstances or patient or
evidence observational studies societal values
2C
Weak Uncertainty in the estimates of Observational studies or case series Very weak recommendation; other
recommendation, benefits, risks, and burden; benefits, alternatives may be equally
Low-quality or very risk and burden may be closely reasonable
low-quality evidence balanced
Reprinted with permission from the American College of Chest Physicians.
RCTs, randomised controlled trials.
3. Rossaint et al. Critical Care 2010, 14:R52 Page 3 of 29
http://ccforum.com/content/14/2/R52
subheadings for each question were selected and modi- updated document was approved by the endorsing orga-
fied based on search results. The scientific questions nisations between October 2009 and January 2010. An
posed that led to each recommendation and the MeSH updated version of the guideline is anticipated in due
headings applied to each search are listed in Additional time.
file 1. Searches were limited to English language In the GRADE system for assessing each recommenda-
abstracts and human studies, and gender and age were tion, the letter attached to the grade of recommendation
not limited. The time period was limited to the past reflects the degree of literature support for the recom-
three years for questions addressed in the 2007 version mendation, whereas the number indicates the level of
of the guideline, but no time-period limits were imposed support for the recommendation assigned by the com-
on new searches. Original publications were evaluated mittee of experts. Recommendations are grouped by
for abstracts that were deemed relevant. Original publi- category and somewhat chronologically in the treatment
cations were graded according to the levels of evidence decision-making process, but not by priority or hierarchy.
developed by the Oxford Centre for Evidence-Based
Medicine (Oxford, Oxfordshire, UK) [17]. Results
The selection of the scientific enquiries to be I. Initial resuscitation and prevention of further bleeding
addressed in the guideline, screening and grading of the Minimal elapsed time
literature to be included and formulation of specific Recommendation 1 We recommend that the time
recommendations were performed by members of the elapsed between injury and operation be minimised for
Task Force for Advanced Bleeding Care in Trauma, a patients in need of urgent surgical bleeding control
multidisciplinary, pan-European group of experts with (Grade 1A).
specialties in surgery, anaesthesia, emergency medicine, Rationale Trauma patients in need of emergency sur-
intensive care medicine and haematology. The core gery for ongoing haemorrhage have increased survival if
group was formed in 2004 to produce educational mate- the elapsed time between the traumatic injury and
rial on the care of the bleeding trauma patient [18], on admission to the operating theatre is minimised. More
which an update (in 2006) and subsequent review article than 50% of all trauma patients with a fatal outcome die
were based [19]. The task force consisted of the core within 24 hours of injury [2]. Despite a lack of evidence
group, additional experts in haematology and guideline from prospective RCTs, well-designed retrospective stu-
development, and representatives of relevant European dies provide evidence for early surgical intervention in
professional societies, including the European Society of patients with traumatic haemorrhagic shock [21-23].
Anaesthesiology, the European Society of Intensive Care In addition, studies that analyse trauma systems indir-
Medicine, the European Shock Society, the European ectly emphasise the importance of minimising the time
Society of Trauma and Emergency Surgery and the Eur- between admission and surgical bleeding control in
opean Society for Emergency Medicine. The European patients with traumatic haemorrhagic shock [24,25]. At
Hematology Association declined the invitation to desig- present, the evidence base for the impact of the imple-
nate a representative to join the task force. mentation of the Advanced Trauma Life Support
As part of the guideline development process that led (ATLS) protocol on patient outcome is very poor,
to the 2007 guideline, task force members participated because the available literature focuses primarily on the
in a workshop on the critical appraisal of medical litera- effectiveness of ATLS as an educational tool [26]. Future
ture. The nominal group process for the updated guide- studies are needed to define the impact of the ATLS
line included several remote (telephone and web-based) program within trauma systems at the hospital and
meetings and one face-to-face meeting supplemented by health system level in terms of controlled before-and-
several Delphi rounds [20]. The guideline development after implementation designed to assess post-injury
group participated in a web conference in March 2009 mortality as the primary outcome parameter.
to define the scientific questions to be addressed in the Tourniquet use
guideline. Selection, screening and grading of the litera- Recommendation 2 We recommend adjunct tourniquet
ture and formulation of recommendations were accom- use to stop life-threatening bleeding from open extre-
plished in subcommittee groups consisting of at least mity injuries in the pre-surgical setting (Grade 1C).
three members via electronic or telephone communica- Rationale Much discussion has been generated recently
tion. After distribution of the recommendations to the regarding the use of tourniquets for acute external hae-
entire group, a face-to-face meeting of the task force morrhage control. Pressure bandages rather than tourni-
was held in June 2009 with the aim of reaching a con- quets should be applied in the case of minor bleeding
sensus on the draft recommendations from each sub- from open wounds in extremity injuries. When uncon-
committee. After final refinement of the rationale for trolled arterial bleeding occurs from mangled extremity
each recommendation and the complete manuscript, the injuries, including penetrating or blast injuries or
4. Rossaint et al. Critical Care 2010, 14:R52 Page 4 of 29
http://ccforum.com/content/14/2/R52
traumatic amputations, a tourniquet represents a simple systems, and the patient’s physiological presentation and
and efficient method to acutely control haemorrhage response to resuscitation should further guide the deci-
[27-31]. Several publications from military settings sion to initiate early surgical bleeding control as out-
report the effectiveness of tourniquets in this specific lined in the ATLS protocol [34-37]. Table 2 summarises
setting [27-30]. A study of volunteers showed that any estimated blood loss based on intitial presentation.
tourniquet device presently on the market works effi- Table 3 characterises the three types of response to
ciently [31]. The study also showed that ‘pressure point initial fluid resuscitation, whereby the transient respon-
control’ was ineffective because collateral circulation ders and the non-responders are candidates for immedi-
was observed within seconds. Tourniquet-induced pain ate surgical bleeding control.
was not an important consideration. Ventilation
Tourniquets should be left in place until surgical con- Recommendation 4 We recommend initial normoventi-
trol of bleeding is achieved [28,30]; however, this time- lation of trauma patients if there are no signs of immi-
span should be kept as short as possible. Improper or nent cerebral herniation (Grade 1C).
prolonged placement of a tourniquet can lead to com- Rationale Ventilation can affect the outcome of severe
plications such as nerve paralysis and limb ischaemia trauma patients. There is a tendency for rescue person-
[32]. Some publications suggest a maximum time of nel to hyperventilate patients during resuscitation
application of two hours [32]. Reports from military set- [38,39], and hyperventilated trauma patients appear to
tings report cases in which tourniquets have remained have increased mortality when compared with non-
in place for up to six hours with survival of the extre- hyperventilated patients [39].
mity [28]. A high percentage of severely injured patients with
ongoing bleeding have traumatic brain injury (TBI).
II. Diagnosis and monitoring of bleeding Relevant experimental and clinical data have shown that
Initial assessment routine hyperventilation is an important contributor to
Recommendation 3 We recommend that the physician adverse outcomes in patients with head injuries; how-
clinically assess the extent of traumatic haemorrhage ever, the effect of hyperventilation on outcome in
using a combination of mechanism of injury, patient patients with severe trauma but no TBI is still a matter
physiology, anatomical injury pattern and the patient’s of debate. A low partial pressure of arterial carbon diox-
response to initial resuscitation (Grade 1C). ide on admission to the emergency room is associated
Rationale The mechanism of injury represents an with a worse outcome in trauma patients with TBI
important screening tool to identify patients at risk for [40-43].
significant traumatic haemorrhage. For example, the There are several potential mechanisms for the
American College of Surgeons defined a threshold of adverse effects of hyperventilation and hypocapnia,
6 m (20 ft) as a ‘critical falling height’ associated with including increased vasoconstriction with decreased cer-
major injuries [33]. Further critical mechanisms include ebral blood flow and impaired tissue perfusion. In the
blunt versus penetrating trauma, high-energy decelera- setting of absolute or relative hypovolaemia, an excessive
tion impact, low-velocity versus high-velocity gunshot ventilation rate of positive-pressure ventilation may
injuries, etc. The mechanism of injury in conjunction further compromise venous return and produce hypo-
with injury severity, as defined by trauma scoring tension and even cardiovascular collapse [41,42]. It has
Table 2 American College of Surgeons Advanced Trauma Life Support (ATLS) classification of blood loss based on
initial patient presentation
Class I Class II Class III Class IV
Blood loss* (ml) Up to750 750-1500 1500-2000 >2000
Blood loss (% blood volume) Up to 15% 15%-30% 30%-40% >40%
Pulse rate <100 100-120 120-140 >140
Blood pressure Normal Normal Decreased Decreased
Pulse pressure (mmHg) Normal or increased Decreased Decreased Decreased
Respiratory rate 14-20 20-30 30-40 >35
Urine output (ml/h) >30 20-30 5-15 Negligible
Central nervous system/mental status Slightly anxious Mildly anxious Anxious, confused Confused, lethargic
Fluid replacement Crystalloid Crystalloid Crystalloid and blood Crystalloid and blood
Table reprinted with permission from the American College of Surgeons [37].
*for a 70 kg male.
5. Rossaint et al. Critical Care 2010, 14:R52 Page 5 of 29
http://ccforum.com/content/14/2/R52
Table 3 American College of Surgeons Advanced Trauma Life Support (ATLS) responses to initial fluid resuscitation*
Rapid response Transient response Minimal or no response
Vital signs Return to normal Transient improvement, recurrence Remain abnormal
of decreased blood pressure and
increased heart rate
Estimated blood loss Minimal (10%-20%) Moderate and ongoing (20%-40%) Severe (>40%)
Need for more crystalloid Low High High
Need for blood Low Moderate to high Immediate
Blood preparation Type and crossmatch Type-specific Emergency blood release
Need for operative intervention Possibly Likely Highly likely
Early presence of surgeon Yes Yes Yes
* 2000 ml of isotonic solution in adults; 20 ml/kg bolus of Ringer’s lactate in children.
Table reprinted with permission from the American College of Surgeons [37].
also been shown that cerebral tissue lactic acidosis combined with signs of severe hypovolaemic shock spe-
occurs almost immediately after induction of hypocapnia cifically require early surgical bleeding control. This
in children and adults with TBI and haemorrhagic shock observation is also true but to a lesser extent for
[44]. In addition, even a modest level of hypocapnia abdominal stab wounds [50]. Data on injuries caused by
(<27 mmHg) may result in neuronal depolarisation with penetrating metal fragments from explosives or gunshot
glutamate release and extension of the primary injury wounds in the Vietnam War confirm the need for early
via apoptosis [45]. surgical control when patients present in shock [51]. In
Ventilation with low tidal volume is recommended in blunt trauma, the mechanism of injury can determine to
patients with acute lung injury. In patients with normal a certain extent whether the patient in haemorrhagic
lung function, the evidence is scarce, but some obser- shock will be a candidate for surgical bleeding control.
vational studies show that the use of a high tidal Only a few studies address the relation between the
volume is an important risk factor for the development mechanism of injury and the risk of bleeding, and none
of lung injury [46,47]. The injurious effect of high tidal of these publications is a randomised prospective trial of
volume may be initiated very early. Randomised studies high evidence [52]. We have found no objective data
demonstrate that short-time ventilation (<five hours) describing the relation between the risk of bleeding and
with high tidal volume (12 ml/kg) without positive the mechanism of injury of skeletal fractures in general
end-expiratory pressure (PEEP) may promote pulmon- or of long-bone fractures in particular.
ary inflammation and alveolar coagulation in patients Traffic accidents are the leading cause of pelvic injury.
with normal lung function [48]. Although more studies Motor vehicle crashes cause approximately 60% of pelvic
are needed, the early use of protective ventilation with fractures followed by falls from great heights (23%).
low tidal volume and moderate PEEP is recommended, Most of the remainder result from motorbike collisions
particularly in bleeding trauma patients at risk of acute and vehicle-pedestrian accidents [53,54]. There is a cor-
lung injury. relation between ‘unstable’ pelvic fractures and intra-
Immediate intervention abdominal injuries [53,55]. An association between
Recommendation 5 We recommend that patients pre- major pelvic fractures and severe head injuries, conco-
senting with haemorrhagic shock and an identified mitant thoracic, abdominal, urological and skeletal inju-
source of bleeding undergo an immediate bleeding con- ries is also well described [53]. High-energy injuries
trol procedure unless initial resuscitation measures are produce greater damage to both the pelvis and organs.
successful (Grade 1B). Patients with high-energy injuries require more transfu-
Rationale The source of bleeding may be immediately sion units, and more than 75% have associated head,
obvious, and penetrating injuries are more likely to thorax, abdominal or genitourinary injuries [56]. It is
require surgical bleeding control. In a retrospective well documented that ‘unstable’ pelvic fractures are
study of 106 abdominal vascular injuries, all 41 patients associated with massive haemorrhage [55,57], and hae-
arriving in shock following gunshot wounds were candi- morrhage is the leading cause of death in patients with
dates for rapid transfer to the operating theatre for sur- major pelvic fractures.
gical bleeding control [49]. A similar observation in a Further investigation
study of 271 patients undergoing immediate laparotomy Recommendation 6 We recommend that patients pre-
for gunshot wounds indicates that these wounds senting with haemorrhagic shock and an unidentified
6. Rossaint et al. Critical Care 2010, 14:R52 Page 6 of 29
http://ccforum.com/content/14/2/R52
source of bleeding undergo immediate further investiga- Huber-Wagner and colleagues [62] also showed the ben-
tion (Grade 1B). efit of integration of the whole-body CT into early
Rationale A patient in haemorrhagic shock with an uni- trauma care. CT diagnosis significantly increases the
dentified source of bleeding should undergo immediate probability of survival in patients with polytrauma.
further assessment of the chest, abdominal cavity and Whole-body CT as a standard diagnostic tool during the
pelvic ring, which represent the major sources of acute earliest resuscitation phase for polytraumatised patients
blood loss in trauma. Aside from a clinical examination, provides the added benefit of identifying head and chest
X-rays of chest and pelvis in conjunction with focused injuries and other bleeding sources in patients with mul-
abdominal sonography for trauma (FAST) [58] or diag- tiple injuries.
nostic peritoneal lavage (DPL) [59] are recommended Some authors have shown the benefit of contrast
diagnostic modalities during the primary survey medium enhanced CT scanning. Anderson and collea-
[37,60,61]. In selected centres, readily available com- gues [82,83] found high accuracy in the evaluation of
puted tomography (CT) scanners [62] may replace con- splenic injuries resulting from trauma after administra-
ventional radiographic imaging techniques during the tion of intravenous contrast material. Delayed phase CT
primary survey. may be used to detect active bleeding in solid organs.
Imaging Fang and colleagues [84] demonstrated that the pooling
Recommendation 7 We recommend early imaging of contrast material within the peritoneal cavity in blunt
(FAST or CT) for the detection of free fluid in patients liver injuries indicates active and massive bleeding.
with suspected torso trauma (Grade 1B). Patients with this finding showed rapid deterioration of
Recommendation 8 We recommend that patients with haemodynamic status and most of them required emer-
significant free intra-abdominal fluid and haemodynamic gent surgery. Intraparenchymal pooling of contrast
instability undergo urgent intervention (Grade 1A). material with an unruptured liver capsule often indicates
Recommendation 9 We recommend further assessment a self-limited haemorrhage, and these patients respond
using CT for haemodynamically stable patients who are well to non-operative treatment.
either suspected of having torso bleeding or have a Compared with MSCT, all traditional techniques of
high-risk mechanism of injury (Grade 1B). diagnostic and imaging evaluation are associated with
Rationale Blunt abdominal trauma represents a major some limitations. The diagnostic accuracy, safety and
diagnostic challenge and an important source of internal effectiveness of immediate MSCT are dependent on
bleeding. FAST has been established as a rapid and sophisticated pre-hospital treatment by trained and
non-invasive diagnostic approach for the detection of experienced emergency personnel and short transporta-
intra-abdominal free fluid in the emergency room tion times [85,86]. If an MSCT is not available in the
[63-65]. Large prospective observational studies deter- emergency room, the realisation of CT scanning implies
mined a high specificity and accuracy but low sensitivity transportation of the patient to the CT room, and there-
of initial FAST examination for detecting intra-abdom- fore the clinician must evaluate the implications and
inal injuries in adults and children [66-72]. Liu and col- potential risks and benefits of the procedure. During
leagues [73] found a high sensitivity, specificity and transport, all vital signs should be closely monitored and
accuracy of initial FAST examination for the detection resuscitation measures continued. For those patients in
of haemoperitoneum. Although CT scans and DPL were whom haemodynamic stability is questionable, imaging
shown to be more sensitive than sonography for the techniques such as ultrasound and chest and pelvic
detection of haemoperitoneum, these diagnostic modal- radiography may be useful. Peritoneal lavage is rarely
ities are more time-consuming (CT and DPL) and inva- indicated if ultrasound or CT is available [87]. Transfer
sive (DPL) [73]. times to and from all forms of diagnostic imaging need
The role of CT scanning of acute trauma patients is to be considered carefully in any patient who is haemo-
well documented [74-81], and in recent years imaging dynamically unstable. In addition to the initial clinical
for trauma patients has migrated towards multi-slice CT assessment, near patient testing results, including full
(MSCT). The integration of modern MSCT scanners in blood count, haematocrit (Hct), blood gases and lactate,
the emergency room area allows the immediate assess- should be readily available under ideal circumstances.
ment of trauma victims following admission [76,77]. Hypotensive patients (systolic blood pressure below
Using modern MSCT scanners, total whole-body scan- 90 mmHg) presenting with free intra-abdominal fluid
ning time may be reduced to less than 30 seconds. In a according to FAST or CT are potential candidates for
retrospective study comparing 370 patients in two early surgery if they cannot be stabilised by initiated
groups, Weninger and colleagues [77] showed that faster fluid resuscitation [88-90]. A retrospective study by
diagnosis using MSCT led to shorter emergency room Rozycki and colleagues [91] of 1540 patients (1227
and operating room time and shorter ICU stays [77]. blunt, 313 penetrating trauma) assessed with FAST as
7. Rossaint et al. Critical Care 2010, 14:R52 Page 7 of 29
http://ccforum.com/content/14/2/R52
an early diagnostic tool showed that the ultrasound This study showed that changes in lactate concentra-
examination had a sensitivity and specificity close to tions provide an early and objective evaluation of a
100% when the patients were hypotensive. patient’s response to therapy and suggested that
A number of patients who present with free intra- repeated lactate determinations represent a reliable
abdominal fluid according to FAST can safely undergo prognostic index for patients with circulatory shock
further investigation with MSCT. Under normal circum- [103]. Abramson and colleagues [104] performed a pro-
stances, adult patients need to be haemodynamically spective observational study in patients with multiple
stable when MSCT is performed outside of the emer- trauma to evaluate the correlation between lactate clear-
gency room [91]. Haemodynamically stable patients with ance and survival. All patients in whom lactate levels
a high risk mechanism of injury, such as high-energy returned to the normal range (≤2 mmol/l) within
trauma or even low-energy injuries in the elderly popu- 24 hours survived. Survival decreased to 77.8% if nor-
lation, should be scanned after FAST for additional inju- malisation occurred within 48 hours and to 13.6% in
ries using MSCT. As CT scanners are integrated in those patients in whom lactate levels were elevated
resuscitation units, whole-body CT diagnosis may above 2 mmol/l for more than 48 hours [104]. These
replace FAST as a diagnostic method. findings were confirmed in a study by Manikis and col-
Haematocrit leagues [105] who showed that the initial lactate levels
Recommendation 10 We do not recommend the use of were higher in non-survivors after major trauma, and
single Hct measurements as an isolated laboratory mar- that the prolonged time for normalisation of lactate
ker for bleeding (Grade 1B). levels of more than 24 hours was associated with the
Rationale Hct assays are part of the basic diagnostic development of post-traumatic organ failure [105].
work up for trauma patients. The diagnostic value of Similar to the predictive value of lactate levels, the
the Hct for detecting trauma patients with severe injury initial base deficit has been established as a potent inde-
and occult bleeding sources has been a topic of debate pendent predictor of mortality in patients with trau-
in the past decade [92-94]. A major limit of the diagnos- matic hemorrhagic shock [106]. Davis and colleagues
tic value of Hct is the confounding influence of resusci- [107] stratified the extent of base deficit into three cate-
tative measures on the Hct due to administration of gories, mild (-3 to -5 mEq/l), moderate (-6 to -9 mEq/l)
intravenous fluids and red cell concentrates [94-97]. and severe (<-10 mEq/l), and established a significant
A retrospective study of 524 trauma patients determined correlation between the admission base deficit and
a low sensitivity (0.5) of the initial Hct on admission for transfusion requirements within the first 24 hours and
detecting those patients with traumatic haemorrhage the risk of post-traumatic organ failure or death [107].
requiring surgical intervention [94]. Two prospective The same group of authors showed that the base deficit
observational diagnostic studies determined the sensitiv- is a better prognostic marker of death than the pH in
ity of serial Hct measurements for detecting patients arterial blood gas analyses [108]. Furthermore, the base
with severe injury [92,93]. Decreasing serial Hct mea- deficit was shown to represent a highly sensitive marker
surements may reflect continued bleeding, but the for the extent of post-traumatic shock and mortality,
patient with significant bleeding may maintain his or both in adult and paediatric patients [109,110].
her serial Hct. In contrast to the data on lactate levels in haemorrha-
Serum lactate and base deficit gic shock, reliable large-scale prospective studies on the
Recommendation 11 We recommend both serum lac- correlation between base deficit and outcome are still
tate and base deficit measurements as sensitive tests to lacking. Although both the base deficit and serum lac-
estimate and monitor the extent of bleeding and shock tate levels are well correlated with shock and resuscita-
(Grade 1B). tion, these two parameters do not strictly correlate with
Rationale Serum lactate has been used as a diagnostic each other in severely injured patients [111]. Therefore,
parameter and prognostic marker of haemorrhagic the independent assessment of both parameters is
shock since the 1960s [98]. The amount of lactate pro- recommended for the evaluation of shock in trauma
duced by anaerobic glycolysis is an indirect marker of patients [99,101,111,112]. Composite scores that assess
oxygen debt, tissue hypoperfusion and the severity of the likelihood of massive transfusion and include base
haemorrhagic shock [99-102]. Similarly, base deficit deficit and other clinical parameters have been devel-
values derived from arterial blood gas analysis provide oped but require further validation [112,113]. Callaway
an indirect estimation of global tissue acidosis due to and colleagues [114] performed a seven-year retrospec-
impaired perfusion [99,101]. tive analysis of a prospective trauma registry from a
Vincent and colleagues [103] showed the value of level I trauma centre to determine predictors of mortal-
serial lactate measurements for predicting survival in a ity in elderly patients 65 years or older who sustained
prospective study in patients with circulatory shock. blunt trauma and presented with a normal initial
8. Rossaint et al. Critical Care 2010, 14:R52 Page 8 of 29
http://ccforum.com/content/14/2/R52
systolic blood pressure (≥90 mmHg). The odds ratio for outcomes. There is insufficient evidence at present to
death was increased more than four-fold in those support the utility of thrombelastometry in the detection
patients who had either elevated serum lactate levels of post-traumatic coagulopathy. More research is
above 4 mmol/l or a base deficit below -6 mEq/l, required in this area, and in the meantime physicians
compared with patients with normal lactate levels should make their own judgement when developing
(<2.5 mmol/l) or a base excess (>0 mEq/l). Paladino and local policies.
colleagues [115] assessed the prognostic value of a com- It is theoretically possible that the pattern of change in
bination of abnormal vital signs (heart rate >100 beats/ measures of coagulation such as D-dimers may help to
min or a systolic blood pressure <90 mmHg) in con- identify patients with ongoing bleeding. However, there
junction with serum lactate and base deficit for identify- are no publications relevant to this question, so tradi-
ing trauma patients with major injuries, using cut-off tional methods of detection for ongoing bleeding, such
values for lactate at more than 2.2 mmol/l and base def- as serial clinical evaluation of radiology (ultrasound, CT
icit at less than -2.0 mEq/l, respectively. The authors or angiography) should be used.
found that the addition of the metabolic parameters to
the vital signs increased the sensitivity for identifying III. Rapid control of bleeding
major injury from 40.9% to 76.4%, implying that the Pelvic ring closure and stabilisation
addition of lactate and base deficit to triage vital signs Recommendation 13 We recommend that patients with
increases the ability to distinguish major from minor pelvic ring disruption in haemorrhagic shock undergo
injury. immediate pelvic ring closure and stabilisation (Grade
Coagulation monitoring 1B).
Recommendation 12 We recommend that routine prac- Packing, embolisation and surgery
tice to detect post-traumatic coagulopathy include the Recommendation 14 We recommend that patients with
measurement of international normalised ratio (INR), ongoing haemodynamic instability despite adequate pel-
activated partial thromboplastin time (APTT), fibrinogen vic ring stabilisation receive early preperitoneal packing,
and platelets. INR and APTT alone should not be used angiographic embolisation and/or surgical bleeding con-
to guide haemostatic therapy (Grade 1C). We suggest trol (Grade 1B).
that thrombelastometry also be performed to assist in Rationale The mortality rate of patients with severe
characterising the coagulopathy and in guiding haemo- pelvic ring disruptions and haemodynamic instability
static therapy (Grade 2C). remains unacceptably high [119-122]. The early detec-
Rationale Little evidence supports a recommendation tion of these injuries and initial efforts to reduce disrup-
for the best haemostatic monitoring tool(s). Standard tion and stabilise the pelvis as well as containing
monitoring comprises INR, APTT, platelets and fibrino- bleeding is therefore crucial. Markers of pelvic haemor-
gen, although there is little direct evidence for the effi- rhage include anterior-posterior and vertical shear
cacy of these measures. Increasing emphasis focuses on deformations, CT ‘blush’ (active arterial extravasation),
the importance of fibrinogen and platelet measurements. bladder compression pressure, pelvic haematoma
It is often assumed that the conventional coagulation volumes of more than 500 ml evident by CT and
screens (INR and APTT) monitor coagulation; however, ongoing haemodynamic instability despite adequate frac-
these tests monitor only the initiation phase of blood ture stabilisation [123-125].
coagulation and represent only the first 4% of thrombin The initial therapy of pelvic fractures includes control
production [116]. It is therefore possible that the con- of venous and/or cancellous bone bleeding by pelvic clo-
ventional coagulation screen appears normal, while the sure. Some institutions use primarily external fixators to
overall state of blood coagulation is abnormal. There- control haemorrhage from pelvic fractures [124,125] but
fore, a more complete monitoring of blood coagulation pelvic closure may also be achieved using a bed sheet,
and fibrinolysis, such as thrombelastometry, may facili- pelvic binder or a pelvic C-clamp [126-128]. In addition
tate more accurate targeting of therapy. Case series to the pelvic closure, fracture stabilisation and the tam-
using thrombelastometry to assess trauma patients have ponade effect of the haematoma, pre, extra or retroperi-
been published. One study applied thrombelastometry toneal packing will reduce or stop the venous bleeding
to 23 patients, but without a comparative standard [122,129-131]. Preperitoneal packing decreases the need
[117]. Another study found a poor correlation between for pelvic embolisation and may be performed simulta-
thrombelastometry and conventional coagulation para- neously or soon after initial pelvic stabilisation
meters [10]. Johansson [118] implemented a haemostatic [122,129,131]. The technique can be combined with a
resuscitation regime (early platelets and fresh frozen consecutive laparotomy if deemed necessary [122,129].
plasma (FFP)) guided using thrombelastometry in This may decrease the high mortality rate observed in
a before-and-after study which showed improved patients with major pelvic injuries who underwent
9. Rossaint et al. Critical Care 2010, 14:R52 Page 9 of 29
http://ccforum.com/content/14/2/R52
laparotomy as the primary intervention. As a conse- injury, a need for time-consuming procedures or conco-
quence, it was recommended that non-therapeutic lapar- mitant major injury outside the abdomen (Grade 1C).
otomy should be avoided [132]. Rationale The severely injured patient arriving to the
Angiography and embolisation is currently accepted as hospital with continuous bleeding or deep haemorrhagic
a highly effective means with which to control arterial shock generally has a poor chance of survival unless
bleeding that cannot be controlled by fracture stabilisa- early control of bleeding, proper resuscitation and blood
tion [122-126,131-140]. The presence of sacroiliac joint transfusion are achieved. This is particularly true for
disruption, female gender and duration of hypotension patients who present with uncontrolled bleeding due to
can reliably predict patients who would benefit from the multiple penetrating injuries or patients with multiple
procedure [138]. Controversy exists about the indica- injuries and unstable pelvic fractures with ongoing
tions and optimal timing of angiography in haemodyna- bleeding from fracture sites and retroperitoneal vessels.
mically unstable patients [131]. Institutional differences The common denominator in these patients is the
in the capacity to perform timely angiography and exhaustion of physiological reserves with resulting pro-
embolisation may explain the different treatment algo- found acidosis, hypothermia and coagulopathy, also
rithms suggested by many authors [119-122,125,129, known as the ‘bloody vicious cycle’. In 1983, Stone and
131,132,140]. Nevertheless, the general consensus is that colleagues described the techniques of abbreviated lapar-
a multidisciplinary approach to these severe injuries is otomy, packing to control haemorrhage and of deferred
required. definitive surgical repair until coagulation has been
Early bleeding control established [156]. Since then, a number of authors have
Recommendation 15 We recommend that early bleed- described the beneficial results of this concept, now
ing control of the abdomen be achieved using packing, called ‘damage control’ [50,54,121,134,151,156-158].
direct surgical bleeding control and the use of local hae- Damage control surgery of the abdomen consists of
mostatic procedures. In the exsanguinating patient, aor- three components: the first component is an abbreviated
tic cross-clamping may be employed as an adjunct resuscitative laparotomy for control of bleeding, the res-
(Grade 1C). titution of blood flow where necessary and the control
Rationale Abdominal resuscitative packing is an early of contamination. This should be achieved as rapidly as
part of the post-traumatic laparotomy to identify major possible without spending unnecessary time on tradi-
injuries and sources of haemorrhage [141,142]. If bleed- tional organ repairs that can be deferred to a later
ing cannot be controlled using packing and conventional phase. The abdomen is packed and temporary abdom-
surgical techniques when the patient is in extremis or inal closure is performed. The second component is
when proximal vascular control is deemed necessary intensive care treatment, focused on core re-warming,
before opening the abdomen, aortic cross clamping may correction of the acid-base imbalance and coagulopathy
be employed as an adjunct to reduce bleeding and redis- as well as optimising the ventilation and the haemody-
tribute blood flow to the heart and brain [143-145]. namic status. The third component is the definitive sur-
When blood losses are important, when surgical mea- gical repair that is performed only when target
sures are unsuccessful and/or when the patient is cold, parameters have been achieved [159-162]. Although the
acidotic and coagulopathic, definitive packing may also concept of ‘damage control’ intuitively makes sense, no
be the first surgical step within the concept of damage RCTs exist to support it. Retrospective studies support
control [146-155]. Packing aims to compress liver rup- the concept showing reduced morbidity and mortality
tures or exert direct pressure on the sources of bleeding rates in selective populations [50,151,157,161].
[141,142,146-150,152-154]. The definitive packing of the The same ‘damage control’ principles have been
abdomen may allow further attempts to achieve total applied to orthopaedic injuries in severely injured
haemostasis through angiography and/or correction of patients [134,163-166]. Scalea was the first to coin the
coagulopathy [155]. The removal of packs should prefer- term ‘damage control orthopaedics’ [166]. Relevant frac-
ably be performed only after 48 hours to lower the risk tures are primarily stabilised with external fixators
of rebleeding [152,153]. rather than primary definitive osteosynthesis [134,163].
Damage control surgery The less traumatic and shorter duration of the surgical
Recommendation 16 We recommend that damage con- procedure aims to reduce the secondary trauma load.
trol surgery be employed in the severely injured patient Definitive osteosynthesis surgery can be performed after
presenting with deep haemorrhagic shock, signs of 4 to 14 days when the patient has recovered sufficiently.
ongoing bleeding and coagulopathy. Additional factors Retrospective clinical studies and prospective cohort stu-
that should trigger a damage control approach are dies seem to support the concept of damage control
hypothermia, acidosis, inaccessible major anatomical [134,163-165]. The only available randomised study
10. Rossaint et al. Critical Care 2010, 14:R52 Page 10 of 29
http://ccforum.com/content/14/2/R52
shows an advantage for this strategy in ‘borderline’ and diatoms and microporous polysaccharide haemo-
patients [164]. spheres produced from potato starch. The mechanism of
Local haemostatic measures action is complex and depends on the purity or combina-
Recommendation 17 We recommend the use of topical tion with other substances such as cellulose or fibrin.
haemostatic agents in combination with other surgical A number of different products are currently available
measures or with packing for venous or moderate arterial and have been shown to be efficient for external use.
bleeding associated with parenchymal injuries (Grade 1B). An observational study showed that haemorrhage control
Rationale A wide range of local haemostatic agents are was achieved using an N-acetylglucosamine-based ban-
currently available for use as adjuncts to traditional dage applied to 10 patients with severe hepatic and
surgical techniques to obtain haemorrhage control. abdominal injuries, acidosis and clinical coagulopathy
These topical agents can be particularly useful when [180].
access to the bleeding area is difficult. Local haemo- vi) The inorganic haemostatics based on minerals such
static agents include collagen, gelatin or cellulose- as zeolite or smectite have been used and studied mainly
based products, fibrin and synthetic glues or adhesives on external bleeding [167,168].
that can be used for both external and internal bleed-
ing while polysaccharide-based and inorganic haemo- IV. Tissue oxygenation, fluid and hypothermia
statics are still mainly used and approved for external Volume replacement
bleeding. The use of topical haemostatic agents should Recommendation 18 We recommend a target systolic
consider several factors such as the type of surgical blood pressure of 80 to 100 mmHg until major bleeding
procedure, cost, severity of bleeding, coagulation status has been stopped in the initial phase following trauma
and each agent’s specific characteristics. Some of these without brain injury (Grade 1C).
agents should be avoided when autotransfusion is Rationale In order to maintain tissue oxygenation, tra-
used and several other contraindications need to be ditional treatment of trauma patients uses early and
considered [167,168]. The capacity of each agent to aggressive fluid administration to restore blood volume.
control bleeding was initially studied in animals but This approach may, however, increase the hydrostatic
increasing experience from humans is now available pressure on the wound, cause a dislodgement of blood
[167-180]. clots, a dilution of coagulation factors and undesirable
The different types of local haemostatics are briefly cooling of the patient. The concept of low-volume fluid
presented according to their basis and haemostatic resuscitation, so-called ‘permissive hypotension’, avoids
capacity: the adverse effects of early aggressive resuscitation while
i) Collagen-based agents trigger platelet aggregation maintaining a level of tissue perfusion that, although
resulting in clot formation when in contact with a lower than normal, is adequate for short periods [130].
bleeding surface. They are often combined with a pro- A controlled hypotensive fluid resuscitation should aim
coagulant substance such as thrombin to enhance the to achieve a mean arterial pressure of 65 mmHg or
haemostatic effect. A positive haemostatic effect has more [181]. Its general effectiveness remains to be con-
been shown in several human studies [169-172]. firmed in RCTs; however, studies have demonstrated
ii) Gelatin-based products can be used alone or in increased survival when a low volume fluid resuscitation
combination with a procoagulant substance [167]. Swel- concept was used in penetrating trauma [182,183]. In
ling of the gelatin in contact with blood reduces the contrast, no significant difference in survival was found
blood flow and, in combination with a thrombin-based in patients with blunt trauma [184]. One study con-
component, enhances haemostasis. A similar or superior cluded that mortality was higher after on-site resuscita-
haemostatic effect has been observed compared with tion compared with in-hospital resuscitation [185]. It
collagen-based agents [173-175]. seems that greater increases in blood pressure are toler-
iii) The effect of cellulose-based haemostatic agents on ated without exacerbating haemorrhage when they are
bleeding has been less well studied and only case reports achieved gradually and with a significant delay following
that support their use are available. the initial injury [186]. All the same, a recent Cochrane
iv) Fibrin and synthetic glues or adhesives have both systematic review concluded that there is no evidence
haemostatic and sealant properties and their significant from RCTs for or against early or larger volume intrave-
effect on haemostasis have been shown in several nous fluids to treat uncontrolled haemorrhage [187].
human RCTs involving vascular, bone, skin and visceral However, a recent retrospective analysis demonstrated
surgery [176-178]. that aggressive resuscitation techniques, often initiated
v) Polysaccharide-based haemostatics can be divided in the prehospital setting, appear to increase the likeli-
into two broad categories [167]: N-acetyl-glucosamine- hood that patients with severe extremity injuries develop
containing glycosaminoglycans purified from microalgae secondary abdominal compartment syndrome (ACS)
11. Rossaint et al. Critical Care 2010, 14:R52 Page 11 of 29
http://ccforum.com/content/14/2/R52
[188]. In this study, early, large-volume crystalloid therapy strategies, limited quality of analysed studies,
administration was the greatest predictor of secondary mortality not always being the primary outcome, and
ACS. Moreover, a retrospective analysis of the German different, often short, observation periods. It is therefore
Trauma Registry database including 17,200 multiply difficult to reach a definitive conclusion as to the advan-
injured patients showed that the incidence of coagulopa- tage of one type of resuscitation fluid over the other.
thy increased with increasing volume of intravenous The Saline versus Albumin Fluid Evaluation study com-
fluids administered pre-clinically. Coagulopathy was pared 4% albumin with 0.9% sodium chloride in 6997
observed in more than 40% of patients with more than ICU patients and showed that albumin administration
2000 ml, in more than 50% with more than 3000 ml, was not associated with worse outcomes; however, there
and in more than 70% with more than 4000 ml adminis- was a trend towards higher mortality in the brain
tered [3]. trauma subgroup that received albumin (P = 0.06) [196].
The low-volume approach is contraindicated in TBI Promising results have been obtained with hypertonic
and spinal injuries, because an adequate perfusion pres- solutions. Recently, a double-blind, RCT in 209 patients
sure is crucial to ensure tissue oxygenation of the with blunt traumatic injuries analysed the effect of the
injured central nervous system. In addition, the concept treatment with 250 ml of 7.5% hypertonic saline and 6%
of permissive hypotension should be carefully consid- dextran 70 compared with lactated Ringer solution on
ered in the elderly patient and may be contraindicated if organ failure. The intent-to-treat analysis demonstrated
the patient suffers from chronic arterial hypertension. no significant difference in organ failure and in acute
A recent analysis from an ongoing multi-centre pro- respiratory disress syndrome (ARDS)-free survival. How-
spective cohort study suggests that the early use of vaso- ever, there was improved ARDS-free survival in the sub-
pressors for haemodynamic support after haemorrhagic set (19% of the population) requiring 10 U or more of
shock in comparison to aggressive volume resuscitation packed red blood cells (RBCs) [197]. One study showed
may be deleterious and should be used cautiously [189]. that the use of hypertonic saline was associated with
However, this study has several limitations: the study is lower intracranial pressure than with normal saline in
a secondary analysis of a prospective cohort study, and brain-injured patients [198] and a meta-analysis compar-
was not designed to answer the specific hypothesis ing hypertonic saline dextran with normal saline for
tested. Thus, it is not possible to separate vasopressor resuscitation in hypotension from penetrating torso
from the early management of trauma patients. In addi- injuries showed improved survival in the hypertonic sal-
tion, although the use of a vasopressor helps to rapidly ine dextran group when surgery was required [199].
restore arterial pressure, it should not be viewed as a A clinical trial with brain injury patients found that
substitute for fluid resuscitation and the target blood hypertonic saline reduced intracranial pressure more
pressure must be respected. effectively than dextran solution with 20% mannitol
Fluid therapy when compared in equimolar dosing [200]. However,
Recommendation 19 We recommend that crystalloids Cooper and colleagues found almost no difference in
be applied initially to treat the bleeding trauma patient neurological function six months after TBI in patients
(Grade 1B). We suggest that hypertonic solutions also who had received pre-hospital hypertonic saline resusci-
be considered during initial treatment (Grade 2B). We tation compared with conventional fluid [201]. In con-
suggest that the addition of colloids be considered clusion, the evidence suggests that hypertonic saline
within the prescribed limits for each solution in haemo- solutions are safe, and will improve haemodynamics
dynamically unstable patients (Grade 2C). during hypovolaemic resuscitation. The evidence for
Rationale It is still unclear what type of fluid should be increased survival with use of hypertonic saline solutions
employed in the initial treatment of the bleeding trauma is inconclusive. It is possible that certain subgroups
patient. Although several meta-analyses have shown an might benefit from hypertonic saline solutions, but
increased risk of death in patients treated with colloids further research is required [202].
compared with patients treated with crystalloids Normothermia
[190-194] and three of these studies showed that the Recommendation 20 We recommend early application
effect was particularly significant in a trauma subgroup of measures to reduce heat loss and warm the hypother-
[190,193,194], a more recent meta-analysis showed no mic patient in order to achieve and maintain nor-
difference in mortality between colloids and crystalloids mothermia (Grade 1C).
[195]. If colloids are used, modern hydroxyethyl starch Rationale Hypothermia, defined as a core body tem-
or gelatin solutions should be used because the risk:ben- perature below 35°C, is associated with acidosis, hypo-
efit ratio of dextran is disadvantageous. Problems in tension and coagulopathy in severely injured patients. In
evaluating and comparing the use of different resuscita- a retrospective study with 122 patients, hypothermia
tion fluids include the heterogeneity of populations and was an ominous clinical sign, accompanied by high
12. Rossaint et al. Critical Care 2010, 14:R52 Page 12 of 29
http://ccforum.com/content/14/2/R52
mortality and blood loss [203]. The profound clinical arrested. If mild hypothermia is applied in TBI, cooling
effects of hypothermia ultimately lead to higher morbid- should take place within the first three hours following
ity and mortality, and hypothermic patients require injury, preferably using selective brain cooling by cool-
more blood products [204]. ing the head and neck, be maintained for at least
Hypothermia is associated with an increased risk of 48 hours [218], rewarming should last 24 hours and the
severe bleeding, and hypothermia in trauma patients cerebral perfusion pressure should be maintained above
represents an independent risk factor for bleeding and 50 mmHg (systolic blood pressure ≥70 mmHg). Patients
death [205]. The effects of hypothermia include altered most likely to benefit from hypothermia are those with
platelet function, impaired coagulation factor function a GCS at admission between 4 and 7 [219]. Possible
(a 1°C drop in temperature is associated with a 10% side effects are hypotension, hypovolaemia, electrolyte
drop in function), enzyme inhibition and fibrinolysis disorders, insulin resistance and reduced insulin secre-
[206,207]. Body temperatures below 34°C compromise tion and increased risk of infection [220]. Further stu-
blood coagulation, but this has only been observed dies are warranted to investigate the postulated benefit
when coagulation tests (prothrombin time (PT) and of hypothermia in TBI taking these important factors
APTT) are carried out at the low temperatures seen in into account.
patients with hypothermia, and not when assessed at
37°C as is routine practice for such tests. Steps to pre- V. Management of bleeding and coagulation
vent hypothermia and the risk of hypothermia-induced Erythrocytes
coagulopathy include removing wet clothing, covering Recommendation 21 We recommend a target haemo-
the patient to avoid additional heat loss, increasing the globin (Hb) of 7 to 9 g/dl (Grade 1C).
ambient temperature, forced air warming, warm fluid Rationale Erythrocytes contribute to haemostasis by
therapy and, in extreme cases, extracorporeal re-warm- influencing the biochemical and functional responsive-
ing devices [208,209]. ness of activated platelets via the rheological effect on
Animal and human studies of controlled hypothermia platelet margination and by supporting thrombin gen-
in haemorrhage have shown some positive results com- eration [221]; however, the optimal Hct or Hb concen-
pared with normothermia [210,211]. Contradictory tration required to sustain haemostasis in massively
results have been observed in meta-analyses that examine bleeding patients is unclear. Further investigations into
mortality and neurological outcomes associated with the role of the Hb concentration on haemostasis in mas-
mild hypothermia in patients with TBI, possibly due to sively transfused patients are therefore warranted.
the different exclusion and inclusion criteria for the stu- The effects of the Hct on blood coagulation have not
dies used for the analysis [212-214]. The speed of induc- been fully elucidated [222]. An acute reduction of the
tion and duration of hypothermia, which may be very Hct results in an increase in the bleeding time [223,224]
important factors that influence the benefit associated with restoration upon re-transfusion [223]. This may
with this treatment. It has been shown that five days of relate to the presence of the enzyme elastase on the sur-
long-term cooling is more efficacious than two days of face of RBC membranes, which may activate coagulation
short-term cooling when mild hypothermia is used to factor IX [225,226]. However, a moderate reduction of
control refractory intracranial hypertension in adults the Hct does not increase blood loss from a standard
with severe TBI [215]. Obviously, the time span of spleen injury [224], and an isolated in vitro reduction of
hypothermia is crucial, because a recent prospective RCT the Hct did not compromise blood coagulation as
in 225 children with severe TBI showed that hypothermic assessed by thrombelastometry [227].
therapy initiated within 8 hours after injury and contin- No prospective RCT has compared restrictive and liberal
ued for 24 hours did not improve the neurological out- transfusion regimens in trauma, but 203 trauma patients
come and may increase mortality [216]. Furthermore, the from the Transfusion Requirements in Critical Care trial
mode of inducing cerebral hypothermia induction may [228] were re-analysed [229]. A restrictive transfusion regi-
influence its effectiveness. In a RCT comparing non-inva- men (Hb transfusion trigger <7.0 g/dl) resulted in fewer
sive selective brain cooling (33 to 35°C) in 66 patients transfusions as compared with the liberal transfusion regi-
with severe TBI and mild systemic hypothermia (rectal men (Hb transfusion trigger <10 g/dl) and appeared to be
temperature 33 to 35°C) and a control group not exposed safe. However, no statistically significant benefit in terms
to hypothermia, natural rewarming began after three of multiple organ failure or post-traumatic infections was
days. Mean intracranial pressure 24, 48 or 72 hours after observed. It should be emphasised that this study was
injury was significantly lower in the selective brain cool- neither designed nor powered to answer these questions
ing group than in the control group [217]. with precision. In addition, it cannot be ruled out that the
Prolonged hypothermia may be considered in patients number of RBC units transfused merely reflects the sever-
with isolated head trauma after haemorrhage has been ity of injury. Nevertheless, RBC transfusions have been
13. Rossaint et al. Critical Care 2010, 14:R52 Page 13 of 29
http://ccforum.com/content/14/2/R52
shown in multiple studies to be associated with increased and persistent coagulopathy at ICU entry has been
mortality [230-234], lung injury [234-236], increased infec- shown to be associated with a increased mortality [248].
tion rates [237,238] and renal failure in trauma victims Therefore, early aggressive treatment is likely to improve
[233]. This ill effect may be particularly important with the outcome of severely injured patients [249]. However,
RBC transfusions stored for more than 14 days [233]. there are also studies in which no survival benefit could
Despite the lack of high-level scientific evidence for a be shown [247,250].
specific Hb transfusion trigger in patients with TBI, Calcium
these patients are currently transfused in many centres Recommendation 23 We recommend that ionised cal-
to achieve an Hb of approximately 10 g/dl [239]. This cium levels be monitored during massive transfusion
might be justified by the recent finding that increasing (Grade 1C). We suggest that calcium chloride be admi-
the Hb from 8.7 to 10.2 g/dl improved local cerebral nistered during massive transfusion if ionised calcium
oxygenation in 75% of patients [158]. In another preli- levels are low or electrocardiographic changes suggest
minary study in patients with TBI, one to two RBC hypocalcaemia (Grade 2C).
transfusions at a Hb of approximately 9 g/dl transiently Rationale Calcium in the extracellular plasma exists
(three to six hours) increased cerebral oxygenation, either in a free ionised state (45%) or bound to proteins
again in approximately 75% of patients [240,241]. A sto- and other molecules in a biologically inactive state
rage time of more than 19 days precluded this effect (55%). The normal concentration of the ionised form
[240]. In another recent study, cerebral tissue oxygena- ranges from 1.1 to 1.3 mmol/l and is influenced by the
tion, on average, did not increase due to an increase in pH. A 0.1 unit increase in pH decreases the ionised cal-
Hb from 8.2 to 10.1 g/dl [242]. Nevertheless, the authors cium concentration by approximately 0.05 mmol/l [181].
came to the conclusion based on multivariable statistical The availability of ionised calcium is essential for the
models that the changes in cerebral oxygenation corre- timely formation and stabilisation of fibrin polymerisa-
lated significantly with Hb concentration [242]. This tion sites, and a decrease in cytosolic calcium concentra-
conclusion, however, was questioned in the accompany- tion precipitates a decrease in all platelet-related
ing editorial [243]. activities [181]. In addition, contractility of the heart
In an initial outcome study the lowest Hct was corre- and systemic vascular resistance are compromised at
lated with adverse neurological outcome and RBC trans- low ionised calcium levels. Combining beneficial cardio-
fusions were also found to be an independent factor vascular and coagulation effects, the level for ionised
predicting adverse neurological outcome [244]. Interest- calcium concentration should therefore be maintained
ingly, the number of days with a Hct below 30% was above 0.9 mmol/l [181].
found to be correlated with an improved neurological Early hypocalcaemia following traumatic injury shows
outcome [244]. In a more recent outcome study in 1150 a significant correlation with the amount of infused col-
patients with TBI, RBC transfusions were found to be loids, but not with crystalloids, and may be attributable
associated with a two-fold increased mortality and a to colloid-induced haemodilution [251]. Also, hypocal-
three-fold increased complication rate [138]. Therefore, caemia develops during massive transfusion as a result
patients with severe TBI should not have an Hb transfu- of the citrate employed as an anticoagulant in blood
sion threshold different than that of other critically ill products. Citrate exerts its anticoagulant activity by
patients. binding ionised calcium, and hypocalcaemia is most
Coagulation support common in association with FFP and platelet transfu-
Recommendation 22 We recommend that monitoring sion because these products contain high citrate concen-
and measures to support coagulation be initiated as trations. Citrate undergoes rapid hepatic metabolism,
early as possible (Grade 1C). and hypocalcaemia is generally transient during standard
Rationale Major trauma results not only in bleeding transfusion procedures. Citrate metabolism may be dra-
from anatomical sites but also frequently in coagulopa- matically impaired by hypoperfusion states, hypothermia
thy, which is associated with a several-fold increase in and in patients with hepatic insufficiency [252].
mortality [3,5,8,9,245]. This early coagulopathy of Fresh frozen plasma
trauma is mainly found in patients with hypoperfusion Recommendation 24 We recommend early treatment
(base deficit >6 mE/l) [8,245] and is characterised by an with thawed FFP in patients with massive bleeding
up-regulation of endothelial thrombomodulin, which (Grade 1B). The initial recommended dose is 10 to
forms complexes with thrombin [246]. 15 ml/kg. Further doses will depend on coagulation
Early monitoring of coagulation is essential to detect monitoring and the amount of other blood products
trauma-induced coagulopathy and to define the main administered (Grade 1C).
causes, including hyperfibrinolysis [10,117]. Early thera- Rationale The clinical efficacy of FFP is largely unpro-
peutic intervention does improve coagulation tests [247] ven [253]. Nevertheless, most guidelines recommend the
14. Rossaint et al. Critical Care 2010, 14:R52 Page 14 of 29
http://ccforum.com/content/14/2/R52
use of FFP either in massive bleeding or in significant consideration only patients alive upon ICU admission,
bleeding complicated by coagulopathy (PT or APTT and another defined massive transfusion as 10 units or
more than 1.5 times control) [7,254,255]. Patients trea- more prior to ICU admission. One report [247] defined
ted with oral anticoagulants (vitamin K antagonists) pre- massive transfusion as more than 10 units over 6 hours.
sent a particular challenge, and FFP is recommended Two of the studies are based on data collected in a
[255] only when prothrombin complex concentrate combat setting, while the other eight were performed
(PCC) is not available [254]. The most frequently based on data collected at civilian trauma centres. The
recommended dose is 10 to 15 ml/kg [254,255], and majority of the studies are single centre; one study is
further doses may be required [256]. As with all pro- multi-centre [261] and one is a retrospective analysis of
ducts derived from human blood, the risks associated the German Trauma Registry [3].
with FFP treatment include circulatory overload, ABO Seven studies showed better outcomes using a high
incompatibility, transmission of infectious diseases FFP:RBC ratio [3,261-266] and two did not [250,267].
(including prion diseases), mild allergic reactions and One study may be classified as indeterminate because a
transfusion-related acute lung injury [254,257,258]. FFP high FFP:RBC ratio (average 1:2) was associated with a
and platelet concentrates appear to be the most fre- better survival than a low ratio (average 1:4), but the
quently implicated blood products in transfusion-related survival curve was U-shaped, with the lowest mortality
acute lung injury [257-260]. Although the formal link at a 1:2 to 1:3 ratio [247]. The two combat studies
between the administration of FFP, control of bleeding showed better outcomes using a high ratio [265,266].
and an eventual improvement in the outcome of bleed- Early empirical infusion of FFP may increase the fre-
ing patients is lacking, most experts would agree that quency of delayed traumatic intracerebral haematoma
FFP treatment is beneficial in patients with massive and the mortality in patients with severe head injury
bleeding or significant bleeding complicated by [268]. Most of the studies calculate FFP:RBC ratio at
coagulopathy. 24 hours after admission. When Snyder and colleagues
There are very few well-designed studies that explore [267] used the FFP:RBC ratio at 24 hours as a fixed
massive transfusion strategy. The need for massive value, patients who received a higher ratio had signifi-
transfusion is relatively rare, occurring in less than 2% cantly better outcomes, but if the timing of component
of civilian trauma patients, but higher (7%) in the mili- product transfusion was taken into account, the differ-
tary setting. Massive transfusion management has been ence was no longer statistically significant.
based on the concept that coagulopathy associated with These combat data are retrospective, refer to young,
severe trauma was primarily consumptive due to the previously healthy male patients with penetrating inju-
dilution of blood clotting factors and the consumption ries and may be confounded to some extent by treat-
of haemostasis factors at the site of injury. ment biases. Because FFP requires a significant amount
FFP was recommended when PT or APTT was 1.5 of time before it is thawed and available for transfusion
times normal or after 10 RBC units had been transfused. and many trauma deaths occur soon after hospital
Many massive transfusion protocols stipulated one unit admission, patients who die early may receive RBC units
of FFP for every four units of RBCs. In recent years, ret- but die before FFP therapy has begun. These cases may
rospective data from the US Army combat support hospi- therefore be included in the low ratio group even if a
tals have shown an association between survival and a 1:1 strategy was intended. One further ground for criti-
higher ratio of transfused FFP and RBC units. These data cism of many of these studies is that the number of
show that casualties who received FFP and RBCs at a RBCs units transfused is an indicator of severity of
ratio of 1:4 or lower, had a three-fold higher mortality injury that cannot be completely adjusted for by regres-
than those who received a massive transfusion with a 2:3 sion analysis. All of these limitations must be kept in
ratio. These data have induced many civilian trauma cen- mind when analysing the available recent literature and
tres to modify their transfusion approach to incorporate emphasises the need for prospective trials.
the early use of thawed FFP in ratios approaching 1:1. Platelets
Ten relevant studies addressing FFP:RBC ratio have Recommendation 25 We recommend that platelets be
been identified, all of which were retrospective studies, administered to maintain a platelet count above 50 ×
although some are based on data collected prospectively 109/l (Grade 1C). We suggest maintenance of a platelet
for other reasons. None of the studies were clinical count above 100 × 10 9 /l in patients with multiple
RCTs. The majority of the authors used massive transfu- trauma who are severely bleeding or have TBI (Grade
sion (10 RBC units within 24 hours) as the entry criter- 2C). We suggest an initial dose of four to eight platelet
ion; however, to limit bias due to FFP unavailability, concentrates or one aphaeresis pack (Grade 2C).
one study [261] excluded patients who died within the Rationale In medical conditions leading to thrombocy-
first 30 minutes. One of the studies [262] took into topaenia, haemorrhage does not often occur until the
15. Rossaint et al. Critical Care 2010, 14:R52 Page 15 of 29
http://ccforum.com/content/14/2/R52
platelet count falls below 50 × 109/l, and platelet func- local collection practice, and physicians should be cogni-
tion decreases exponentially below this point [269-272]. sant of the doses provided locally. A pool of four to
There is no direct evidence to support a particular pla- eight platelet concentrates or a single-donor aphaeresis
telet transfusion threshold in the trauma patient. A con- unit is usually sufficient to provide haemostasis in a
sensus development conference sponsored by the thrombocytopaenic, bleeding patient. If required, the
National Institutes of Health (NIH; Bethesda, MD, USA) dose of platelets (× 109) can be calculated in more detail
in 1986 determined that bleeding is unlikely to be from the desired platelet increment, the patient’s blood
caused by thrombocytopaenia at platelet counts of 50 × volume in litres (estimated by multiplying the patient’s
109/l or greater and agreed that platelet transfusion is body surface area by 2.5, or 70 ml/kg in an adult), and a
appropriate to prevent or control bleeding associated correction factor of 0.67 to allow for pooling of approxi-
with deficiencies in platelet number or function mately 33% of transfused platelets in the spleen.
[273,274]. The NIH consensus did not consider trauma, Fibrinogen and cryoprecipitate
but it seems reasonable to recommend that a platelet Recommendation 26 We recommend treatment with
count of at least 50 × 10 9 /l be maintained following fibrinogen concentrate or cryoprecipitate if significant
injury. bleeding is accompanied by thrombelastometric signs of a
An argument can be made for maintaining a higher functional fibrinogen deficit or a plasma fibrinogen level of
level of platelets, perhaps up to 100 × 109/l, following less than 1.5 to 2.0 g/l (Grade 1C). We suggest an initial
injury. If a patient has increased fibrin degradation pro- fibrinogen concentrate dose of 3 to 4 g or 50 mg/kg of
ducts due to disseminated intravascular coagulation cryoprecipitate, which is approximately equivalent to 15 to
and/or hyperfibrinolysis, this will interfere with platelet 20 units in a 70 kg adult. Repeat doses may be guided by
function and a higher threshold of 75 × 109/l has been thrombelastometric monitoring and laboratory assessment
suggested by consensus groups [275,276]. Moreover, of fibrinogen levels (Grade 2C).
platelet-rich concentrate is an autologous concentration Rationale The formation of fibrin is a key step in
of platelets and growth factors (e.g. transforming growth blood coagulation [222,279], and hypofibrinogenemia is
factor-beta, vascular endothelial growth factor and plate- a usual component of complex coagulopathies asso-
let-derived growth factor), and due to the increased con- ciated with massive bleeding. Coagulopathic civilian
centration and release of these factors, platelet-rich trauma patients had a fibrinogen concentration of
concentrates could potentially enhance bone and soft 0.9 g/l (interquartile ratio (IQR) 0.5 to 1.5 g/l) in con-
tissue healing [277]. Transfusion threshold levels of up junction with a maximum clot firmness of 6 mm (IQR
to 100 × 10 9 /l have been suggested for treatment of 0 to 9 mm) using thrombelastometry, whereas only
severe brain injury and massive haemorrhage, but the 2.5% of healthy volunteers had a maximum clot firm-
evidence for the higher threshold is weak [275,276]. One ness of 7 mm or less [10]. In trauma patients, a maxi-
group showed that trauma patients receiving platelets mum clot firmness of 7 mm was associated with a
and RBCs at a ratio of 1:5 or greater had a lower 30-day fibrinogen level of approximately 2 g/l [10]. During
mortality when compared with those with who received massive blood loss replacement, fibrinogen may be the
less than this ratio (38% vs. 61%, P = 0.001) [264]. first coagulation factor to decrease critically [280].
Another study of massively transfused trauma patients During postpartum haemorrhage, fibrinogen plasma
has pointed to an early aggressive correction of coagulo- concentration is the only coagulation parameter inde-
pathy with platelet transfusion as a possible contributing pendently associated with progress toward severe
factor to good outcome [278]. In this retrospective bleeding, with a level less than 2 g/l having a positive
cohort study, survivors received one platelet transfusion predictive value of 100% [281]. Blood loss and blood
for every 7.7 units of blood transfused whereas nonsur- transfusion needs were also found to inversely corre-
vivors received only one platelet transfusion for every late with preoperative fibrinogen levels in coronary
11.9 units of blood transfused (P = 0.03). artery bypass graft surgery [282].
When platelet transfusion was introduced in the During serious perioperative bleeding, fibrinogen
1950s, no clinical trials were employed to assess the uti- treatment (2 g, range 1 to 5 g) was associated with a
lity of platelet therapy compared with placebo, and such reduction in allogeneic blood product transfusion [283].
trials today might be considered unethical. The appro- The fibrinogen concentration before treatment was
priate dose of platelets is therefore uncertain. Platelet 1.4 g/l (IQR 1.0 to 1.8 g/l) rising to 2.4 g/l (IQR 2.1 to
concentrate produced from a unit of whole blood con- 2.6 g/l) after fibrinogen substitution [283]. An observa-
tains 7.5 × 1010 platelets on average and should increase tional study suggests that fibrinogen substitution can
the platelet count by 5 to 10 × 109/l in a 70 kg recipient. improve survival in combat-related trauma [284]. An
Aphaeresis platelet concentrates generally contain RCT in patients undergoing radical cystectomy with
approximately 3 to 6 × 10 11 platelets, depending on excessive blood loss has shown that postoperative blood
16. Rossaint et al. Critical Care 2010, 14:R52 Page 16 of 29
http://ccforum.com/content/14/2/R52
transfusions could be reduced by the administration of 15 mg/kg followed by an infusion of 1 to 5 mg/kg per
45 mg/kg fibrinogen at a mean pre-treatment fibrinogen hour or ε-aminocaproic acid 100 to 150 mg/kg followed
level of 1.7 ± 0.3 g/l rising to 2.4 ± 0.1 g/l following by 15 mg/kg/h. Antifibrinolytic therapy should be
fibrinogen substitution [285]. guided by thrombelastometric monitoring if possible
Fibrinogen administration using thrombelastometry as and stopped once bleeding has been adequately con-
guidance may be preferable to measuring fibrinogen trolled (Grade 2C).
levels in the laboratory. Some methodological issues in Rationale Tranexamic acid (trans-4-aminomethylcyclo-
the various laboratory methods to measure fibrinogen hexane-1-carboxylic acid) is a synthetic lysine analogue
concentration remain [286,287], and in the presence of that is a competitive inhibitor of plasmin and plasmino-
artificial colloids such as hydroxyethyl starch, even the gen. Tranexamic acid is distributed throughout all tis-
most frequently recommended method [287], the Clauss sues and the plasma half-life is 120 minutes. There is
method, significantly overestimates the actual fibrinogen large variation in the dose employed. In vitro studies
concentration [288]. have suggested that a dose of 10 μg/ml is required to
It is not known whether the administration of fibrino- inhibit fibrinolysis [297]. Studies of plasma levels [298]
gen via factor concentrate, cryoprecipitate or FFP is confirmed that the Horrow regimen (10 mg/kg followed
associated with a post-traumatic venous thrombotic risk. by 1 mg/kg per hour) [299], shown to reduce blood loss
However, fibrinogen levels are expected to rise to a level in cardiac surgery, attained these levels. Other studies
of approximately 7 g/l after major surgery and trauma have used boluses of up to 5 g per patient with no ill
[289,290] even without intra-operative fibrinogen effect [300].
administration, and the effect of intra-operative fibrino- ε-aminocaproic acid is also a synthetic lysine analogue
gen administration on postoperative fibrinogen levels is that has a potency 10-fold weaker than that of tranexa-
unknown at the present time. Interestingly, intra-opera- mic acid. It is therefore administered in a loading dose
tive administration of 45 mg/kg fibrinogen concentrate of 150 mg/kg followed by a continuous infusion of
in patients undergoing cystectomy resulted in higher 15 mg/kg/h. The initial elimination half-life is 60 to
early postoperative fibrinogen levels but already at 75 minutes and it must therefore be administered by
24 hours post-operation fibrinogen levels were identical continuous infusion in order to maintain therapeutic
in patients with and without intra-operative fibrinogen drug levels until the bleeding risk has diminished.
administration [285]. Similarly, 24 hour fibrinogen levels The clear efficacy of antifibrinolytic agents in reducing
were identical in patients who received and those who bleeding in elective surgery and especially in cardiac sur-
did not receive 2 g of fibrinogen prior to coronary artery gery has been shown in numerous clinical trials
bypass graft surgery [291]. This result is in keeping with [301-305]. The benefits of antifibrinolytics in these
the study by Weinkove and Rangarajan, who found no situations where hyperfibrinolysis is not usually seen,
thrombotic risk in patients treated with fibrinogen con- suggests that under normal circumstances when a
centrate due to acquired hypofibrinogenemia (fibrinogen patient has a bleeding vessel, there is low-grade fibrino-
<1.5 g/l) [292]. lytic turnover that exacerbates bleeding. Thus, fibrinoly-
Pharmacological agents sis is ‘switched off’ and less bleeding results. It may be
An increasingly large body of evidence supports the use possible to extrapolate the benefits of antifibrinolytic
of antifibrinolytic agents for the management of bleed- agents to bleeding secondary to trauma, although this
ing in elective surgery and cardiac surgery patients. For assumption is not backed by any published data that
the purpose of these guidelines, we have assumed that suggest that the haemostatic response to trauma is simi-
these effects are transferable to trauma patients, and our lar to the haemostatic response to elective surgery.
recommendations are based upon this unproven There is insufficient evidence from RCTs of antifibrino-
assumption. Since the last guidelines were written, apro- lytic agents in trauma patients to either support or
tinin has been associated with patient safety issues, with refute a clinically important treatment effect. The effi-
an increased rate of renal disease and mortality when cacy of tranexamic acid in trauma has been assessed by
compared with the lysine analogues in a large clinical the Clinical Randomisation of an Antifibrinolytic in Sig-
trial and is therefore no longer recommended [293-296]. nificant Haemorrhage (CRASH) II study, in which
Antifibrinolytic agents 20,000 trauma patients worldwide were randomly
Recommendation 27 We suggest that antifibrinolytic assigned to 1 g of tranexamic acid for a period of
agents be considered in the bleeding trauma patient 10 minutes followed by 1 g infused for a period of eight
(Grade 2C). We recommend monitoring of fibrinolysis hours. This results of this trial are due to be published
in all patients and administration of antifibrinolytic in 2010 [306].
agents in patients with established hyperfibrinolysis The risk of precipitated thrombosis with the use of the
(Grade 1B). Suggested dosages are tranexamic acid 10 to lysine analogues tranexamic acid and ε-aminocaproic