Background

Haemorrhage remains a major cause of potentially preventable deaths worldwide. Trauma and massive transfusion is associated with coagulopathy secondary to tissue injury, hypoperfusion, dilution and consumption of clotting factors and platelets [1–9]. Patients undergoing cardiac surgery accompanied by cardiopulmonary bypass (CPB) stand a high risk of dying due to microvascular bleeding and 11% have excessive bleeding after CPB – in most cases found to be nonsurgical [10, 11]. The non-surgical bleeding risk in these patients originates in coagulopathy arisen from distortion of the haemostatic system [12, 13]. Concepts of damage control surgery in trauma have evolved, prioritizing early control of the cause of bleeding by non-definitive means, while haemostatic resuscitation seeks early control of coagulopathy [14, 15]. Haemostatic resuscitation provides transfusions with fresh frozen plasma (FFP) and platelets in addition to red blood cells (RBCs) in an immediate and sustained manner as part of the transfusion protocol for critically bleeding patients. Transfusion of RBCs, FFP and platelets in a similar proportion as in whole blood prevents both hypovolemia and coagulopathy [16, 17]. Although an early and effective reversal of coagulopathy is documented [16, 18], the most effective means of preventing coagulopathy of massive transfusion remains debated. Results from recent before-and-after studies in massively bleeding patients and one randomised clinical trial (RCT) indicate that trauma exsanguination protocols involving the early administration of plasma and platelets are associated with improved survival [19–22]. Furthermore, viscoelastic haemostatic assays (VHAs), such as thrombelastography (TEG)/rotational thromboelastometry (ROTEM), appear advantageous for identifying coagulopathy in patients with severe haemorrhage, as opposed to conventional coagulation tests (CCTs) [23–25]. Current views recommend that patients with uncontrolled bleeding, regardless of its cause, should be treated with goal-directed haemostatic resuscitation involving the early administration of plasma and platelets and the use of VHAs should be considered. The aim of goal-directed therapy should be to maintain a normal haemostatic competence until surgical haemostasis is achieved, as this appears to be associated with reduced mortality [4, 6, 12, 20].

The aim of the present study was to perform a systematic review and meta-analysis of all published RCTs comparing the effect of VHAs versus CCTs on blood loss, transfusion requirements and mortality.

Materials and methods

An electronic search was conducted by one of the authors (MF) in the PubMed and EMBASE database using the following search strategy: (Thrombelastography OR Thromb?elastograph* OR thromboelastograph OR ROTEM OR TEG OR ROTEG OR Thromboelastometry OR (algorithm AND bleeding)) AND ((randomized controlled trial OR controlled clinical trial) OR (randomized OR placebo OR trial)), to identify all RCTs done on bleeding patients using treatment algorithms based on results from either TEG or ROTEM. The search identified 1245 references in PubMed and 1835 references in EMBASE. 222 duplicate findings were discarded, leaving a total of 2858 references for further assessment. References were assessed by one of the authors (MF) and discussed and consensus reached with all authors in doubt cases. Only published RCTs were eligible for this analysis. Inclusion criteria were 1) trial designs in which patients were randomly allocated to receive transfusions based on either a VHA-guided (TEG or ROTEM) algorithm (intervention group) or at the clinician’s discretion and/or based on laboratory coagulation tests (control group) and 2) references had to adequately report the outcomes bleeding and/or transfusions and/or mortality. Studies written in other languages than English were also eligible for inclusion. Trials were excluded immediately based on title or abstract, if they did not meet the inclusion criteria. Moreover, trials that were not performed on humans and paediatric studies were also excluded. The remaining studies were evaluated and assessed for relevance by all authors. Reference lists of the included studies were searched for subsequent relevant studies not identified by search engines. Corresponding authors were contacted to retrieve inadequately reported or missing data. Primary outcomes for data extraction were all-cause mortality, total amount of bleeding expressed either as bleeding at 12 h, 24 h or perioperative amount of bleeding and amount of total RBC transfusions, FFP transfusions and platelet transfusions. When amount of blood transfusions was given in mL, calculations of the corresponding number of units were done using the conversion factors illustrated in table 1. The volume per unit was an estimate of the standard volume of the given allogeneic blood product over the last years in the Capital Region Blood Bank, Rigshospitalet, Copenhagen. The latest follow up data on mortality were used in the analysis of all-cause mortality.

Table 1 Conversion factors from mL to units
Full size table

Statistics

Statistical meta-analyses were conducted using Review Manager (RevMan) Version 5.3 (The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). Pooled estimates and their 95% confidence intervals (CI) were calculated using the inverse variance method. The random-effects model was used in anticipation of significant heterogeneity [26]. Statistical heterogeneity was explored using the inconsistency (I2) measure [27]. For all calculations, two-tailed P values of less than 0.05 were considered statistically significant.

Results

Study characteristics

We identified a total of 2858 references. All references were screened by their title and abstract and 2812 references were found not to be relevant for this meta-analysis and were therefore excluded immediately, leaving 46 references for further scrutiny (fig. 1). Another 31 references were excluded due to the reasons explained in table 2. This left 15 RCTs with a total of n = 1238 patients to be included in this analysis. Of these trials, 9 referred to cardiothoracic patients [28–36] and one each to liver transplantation [37], surgical excision of burn wounds [38], trauma [22], cirrhotic patients [39], scoliosis surgery [40] and post-partum haemorrhage [41]. In twelve studies the intervention group was guided by TEG [22, 28, 29, 31–35, 37, 39–41] and in the remaining three by ROTEM [30, 36, 38]. Seven trials applied both results from CCTs and the discretion of the attending physician to guide the transfusions of the control group [28, 31, 32, 35, 38, 40, 41], while the control groups of eight trials were guided only by CCTs [22, 29, 30, 33, 34, 36, 37, 39] with the first transfused blood products being guided solely at the clinician’s discretion before blood analyses were available in two trials [22, 30]. Eleven trials reported on bleeding [28–35, 37, 40, 41], nine reported on mortality [22, 28, 30, 33, 34, 37, 39–41] and all studies reported on transfusion requirements. The transfusion triggers for RBCs, FFP and platelet concentrates for each study are demonstrated in table 3 and the individual study characteristics are presented in table 4.

Fig. 1
figure 1

Process of inclusion of trials into meta-analysis

Full size image
Fig. 2
figure 2

Forest plots a All-cause mortality b Perioperative, 24 h and 12 h bleeding c Total transfusion need – RBC d Total transfusion need – FFP e Total transfusion need – Platelets

Full size image
Table 2 Author and year, type of patients examined and reason for exclusion in excluded scrutinized references
Full size table
Table 3 Transfusion algorithm trigger values. Table explaining individual transfusion trigger values in the respective trials included in the meta-analyses
Full size table
Table 4 Study characteristics Author and year, number of patients allocated to control or intervention group and the type of patients and/or procedures performed during the study
Full size table

Meta-analyses

All-cause mortality

Six trials were included in the meta-analysis of all-cause mortality with a total of 579 patients of whom 291 patients were allocated to the intervention. Three trials concerned patients undergoing cardiothoracic surgery [28, 30, 34] one trial concerned orthotopic liver transplantation [37], one studied cirrhotic liver patients [39] and one studied trauma patients [22]. The meta-analysis demonstrated no difference in survival between the groups with an OR of 0.60 (95% CI 0.34 to 1.07; p = 0.08) (figure 2a).

Bleeding volume

Eleven RCTs reported on bleeding while only five of these studies expressed perioperative, 24 or 12-h bleeding as mean ± SD and were therefore eligible for meta-analysis [28, 37, 40–42]. Comparison of the bleeding volume in 538 patients (305 in the intervention groups) resulted in significantly reduced bleeding in the VHA treated patients (standardized mean difference −1.40 [95% CI 2.57 to −0,23]; p = 0.02) (figure 2b).

Transfusion requirements

The analysis for transfusion requirements was limited to six trials concerning RBC transfusions [22, 37, 38, 40–42] and five trials were eligible for the meta-analysis on transfusions of FFP and platelets, respectively [22, 37, 40–42]. All fifteen trials included in this analysis reported on transfusions, while only the above mentioned described the mean transfused amount per patient ± SD as required for meta-analysis. Isolating RBC-transfusion requirements, 260 out of 453 patients were in the intervention group. Random effects analysis resulted in a standardized mean difference of −0.64 (95% CI −1.12 to −0.15; p = 0.01), being statistically significant (figure 2c). Differences in FFP-transfusions were calculated in 423 patients (246 in intervention group) and resulted in a standardized mean difference of −1.98 (95% CI −3.41 to −0.54; p = 0.007), showing a significant reduction in transfused FFP in the intervention group (figure 2d). Numbers for transfused units of platelets were available from the same 423 patients as with FFP-transfusion requirements, however meta-analysis did not reach statistical significance (standardized mean difference −0.34 [95% CI −0.92 to 0.24; p = 0.25]) (figure 2e).

Discussion

We found the total bleeding volume and the amount of transfused RBCs and FFP to be significantly reduced in the VHA-guided intervention groups compared to CCT-guided control groups. Considering that most trials used the same transfusion trigger for RBCs in both groups, the difference in RBC requirements may be explained by a better haemostatic competence in TEG/ROTEM-guided groups accomplished through timely administration of plasma and platelets, further supported by the reduction of bleeding in the VHA-guided group of patients. In our meta-analysis no statistically significant difference was found between groups regarding all cause-mortality and required amounts of platelets. The sizes of the respective trial populations were small and a lack of cohesion in permission of platelet inhibitors, anticoagulants, antifibrinolytics and triggers used to guide resuscitation with blood products was observed. The control groups were managed either by clinical judgement combined with CCTs or by the sole use of algorithms applying only CCT-triggers for transfusion. The decision to transfuse potentially encompasses a bias to a greater number of transfusions between clinicians with a different background and clinical practice, in alignment with Avidan et al. [29] finding a reduction in transfusions administered with CTT-algorithm guided perioperative management versus transfusion guidance based only on the physician’s discretion. Although only a difference in amount of FFP and no statistical difference in the amount of platelets transfused between groups was detected, the timing of these transfusions may differ with VHA-analyses having shorter turn-around time than conventional coagulation tests [43]. This accentuates the importance of early administration of the appropriate blood products as also emphasized by Cotton et al. [20] who found reduced odds of mortality (74%) and transfusions in a group of trauma patients managed with early and aggressive resuscitation on admittance to the emergency department. Although 24-h transfusion requirements were reduced in patients treated according to the exsanguination protocol, amounts of intraoperative transfusions were found to be larger in this cohort in comparison with the conventionally treated controls, illustrating the importance of early resuscitation with blood products. Also Johansson et al. [21] found similar results in patients undergoing surgery for ruptured abdominal aortic aneurysm (rAAA) with a proactive intraoperative administration of platelets and FFP yielding an increase in survival in massively bleeding rAAA patients. They found a significant reduction in postoperative transfusions, indicating that early blood product administration plays a pivotal role in improving haemostasis in massive bleeders. Gonzalez et al. [22] have conducted the first RCT to evaluate VHA-guided transfusion therapy in trauma. They found a survival benefit in the TEG-guided group especially with regards to less haemorrhagic and early deaths. Additionally, they argued that the administration of more platelets and FFP does not necessarily increase survival chances but highlight the effect of the appropriate treatment being given at the optimal time rather than the amount of blood product administered. Moreover, in patients undergoing surgery with extracorporeal circulation, the use of TEG/ROTEM heparinase analyses, where coagulopathy can be identified despite patient being heparinized, may provide an even earlier assessment of coagulation status and thereby enable an earlier correction of coagulopathies, exemplified by Royston et al. [32] and Girdauskas et al. [30].

Weber et al. [34] report a notably higher mortality among their patients than usually seen in cardiac surgery. Despite this, we did not find a significant difference in mortality in the VHA-guided groups compared to conventionally treated groups. However, our meta-analysis suggested clinical difference in survival in patients having treatment based on VHA-results, in congruence with a before- and after study conducted on trauma patients by Johansson et al., showing a reduction in mortality of approximately 30% in a group resuscitated using TEG results in patients requiring massive transfusions [19]. Furthermore, a Cochrane review from Wikkelsø et al. [44] found the use of TEG or ROTEM in guiding resuscitation of bleeding patients to reduce all-cause mortality and the number of patients transfused with blood products, although no difference was found with regard to excessive bleeding events and proportion of massively transfused, in agreement with our results. Also, NICE-report done by Whiting et al. [45] finds a tendency to fewer transfusions of allogeneic blood products being administered in cardiac surgery patients treated according to VHA-results when comparing to patients managed with CCT-results, while no difference was found with regard to trauma patients and post-partum bleeding. The discrepancies in study selection with the review from Whiting et al. [45] are explained in table 5.

Table 5 Explanation for discrepancies with RTCs included by Whiting et al. [45] (NICE-report)
Full size table

Limitations

A limited number of adequately reported trials were eligible for our meta-analyses. Out of the 15 included trials in this analysis, five did not report sufficient information to be included in any of the meta-analyses performed [29, 31, 32, 35, 36]. This meta-analysis has an overweight of trials concerning cardio-thoracic patients, while other patient groups are only represented by a single RCT each, limiting comparability of results. Furthermore, the studies included present patients with bleeding originating from different aetiologies. This can potentially be problematic in that the severity of bleeding may vary.

Conclusions

In conclusion, the performed meta-analyses demonstrated trends towards the superiority of treating haemorrhaging patients under the guidance of VHA-algorithms. There is, however, a need for larger RCTs, such as the ongoing trials “implementing Treatment Algorithms for the Correction of Trauma Induced Coagulopathy (iTACTIC)” NCT02593877.