The European guideline on management of major bleeding and coagulopathy following trauma: fifth edition

Abstract

Background

Severe traumatic injury continues to present challenges to healthcare systems around the world, and post-traumatic bleeding remains a leading cause of potentially preventable death among injured patients. Now in its fifth edition, this document aims to provide guidance on the management of major bleeding and coagulopathy following traumatic injury and encourages adaptation of the guiding principles described here to individual institutional circumstances and resources.

Methods

The pan-European, multidisciplinary Task Force for Advanced Bleeding Care in Trauma was founded in 2004, and the current author group included representatives of six relevant European professional societies. The group applied a structured, evidence-based consensus approach to address scientific queries that served as the basis for each recommendation and supporting rationale. Expert opinion and current clinical practice were also considered, particularly in areas in which randomised clinical trials have not or cannot be performed. Existing recommendations were re-examined and revised based on scientific evidence that has emerged since the previous edition and observed shifts in clinical practice. New recommendations were formulated to reflect current clinical concerns and areas in which new research data have been generated.

Results

Advances in our understanding of the pathophysiology of post-traumatic coagulopathy have supported improved management strategies, including evidence that early, individualised goal-directed treatment improves the outcome of severely injured patients. The overall organisation of the current guideline has been designed to reflect the clinical decision-making process along the patient pathway in an approximate temporal sequence. Recommendations are grouped behind the rationale for key decision points, which are patient- or problem-oriented rather than related to specific treatment modalities. While these recommendations provide guidance for the diagnosis and treatment of major bleeding and coagulopathy, emerging evidence supports the author group’s belief that the greatest outcome improvement can be achieved through education and the establishment of and adherence to local clinical management algorithms.

Conclusions

A multidisciplinary approach and adherence to evidence-based guidance are key to improving patient outcomes. If incorporated into local practice, these clinical practice guidelines have the potential to ensure a uniform standard of care across Europe and beyond and better outcomes for the severely bleeding trauma patient.

Key messages

• Traumatically injured patients should be transported quickly and treated by a specialised trauma centre whenever possible.

• Measures to monitor and support coagulation should be initiated as early as possible and used to guide a goal-directed treatment strategy.

• A damage-control approach to surgical intervention should guide patient management.

• Coagulation support and thromboprophylactic strategies should consider trauma patients who have been pre-treated with anticoagulants or platelet inhibitors.

• Local adherence to a multidisciplinary, evidence-based treatment protocol should serve as the basis of patient management and undergo regular quality assessment.

Background

Severe trauma is a major global public health issue, contributing to about 1 in 10 mortalities and resulting in the annual worldwide death of more than 5.8 million people [1, 2]. According to the World Health Organization (WHO), road traffic accidents, suicides and homicides are the three leading causes of injury and violence-related deaths [3]. In recent years, sudden mass casualties due to bombing and assaults have become an new phenomenon in Europe and other regions, resulting in hundreds of severely injured and bleeding patients within a very short period of time, thereby posing huge challenges for local healthcare systems [4,5,6].

Uncontrolled post-traumatic bleeding is still the leading cause of potentially preventable death among injured patients [7,8,9] and one third of all bleeding trauma patients show signs of coagulopathy at hospital admission [10,11,12,13,14,15,16,17]. These patients develop multiple organ failure and experience death more frequently than patients with similar injury patterns in the absence of coagulopathy [11, 13, 14, 18, 19]. The early acute coagulopathy associated with traumatic injury has recently been recognised as a multifactorial primary condition that results from a combination of bleeding-induced shock, tissue injury-related thrombomodulin upregulation, thrombin-thrombomodulin-complex generation and the activation of anticoagulant and fibrinolytic pathways (Fig. 1) [8, 10, 13,14,15, 20,21,22,23,24,25,26]. The severity of the coagulation disorder is influenced by environmental and therapeutic factors that result in acidaemia, hypothermia, dilution, hypoperfusion and consumption of coagulation factors [10, 14, 24, 27,28,29,30,31,32]. Moreover, the coagulopathy is modified by trauma-related factors such as brain injury [33] and individual patient-related factors that include age, genetic background, co-morbidities, inflammation and medication administered prior to becoming injured, especially oral anticoagulants, and pre-hospital fluid administration [28, 34, 35].

Fig. 1

Schematic drawing of the factors, including those that are preexisting as well as those related to both trauma and resuscitation measures, that contribute to traumatic coagulopathy. Adapted from [20, 24, 30,31,32, 38]

This European clinical practice guideline, originally published in 2007 [36] and updated in 2010 [37], 2013 [38] and 2016 [39], represents the fifth edition of the guideline and is part of the European “STOP the Bleeding Campaign”, an international initiative launched in 2013 to reduce morbidity and mortality associated with bleeding following traumatic injury [40]. In the last 3 years, a multitude of studies were published that enhance understanding of the pathophysiology of trauma-induced coagulopathy, fill important knowledge gaps about the mechanism and efficacy of trauma treatment strategies and provide evidence that individualised goal-directed trauma treatment improves the outcome of severely injured patients. This new information has been integrated in the current version of the guideline.

Although this set of recommendations outlines corridors for diagnosis and treatment, the author group believes that the greatest outcome improvement can be achieved through education and the establishment of local clinical management guidelines or algorithms. We believe that adherence to local management guidelines or algorithms should be assessed on a regular basis and will lead to greater adherence. If incorporated into local practice, these clinical practice guidelines have the potential to ensure a uniform standard of care across Europe and beyond and better outcomes for the severely bleeding trauma patient, as has indeed be found in three recent studies [41,42,43].

Methods

The recommendations made in this guideline are graded according to the Grading of Recommendations Assessment, Development and Evaluation (GRADE) system [44], summarised in Table 1. According to the GRADE scheme, the number associated with each recommendation reflects the strength of the recommendation by the author group, with “we recommend” (Grade 1) being stronger and “we suggest” (Grade 2) being weaker, while the associated letter (A, B or C) reflects the quality of the scientific evidence. Comprehensive, structured, computer-based literature searches were performed using the indexed online database MEDLINE/PubMed, supplemented by screening of reference lists within relevant publications. The aim of each search strategy was to identify randomised controlled trials (RCTs), non-RCTs and systematic reviews that addressed specific scientific queries. In the absence of high-quality scientific support, case reports, observational studies and case control studies were also considered and the literature support for each recommendation graded accordingly.

Table 1 Grading of recommendations after [44]. RCT, randomised controlled trial. Table reprinted with permission

Boolean operators, medical subject headings (MeSH) and key terms were applied to structure each literature search. Searches were limited to a uniform human patient population defined by the search terms and the time period since 01 February 2015. The structured literature search strategies applied to each section of the guideline are listed in Additional file 1. Abstracts identified by each search strategy were screened by a subset of authors and if considered relevant, full publications evaluated. Evaluation of literature chosen for citation in the guideline was performed according to the 2011 Oxford Centre for Evidence-Based Medicine (OCEBM) working group levels of evidence (Table 2) [45]. Each literature citation included in this version of the guideline and the corresponding grading according to the OCEBM levels of evidence (Table 2) are listed in Additional file 2.

Table 2 Oxford Centre for Evidence-based Medicine (OCEBM) levels of evidence (2011) [45]

Selection of the scientific queries addressed, screening and evaluation of the literature, formulation of the recommendations and the supporting rationales was performed by members of the Task Force for Advanced Bleeding Care in Trauma, which was founded in 2004. The Task Force comprises a multidisciplinary team of pan-European experts representing the fields of emergency medicine, surgery, anaesthesiology, haematology and intensive care medicine. Among the authors are representatives of the European Society for Trauma and Emergency Surgery (ESTES), the European Society of Anaesthesiology (ESA), the European Shock Society (ESS), the European Society for Emergency Medicine (EuSEM), the Network for the Advancement of Patient Blood Management, Haemostasis and Thrombosis (NATA) and the European Society of Intensive Care Medicine (ESICM).

The guideline update process involved several remote (telephone and/or internet-based) meetings, extensive electronic communication and one face-to-face consensus conference. In December 2017, the authors participated in a web conference during which the queries to be addressed in the updated guideline were defined. Screening and evaluation of abstracts and full publications identified by the structured searches and formulation of draft recommendations and rationales was performed by working subgroups. Each chapter was reviewed by an assigned working subgroup and then the entire author group. The wording of each recommendation was finalised during a face-to-face consensus conference that took place in April 2018. Following revisions and approval by the author group, the manuscript was approved by the endorsing societies between August and November 2018. An update of this manuscript is anticipated in due time.

Results

I. Initial resuscitation and prevention of further bleeding

Minimal elapsed time

Recommendation 1

We recommend that severely injured patients be transported directly to an appropriate trauma facility. (Grade 1B)

We recommend that the time elapsed between injury and bleeding control be minimised. (Grade 1A)

Rationale

Because relatively few hospitals provide all of the services required to treat patients with multiple injuries, many healthcare systems have developed trauma networks or processes. The underlying aim of trauma care organisation is to move patients to multi-specialist care as early as possible, yet still provide immediate critical interventions. These aims can come into conflict, and there are a number of different means with which to resolve these issues, resulting in large variations in trauma care systems both between and within countries and a consequent significant heterogeneity in the literature. The evidence is weak, but there is a general consensus that the organisation of a group of hospitals into a “trauma system” leads to about a 15% reduction in trauma death, with about a 50% reduction in “preventable death” [46]. Inter-hospital transfer of patients does not seem to change overall mortality [47], and the evidence neither supports nor refutes direct transport from the accident scene to a major trauma centre [48]. However, there is some evidence that a lower threshold for trauma centre care should be used in patients aged > 65 years [49]. No definitive conclusion can be drawn about the relationship between a hospital’s trauma patient volume and outcomes [50, 51]. Despite a lack of evidence, there is a consensus that “systemised” trauma care that matches each patient to the most appropriate treatment facility in a timely manner is advantageous, whereby the definition of “appropriate” will depend on the patient profile, the nature of the injuries and the hospital facilities available [52].

Trauma patients in need of emergency surgery for ongoing haemorrhage have increased survival if the elapsed time between the traumatic injury and admission to the operating theatre is minimised. More than 50% of all trauma patients with a fatal outcome die within 24 h of injury [7]. Despite a lack of evidence from prospective RCTs, well-designed retrospective studies provide evidence for early surgical intervention in patients with traumatic haemorrhagic shock [53, 54]. In addition, studies that analyse trauma systems indirectly emphasise the importance of minimising the time between admission and surgical bleeding control in patients with traumatic haemorrhagic shock [55]. Minimisation of time to surgery is an accepted principle of trauma care and is unlikely to ever be tested in a clinical trial due to lack of equipoise.

Local bleeding management

Recommendation 2

We recommend local compression to limit life-threatening bleeding. (Grade 1A)

We recommend adjunct tourniquet use to stop life-threatening bleeding from open extremity injuries in the pre-surgical setting. (Grade 1B)

We recommend the adjunct use of a pelvic binder to limit life-threatening bleeding in the presence of a suspected pelvic fracture in the pre-surgical setting. (Grade 1B)

Rationale

Most life-threatening bleeding from extremities observed in the civilian setting can be controlled by local compression, by either manual compression or pressure bandages applied to the wounds. Extra local compression to the source of bleeding can also be achieved in certain penetrating injuries by Foley catheter insertion directly into the wound [56]. Foley catheter balloon tamponade was initially described in bleeding penetrating injuries of the neck [57, 58]. In addition, the use of topical haemostatic agents in combination with direct pressure enhances bleeding control in the pre-hospital setting [59] (see also R21).

When uncontrolled arterial bleeding occurs as a result of mangled extremity injuries, including penetrating or blast injuries or traumatic amputations, a tourniquet is a simple and efficient method with which to acutely control haemorrhage [60,61,62,63,64]. Tourniquet application has become the standard of care for the control of severe external haemorrhage following military combat injuries, and several publications report the effectiveness of tourniquets in this specific setting in adults [60,61,62,63, 65] and children [66]. A study of volunteers showed that any tourniquet device presently on the market works efficiently [64]. The study also showed that “pressure point control” was ineffective because collateral circulation was observed within seconds. Tourniquet-induced pain was not often reported by patients. No evidence or opinion supports the use of tourniquets in the context of closed injuries.

Tourniquets should be left in place until surgical control of bleeding is achieved [61, 63]; however, this time span should be restricted as much as possible. Improper or prolonged placement of a tourniquet can lead to complications such as nerve paralysis and limb ischaemia [67]; however, these effects are rare [65]. Some publications suggest a maximum application time of 2 h [67]. Reports from military settings describe cases in which tourniquets have remained in place for up to 6 h with survival of the extremity [61]. Much recent discussion has centred on the translation of this evidence to civilian practice, as little published evidence exists. Bleeding from most civilian wounds can be controlled using local pressure; however, uncontrolled external bleeding from either blunt [59] or penetrating [68] limb injury should be controlled with a tourniquet.

Patients with severe high-energy and complex pelvic trauma, haemodynamic instability and massive blood loss belong to the most severe and highly lethal group of trauma patients, and their management is time-sensitive and challenging [69]. Global mortality in polytraumatised patients presenting with pelvic ring fractures remains high (33%) despite improvements in management and treatment algorithms [70]. The pelvis can create a multifocal haemorrhage, including significant retroperitoneal haematoma, which may not be easily compressible or possible to manage using traditional surgical methods [71]. Treatment of pelvic ring fractures requires re-approximation of bony structures to address mechanical instability, damage-control resuscitation (DCR) to restore haemostasis, assessment for associated injuries and triage of investigations. In addition, multimodal haemorrhage control [external fixation and compression (damage-control orthopaedics), retroperitoneal packing (damage-control surgery), urgent radiologic angioembolisation or resuscitative endovascular balloon occlusion of the aorta (REBOA)] by multidisciplinary trauma specialists (general surgeons, orthopaedic surgeons, endovascular surgeons/interventional radiologists) is required [69, 72,73,74,75].

Correctly placed pelvic binders lead to anatomical closure of the pelvic ring, with a favourable haemodynamic effect. These devices are increasingly being used in the pre-hospital setting if a pelvic fracture is suspected [76, 77]. Unstable pelvic ring fractures may be clinically and radiologically overlooked during initial assessment, especially in unconscious patients, and the time point for opening and/or removal remains controversial. In-hospital external fixation stabilises anterior pelvic ring lesions and can be combined with posterior stabilisation using percutaneous sacro-iliac screws in the presence of associated lesions to the posterior ring. The external fixator is especially useful in the acute phase, acquiring an acceptable reduction and an adequate stability in the partially unstable lesions and also reduces pelvic volume and bleeding [78]. In a small quasi-randomised study, pelvic packing achieved more rapid control of severe pelvic trauma than angioembolisation [79]. The median time from admission to angiography was 102 min (range 76−214), and longer than 77 min (range 43–125) from admission to pelvic packing (p < 0.01). The procedure time for angioembolisation was 84 min (range 62–105), while the surgical time was 60 min (range 41–92; p < 0.001). Nine patients had to undergo pelvic packing for persistent bleeding after embolisation. If haemodynamic instability persists, a laparotomy for haemostasis according to damage-control principles to all potentially involved systems (digestive, vascular, urinary and bone) should be performed [80].

Ventilation

Recommendation 3

We recommend the avoidance of hypoxaemia. (Grade 1A)

We recommend normoventilation of trauma patients. (Grade 1B)

We suggest hyperventilation in the presence of signs of imminent cerebral herniation. (Grade 2C)

Rationale

Tracheal intubation of severely injured patients is a delicate procedure that involves risks and requires skill and proper training of the operator. The fundamental objective of intubation is to ensure adequate ventilation and oxygenation and to guarantee the patency of the airway. There are well-defined situations in which intubation is mandatory, for example, in the presence of airway obstruction, altered consciousness [Glasgow Coma Scale (GCS) ≤ 8], haemorrhagic shock, hypoventilation or hypoxaemia [81]; however, other aspects should also be considered. For example, the introduction of positive pressure can induce potentially life-threatening hypotension in hypovolaemic patients [82], and some authors have reported increased mortality associated with pre-hospital intubation [83].

Several factors influence the success of intubation and therefore patient prognosis. Rapid sequence induction appears to be the best method [84]; however, several aspects remain to be clarified, such as who is best suited to make the decision to intubate, which drugs and which rescue device to use and the ideal infrastructure of emergency services. Most of the available data come from retrospective studies, which are open to bias; therefore, controversy remains about the appropriate use of tracheal intubation in patients following traumatic injury [85].

The negative effects of hypoxaemia are well known, particularly in patients with traumatic brain injury (TBI) [86, 87]; therefore, high oxygen concentrations are generally targeted during the initial management of these patients to ensure oxygen delivery to ischaemic areas. Some studies, however, have suggested that prolonged hyperoxia is associated with increased mortality [88, 89]. A recent meta-analysis based on high-quality evidence [90] showed that prolonged liberal oxygen therapy in acutely ill adults increased mortality without improving other patient-important outcomes. Extreme hyperoxia (PaO₂ > 487 mmHg [> 65 kPa]) should therefore be avoided in patients with TBI [91]. Another recent study showed that the mortality increase was related to the duration and extent of hyperoxia [92]. On the other hand, mechanical ventilation using settings that targeted an oxygen saturation of 88–92% compared with > 95% did not negatively influence survival in critical care patients [93]. The negative effects of hyperoxia are likely related to altered microcirculation associated with high PaO₂ [94] and increased production of oxygen-free radicals [95] and patients with severe brain injury may be at particular risk [96]. Therefore, although hyperoxia may increase the oxygen content and delivery in an extremely anaemic trauma patient and be associated with a benefit in this specific situation, hyperoxia should be returned to normoxia as soon as the haemoglobin (Hb) level allows [91].

While adequate ventilation can affect the outcome of severe trauma patients, there is a tendency for rescue personnel to hyperventilate patients during initial resuscitation [97, 98]. Hyperventilated trauma patients appear to have increased mortality when compared with non-hyperventilated patients [88]. Target PaCO₂ should be 5.0–5.5 kPa (35–40 mmHg).

The effect of hyperventilation on bleeding and outcome in patients with severe trauma without TBI is not known. There are several potential mechanisms by which the adverse effects of hyperventilation and hypocapnia could be mediated, including increased vasoconstriction with decreased cerebral blood flow and impaired tissue perfusion. Cerebral tissue lactic acidosis has been shown to occur almost immediately after induction of hypocapnia in children and adults with TBI and haemorrhagic shock [99]. In addition, even a modest level of hypocapnia [< 27 mmHg (3.6 kPa)] may result in neuronal depolarisation with glutamate release and exacerbation of the primary injury via apoptosis [100]. In the setting of absolute or relative hypovolaemia, an excessive rate of positive pressure ventilation may further compromise venous return and produce hypotension and even cardiovascular collapse [101, 102].

The only situation in which hyperventilation-induced hypocapnia may play a potential role is imminent cerebral herniation. The decrease in cerebral blood flow produced by acute hypocapnia during hyperventilation causes a decrease in intracranial pressure that can be used for short periods of time and in selected cases such as imminent brain herniation. The presence of signs such as unilateral or bilateral pupillary dilation or decerebrate posturing are indicators for an extreme risk of imminent death or irreversible brain damage. Hyperventilation may be used under these circumstances to try to gain time until other measures are effective [103, 104]. There are no clinical studies that evaluate this practice; however, there is a clear physiological rationale. Given the extreme risk of death if no measures are undertaken, the risk–benefit balance seems favourable; however, it is important to normalise PaCO₂ as soon as feasible.

Ventilation with low tidal volume (around 6 mL/kg) is now recommended in all patients treated with mechanical ventilation, even during surgery [105]. Randomised studies demonstrate that short-term ventilation (< 5 h) with high tidal volume (12 mL/kg) without positive end-expiratory pressure (PEEP) may promote pulmonary inflammation and alveolar coagulation in patients with normal lung function [106]. The early use of protective ventilation with low tidal volume and moderate PEEP is recommended, particularly in bleeding trauma patients, who are all at risk of acute respiratory distress syndrome (ARDS).

II. Diagnosis and monitoring of bleeding

Initial assessment

Recommendation 4

We recommend that the physician clinically assess the extent of traumatic haemorrhage using a combination of patient physiology, anatomical injury pattern, mechanism of injury and the patient response to initial resuscitation. (Grade 1C)

We suggest that the shock index (SI) be used to assess the degree of hypovolaemic shock. (Grade 2C)

Rationale

Trauma physicians must quickly and accurately assess and predict when a massive transfusion protocol, including corresponding logistics, should be activated [107] and terminated [108]. While blood loss may sometimes be obvious, neither visual estimation nor physiological parameters are satisfactory guides to estimate the degree of bleeding [109]. Knowledge about the mechanism of injury provides useful information to identify patients at risk of significant haemorrhage at an early stage. For example, the American College of Surgeons defined a threshold of 6 m (20 ft) as a “critical falling height” associated with major injuries, including haemorrhage [110]. Further critical mechanisms include high-energy deceleration impact as well as low-velocity versus high-velocity gunshot injuries. The mechanism of injury combined with injury severity and the patient’s physiological presentation should further guide the decision to initiate early surgical bleeding control as outlined in the Advanced Trauma Life Support (ATLS) survey [111,112,113,114]. Table 3 summarises estimated blood loss based on initial presentation according to the ATLS classification system of hypovolaemic shock. This classification has been shown to be useful as a rough estimation of sustained blood loss in patients with haemorrhagic shock [115]. However, several groups have highlighted discrepancies associated with the weight assigned to each parameter when assessing blood loss that makes it challenging to classify patients using this system. Mutschler and co-workers have analysed the adequacy of this classification and found that > 90% of all trauma patients could not be categorised according to the ATLS classification of hypovolaemic shock [116]. The same group analysed the validity of the ATLS classification and concluded that this system may underestimate mental disability in the presence of hypovolaemic shock, while overestimating the degree of tachycardia associated with hypotension [117]. A retrospective analysis of the validity of the ATLS classification showed that increasing blood loss produces an increase in heart rate and a decrease in blood pressure, but to a lesser degree than suggested by the ATLS classification. In addition, there are no significant changes in respiratory rate or in level of consciousness with bleeding [118]. Other parameters used for this classification, such as pulse pressure and urinary output, may not be adequately assessed during the initial phase of care. The individual response to fluid challenge as suggested by the ATLS survey should be viewed critically in the context of low-volume resuscitation and “permissive hypotension”, which is currently advocated in bleeding trauma patients.

Table 3 American College of Surgeons Advanced Trauma Life Support (ATLS) classification of blood loss based on initial patient presentation. Signs and symptoms of haemorrhage by class. Table reprinted with permission from the American College of Surgeons [111]

Isolated vital signs, such as heart rate or systolic blood pressure, have been shown to be unreliable in the assessment of hypovolaemic shock. Heart rate alone has not been shown to predict the need for massive transfusion, in particular not in the geriatric trauma population [119]. In contrast, the SI, defined as the ratio of heart rate to systolic blood pressure, has been advocated to better risk-stratify patients for critical bleeding, increased transfusion requirements and early mortality [120, 121]. Paladino and co-workers found that this index may be useful to draw attention to abnormal values, but may be too insensitive to exclude disease and should not lower the suspicion of major injury [122]. Mutschler and co-workers have suggested a novel and clinically reliable classification of hypovolaemic shock based on four classes of worsening base deficit. The objective of this study was to correlate this classification with corresponding SI strata for the rapid assessment of trauma patients in the absence of laboratory parameters. Twenty-one thousand eight hundred fifty-three adult trauma patients were retrieved from the TraumaRegister DGU® database and divided into four strata of worsening SI at emergency department arrival (group I, SI < 0.6; group II, SI ≥ 0.6 to < 1.0; group III, SI ≥ 1.0 to < 1.4; and group IV, SI ≥ 1.4), and demographics, injury characteristics, transfusion requirements, fluid resuscitation and outcomes were assessed [123]. Worsening of SI was associated with increasing injury severity scores (ISS) from 19.3 (± 12.0) in group I to 37.3 (± 16.8) in group IV, while mortality increased from 10.9 to 39.8%. Increments in SI paralleled increasing fluid resuscitation, vasopressor use and decreasing Hb, platelet counts and Quick values. The number of blood units transfused increased from 1.0 (± 4.8) in group I to 21.4 (±2 6.2) in group IV patients. Of patients, 31% in group III and 57% in group IV required ≥ 10 red blood cell (RBC) units prior to intensive care unit (ICU) admission. Another retrospective database analysis of 10,234 patients has confirmed the role of SI either upon arrival or at departure from the emergency department as a detrimental sign of poor outcome in adult trauma patients [124].

A number of scoring systems that predict the risk of ongoing bleeding, transfusion requirements and coagulopathy have been introduced, but all of these lack prospective validation [108, 125,126,127,128,129,130,131]. Each scoring system has its unique advantages and disadvantages, and specific aspects of each scoring system may affect widespread applicability and statistical performance.

Immediate intervention

Recommendation 5

We recommend that patients with an obvious bleeding source and those presenting with haemorrhagic shock in extremis and a suspected source of bleeding undergo an immediate bleeding control procedure. (Grade 1C)

Rationale

The patient who presents in extremis is a patient who has already lost a large amount of blood and is in a severe shock. If bleeding continues, death in shock is an imminent risk. The source of bleeding may be immediately obvious, and penetrating injuries are more likely to require surgical bleeding control. In a retrospective study of 106 abdominal vascular injuries, all 41 patients arriving in shock following gunshot wounds were candidates for rapid transfer to the operating theatre for surgical bleeding control [132]. A similar observation in a study of 271 patients undergoing immediate laparotomy for gunshot wounds indicates that these wounds combined with signs of severe hypovolaemic shock specifically require early surgical bleeding control. This observation is true to a lesser extent for abdominal stab wounds [133]. Data on injuries caused by penetrating metal fragments from explosives or gunshot wounds during the Vietnam War confirm the need for early surgical control when patients present in shock [134]. Following blunt trauma, the mechanism of injury can to a certain extent determine whether the patient in haemorrhagic shock will be a candidate for surgical bleeding control. Only a few studies address the relationship between the mechanism of injury and the risk of bleeding, however, and none of these publications describes a randomised prospective trial with high-level evidence [135]. We have found no objective data describing the relationship between the risk of bleeding and the mechanism of injury resulting in skeletal fractures in general or of long-bone fractures in particular.

Traffic accidents are the leading cause of pelvic injury. Motor vehicle crashes cause approximately 60% of pelvic fractures followed by falls from great height (23%). Most of the remainder result from motorbike collisions and vehicle-pedestrian accidents [136, 137]. There is a correlation between “unstable” pelvic fractures and intra-abdominal injuries [136, 138]. An association between major pelvic fractures and severe head injuries, concomitant thoracic, abdominal, urological and skeletal injuries is also well described [136]. High-energy injuries produce greater damage to both the pelvis and organs. Patients with high-energy injuries require more transfusion units, and more than 75% have associated head, thorax, abdominal or genitourinary injuries [139]. It is well documented that “unstable” pelvic fractures are associated with massive haemorrhage [138, 140], and haemorrhage is the leading cause of death in patients with major pelvic fractures. Vertical shear pelvic ring fractures with caudal displacement of the hemi-pelvis may disrupt the pelvic floor and pelvic vasculature far more than standard vertical shear injuries. Inferior displacement of the hemi-pelvis using X-ray imaging should therefore alert the surgeon to the possible presence of severe arterial injuries [141].

In blunt chest trauma haemothoraces, > 500 mL should trigger chest tube insertion. Thoracotomy is indicated for ongoing bleeding and chest tube output > 1500 mL within 24 h or > 200 mL for three consecutive hours. Acute damage-control thoracotomy should be performed for refractive haemorrhagic shock due to persistent chest bleeding enhanced by initial chest tube output > 1500 mL [142, 143].

Further investigation

Recommendation 6

We recommend that patients without a need for immediate bleeding control and an unidentified source of bleeding undergo immediate further investigation. (Grade 1C)

Rationale

Haemodynamically stable patients, or patients who can be stabilised during initial resuscitation, with an unidentified bleeding source, but not in need of immediate bleeding control, should undergo further investigation of the chest, abdominal cavity and pelvic ring, which can be major sources of acute blood loss following traumatic injury. Besides clinical examination, imaging studies, including ultrasonography and computed tomography (CT) [144], as well as laboratory tests, including blood gas analysis and coagulation profiles, together with functional assays, are recommended diagnostic modalities during the primary survey [111, 145, 146].

As CT scanners are increasingly being advocated and integrated into modern resuscitation units and emergency departments, this technique may replace conventional radiographic imaging and ultrasound as diagnostic measures during the primary survey [147]. The diagnostic accuracy, safety and effectiveness of these immediate measures are dependent on sophisticated pre-hospital treatment by trained and experienced emergency personnel and short transportation times [148, 149]. Proximity of the CT scanner to the resuscitation room in the emergency department has been shown to have a significant positive effect on the probability of survival for the severely injured patient [150]. Distances of more than 50 m had a significant negative effect on outcome and should be considered when new emergency departments are planned and constructed. If a CT scanner is not available in the emergency department, CT scanning implies transportation of the patient to the CT room; therefore, the clinician must evaluate the implications and potential risks and benefits of the procedure. Transfer times to and from all forms of diagnostic imaging need to be considered in the context of haemodynamic stability. During transport, all vital signs should be closely monitored and resuscitation measures continued. If performed quickly within a well-structured environment and by a well-organised trauma team, CT seems to be safe, feasible and justified, even in severely injured haemodynamically unstable patients [151]. Among haemodynamically unstable haemoperitoneum patients, 17.2% had no documented intraperitoneal injury and over half of the patients were treated without emergent surgical intervention [152].

Imaging

Recommendation 7

We recommend the use of focused assessment with sonography in trauma (FAST) ultrasound for the detection of free fluid in patients with torso trauma. (Grade 1C)

We recommend early imaging using contrast-enhanced whole-body CT (WBCT) for the detection and identification of type of injury and potential source of bleeding. (Grade 1B)

Rationale

Focused assessment with sonography in trauma (FAST)

The FAST examination has developed into a key instrument in the acute evaluation of trauma patients with suspected abdominal and thoraco-abdominal injuries [153]. FAST techniques are being used with reduced examination times and a focused assessment of specific clinical issues using only a few standardised cross-sectional planes [154]. As a rapid and non-invasive diagnostic approach to the detection of haemorrhages in the peritoneal, pleural and pericardial cavities in the emergency department, FAST represents a cornerstone of the primary ATLS survey [153, 155,156,157]. Volume status can be assessed non-invasively using ultrasound of the inferior vena cava. Several studies have indicated the specificity and accuracy, but low sensitivity, of initial FAST for detecting and excluding free intraperitoneal fluid as well as intra-abdominal injuries [158,159,160,161,162,163,164] in both penetrating [165] and blunt abdominal trauma [166, 167]. Liu and colleagues [168] found a high sensitivity, specificity and accuracy of initial ultrasound examination for the detection of haemoperitoneum. In a retrospective registry study, free fluid or organ injury was detected in 72.4% of patients using FAST versus 84.3% using CT, yielding a sensitivity of 92% for initial FAST [169]. In another retrospective study that included 1540 hypotensive patients (1227 blunt, 313 penetrating trauma), ultrasound examination had a sensitivity and specificity close to 100% for free intra-abdominal fluid [170]. The double-line sign, which has been described as a wedge-shaped hypoechoic area in the Morison pouch, bounded on both sides by echogenic lines during FAST, may represent a false-positive finding for free intraperitoneal fluid with an overall prevalence of 27% [171].

A recent retrospective review examined the role of FAST as a screening tool for identifying intra-abdominal injuries [172]. A total of 1671 blunt-trauma patients were assessed over 1.5 years, and intra-abdominal injuries were confirmed in 146 patients using CT and/or laparotomy. Intraoperative findings included injuries to the liver, spleen, kidneys and bowels. Among 114 haemodynamically stable patients, FAST was positive in 25 patients, with a sensitivity of 22%. FAST was positive in 9 of 32 haemodynamically unstable patients, with a sensitivity of 28%. Free peritoneal fluid and splenic injury were associated with a positive FAST on univariate analysis and were independent predictors of a positive FAST on multiple logistic regression. An updated Cochrane review from 2015, including RCTs, assessed the effect of diagnostic algorithms using ultrasonography, including FAST examinations, in the emergency department relative to early, late and overall mortality of patients with suspected blunt abdominal trauma [173]. Four studies were identified, but the trials were of overall poor to moderate methodological quality. Mortality data were pooled from three trials involving 1254 patients; the risk ratio (RR) in favour of the FAST arm was 1.00 [95% confidence interval (CI) 0.50–2.00]. FAST-based pathways reduced the number of CT scans [random-effects model risk difference (RD) − 0.52, 95% CI − 0.83 to − 0.21], but the meaning of this result remained unclear. It is unlikely that FAST will ever be investigated by means of a confirmatory, large-scale RCT; therefore, this review may provide the best available evidence for clinical practice guidelines and management recommendations. From the few published head-to-head studies, it appears that negative ultrasound scans are likely to reduce the incidence of multidetector CT (MDCT) scans, which, given the low sensitivity of FAST (or reliability of negative results), may adversely affect the diagnostic yield of the trauma survey. At best, ultrasound has no negative impact on mortality or morbidity.

In haemodynamically stable patients, a negative FAST without a CT scan may result in missed intra-abdominal injuries and should direct further diagnostic investigations. A number of patients who present with free intra-abdominal fluid according to ultrasound can safely undergo further investigation using multislice CT (MSCT). Under normal circumstances, adult patients need to be haemodynamically stable when MSCT is performed outside of the emergency department [170]. Haemodynamically stable patients with a high-risk mechanism of injury, such as high-energy trauma, or even low-energy injuries in elderly individuals, should be scanned after ultrasound for additional injuries using MSCT. As CT scanners are integrated into resuscitation units, WBCT diagnosis may replace ultrasound as a diagnostic method. In haemodynamically unstable blunt-trauma patients with clear physical findings on examination, the decision to perform exploratory laparotomy should not be discouraged by a negative FAST [169, 172].

Follow-up sonography as part of secondary or tertiary surveys in patients without abdominal parenchymal organ lesions or free intra-abdominal fluid on initial WBCT is not routinely required, but should be performed if indicated on a clinical or laboratory basis due to its rapid and non-invasive character [174]. New ultrasound techniques using second-generation contrast agents [contrast-enhanced ultrasound (CEUS)] have been developed, allowing all of the vascular phase to be performed in real time, increasing ultrasound capability to detect parenchymal injuries, enhancing some qualitative findings, such as lesion extension, margins and its relationship with capsule and vessels [175]. These techniques are currently under investigation.

Computed tomography (CT)

The advantages of MSCT, including WBCT, among severely injured patients in time savings, diagnostic accuracy and potentially also survival have been documented [151, 176,177,178,179,180,181,182,183,184]. The integration of modern MSCT scanners in the emergency department area prompts immediate assessment of any trauma victim likely to survive the assessment following admission [177, 184], thereby allowing timely diagnosis, differentiation between various types of major vascular injury, identification of associated findings, specific localisation of the source of bleeding and planning for bleeding control [80, 185, 186]. A 1-year review of early management of pelvic fracture patients documented a significant delay in the recognition of (major) pelvic fractures, including those associated with hip dislocations and (potential) pelvic bleeding with selective pelvic X-ray versus CT scanning [187]. More than one third of patients with thoracic stab wounds presented with negative chest X-ray, but pathologies using CT [188].

MDCT is currently considered the “gold standard” in the assessment of intra-abdominal blunt-traumatic injury [189]. Mesenteric active bleeding, adjacent interloop-free fluid and bowel wall perfusion defects have been associated with surgically significant bowel injuries and an overall accuracy, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for 64-slice MDCT of 73.8%, 80.0%, 73.0%, 28.6%, and 96.4%, respectively [190]. Advancements in modern MDCT technology and an improved understanding of optimal protocols have enabled full-body scanning of adequate image quality and in less than 30 s. In a retrospective study comparing 370 patients in two groups, Weninger and colleagues [184] showed that faster diagnosis using MSCT led to shorter emergency department and operating room time and shorter ICU stays [184]. Huber-Wagner et al. also showed the benefit of WBCT integration into early trauma care as CT diagnosis significantly increased the probability of survival in patients with polytrauma [147, 150]. WBCT as a standard diagnostic tool during the earliest resuscitation phase provides the added benefit of identifying head and chest injuries and other bleeding sources in multiply injured patients. Nonselective throracic CT was superior to selective CT in detecting thoracic injuries in blunt trauma [191], and thoracic CT showed a NPV value of 99% in triaging haemodynamically normal patients with penetrating chest trauma [192]. A comparison between emergency physicians and on-call radiologists on the accuracy of CT interpretations showed that emergency physicians were successful in identifying fatal injuries on trauma scans following a short-term interpretation training [193].

A series of systematic reviews has assessed the benefits of WBCT in the early management of severely injured patients and all showed a survival benefit with the use of WBCT in trauma patients [194,195,196,197]. In contrast, the only prospective RCT conducted to date in this area compared immediate WBCT scanning versus conventional imaging and selective CT scanning in patients with severe trauma [A Multicenter, Randomised Study of Early Assessment by CT Scanning in Severely Injured Trauma Patients (REACT-2)] in four centres in the Netherlands and one in Switzerland and found no survival benefit with WBCT [198]. A total of 1403 trauma patients aged ≥ 18 years with compromised vital parameters, clinical suspicion of life-threatening injuries or severe injury were randomly assigned (1:1) to immediate WBCT scanning or to a standard work-up with conventional imaging supplemented with selective CT scanning. The primary analysis included 541 patients in the immediate WBCT scanning group and 542 in the standard work-up group. In-hospital mortality did not differ between groups (WBCT 86 [16%] of 541 vs standard work-up 85 [16%] of 542; p = 0.92). In-hospital mortality also did not differ in subgroup analyses among patients with polytrauma (WBCT 81 [22%] of 362 vs standard work-up 82 [25%] of 331; p = 0.46) and TBI (68 [38%] of 178 vs 66 [44%] of 151; p = 0.31).

The WBCT protocol usually includes a non-contrast scan of the brain and neck followed by a contrast-enhanced scan of the chest, abdomen and pelvis. Several authors have emphasised the benefit of contrast medium-enhanced CT scanning. MSCT is the “gold standard” for the identification of retroperitoneal haemorrhage (RPH). After injection of intravenous (i.v.) contrast media, CT identified RPH in all cases (100%) and may detect the source of bleeding (40%) by extravasation of contrast media [199]. Dual-phase contrast-enhanced CT (CECT) without CT angiography showed a high sensitivity (93.9%) and PPV (88.6%) compared with digital subtraction angiography for the detection of active haemorrhage in patients with blunt abdominopelvic trauma [200]. Anderson et al. [201, 202] found high accuracy in the evaluation of splenic injuries resulting from trauma after administration of an i.v. contrast medium. Delayed-phase CT may be used to detect active bleeding in solid organs. Fang et al. [203] demonstrated that the pooling of contrast material within the peritoneal cavity in blunt liver injuries indicates active and massive bleeding. Patients with this finding showed rapid deterioration of haemodynamic status, and most required emergent surgery. Intra-parenchymal pooling of contrast material with an unruptured liver capsule often indicates a self-limited haemorrhage, and these patients respond well to non-operative treatment. Tan and colleagues [204] found that patients with hollow viscus and mesenteric injuries following blunt abdominal trauma exhibited an abnormal preoperative CT scan. Wu et al. [205] confirmed the accuracy of CT in identifying severe, life-threatening mesenteric haemorrhage and blunt bowel injuries. Although contrast extravasation (CE) in CT scans of pelvises with blunt trauma may be common, many patients will not require intervention such as angioembolisation [206]. The negative predicted value of 100% should be reassuring to trauma surgeons such that if a modern CT scanner is used, and no CE is detected using CT, then the pelvis is unlikely to be a source of haemorrhagic shock. All of these findings are attributable to both increased comfort with observing CEs and the increased sensitivity of modern CT scanners.

The issue of radiation is still debated, but iterative as well as split-bolus protocols can now significantly reduce radiation exposure [207]. Imaging algorithms including WBCT in multi-trauma patients are standardised but may vary substantially between centres [208]. An online survey among level-1 trauma centres in Switzerland revealed radiation doses ranging from 1268 to 3988 mGy × cm per WBCT. Including WBCT in the initial work-up of trauma patients results in higher radiation doses, but fewer additional CT examinations are needed, and the time to complete trauma-related imaging is shorter [209]. Risk-stratification criteria based upon documented suspected injuries during the primary survey at the site of the accident or the emergency department may identify high-energy trauma patients not in need of extended radiological imaging, including WBCT [210]. To a large extent, WBCT in high-energy trauma patients does not affect patient care if the patient is mentally alert, not intoxicated or showing signs of more than minor injuries when clinically evaluated. The risk of missing important traumatic findings in these patients is very low. Observation of the patient with re-examination instead of imaging may be considered in this group, often young patients, for whom radiation dose is an issue [210]. Davies and co-workers have developed a scoring system with a sensitivity of 97% (95% CI 88–99%) and a specificity of 56% (95% CI 49–64%) for significant injury to stratify the use of trauma radiographs, focused on CT and WBCT, and which may add an objective component to decision-making to reduce unnecessary scans [211]. Regression modelling identified clinical signs in more than one body region, reduced GCS, haemodynamic abnormality, respiratory abnormality and mechanism of injury as independent predictors of polytrauma.

Haemoglobin

Recommendation 8

We recommend that a low initial Hb be considered an indicator for severe bleeding associated with coagulopathy. (Grade 1B)

We recommend the use of repeated Hb measurements as a laboratory marker for bleeding, as an initial Hb value in the normal range may mask bleeding. (Grade 1B)

Rationale

Hb or haematocrit (Hct) assays are part of the basic diagnostic work-up for trauma patients. Recently, non-invasive Hb monitoring has also been tested and showed high precision compared with laboratory measurements [212, 213]. Currently, the use of Hb rather than Hct is widespread, and the latter is a calculated parameter derived from the Hb. However, most studies on which these recommendations are based analysed Hct rather than Hb. Because both parameters are used interchangeably in clinical practice, in these guidelines, we refer to both parameters according to the parameter described by the literature.

The diagnostic value of the Hb or Hct for detecting trauma patients with severe injury and occult bleeding sources has been a topic of debate [214,215,216]. A major limitation of the diagnostic value is the confounding influence of resuscitation measures on the Hb/Hct due to administration of i.v. fluids and erythrocyte concentrates [217,218,219]. In addition, initial Hb or Hct measurements may not accurately reflect blood loss, because patients bleed whole blood and compensatory mechanisms that move fluids from interstitial spaces require time. The suggestion that initial Hb/Hct for the detection of severe bleeding is associated with low sensitivity has been challenged. In a retrospective study of 196 trauma patients, Ryan et al. [220] found that Hct at admission closely correlates with haemorrhagic shock. Knottenbelt et al. evaluated 1000 trauma patients and found lower initial Hb level in moderately and severely shocked patients [221]. Other authors also recommended that the initial Hct play a greater role in the assessment of blood loss in trauma patients. In a retrospective analysis of 1492 consecutive trauma patients, Thorson et al. found that the initial Hct is associated more closely with the need for transfusion than other parameters such as heart rate, blood pressure or acidaemia, suggesting that fluid shifts are rapid following traumatic injury and imply a more important role for Hct in the initial assessment of trauma victims [222]. An initial low Hb level is one of the predictive criteria for massive transfusion using the trauma-associated severe haemorrhage (TASH) [126] and Vandromme [223] scores.

Thorson et al. [224] analysed changes in Hct in two successive determinations and concluded that the change in Hct is a reliable parameter with which to detect blood loss. Two prospective observational diagnostic studies also showed the sensitivity of serial Hct measurements for the detection of patients with severe injury [214, 216]. Holstein and co-workers showed that a Hb level below 80 g/L in patients with pelvic trauma was associated with non-survival [225]. Decreasing serial Hct measurements may reflect continued bleeding. However, a patient with significant bleeding may maintain the serial Hct in the context of ongoing resuscitation and physiological compensatory mechanisms. Acute anaemia may play an adverse role in the clotting process, because a low Hct may reduce platelet marginalisation, with a potentially negative impact on platelet activation. Moreover, Schlimp et al. [226] demonstrated strong correlation between fibrinogen levels and Hb.

Serum lactate and base deficit

Recommendation 9

We recommend serum lactate and/or base deficit measurements as a sensitive test to estimate and monitor the extent of bleeding and shock. (Grade 1B)

Rationale

Serum lactate has been used as a diagnostic parameter and prognostic marker of haemorrhagic shock since the 1960s [227]. The amount of lactate produced by anaerobic glycolysis is an indirect marker of oxygen debt, tissue hypoperfusion and the severity of haemorrhagic shock [228,229,230,231]. Similarly, base deficit values derived from arterial blood gas analysis provide an indirect estimation of global tissue acidosis due to impaired perfusion [230, 231]. Vincent and colleagues [232] showed the value of serial lactate measurements for predicting survival in a prospective study in patients with circulatory shock. This study showed that changes in lactate concentration provide an early and objective evaluation of patient response to therapy and suggested that repeated lactate determinations represent a reliable prognostic index for patients with circulatory shock [232]. Abramson and colleagues [233] performed a prospective observational study in patients with multiple traumatic injuries to evaluate the correlation between the time course of blood lactate levels and survival. All patients in whom lactate levels returned to the normal range (≤ 2 mmol/L) within 24 h survived. Survival decreased to 77.8% if normalisation occurred within 48 h and to 13.6% in those patients in whom lactate levels were elevated above 2 mmol/L for more than 48 h [233]. These findings were confirmed in a study by Manikis et al., who showed that initial lactate levels were higher in non-survivors after major trauma and that prolongation of time to normalisation of lactate levels of more than 24 h was associated with the development of post-traumatic organ failure [234]. The determination of lactate and/or base deficit may be particularly important in penetrating trauma. Following this type of injury, triage vital signs, such as blood pressure, heart rate and respiratory rate, do not reflect the severity of injury and are not related to lactate or base deficit levels [235]. A systemic review on the value of blood lactate kinetics in critically ill patients has been published recently [236].

The reliability of lactate determination may be lower when traumatic injury is associated with alcohol consumption. Ethanol metabolism induces the conversion of pyruvate to lactate via lactate dehydrogenase, causing an increase in the level of lactate in the blood. In alcohol-associated trauma, therefore, base deficit may be a better predictor of prognosis than lactate [237], although some authors suggest that ethanol-induced acidosis may also affect base deficit, masking the prognosis of trauma patients [238]. Therefore, in the case of traumatic injury associated with alcohol consumption, the results of the lactate measurements should be interpreted with caution.

Similar to the predictive value of lactate levels, the initial base deficit, obtained either from arterial or peripheral venous blood [239] has been established as a potent independent predictor of mortality in patients with traumatic haemorrhagic shock [237]. Davis and colleagues stratified the extent of base deficit into three categories: mild (− 3 to − 5 mEq/L), moderate (− 6 to − 9 mEq/L) and severe (<− 10 mEq/L) and established a significant correlation between the admission base deficit, transfusion requirements within the first 24 h and the risk of post-traumatic organ failure or death [240]. The same group of authors showed that the base deficit is a better prognostic marker of death than the pH in arterial blood gas analyses [241]. Mutschler et al. [123] analysed a cohort of 16,305 severely injured patients derived from the German Trauma Registry database and concluded that the determination of base deficit upon emergency department admission predicts transfusion requirements and mortality better than ATLS classification [123]. Furthermore, the base deficit was shown to represent a highly sensitive marker for the extent of post-traumatic shock and mortality, both in adult and paediatric patients [242, 243].

Although both the base deficit and serum lactate levels are well correlated with shock and resuscitation, these two parameters do not strictly correlate with each other in severely injured patients [244], and lactate levels more specifically reflect the degree of tissue hypoperfusion [230, 231, 244].

Coagulation monitoring

Recommendation 10

We recommend that routine practice include the early and repeated monitoring of haemostasis, using either a combined traditional laboratory determination [prothrombin time (PT), platelet counts and Clauss fibrinogen level] and/or point-of-care (POC) PT/international normalised ratio (INR) and/or a viscoelastic method (VEM). (Grade 1C)

We recommend laboratory screening of patients treated or suspected of being treated with anticoagulant agents. (Grade 1C)

Rationale

Standard coagulation monitoring comprises early and repeated determination of PT, platelet counts and Clauss fibrinogen level. The PT measures the activity of the extrinsic coagulation pathway (factors II, VII, and X), resulting in a prolonged PT value when any of these factors is low. There is frequently confusion in the literature over the terms PT and INR, because they are often used interchangeably, despite being based on different comparative values. Strictly speaking, PT is the ratio of the patient’s PT compared with a PT performed using pooled plasma from healthy individuals. Conventionally, PT testing has been used for all patients except those treated with a vitamin K antagonist (VKA). The INR, on the other hand, represents a PT in which the activating tissue factor used in the assay is assigned a value such that the effect of the VKA is consistent across laboratories.

Because the definition of traumatic coagulopathy is equivalent to a prolongation of the PT [11], PT values on admission have been shown to correlate with the degree of shock and to be predictive of clinical outcome in the presence of traumatic haemorrhage. Peltan et al., for example, found that acute traumatic coagulopathy affected 50% of patients with traumatic bleeding, defined as a PT:INR ratio > 1.2 and 21% of subjects if traumatic coagulopathy was defined as an INR > 1.5 [245]. The latter was significantly associated with all-cause death, haemorrhagic shock-associated death, venous thromboembolism (VTE) and multiple organ failure. As a result, PT/INR is used to assess the severity of traumatic coagulopathy and the need for transfusion.

Recently, POC monitors (portable coagulometers) that assess the INR have improved in quality and ease of use. They are widely applied by professionals in anticoagulant clinics and at home by patients to monitor the effect of VKAs. Use may be more common in the emergency department to identify patients with significant coagulopathy compared with laboratory-based methods [246, 247]. It is, however, important to note that variation between these devices and a laboratory-based PT may be 15% [246, 248]. David et al. suggest that a near-patient INR value of 1.5 could be used to guide fresh frozen plasma (FFP) or prothrombin complex concentrate (PCC) administration [247]. Goodman et al. demonstrated that POC INR testing was more rapid and cheaper than a modified thrombelastography [TEG®; rapid TEG® (r-TEG®)] and correlated not only with r-TEG® values, but also with blood product transfusion [249].

It is often misunderstood that the conventional coagulation screens [PT and activated partial thromboplastin time (APTT)] only provide information on levels of coagulation factor [250]. These values, therefore, will typically appear normal during early blood loss, despite the potential for an underlying activation of coagulation and thrombus formation [251,252,253,254]. The turnaround time for results of VEM [TEG®, rotational thromboelastometry (ROTEM®)], as for POC PT/INR, has been shown to be significantly shorter than conventional laboratory testing, with a time saving of 30–60 min [251, 255, 256]. VEM may also be useful in the detection of coagulation abnormalities associated with the use of direct thrombin inhibitors such as dabigatran, argatroban, bivalirudin or hirudin, although these tests cannot discriminate between the effects of inhibitors and the impact of traumatic coagulopathy [257].

VEM provides a rapid assessment of haemostasis to support clinical decision-making. This in turn has generated a growing confidence in these methods and increased use in children, adolescent and adult patients [29, 256, 258]. To date, however, only one open randomised controlled study has been completed, which involved 111 injured patients from an academic level-1 trauma centre meeting criteria for massive transfusion protocol activation [259]. Patients were randomised to receive either a massive transfusion protocol goal-directed using TEG® or by conventional coagulation assays (CCA). Survival at 28 days in the TEG® group was significantly higher than the CCA group, with 20 deaths in the CCA group (36.4%) compared with 11 in the TEG® group (19.6%) (p = 0.049). Most bleeding deaths occurred within the first 6 h following patient arrival at the clinic (21.8% CCA group vs 7.1% TEG® group) (p = 0.032). CCA patients required a similar number of RBC units as the TEG® patients but more plasma units [CCA, 2.0 (0–4); TEG®, 0.0 (0–3)] (p = 0.022), and more platelet units [CCA, 0.0 (0–1); TEG®, 0.0 (0–0)] (p = 0.041) in the first 2 h of resuscitation. Despite these very promising results, it should be noted that this study was open, unblinded, and that randomisation into either of the two treatment modalities was based on alternating weeks, which potentially introduces a bias into the care of the patients.

r-TEG® is a new variant of VEM in which coagulation is initiated by the addition of kaolin and tissue factor, which appears to reduce the measurement time compared with conventional TEG® in adults [260, 261] and children [262, 263]. One of several validation studies included 808 adult trauma patients in a prospective international multicentre cohort study from four major trauma centres. The authors demonstrated that a ROTEM® clot amplitude of 5 mm was a valid marker for acute traumatic coagulopathy and a predictor of massive transfusion [22]. Meyer et al. evaluated fibrinogen levels in trauma patients determined using two whole-blood VEM, TEG® functional fibrinogen (FF) and ROTEM® FIBTEM (FIBTEM, fibrin-based extrinsically activated test) and compared these with the plasma-based Clauss method. Both methods correlated with the Clauss fibrinogen level, without variation in the strength of these correlations [264].

Recent discussion has focused on the specific usefulness of VEM in the detection of early fibrinolysis. On the one hand, Moore et al. found that VEM only demonstrates hyperfibrinolytic traces in a minority of those with traumatic bleeding [265]. On the other hand, Brohi et al. have shown that VEM is a poor detector of fibrinolytic activation, which they suggest may be due to the production of soluble S100A10 from the endothelium, thereby blocking detection of tissue plasminogen activator by VEM [266]. The widespread use of tranexamic acid (TXA) in trauma patients may be expected to counteract acute fibrinolysis in these patients. At this time, therefore, it is not possible to support the use of VEM as a superior option over conventional coagulation tests. Results from the global multicentre Implementing Treatment Algorithms for the Correction of Trauma Induced Coagulopathy (iTACTIC) study are expected to reveal how the use of VEM might impact clinical outcomes [267].

Despite the widespread use of VEM, their usefulness is still being evaluated. In a recent systematic Cochrane review, Hunt et al. [268] found no evidence for the accuracy of TEG®, and very little evidence to support the accuracy of ROTEM®, therefore were unable to offer any advice about the use of these methods [268]. In another systematic review, Da Luz et al. [269] concluded that only limited evidence from observational studies was available to support the use of VEM in the diagnosis of early traumatic coagulopathy. While these tests may be used to predict blood product transfusion, mortality and other important patient outcomes may be unaffected [269]. A number of other limitations associated with the use of VEM have been described elsewhere. TEG® may lead to unnecessary transfusion with platelets, whereas the application of ROTEM® may result in goal-directed fibrinogen substitution. Although use is rapidly increasing, controversy remains at present regarding the utility of VEM for the detection of posttraumatic coagulopathy.

Agreement between the results of VEM and standard coagulation tests also remains a matter of debate. Some studies find acceptable agreement between results [261, 263, 270], while a number of other studies show significant discrepancies, even among different VEM (TEG® and ROTEM®) [29, 248, 271, 272]. In one instance, Agren et al. suggest that TEG® functional analyses may have overestimated fibrinogen levels (by more than one gram per litre) [272]. Elsewhere, Hagemo et al. found that the correlation was highly variable at different stages of the clotting process and between centres [273], highlighting the need for clarification and standardisation of these techniques. One additional potential limitation of VEM may be the lack of sensitivity in detecting and monitoring platelet dysfunction due to antiplatelet drugs. If platelet dysfunction is expected, POC platelet function tests, for example whole-blood impedance aggregometry, should be used in addition to VEM. More research is required to understand these variations, and in the meantime, physicians should use their own judgement when developing local policies.

Eventually, new POC devices to measure fibrinogen concentration could represent a new means with which to assess traumatic coagulopathy. Several monitors are in development [274] and may compete with VEM in the near future.

The increasing use of pre-injury anticoagulants and, in particular, the so-called direct (non-vitamin K-dependent) oral anticoagulants (DOACs) pose an increasing challenge in the setting of trauma haemorrhage, as these agents can substantially complicate the extent and dynamics of bleeding [275]. Retrospectively, preexisting coagulation disorders, either congenital or acquired, e.g. due to anticoagulant intake, were associated with an elevated mortality in trauma patients with and without head injury (43% versus 17% [276,277,278,279]). While VKAs and antiplatelet agents (APA) can be assessed using INR measurements and platelet function assays, to date there is no universally available and validated (rapid) test system for any of the DOACs that is associated with meaningful sensitivity and specificity [275]. The standard PT (preferably the INR) is prolonged in VKA-treated patients. If time and amount of the most recent dose of dabigatran are unknown, normal values for thrombin time, ecarin clotting time and diluted thrombin time suggest the absence of dabigatran in clinically relevant concentrations [275]. A normal standard anti-Xa test may also exclude intake (or efficacy) of an anti-Xa agent (rivaroxaban, apixaban, edoxaban, betrixaban). If these tests are prolonged, a diluted thrombin time (Hemoclot® for dabigatran) or a specific anti-Xa test (for anti-Xa agents) should be performed [280]. Chromogenic anti-factor-Xa-activity tests can be used to estimate the plasma concentrations of factor Xa-inhibitors (apixaban, edoxaban, rivaroxaban), but require calibration with substance-specific reagents [275, 281, 282].

Platelet function monitoring

Recommendation 11

We suggest the use of POC platelet function devices as an adjunct to standard laboratory and/or POC coagulation monitoring in patients with suspected platelet dysfunction. (Grade 2C)

Rationale

Traumatic injury has been associated with platelet dysfunction [283,284,285]. Unfortunately, neither CCAs nor standard VEM reliably reflect platelet function status [286, 287]. Light transmission aggregometry (LTA), considered the gold standard for the assessment of platelet function, is inadequate in the acute setting [288]. Several POC platelet function devices are available, such as the platelet function analyser (PFA-100®), whole-blood multiple electrode impedance aggregometry (MEA), platelet reactivity assay (e.g. VerifyNow®), vasodilator-stimulated phosphoprotein (VASP) or VEM devices with channels for measuring platelet function. Different POC tests capture different aspects of platelet function and are therefore not interchangeable in the assessment of platelet reactivity. However, these devices may be of value in detecting pharmacologically induced platelet inhibition in trauma patients for whom prior intake of antiplatelet agents (APA) is unknown, for example in unconscious or confused patients, and in patients with uncertain treatment compliance.

The VerifyNow® platelet reactivity test for aspirin (VN®-ASA) successfully identified TBI patients who reported using aspirin therapy [289, 290]. The MEA device allowed rapid assessment of APA activity in patients admitted for intracranial haemorrhage (ICH) requiring urgent neurosurgical intervention [291] and in TBI [283, 292,293,294]. The thrombelastography platelet mapping (TEG®-PM®) assay also identified APA use in trauma patients [286, 295]; however, PFA-100 showed low sensitivity and PPVs (48.6% and 63.4%, respectively) for detecting pharmacologically induced platelet dysfunction in trauma patients on APA [296]. In one study comparing MEA, VerifyNow® and TEG®-PM® in adult trauma patients, specific tests for the arachidonic acid (AA) pathway in all three devices accurately identified any APA use (either aspirin or clopidogrel) [286]. AA tests to identify platelet dysfunction performed with TEG®-PM® and VerifyNow® devices correlated well with MEA [area under the curve (AUC) 0.78, 0.89, respectively]. However, MEA and VerifyNow® had superior AUCs compared with the TEG®-PM® percent inhibition AUC (both 0.90 vs 0.77). The adenosine diphosphate (ADP)-specific assays on these three devices did not correlate with APA use; however, the number of patients pre-treated with clopidogrel was small [286]. Trauma patients with normal platelet activity despite a positive history of APA intake (“non-responders”) or patients with high on-treatment platelet reactivity (HTPR) can also be identified using VerifyNow® [289, 290, 293, 297, 298]. In these patients, empiric administration of haemostatic substances would unnecessarily increase the risk of thrombotic events.

VerifyNow® [286, 289, 290, 293, 297, 298], MEA [283, 284, 286, 292, 294, 299,300,301] and TEG®-PM® [286, 287, 295, 302,303,304,305] can also be used to detect platelet dysfunction in trauma patients in the absence of APA intake. A coagulopathy POC panel consisting of r-TEG® and VN®-ASA successfully identified a subset of TBI patients with an occult coagulopathy that would otherwise have been missed [290]. Platelet dysfunction, as indicated by MEA, exhibits a temporal profile whereby MEA values are low initially and subsequently increase during the days following TBI [286, 289, 290, 292, 293, 297], similar to the changes observed perioperatively in elective hip arthroplasty [306]. Interestingly, both the ADP pathway and the thrombin receptor pathway measured using a thrombin receptor activating peptide (TRAP) test are significantly affected in trauma patients [301].

Distinguishing pharmacologic from trauma-induced platelet receptor hypofunction is not easy, as both conditions are associated with assay values below the reference interval. Moreover, diagnostic cut-offs for pathologic platelet dysfunction after traumatic injury have not been established. For example, ADP inhibition measured by TEG®-PM® was 42.5% in one study [287] and as high as 86% in another [285], compared with only 4% in healthy volunteers [287]. Over 75% of the TBI patients had impairment of the ADP pathway in one study [265] and the severity of brain injury appeared to correlate with ADP inhibition on TEG®-PM® (severe TBI 93.1%, mild TBI 56.5%, control 15.5%; p < 0.01) [302]. When TEG®-PM® and MEA were compared in severely injured trauma patients, results correlated poorly with the ADP pathway and moderately with the AA pathway [299].

The utility of POC platelet function assays to predict outcome or stratify trauma patients at a higher risk of bleeding who may benefit subsequently from transfusion is uncertain. By using a composite outcome, one study found no difference in bleeding complications in trauma patients on clopidogrel who presented with high or low platelet inhibition measured using VerifyNow® [298]. Similarly, progression of ICH and the need for neurosurgical intervention was independent of platelet activity assessed using VerifyNow® [307]. MEA values were also not predictive of haemorrhagic progression [292] or outcome [289, 292, 294] in some studies in trauma patients; however, 87% of patients received haemostatic therapy following detection of impaired platelet function, and this strategy could have influenced the results in one study [294]. In contrast, the MEA TRAP [283] and the AA receptor aspirin inhibition (ASPI) test [299] were independent predictors of mortality. In another study that included a mixed trauma population, which was not adjusted for confounders, ADP and TRAP values were also different between survivors and non-survivors [284]. Others have found ADP, but not the AA test, to be a predictor of mortality [303].

TEG®-PM® was found to be a superior indicator of haemorrhagic shock in trauma patients compared with MEA [299]. TEG®-PM® AA-induced platelet activity reduction identified TBI patients with a high risk of bleeding complications [304] and TEG®-PM® ADP-induced platelet activity reduction [285] or inhibition in both pathways [299] was predictive of blood product transfusion in severe trauma. Another study demonstrated that the MEA and VerifyNow® AA tests were not predictive of ICH, whereas the TEG®-PM® AA percent inhibition may be associated with ICH progression, with 71% specificity at 32% inhibition [286]. Studies reporting ADP receptor inhibition measured using TEG®-PM® also showed an association between this parameter and mortality [287] and significant correlations between the severity of TBI, the degree of ADP inhibition and increased risk of mortality [302,303,304]. In one study, platelet ADP inhibition exceeding 60% independently predicted in-hospital mortality amongst patients with TBI, while controlling for age, gender, the presence of hypotension, pre-injury APA, GCS and ISS [295]. In contrast, others found no correlation between TEG®-PM® values and ISS, length of hospital stay or mortality in trauma patients with or without TBI [305].

The role of POC platelet function devices in guiding haemostatic therapy is not established. One study showed no impact of platelet transfusion on platelet activity in patients with traumatic ICH with pre-injury aspirin treatment assessed using the VerifyNow® assay. There was also no difference in ICH progression or neurosurgical intervention in functional and non-functional platelet groups after platelet transfusion [307]. Further studies using the VerifyNow® assay showed that a single platelet apheresis unit was not sufficient to reverse platelet inhibition in almost half of patients [297] and a trend toward increased mortality in patients whose platelet function failed to normalise with transfusion [289]. A dose–response relationship between the quantity of platelets transfused and reversal of VN®-ASA inhibition was observed [289].

In contrast, haemostatic measures significantly increased AA-induced platelet activity measured using MEA by 100 ± 66% [291]. Others showed that platelet transfusion improved aspirin-induced platelet dysfunction but did not recover traumatic platelet dysfunction measured using MEA [308]. TEG®-PM® was also not supported as a solitary tool to guide platelet transfusions in trauma patients [287, 305]. It seems that although platelet transfusion may improve platelet function via AA receptor-mediated pathways, it has little, if any, impact on ADP receptor-mediated pathways [305]. Moreover, TBI patients who received platelet transfusion had significant reductions in the degree of platelet inhibition detected using the AA TEG®-PM® assay, but no change in outcomes [309].

The lack of congruency among the studies summarised above indicates that there is a pressing need for future prospective studies that investigate the potential benefit of platelet function monitoring in trauma patients. Although these devices are capable of measuring platelet receptor inhibition to detect pre-treatment with APA, their role in identifying trauma-induced platelet dysfunction and in guiding haemostatic therapy remains unclear and their use can only be recommended as an adjunct to standard laboratory monitoring.

III. Tissue oxygenation, volume, fluids and temperature

Tissue oxygenation

Recommendation 12

We recommend permissive hypotension with a target systolic blood pressure of 80–90 mmHg (mean arterial pressure 50–60 mmHg) until major bleeding has been stopped in the initial phase following trauma without brain injury. (Grade 1C)

In patients with severe TBI (GCS ≤ 8), we recommend that a mean arterial pressure ≥ 80 mmHg be maintained. (Grade 1C)

Restricted volume replacement

Recommendation 13

We recommend use of a restricted volume replacement strategy to achieve target blood pressure until bleeding can be controlled. (Grade 1B).

Vasopressors and inotropic agents

Recommendation 14

In the presence of life-threatening hypotension, we recommend administration of vasopressors in addition to fluids to maintain target arterial pressure. (Grade 1C)

We recommend infusion of an inotropic agent in the presence of myocardial dysfunction. (Grade 1C)

Rationale

At present, the initial treatment of trauma-induced hypotension uses the concept of DCR, with restricted volume replacement and permissive hypotension. Although the general effectiveness of such a restricted volume replacement, resulting in permissive hypotension, remains to be confirmed in RCTs, two studies published in the 1990s demonstrated increased survival when a low and delayed fluid volume resuscitation concept was used in penetrating [310] or penetrating and blunt [311] trauma. A further small pilot RCT published in 2015 demonstrated a 24-h survival benefit for hypotensive patients with blunt trauma initially treated with a restrictive volume administration when compared with standard volume replacement [312]. However, in contrast to these studies, no significant differences in survival were found in two non-randomised controlled trials examining patients with either penetrating and blunt trauma [313] or blunt trauma alone [314].

Moreover, future RCTs must also confirm whether the present more or less arbitrary recommendation for systolic and mean arterial blood pressures for permissive hypotension are safe for all trauma patients or whether target blood pressures should be different in specific subgroups, e.g. in blunt or penetrating trauma patients. Existing data already show that the concept of permissive hypotension should be carefully considered in the elderly patient [315] and may be contraindicated if the patient suffers from chronic arterial hypertension [316].

Nevertheless, the concept of DCR is supported by several retrospective studies demonstrating that aggressive resuscitation techniques, often initiated in the pre-hospital setting, may be detrimental for trauma patients [14, 34, 317,318,319,320,321,322,323,324,325]. It has been shown that aggressive volume administration increased the incidence of secondary abdominal compartment syndrome (ACS) [324], damage-control laparotomy [322], coagulopathy [14], multiple organ failure [323], nosocomial infections [323], the number of blood as well as mass transfusions [319, 323] and prolonged the length of ICU and hospital stays [323]. At the same time, increased volume administration decreased the likelihood of survivial [34, 320, 321, 323].

The timing and volume of i.v. fluid administration in bleeding trauma patients was assessed in a meta-analysis by Kwan et al. [326]. Three trials, including a total of 1957 patients, were identified that addressed the timing of administration, and three other studies investigated volume load, but included only 171 patients. In contrast to the retrospective analysis described above, the meta-analysis failed to demonstrate an advantage associated with delayed compared to early fluid administration, nor of smaller compared to larger volume fluid administration in this small group of prospective studies that included only a very limited number of patients. A further meta-analysis that assessed seven retrospective observational studies that included a total of 13,687 patients and three prospective studies that included 798 patients estimated a small benefit in favour of a restricted volume replacement strategy [327]. However, the authors cautioned that the available studies were subject to a high risk of selection bias and clinical heterogeneity.

It should be noted that DCR strategies using restrictive volume replacement affecting hypotensive blood pressure are contraindicated in patients with TBI and spinal injuries. This is because an adequate perfusion pressure is crucial to ensure tissue oxygenation of the injured central nervous system [328]. However, it remains unclear how to attain the best balance between volume resuscitation and vasopressor administration in order to achieve an adequate perfusion pressure. [315, 316]

In conclusion, a DCR strategy using a concept of restricted fluid replacement that aims to achieve a lower than normal systolic blood pressure of 80–90 mmHg in patients without TBI and/or spinal injury is supported by the literature. However, strong evidence from sufficiently robust RCTs is lacking.

Vasopressors may also be required transiently, even when fluid expansion is in progress and hypovolaemia has not yet been corrected, to sustain life and maintain tissue perfusion in the presence of life-threatening hypotension. Norepinephrine is commonly used to restore arterial pressure in septic and haemorrhagic shock and is now considered by many to be the agent of choice for this purpose during septic shock [329]. Although norepinephrine has some β-adrenergic effects, it acts predominantly as a vasoconstrictor. Arterial α-adrenergic stimulation increases arterial resistance and may increase cardiac afterload, while norepinephrine exerts both arterial and venous α-adrenergic stimulation [330]. Indeed, in addition to its arterial vasoconstrictor effect, norepinephrine induces venoconstriction at the level of the splanchnic circulation in particular, which increases the pressure in capacitance vessels and actively shifts splanchnic blood volume to the systemic circulation [331]. This venous adrenergic stimulation may to some extent recruit blood from the venous unstressed volume, thereby filling the blood vessels without generating intravascular pressure. Moreover, stimulation of β₂-adrenergic receptors decreases venous resistance and increases venous return [331]. Animal studies of uncontrolled haemorrhage have suggested that norepinephrine infusion reduces the amount of fluid resuscitation required to achieve a given arterial pressure target associated with lower blood loss and improved survival [332, 333].

Despite a general paucity of research into the use of vasopressors in hypotensive trauma patients, a double-blind randomised trial has assessed the safety and efficacy of adding vasopressin to resuscitative fluid. Patients were administered fluid alone or fluid plus vasopressin (bolus 4 IU) and i.v. infusion of 200 mL/h (vasopressin 2.4 IU/h) for 5 h. The fluid plus vasopressin group needed a significantly lower total resuscitation fluid volume over 5 days than the control group (p = 0.04). The rates of adverse events, organ dysfunction and 30-day mortality were similar [334].

An interim analysis performed during an ongoing multicentre prospective cohort study has suggested that the early use of vasopressors for haemodynamic support after haemorrhagic shock may be deleterious in comparison to aggressive volume resuscitation and should be used cautiously [335]. However, the study was limited in that it was a secondary analysis of a prospective cohort study and not designed to answer the specific hypothesis tested. Moreover, the group receiving vasopressors had a higher rate of thoracotomy. A second study retrospectively analysed the records from 225 patients who received different vasopressor therapies during emergency trauma surgery [336]. Whereas the use of epinephrine was independently associated with increased mortality, there was no difference in the mortality rate compared with other vasopressors. The most recent paper on vasopressor use in massively transfused trauma patients retrospectively analysed 120 trauma patients and described, not surprisingly, an association between the use of vasopressor and mortality [337]. However, on hospital arrival, the patients receiving a vasopressor had a much lower GCS and a higher lactate level and showed a trend toward the transfusion of more blood products.

In conclusion, the effects of vasopressors have not yet been rigorously investigated in humans during haemorrhagic shock and prospective studies to define the effect of vasopressors on patients during haemorrhagic shock are warranted. Current evidence suggests that vasopressors may be useful if used transiently to sustain arterial pressure and maintain tissue perfusion in the face of life-threatening hypotension. However, if used, it is essential to respect the recommended objectives for systolic arterial pressure (80–90 mmHg) in patients without TBI. Because vasopressors may increase cardiac afterload if the infusion rate is excessive or left ventricular function is already impaired, an assessment of cardiac function during the initial ultrasound examination is essential. Cardiac dysfunction could be altered in the trauma patient following cardiac contusion, pericardial effusion or secondary to brain injury with intracranial hypertension [338]. The presence of myocardial dysfunction requires treatment with an inotropic agent such as dobutamine or epinephrine. In the absence of an evaluation of cardiac function or cardiac output monitoring, as is often the case in the early phase of haemorrhagic shock management, cardiac dysfunction must be suspected if there is a poor response to fluid expansion and norepinephrine.

Type of fluid

Recommendation 15

We recommend that fluid therapy using isotonic crystalloid solutions be initiated in the hypotensive bleeding trauma patient. (Grade 1A)

We recommend the use of balanced electrolyte solutions and the avoidance of saline solutions. (Grade 1B)

We recommend that hypotonic solutions such as Ringer’s lactate be avoided in patients with severe head trauma. (Grade 1B)

We recommend that the use of colloids be restricted due to the adverse effects on haemostasis. (Grade 1C)

Rationale

It is widely accepted that during the initial phase of haemorrhagic trauma shock, a restrictive volume strategy be supported with crystalloid solutions. The main reason for this is that all colloid solutions can alter haemostasis. However, if the bleeding is excessive and if crystalloids in combination with vasopressors are not able to maintain basic tissue perfusion, colloid infusions represent a further, however controversial, option to restore perfusion. If a colloid solution is administered, it is still unclear which colloid solution should be used in the initial treatment of the bleeding trauma patient.

In most trauma studies, 0.9% sodium chloride was used as the crystalloid solution. However, at least seven studies in both non-critically and critically ill patients suggest that the use of this crystalloid solution as the main i.v. fluid source results in harm to patients, e.g. reduced renal blood flow velocity and renal cortical tissue perfusion, hyperchloraemic acidosis, increased incidence of kidney injury or even reduced survival [339,340,341,342,343,344,345,346,347]. In contrast to 0.9% sodium chloride, balanced electrolyte solutions comprise physiological or near-physiological concentrations of chloride and may therefore be advantageous. Similarly, in a retrospective analysis of ICU patients receiving more than 60 mL/kg 0.9% sodium solution over a 24 h period, each 100 mEq increase in chloride load was associated with a 5.5% increase in the risk of death, even after controlling for total fluid volume, age, and severity (p = 0.0015) over a 1-year period [344]. The two most recent RCTs comparing balanced crystalloids vs 0.9% sodium chloride, one including 13,347 non-critically ill adults [342] and the other including 15,802 critically ill patients [343], found variation in the negative side-effects of 0.9% sodium chloride, depending on the health status, and thereby on the physiological ability of each patient to compensate. In one of the studies, non-critically ill patients receiving balanced crystalloid solutions compared with saline were shown to have a lower incidence of major kidney-related adverse events within 30 days, without an influence on the length of hospital stay [342]. In the other study, critically ill patients receiving balanced crystalloid solutions compared with saline were shown to have a lower rate of composite outcome death from any cause, new renal replacement therapy or persistent renal dysfunction [343]. In a small prospective randomised trial involving 46 trauma patients, a balanced electrolyte solution improved acid-base status and resulted in less hyperchloraemia at 24 h post-injury compared with 0.9% sodium chloride [346]. Moreover, a secondary analysis demonstrated that the use of balanced electrolyte solutions resulted in a net cost benefit in comparison to the use of 0.9% saline chloride [345]. On the other hand, another recently published study could not exclude the possibility that an acetate-based balanced crystalloid solution increased patient bleeding during cardiac surgery, which warrants further investigation [348]. In conclusion, for critically ill patients such as trauma patients, a balanced electrolyte solution should be favoured over 0.9% sodium chloride, and if a 0.9% sodium chloride solution is used, it should be limited to a maximum of 1–1.5 L.

Hypotonic crystalloid solutions, such as Ringer’s lactate, should be avoided in patients with TBI in order to minimise a fluid shift into the damaged cerebral tissue. A secondary analysis from the Prospective, Observational, Multicenter, Major Trauma Transfusion (PROMMTT) study revealed that Ringer’s lactate solutions were associated with higher adjusted mortality compared with normal saline (HR 1.78; CI 1.04–3.04; p = 0.035) [349].

A recent study has suggested that solutions with the potential to restore pH may also be advantageous. It was shown that Ringer’s acetate solution ameliorated splanchnic dysoxia more rapidly, as evidenced by gastric tonometry, than Ringer’s lactate [350]. Whether there are benefits associated with the use of certain isotonic balanced crystalloids with respect to a reduced morbidity or mortality, however, is not clear and remains to be evaluated [339, 344, 347].

Colloid solutions have been used more effectively to restore intravascular volume, as would be expected from basic physiologic concept of fluid exchange across the vasculature. A review of RCTs indicated that colloid solutions can result in lower fluid requirements than crystalloids in all types of patient, including trauma victims, with a ratio of 1.5:1 [351]. A large pragmatic study prospectively comparing colloids to crystalloids reported the same 1.5:1 ratio [352].

Particularly in situations in which there is a need for rapid volume replacement due to severe shock, colloids have often been administered. However, it is still unclear whether colloids really have a beneficial effect on morbidity or mortality. The most recent meta-analysis comparing colloids or crystalloids failed to demonstrate that any colloid reduces morbidity or mortality compared to resuscitation with crystalloids in critically ill or elective surgical patients [353, 354]. The authors concluded that there is no evidence that resuscitation with colloids has any beneficial effect on survival [355]. However, neither the time point of fluid resuscitation nor the duration and dosages of fluid resuscitation have been analysed or openly discussed. Nevertheless, at the present time, good data are lacking to demonstrate the survival benefit of colloids compared with other types of solutions.

Conflicting meta-analyses have shown increased kidney injury and increased mortality in critically ill patients treated with hydroxyethyl starch (HES) solutions [355,356,357]. On the other hand, it has also been shown that there is no difference in the incidence of death or acute kidney failure in surgical patients receiving HES solutions [358]. It seems doubtful that any conclusions can be drawn from these studies, which were performed mostly under different conditions than are present in the acute hypovolaemic trauma patient. In addition to these conflicting results, an in vitro study using blood from healthy volunteers demonstrated that coagulation and platelet function are impaired by all HES and gelatine solutions [359]. However, gelatine-induced coagulopathy was reversible with the administration of fibrinogen, whereas HES-induced coagulopathy was not. So far, only one small RCT described a benefit for a HES solution in trauma patients. HES (130/0.4) provided a significantly more rapid decline in blood lactate levels and less renal injury than saline solution in penetrating trauma patients [360]. However, because only 42 blunt trauma patients were included in the study, no differences in these parameters could be demonstrated using the different solutions. At present, other colloids, including gelatine solutions, cannot be recommended without restrictions [361].

A number of studies have investigated hypertonic solutions. In 2008, a double-blinded RCT in 209 patients with blunt traumatic injuries analysed the effect of treatment with 250 mL 7.5% hypertonic saline and 6% dextran 70 compared to lactated Ringer’s solution on organ failure [362]. The intention-to-treat analysis demonstrated no significant difference in organ failure and in ARDS-free survival. However, there was improved ARDS-free survival in the subset (19% of the population) requiring 10 U or more of packed RBC [362]. A relatively small clinical trial involving nine patients with intracranial pressure > 20 mmHg found that hypertonic saline reduced intracranial pressure more effectively than dextran solutions with 20% mannitol when compared in equimolar dosing [363]. However, Cooper et al. found almost no difference in neurological function 6 months after TBI in 229 patients who had received pre-hospital hypertonic saline resuscitation compared to conventional fluid [364]. Moreover, two large prospective randomised multicentre studies reported by Bulger and co-workers [365, 366] analysed the effect of out-of-hospital administration of hypertonic fluids on neurological outcome following severe TBI and survival after traumatic hypovolaemic shock. These studies were not able to demonstrate any advantage compared to normal 0.9% saline among the 2184 patients included. In contrast, a recent retrospective analysis in 34 trauma patients demonstrated that hypertonic solutions interfere with coagulation [367]. Two recently published meta-analyses, one including nine trials with 3490 trauma patients and one including 12 trials including 2932 haemorrhagic shock patients, confirmed that there is no beneficial effect of hypertonic saline with or without dextran in general trauma patients [368, 369].

In conclusion, at least during the initial treatment phase and as part of the restricted volume replacement strategy, administration of crystalloids is advocated. The data published to date demonstrate that balanced crystalloid solutions are preferable to 0.9% saline solution, especially if administered in larger amounts. In patients with TBI, hypotonic solutions, crystalloids as well as colloids, should be avoided. If small-volume resuscitation fails to restore the target blood pressure in spite of additional use of norepinephrine, or if extensive volume resuscitation is necessary in the intra-hospital phase of initial trauma management, this can be achieved either with large-volume balanced crystalloid administration or with colloids. Large-volume balanced crystalloid solutions are not independently associated with multiple organ failure [370]. In contrast, a retrospective study showed that resuscitation with at least 1 L crystalloid per unit RBC seems to be associated with reduced overall mortality [371]. However, at present, it is not clear whether colloids should be used if crystalloids fail to restore target blood pressure. Hypertonic saline solutions do not demonstrate any advantage to other less expensive crystalloids. The evidence suggests that hypertonic saline solutions are safe, but will neither improve survival nor improve neurological outcome after TBI.

Erythrocytes

Recommendation 16

We recommend a target Hb of 70 to 90 g/L. (Grade 1C)

Rationale

Oxygen delivery to tissues is the product of blood flow and arterial oxygen content, which is directly related to the Hb concentration; therefore, decreasing Hb might be expected to increase the risk of tissue hypoxia. However, compensatory responses to acute normovolaemic anaemia occur, including macro- and microcirculatory changes in blood flow and capillary recruitment, so the consequences of low Hb in terms of tissue oxygenation are difficult to predict based on macrocirculatory haemodynamic parameters and Hb levels. This has been demonstrated in haemorrhagic shock patients, in whom RBC transfusion was able to improve microcirculation and tissue oxygenation independent of macrocirculation and Hb level [372, 373]. However, the transfusion of RBCs containing methaemoglobin and thus not participating in oxygen delivery also improved microcirculation [372], most likely due to increased blood viscosity [374].

Erythrocytes are oxygen sensors and modulators of vascular tone and microcirculation. Erythrocytes play a fundamental role in matching microvascular oxygen supply with local tissue oxygen demand. Although a number of theories to explain this critical function have been proposed [transport of nitric oxide (NO) in the form of S-nitrosothiol by erythrocyte, deoxyhaemoglobin acting as a nitrite reductase converting nitrite to NO and release of adenosine triphosphate (ATP) from the erythrocyte, resulting in the production of mediators], none has been either universally accepted or fully tested in the intact microcirculation [375]. In addition, erythrocytes may contribute to haemostasis by influencing the biochemical and functional responsiveness of activated platelets through the rheological effect on platelet margination and by supporting thrombin generation [376].

The effects of the Hct level on blood coagulation have not been fully elucidated [377]. An acute reduction of the Hct level results in an increase in the bleeding time [378], with restoration upon re-transfusion [379]. This may relate to the presence of the enzyme elastase on the surface of RBC membranes, which may activate coagulation factor IX [380, 381]. However, an animal model showed that a moderate reduction in Hct level does not increase blood loss from a standard spleen injury [379], and an isolated in vitro reduction of the Hct level did not compromise blood coagulation as assessed using TEG® [382].

RCTs that have evaluated Hb thresholds for transfusion in critically ill patients have consistently found that restrictive transfusion strategies (Hb thresholds between 70 and 90 g/L) are as safe as, or safer than, liberal strategies (thresholds ≥ 90 g/L) [383,384,385,386,387] with the possible exception of patients with acute coronary syndrome. Recently, in high-risk patients undergoing cardiac surgery, a restrictive strategy regarding red cell transfusion was non-inferior to a liberal strategy with respect to the composite outcome of death from any cause, myocardial infarction, stroke or new-onset renal failure with dialysis, with fewer RBCs transfused [388]. These studies excluded patients with massive bleeding and no prospective RCT has compared restrictive and liberal transfusion regimens in trauma patients. A subset of 203 trauma patients from the Transfusion Requirements in Critical Care (TRICC) trial [384] was re-analysed [389]. A restrictive transfusion regimen (Hb transfusion trigger < 70.0 g/L) resulted in fewer transfusions compared with the liberal transfusion regimen (Hb transfusion trigger < 100 g/L) and appeared to be safe. However, no statistically significant benefit in terms of multiple organ failure or post-traumatic infections was observed. It should be emphasised that this study was neither designed nor powered to answer these questions with precision. In addition, it cannot be ruled out that the number of RBC units transfused merely reflects the severity of injury. Nevertheless, RBC transfusions have been shown in multiple studies to be associated with increased mortality [390,391,392,393,394], lung injury [391, 395, 396], increased infection rates [397, 398] and renal failure in trauma victims [394].

Because anaemia is a possible cause of secondary ischaemic damage, concerns have been raised about the safety of restrictive transfusion strategies in the subpopulation of patients with TBI. Most early clinical information comes from retrospective observational studies with important methodological limitations. These data have yielded inconsistent results on the effects of RBC transfusion on markers of cerebral perfusion and metabolism in patients with isolated severe TBI. Two systematic reviews published in 2012 stressed the lack of high-level scientific evidence for a specific Hb transfusion trigger in this setting [399, 400]. A retrospective review of data collected prospectively in 1158 patients with a GCS ≤ 8 in the absence of haemorrhagic shock found that RBC transfusion was associated with worse outcomes (28-day survival, ARDS-free survival, 6-month neurological outcome) when the initial Hb was > 100 g/L [401]. No relationship between RBC transfusion and outcomes was found in patients with an initial Hb ≤ 100 g/L [401]. In a RCT of 200 patients with TBI at two clinical sites, Robertson et al. compared two Hb transfusion thresholds (70 or 100 g/L), and separately compared administration of erythropoietin or placebo [402]. Patients were enrolled within 6 h of injury and 99 patients were assigned to the 70 g/L transfusion threshold and 101 patients to the 100 g/L threshold. Neither the administration of erythropoietin nor maintenance of Hb concentration > 100 g/L resulted in improved neurological outcome at 6 months, and the 100 g/L threshold was associated with a higher incidence of adverse events [402].

Alternative methods of increasing Hb have been studied. The erythropoietic response is blunted in trauma patients [403]; therefore, the administration of erythropoietin appears an attractive option. In a first prospective randomised trial in ICU patients (n = 1302, 48% being trauma patients), a significant reduction in RBC transfusion percentage from 60.4 to 50.5% (p < 0.001) and reduction in the median number of RBC units transfused from two to one (p < 0.001) was observed [404]. In the subgroup of trauma patients, 28-day mortality was also reduced [odds ratio (OR) 0.43 (0.23–0.81)] [404]. In a subsequent prospective randomised trial in ICU patients (n = 1460, 54% being trauma patients), no significant reduction in RBC transfusions was observed [405]. Thrombotic complications were higher in erythropoietin-treated patients [HR 1.58 (1.09 to 2.28)]; however, this difference was observed exclusively in patients without heparin prophylaxis [405]. Recently, in a double-blind, placebo-controlled trial undertaken in 29 centres within 24 h of moderate or severe TBI, 606 patients were randomly assigned to receive erythropoietin (40,000 units subcutaneously) or placebo once per week for a maximum of three doses. Erythropoietin did not reduce the number of patients with severe neurological dysfunction (GOS-E level 1–4), the transfusion of RBC or increase the incidence of deep venous thrombosis (DVT) of the lower limbs [406]. Mortality at 6 months tended to be lower in patients treated with erythropoietin (11%) than in control patients with a mortality of 16% (RR 0.68; 95% CI 0·44–1·03; p = 0.07) [406]. Interestingly, erythropoietin treatment of critically ill trauma patients resulted in a substantial reduction of mortality (RR 0.63; 0.49–0.79, p = 0.0001) in a recent meta-analysis [407].

The limited effect of erythropoietin treatment on transfusion needs may be surprising given the blunted response in trauma patients [403]. However, iron metabolism is also altered after trauma, with iron not being fully available for haematopoiesis [403]. Neither iron metabolism nor availability are fully understood following traumatic injury and complicated by the fact that certain proteins such as ferritin are massively upregulated after trauma as part of the acute phase response [403]. Intravenous iron may therefore represent another attractive option with which to foster haematopoiesis. Indeed, studies that assess the effect of i.v. iron (with [408, 409] or without [410] concomitant epoetin alpha) showed reduced RBC transfusions [408,409,410], postoperative infections [409, 410], length of hospital stay [409] and mortality in patients with hip fractures [409]. While i.v. iron appears to be promising, oral iron is largely ineffective. However, a recent multicentre, randomised, double-blind, trial during the perioperative period of hip fracture did not find that ferric carboxymaltose with or without erythropoietin induced a reduction of RBC transfusion despite obtaining significant increases in Hb levels at discharge and 60 days after discharge [411]. In a randomised, placebo-controlled, blinded study in anaemic intensive care patients, early administration of low-dose i.v. ferric carboxymaltose, compared with placebo, did not result in a significant lowering of RBC transfusion requirements during hospital stay [412]. Patients who received i.v. iron had a significantly higher Hb concentration at hospital [412].

Temperature management

Recommendation 17

In order to optimise coagulation, we recommend early application of measures to reduce heat loss and warm the hypothermic patient to achieve and maintain normothermia. (Grade 1C)

Rationale

Since coagulopathy following traumatic injury increases mortality [39], it is recommended to target “normothermia”, with a core temperature between 36 and 37 °C in order to create optimal preconditions for coagulation. Hypothermia, a core body temperature <  35 °C, is associated with acidosis, hypotension and coagulopathy in severely affected patients. The effects of hypothermia include altered platelet function, impaired coagulation factor function (a 1 °C drop in temperature is associated with a 10% drop in function), enzyme inhibition and fibrinolysis [413,414,415]. Body temperatures below 34 °C compromise blood coagulation, but this has only been observed when coagulation tests (PT and APTT) are performed at the low temperatures present in patients with hypothermia, but not when assessed at 37 °C, as is routine practice for such laboratory tests.

The profound clinical effects of hypothermia ultimately lead to higher morbidity and mortality [416], and hypothermic patients require more blood products [417]. In a retrospective study of 604 trauma patients who required massive transfusion, a logistic regression analysis demonstrated that a temperature lower than 34 °C was associated with a greater independent risk of mortality of more than 80% after controlling for differences in shock, coagulopathy, injury severity and transfusion requirements [OR 1.87; 95% CI 1.18–3.0; p = 0.007] [418]. A recent study performed a secondary analysis using 10 years of data from the Pennsylvania Trauma Outcome Study (PTOS). It analysed 11,033 patients with severe TBI and demonstrated that spontaneous hypothermia at hospital admission was associated with a significant increase in the risk of mortality [419]. Steps to prevent hypothermia and the risk of hypothermia-induced coagulopathy include removing wet clothing, covering the patient to avoid additional heat loss, increasing the ambient temperature, forced air warming, warm fluid therapy, and, in extreme cases, extracorporeal re-warming devices [420,421,422]. Recently, the use of a hypothermia prevention and management kit has been advocated [423]. This kit is a low-cost, lightweight, low-volume commercial product that sustains 10 h of continuous dry heat with an oxygen-activated, self-heating liner and provides thermal insulation due to the multi-layer composite construction of the outer shell. The kit was designed to prevent hypothermia during tactical casualty evacuation; however, application in the civilian sector for the active re-warming of trauma patients is conceivable. In comparison to other methods and devices, the hypothermia prevention and management kit achieved and maintained significantly higher temperatures than all other methods and controls at 120 min [424].

IV. Rapid control of bleeding

Damage-control surgery

Recommendation 18

We recommend that damage-control surgery be employed in the severely injured patient presenting with deep haemorrhagic shock, signs of ongoing bleeding and coagulopathy. (Grade 1B)

Other factors that should trigger a damage-control approach are hypothermia, acidosis, inaccessible major anatomic injury, a need for time-consuming procedures or concomitant major injury outside the abdomen. (Grade 1C)

We recommend primary definitive surgical management in the haemodynamically stable patient and in the absence of any of the factors above. (Grade 1C)

Rationale

The severely injured patient arriving at the hospital with continuing bleeding or deep haemorrhagic shock generally has a poor chance of survival without early control of bleeding, proper resuscitation and blood transfusion. This is particularly true for patients who present with uncontrolled bleeding due to multiple penetrating injuries or patients with major abdominal injury and unstable pelvic fractures with bleeding from fracture sites and retroperitoneal vessels. The final common pathway in these patients is the exhaustion of physiological reserves, with resulting profound acidosis, hypothermia and coagulopathy, also known as the “bloody vicious cycle” or “lethal triad”.

In 1983, Stone et al. described the techniques of abbreviated laparotomy, packing to control haemorrhage and of deferred definitive surgical repair until coagulation has been established [425]. Several articles have since described the beneficial results of this approach, now referred to as “damage control” [426,427,428]. This approach should be considered in patients with major abdominal injury and a need for adjunctive angioembolisation, major abdominal injury and a need to evaluate other injuries as early as possible, major abdominal injury and traumatic amputation of a limb. Factors that should trigger a damage-control approach in the operating theatre are temperature ≤ 34 °C, pH ≤ 7.2, an inaccessible major venous injury, a need for time-consuming procedures in a patient with suboptimal response to resuscitation or inability to achieve haemostasis due to recalcitrant coagulopathy [429, 430].

Damage-control surgery of the abdomen consists of three components: the first component is an abbreviated resuscitative laparotomy for control of bleeding, the restitution of blood flow where necessary and the control of contamination. This should be achieved as rapidly as possible without spending unnecessary time on traditional organ repairs that can be deferred to a later phase. The abdomen is packed and temporary abdominal closure is performed. Packing aims to compress liver ruptures or exert direct pressure on the sources of bleeding and abdominal packing may permit further attempts to achieve total haemostasis through angiography and/or correction of the “lethal triad”. The removal of packs should preferably be deferred for at least 48 h to lower the risk of re-bleeding.

The second component of damage-control surgery is intensive care treatment, focused on core re-warming, correction of the acid-base imbalance and coagulopathy, as well as optimising the ventilation and the haemodynamic status. If complementary angiography and/or further injury investigation is needed, it should be performed during this phase.

The third component is the definitive surgical repair that is performed only when target parameters have been achieved [133, 426,427,428, 431,432,433]. Although the concept of “damage control” intuitively makes sense, no RCTs exist to support it. Retrospective studies support the concept showing reduced morbidity and mortality rates in selective populations [428].

The same “damage control” principles have been applied to orthopaedic injuries in severely injured patients. Scalea et al. were the first to coin the term “damage control orthopaedics” [434]. Relevant fractures are primarily stabilised with external fixators rather than primary definitive osteosynthesis [434,435,436]. The less traumatic nature and shorter duration of the surgical procedure aims to reduce the secondary procedure-related trauma. Definitive osteosynthesis surgery can be performed after 4–14 days when the patient has recovered sufficiently. Retrospective clinical studies and prospective cohort studies seem to support the concept of damage control. The only available randomised study shows an advantage for this strategy in “borderline” patients [436]. The damage-control concept has also been described for thoracic and neurosurgery [437, 438]. In addition to damage-control surgical approaches, damage-control anaesthesia or resuscitation comprises a number of important measures described in the other recommendations within this document.

Pelvic ring closure and stabilisation

Recommendation 19

We recommend that patients with pelvic ring disruption in haemorrhagic shock undergo immediate pelvic ring closure and stabilisation. (Grade 1B)

Packing, embolisation and surgery

Recommendation 20

We recommend that patients with ongoing haemodynamic instability, despite adequate pelvic ring stabilisation, receive early surgical bleeding control and/or pre-peritoneal packing and/or angiographic embolisation. (Grade 1B)

We suggest that the use of aortic balloon occlusion be considered only under extreme circumstances in patients with pelvic fracture in order to gain time until appropriate bleeding control measures can be implemented. (Grade 2C)

Rationale

The mortality rate for patients with severe pelvic ring disruptions and haemodynamic instability remains high [439, 440]. There is no consensus as to the optimal treatment paradigm for patients presenting with haemorrhage from severe pelvic fractures. Angioembolisation and an external fixator are the most common approaches. REBOA is considered by some practitioners to be an important adjunct in the treatment of patients with severe pelvic fracture and in shock. However, this method is still in the early stages of development and is not currently used widely across trauma centres [441]. Pelvic ring injuries are associated with a high mortality rate within the first 24 h, due mainly to exsanguinations. Injured patients are managed using a multidisciplinary damage-control strategy. Unstable patients should undergo surgical haemostasis control immediately. Arterial embolisation is an effective means of achieving this and justifies the permanent availability of this approach in level-1 trauma centres. Following CT assessment of injuries, stable patients can undergo arterial embolisation if active arterial bleeding or vascular damage is present. The selective or nonselective embolisation methods and agents used depend on the patient’s haemodynamic status and an assessment of the injury whenever possible [442]. The early detection of these injuries and initial efforts to reduce disruption and stabilise the pelvis as well as containing bleeding is therefore crucial.

Severe pelvic trauma is a particularly challenging condition, requiring a multidisciplinary trauma team, including a general surgeon, orthopaedic surgeon, endovascular surgeon/interventional radiologist. The pelvis can harbour a multifocal haemorrhage that is not easily compressible or managed using traditional surgical methods such as tying off a blood vessel or removing an organ. Treatment often requires a triage of multiple investigations that can lead to the re-approximation of bony structures, including DCR, assessment of associated injuries and multimodal haemorrhage control via external fixation, pre-peritoneal packing, angioembolisation and/or REBOA, for example [71]. Markers of pelvic haemorrhage include anterior-posterior and vertical shear deformations on standard roentgenograms, CT “blush” (active arterial extravasation), bladder compression pressure, pelvic haematoma evident using CT and ongoing haemodynamic instability, despite adequate fracture stabilisation [443,444,445].

If the patient is haemodynamically unstable and in haemorrhagic shock, the urgent treatment goal is rapid achievement of haemostasis. An initial strategy, performed while DCR is ongoing and before proceeding to arteriography, relies on the insertion of an intra-aortic occlusion balloon and/or extra-peritoneal pelvic packing. If haemodynamic instability persists, a laparotomy for haemostasis should be performed without delay. In a haemodynamically stable patient, contrast-enhanced systematic CT is required to obtain a comprehensive assessment of the lesions prior to surgery [80].

The initial therapy for pelvic fractures includes control of venous and/or cancellous bone bleeding by pelvic closure as a first step [446]. Some institutions use primarily external fixators to control haemorrhage from pelvic fractures [443], but pelvic closure may also be achieved using a pelvic binder, a pelvic C-clamp or improvised methods such as a bed sheet [446, 447]. Based on the available literature, pelvic circumferential compression devices are widely used in the initial management of patients with suspected pelvic bleeding. There is evidence to suggest that external compression reduces disrupted pelvic rings. However, some complications have been reported following the application of pelvic circumferential compression devices. Until this can be clarified, judicious application of pelvic circumferential compression devices will continue to be recommended [448]. In addition to the pelvic closure, fracture stabilisation and the tamponade effect of the haematoma, pre-, extra- or retroperitoneal packing may reduce or control the venous bleeding [449,450,451]. Pre-peritoneal packing is used to decrease the need for pelvic embolisation and may be performed simultaneously, or soon after, initial pelvic fracture stabilisation. The most commonly embolised vascular bed and therefore the most studied is the pelvis [452]. Pelvic packing could potentially aid in early intra-pelvic bleeding control and provide crucial time for more selective haemorrhage management [449, 451].

Delayed interventions are common in damage-control laparotomy, with abdominal interventions often spread over multiple explorations. In such cases, mortality has been shown to increase in patients undergoing emergent re-exploration, or to delay the repair of major vascular injuries. Ideal treatment of damage-control laparotomy patients may include addressing injuries more completely at the first laparotomy instead of deferring care for other priorities [453].

REBOA has been used in patients with end-stage shock following blunt and penetrating trauma, together with embolisation of the vascular bed in the pelvis. In a military setting with hand-held ultrasound, it has been reported that 7 Fr femoral sheath access ER-REBOA® were positioned and inflated in the aorta without radiography. In all reported cases, REBOA resulted in immediate normalisation of blood pressure and permitted induction of anaesthesia, initiation of whole-blood transfusion, damage-control laparotomy and attainment of surgical haemostasis (range of inflation time 18–65 min). No access- or REBOA-related complications were reported, and all patients survived to achieve transport to the next echelon of care in stable condition. It has been suggested that the use of this device by non-surgeons and surgeons not specially trained in vascular surgery in the non-hospital setting may be useful as a stabilising and damage-control adjunct, allowing time for resuscitation, laparotomy and surgical haemostasis [454]. However, some authors, such as Maruhashi et al. [455], advise the use of REBOA with caution on the basis that it may increase the bleeding of minor thoracic injury in severe multiple trauma patients.

In the case of major pelvic injury, it is nevertheless agreed that damage-control interventional radiology and urgent resuscitative surgery should be initiated early and simultaneously [456]. Adjunct techniques can be combined with a consecutive laparotomy if deemed necessary [451]. This may decrease the high mortality rate observed in patients with major pelvic injuries who have undergone laparotomy as the primary intervention. However, non-therapeutic laparotomy should be avoided [457]. Time to pelvic embolisation for haemodynamically unstable pelvic fractures may impact survival [439, 458].

Angiography and embolisation are currently accepted as highly effective means with which to control arterial bleeding that cannot be controlled by fracture stabilisation [73, 443, 447, 449, 457, 459, 460]. Radiological management can also be usefully applied to abdominal and thoracic bleeding [461,462,463,464,465]. Martinelli et al. [466] reported the use of intra-aortic balloon occlusion to reduce bleeding and permit transport to an angiography theatre. In contrast, Morozumi et al. suggested the use of mobile digital subtraction angiography for arterial embolisation performed in the clinic by trauma surgeons [467]. A number of authors argue that permissive hypotension could achieve better survival by achieving pelvic stabilisation and/or angiography through DCR, hypertonic solutions and controlled hypothermia. Institutional differences in the capacity to perform timely angiography and embolisation may explain the different treatment algorithms suggested by many authors. Reports on transcatheter angiographic embolisation suggest a 100% higher mortality during off-hours due to lack of radiological service [468]. Therefore, a multidisciplinary approach to these severe injuries is required.

Local haemostatic measures

Recommendation 21

We recommend the use of topical haemostatic agents in combination with other surgical measures or with packing for venous or moderate arterial bleeding associated with parenchymal injuries. (Grade 1B)

Rationale

A wide range of local haemostatic agents is currently available for use as adjuncts to traditional surgical techniques to obtain haemorrhagic control. These topical agents can be particularly useful when access to the site of bleeding is difficult. Local haemostatic agents include collagen, gelatine or cellulose-based products, fibrin and synthetic glues or adhesives that can be used for both external and internal bleeding while polysaccharide-based and inorganic haemostatics are still mainly used and approved for external bleeding.

The use of topical haemostatic agents should consider several factors, such as the type of surgical procedure, cost, bleeding severity, coagulation status and each agent’s specific characteristics. Some of these agents should be avoided when auto-transfusion is applied, and several other contraindications need to be considered [469, 470]. The capacity of each agent to control bleeding was initially studied in animals, but increasing experience in humans is now available [469,470,471,472,473,474,475,476,477,478,479,480,481,482,483,484].

The different types of local haemostatic agents are briefly presented according to their basis and haemostatic capacity.

• Collagen-based agents trigger platelet aggregation, resulting in clot formation when in contact with a bleeding surface. They are often combined with a procoagulant substance such as thrombin to enhance the haemostatic effect. A positive haemostatic effect has been shown in several human studies [471, 479, 480, 485].

• Gelatine-based products can be used alone or in combination with a procoagulant substance [470]. Swelling of the gelatine in contact with blood reduces the blood flow and, in combination with a thrombin-based component, enhances haemostasis [476, 477, 482]. These products have been successfully used for local bleeding control in brain or thyroid surgery when electrocautery may cause damage to nerves [481] or to control bleeding from irregular surfaces such as during post-sinus surgery [484].

• Absorbable cellulose-based haemostatic agents have been widely used to treat bleeding for many years, and case reports as well as a prospective observational human study support their effectiveness [483]. The oxidised cellulose-based product can be impregnated with polyethylene glycol and other salts and achieve comparable and more rapid haemostasis compared to the combined products described below [475].

• Fibrin and synthetic glues or adhesives have both haemostatic and sealant properties, and their significant effect on haemostasis has been shown in several randomised controlled human studies involving vascular, bone, skin and visceral surgery [472, 474, 478].

• Polysaccharide-based haemostatics can be divided into two broad categories [470]: N-acetyl-glucosamine-containing glycosaminoglycans purified from microalgae and diatoms, and microporous polysaccharide haemospheres produced from potato starch. Some minerals, such as kaolin, also seem to have haemostatic effects. The mechanism of action is complex and depends on the purity or combination with other substances such as cellulose or fibrin. A number of different products in the form of pads, patches or bandages are currently available and have been shown to be efficient for external use and for splanchnic bleeding. Observational studies have shown that haemorrhage control is achieved using a poly-N-acetyl glucosamine- or kaolin-based bandage applied to patients with severe hepatic and abdominal injuries, acidosis and clinical coagulopathy [480, 486].

V. Initial management of bleeding and coagulopathy

Antifibrinolytic agents

Recommendation 22

We recommend that TXA be administered to the trauma patient who is bleeding or at risk of significant haemorrhage as soon as possible and within 3 h after injury at a loading dose of 1 g infused over 10 min, followed by an i.v. infusion of 1 g over 8 h. (Grade 1A)

We recommend that protocols for the management of bleeding patients consider administration of the first dose of TXA en route to the hospital. (Grade 1C)

We recommend that the administration of TXA not await results from a viscoelastic assessment. (Grade 1B)

Rationale

Tranexamic acid (trans-4-aminomethyl cyclohexane-1-carboxylic acid, TXA) is a synthetic lysine analogue that is a competitive inhibitor of plasminogen. TXA is distributed throughout all tissues, and the plasma half-life is 120 min [487]. The Clinical Randomisation of Antifibrinolytic therapy in Significant Haemorrhage (CRASH-2) trial [488] assessed the effects of early administration of a short course of TXA on death, vascular occlusive events and the administration of blood product transfusion to trauma patients who were bleeding or at risk of significant bleeding. The trial randomised 20,211 adult trauma patients with or at risk of significant bleeding to either TXA (loading dose 1 g over 10 min followed by infusion of 1 g over 8 h) or matching placebo within 8 h of injury. The primary outcome was in-hospital death within 4 weeks of injury. All analyses assessed the intention-to-treat population. All-cause mortality was significantly reduced with TXA by 1.5%; the risk of death due to bleeding was significantly reduced by 0.8% and a reduction in bleeding deaths by one third, mainly through preventing exsanguination within the first 24 h [489, 490]. Paediatric patients were not included in the CRASH-2 study; however, an ongoing study is investigating the use of TXA in children [491] and a study in children undergoing craniosynostosis surgery [492, 493] administered an initial bolus of 15–30 mg/kg followed by 2−10 mg/kg/h. One retrospective study has suggested that TXA is of no benefit in patients with viscoelastic hyperfibrinolysis [494] and another found TXA to reduce multiple organ failure and mortality in severely injured shocked patients [495]. This discrepancy is probably attributable to methodological limitations.

The risk of thrombosis after the use of the lysine analogues TXA and ε-aminocaproic acid had been of major theoretical concern; however, CRASH-2 showed that the rate of VTE was not altered, while arterial thromboses, especially myocardial infarction, were lower with the use of TXA. TXA use to prevent or manage haemorrhage has been studied in approximately one million patients without increased rates of thrombosis [496,497,498,499].

No adverse events were described with the use of TXA in CRASH-2, although an increased rate of seizures has been described in patients undergoing cardiac surgery receiving considerably higher doses of TXA than recommended here [500]. This may reflect the role of fibrinolytic molecules as neurotransmitters.

An unplanned subgroup analysis of the CRASH-2 data [501] showed that early treatment (≤ 1 h from injury) significantly reduced the risk of death due to bleeding by 2.5%. Treatment administered between 1 and 3 h also reduced the risk of death due to bleeding by 1.3%. Treatment given after 3 h increased the risk of death due to bleeding by 1.3%; therefore, we recommend that TXA be administered within 3 h following injury. Further data from over 20,000 patients randomised to TXA versus placebo in post-partum haemorrhage has also shown that benefit is most apparent within the first 3 h. Gayet-Ageron et al. showed that the benefits of TXA were more marked when given as soon as possible after injury and its efficacy decreased by 10% every 15 min from time of injury [502].

In order to ensure that TXA is administered early, TXA administration at the pre-hospital site of injury needs to be planned, and we recommend that protocols for the management of bleeding patients strongly consider administration of the first dose of TXA at the site of injury. El-Menyar et al. looked at the efficacy of pre-hospital TXA in a meta-analysis and showed that it reduced 24-h and 30-day mortality and thromboembolic events. However, the authors found only two studies and concluded that further RCTs are required [503]. This is in keeping with a recent study demonstrating the benefit of on-scene administration of TXA in patients with major trauma. In this study, the normally occurring deterioration of the coagulation status between the site of injury and hospital admission was clearly mitigated by the on-scene TXA administration [504].

If TXA is restricted to massive transfusion protocols or only used in patients clinically judged to be at “high risk”, it is estimated that only 40% of the potential population who would benefit from this treatment will be treated [505]. For all potential patients to receive TXA, TXA should therefore be administered to all patients with trauma and significant bleeding. Thus, TXA should be included as part of each institutional “trauma management protocol” not the “massive blood loss” or “major haemorrhage” protocols. The benefit of on-scene TXA administration has recently been shown to be independent of the severity of injury [504].

Secondly, Moore et al. suggested that TXA be administered only in those patients with hyperfibrinolysis determined using TEG®, as many patients who have traumatic injuries lack a hyperfibrinolytic trace, a so-called “hypofibrinolytic shutdown” [506]. However, Raza et al. have clearly shown that TEG® is poor at detecting fibrinolytic activation when compared with more sensitive assays [507]. Furthermore, support for the unreliability of ROTEM® in detecting hyperfibrinolysis comes from Gall et al., who showed that S100A10, an endothelial receptor for plasminogen, leaches off the endothelium during trauma and interferes with detection of fibrinolysis using ROTEM® [266]. We therefore recommend that TXA be administered as soon as possible, without waiting for viscoelastic results.

The cost-effectiveness of TXA in patients with traumatic injury has been calculated in three countries [508, 509]: Tanzania as an example of a low-income country, India as a middle-income country and the UK as a high-income country. The cost of TXA administration to 1000 patients was US$17,483 in Tanzania, US$19,550 in India and US$30,830 in the UK. The estimated incremental cost of administering TXA per life-year gained was $48, $66 and $64 in Tanzania, India and the UK, respectively.

ε-Aminocaproic acid is also a synthetic lysine analogue that has a potency ten-fold weaker than that of TXA. It is administered at a loading dose of 150 mg/kg, followed by a continuous infusion of 15 mg/kg/h. The initial elimination half-life is 60–75 min and must therefore be administered by continuous infusion in order to maintain therapeutic drug levels until the bleeding risk has diminished. This agent is a potential alternative to TXA if TXA is not available. Due to concerns about safety [510], the use of aprotinin is not advised in bleeding trauma patients, now that TXA has been shown to be efficacious and safe.

Coagulation support

Recommendation 23

We recommend that monitoring and measures to support coagulation be initiated immediately upon hospital admission. (Grade 1B)

Rationale

While several general pathophysiological mechanisms have been described that result in trauma-related coagulopathy, including hyperfibrinolysis [16, 29, 252, 511], it is essential to quickly determine the type and degree of coagulopathy in the individual patient in order to determine the most prominent cause or causes to be treated specifically in a goal-directed manner [512]. Early therapeutic intervention improves coagulation [513], which can reduce the need for transfusion of RBC, FFP and platelets [17, 42, 43, 259, 514], the incidence of post-traumatic multi-organ failure [42] and length of hospital stay [17], as well as improving survival [41, 43, 259, 515, 516]. The success of early algorithm-based and goal-directed coagulation management in reducing transfusions and improving outcomes, including mortality, has also been demonstrated in patients undergoing cardiac surgery [517,518,519]. It is, therefore, expected that early algorithm-based and goal-directed coagulation management treatment would also improve the outcome of severely injured patients [41,42,43, 259, 520, 521]. This has indeed been shown in a prospective randomised study [522] and in two studies assessing the introduction of such a concept in two large Italian and one Swiss trauma centre [43, 523]. In other studies, however, no survival benefit could be shown [513, 524, 525]. These variations may be because studies that failed to show an effect tended to base decisions on traditional laboratory values such as PT, APTT and platelet count, and therapies that were often limited to FFP and platelet transfusions.

Initial coagulation resuscitation

Recommendation 24

In the initial management of patients with expected massive haemorrhage, we recommend one of the two following strategies:

• FFP or pathogen-inactivated FFP in a FFP:RBC ratio of at least 1:2 as needed. (Grade 1C)

• Fibrinogen concentrate and RBC. (Grade 1C)

Rationale

The current concept for the resuscitation of patients with massive bleeding with immediate coagulation support was introduced in May 2005, based on reports from the ongoing conflict in Iraq. The US Army’s Institute of Surgical Research recommended the immediate administration of coagulation components in a 1:1:1 ratio for RBC, plasma and platelets until laboratory measurements to adjust therapy were available [526]. In the following years, the best initial strategy to support coagulation became a matter of debate, and two different strategies were proposed. Based on the results of 37 studies, recent guidelines from the Eastern Association for the Surgery of Trauma recommend the transfusion of equal amounts of RBC, plasma and platelets during the early, empiric phase of resuscitation [527]. However, other authors, mainly in Europe, strongly support the use of factor concentrate as the first line of initial coagulation resuscitation in patients with significant bleeding. A few European studies have tried to compare these two strategies; however, there are no good data to date, and no definitive conclusion can be reached. The Reversal of Trauma Induced Coagulopathy Using Coagulation Factor Concentrates or Fresh Frozen Plasma (RETIC) trial, a small single-centre RCT comparing plasma to factor concentrate-based resuscitation, has recently been terminated early after an interim analysis revealed potential harm to patients randomised to the plasma arm [