Trauma Associated Bleeding: Effectiveness of an Early Coagulation Support Protocol

In severe trauma patients, uncontrolled bleeding is a major cause of death, partly caused by trauma-induced coagulopathy (TIC).TIC represents a severe post-traumatic complication associated with increased transfusion requirements and worsened mortality. Fibrinolysis is a central part of the TIC process. Damage control resuscitation is defined as rapid hemorrhage control through early administration of blood products prevention and immediate correction of coagulopathy, and minimization of crystalloid fluids. Massive Transfusion Protocols (MTPs) were introduced as part of damage control resuscitation, with the aim to facilitate rapid blood product release and to increase adherence to hemostatic resuscitation, following a transfusion ratio of 1:1:1 of fresh frozen plasma (FFP), platelets (PLT) and packed red blood cells (pRBC). Their activation is triggered by clinical parameters, including patients' trauma severity, clinical signs and laboratory data, which are included in scoring systems. In 2013, the Italian Trauma Centers Network (TUN) developed a new treatment algorithm providing early coagulation support (ECS) to control coagulopathy and hemorrhage in major trauma patients with a clinically relevant bleeding risk. The protocol includes the use of fibrinogen concentrate and RBC during initial resuscitation, and the early use of viscoelastic techniques when available (thromboelastometry ROTEM® or thromboelastography TEG).

The aim of this multicenter, before and after study was to assess the effects of a new ECS protocol compared to the standard MTP in terms of blood products' consumption and clinical outcome.

We undertook a before and after study using historical controls. We included all consecutive adult patients with major trauma at risk of bleeding admitted to the emergency department of two referral Trauma Centres in Rome, Italy (i.e. Fondazione Policlinico Universitario A. Gemelli and S. Camillo Hospital). Patients treated in 2011-2012 were compared to patients treated in 2013-2014. In the second period, the ECS protocol was concomitantly implemented as standard of care in both trauma centers. Informed consent was waived as all the procedures were considered the standard of care in both periods.

During the control period (pre-ECS), patients were treated according to a MTP with the following targets: FFP/PRBC ratio ≥ 1:1.5, target platelet count > 100.000 x 109/L. According to the ECS protocol applied in the study period, all included patients received an "initial resuscitation" ,defined as the period between hospital admission and the first available coagulation results (coagulation screen, fibrinogen level and/or viscoelastic monitoring and platelet count), with a fixed initial coagulation support of tranexamic acid (TA) 1 g intravenous bolus over 20 min followed by 1 g in continuous infusion over 8 hours and fibrinogen concentrate (FC) 2 g bolus along with 2 to 4 units of PRBCs. Point of care (POC) tests were systematically used to monitor coagulation and to guide subsequent interventions. FFP was not administered in the early treatment stage. Permissive hypotension (systolic arterial pressure 80-90 mmHg or mean arterial 50-60 mmHg) and a fluid restrictive strategy were applied until surgical bleeding control was achieved. However, when prolonged bleeding occurred and an increasing number of PRBCs (>4 units) and fluid volume was required (>2000 ml), plasma was transfused for hemodynamic and hemostatic resuscitation.

Data from the electronic records were matched with the blood bank registries to confirm the number of blood units transfused within the study's time span. For patients transferred from other hospitals, information regarding transfusions administered before admission were collected and matched with the blood banks' data.

Patients' characteristics at admission are summarized through mean and standard deviation for continuous variables. Median and interquartile range for scales and scores and relative frequencies are used for discrete variables. We tested the null hypothesis of no association between study period and patients' characteristics with the t-test, Wilcoxon test and chi-square test for the three types of variables, respectively. We estimated the mean difference in transfused units and length of stay between ECS and pre-ECS from unadjusted Poisson models, using the delta methods to estimate the 95% confidence interval (CI) for the mean difference. Findings were robust by replacing the Poisson model with a negative binomial distribution. In addition, we defined "multiple transfused patients" as those experiencing four or more RBC units during the first 24 hours. The cut-off of four PRBC units represented the sample 75th percentile.

We investigated the association between clinical features and the probability of multiple transfusion by means of univariate and multivariate logistic regression models. For the mortality outcome, we estimated the Relative Risk (RR) and its' 95% CI of death within the first 24 hours and within 28 days since admission for the post-ECS as compared to pre-ECS patients from Poisson models without adjustment for other variables. Among those surviving the first 24 hours, we replicated the analysis by considering death between the 2nd and the 28th day of hospital stay. Furthermore, we investigated whether differences in resources' use and mortality between ECS and pre-ECS were related to major patients' demographic and clinical features at presentation, including age (18-40, 40-65, 65+ years), severity of traumatic brain injury (head AIS <4 vs. ≥ 4), and trauma severity (number of critical illness criteria; presence/absence of each criterion). These stratified analyses were performed including a study period patients' feature interaction in separated Poisson models and by formally testing the null hypothesis of equal efficacy of the ECS protocol among categories of patients through an F test. To further control for any residual difference in clinical features between pre and ECS groups, we compared the resource consumption, the probability of multiple transfusion and the mortality outcomes among propensity-score matched patients. The propensity score included the following variables: age, sex, ISS, AIS head, worse SBP, base excess, lactates, hemoglobin and platelets. For the statistical analyses, we used the SAS software, 9.4 release (SAS Institute Inc., Cary, NC, US). The propensity score matching was performed using a standard macro.