Correction of Trauma-induced Coagulopathy by Goal-directed Therapy: A Secondary Analysis of the ITACTIC Trial

In the ITACTIC study, adult trauma patients were enrolled if they presented with clinical signs of bleeding activating the local major hemorrhage protocol and if erythrocyte transfusion had been initiated. Participants had to be randomized within 3 h of injury and a maximum of 1 h after admission to the emergency department. There were no additional exclusion criteria. All patients were treated with the local center’s empiric major hemorrhage protocol care consisting of tranexamic acid, and blood products were administered in a 1:1 ratio of erythrocytes and plasma, with some empiric administration of platelet concentrates, cryoprecipitate, or fibrinogen concentrate. In addition, patients were randomized to receive additional goal-directed hemostatic therapy guided by either conventional coagulation test - or viscoelastic hemostatic assayVHA-driven algorithms.⁷  In both arms, blood samples were taken at baseline and after every 4 units of erythrocyte transfusions until hemostasis. Subjects in the conventional coagulation test arm had only laboratory tests performed. In the viscoelastic hemostatic assay arm, viscoelastic and laboratory tests were performed, but only the viscoelastic results were used to guide study interventions. Test results were reviewed as soon as possible after they were available and used to guide administration of additional blood products and tranexamic acid.

Thresholds for intervention were from the previously developed TACTIC algorithms.⁷  Abnormal values triggered administration of goal-directed therapy as follows: 4 g of equivalent fibrinogen as two pools of cryoprecipitate (United Kingdom) or 4 g of fibrinogen concentrate (Norway) was indicated for a laboratory (Clauss) fibrinogen less than 2 g/l or rotational thromboelastometry (ROTEM) FIBTEM A5 less than 10 mm; one additional pool of platelet concentrate for a platelet count less than 100 × 10⁹ or PLTTEM A5 (EXTEM A5 – FIBTEM A5) less than 30 mm; and four additional units of plasma were given with a prothrombin time ratio greater than 1.2 or an EXTEM clot time greater than 80 s with normal fibrinogen thresholds. For the viscoelastic hemostatic assay algorithm, an additional 1 g of tranexamic acid was given if EXTEM lysis index at 30 min was less than 85%.⁷ 

In the ITACTIC trial, a full panel of conventional and viscoelastic coagulation tests was not available for all participants. Patients in the conventional coagulation test arm did not have viscoelastic tests performed, but those in the viscoelastic hemostatic assay arm had both viscoelastic and laboratory tests. A parallel study, Activation of Coagulation and Inflammation in Trauma (ACIT) was active at some of the ITACTIC sites. ACIT is a prospective observational cohort study in which trauma patients are enrolled on arrival to a major trauma center and was already running at the London and Oslo ITACTIC study sites. As part of the ACIT study, blood samples for conventional coagulation tests and viscoelastic tests were taken at baseline and after 4, 8, and 12 units of erythrocyte transfusions.⁸  These are the same sampling intervals used to guide goal-directed interventions in the ITACTIC trial. ITACTIC patients who were coenrolled into ACIT therefore had full coagulation profiles available for analysis.

The specifics of the ACIT study have been reported previously.⁷  In brief, trauma patients arriving in the emergency department within 2 h of injury are eligible for inclusion. Patients were excluded if they had burns covering greater than 5% of the body surface area, were taking anticoagulant medications other than aspirin, received greater than 2,000 ml of IV fluids before arrival in the hospital, had a known bleeding disorder, or had moderate to severe liver disease.⁸  At all times, the ITACTIC clinical teams were blinded to the viscoelastic hemostatic assay results performed in the ACIT study. For this study, only ACIT tests that matched those in the ITACTIC viscoelastic hemostatic assay algorithms were analyzed, specifically EXTEM A5, clot time, and lysis index at 30 min (LI30) and FIBTEM A5. Detailed methods for ROTEM have been described previously.^9,10 

To define trauma-induced coagulopathy, we used the standard viscoelastic definition of ROTEM EXTEM A5 of less than 40 mm and a prolonged prothrombin time ratio (greater than 1.2) for conventional laboratory tests.^4,7,10,11  Goal-directed treatment was defined as administration of hemostatic therapy in response to an abnormal test value in accordance with TACTIC treatment algorithms. Time to intervention was defined as the time from the baseline blood sample to administration of the first targeted treatment for a specific abnormality as defined above.

Our primary outcome was correction of coagulopathy during resuscitation (up to the 12th erythrocyte unit after baseline). Correction of coagulopathy was defined as normalization (to above the algorithm treatment threshold) of the coagulation parameter (e.g., EXTEM A5 or prothrombin time ratio), from the baseline sample to the final sample available for each patient. Secondary outcomes of interest were blood component requirements at 24 h and 28-day mortality.

Both ITACTIC and ACIT studies received ethical approval from local ethical review boards. Initial consent for enrollment in ACIT and ITACTIC was sought from a professional legal representative, usually an independent clinician involved in the patient’s care. Subsequent consent was obtained from personal legal representatives as soon as possible after enrollment when it was possible to identify one and later by the participant if they regained capacity.

The data were analyzed using R version 4.0.3. The data were assumed to be missing at random, and participants with missing data for a given test were excluded from any analysis of that test. The descriptive statistics are presented as median and interquartile range or as number and percentage. The three main comparison groups for this study were 28-day mortality (alive vs. dead), ITACTIC trial arm (conventional coagulation test vs. viscoelastic hemostatic assay), and goal-directed therapy (yes vs. no). For each comparison group, we analyzed the proportion who corrected coagulopathy (EXTEM A5) and the patterns of changes in coagulation test results over time. Differences in proportions of patients who corrected coagulopathy were analyzed using chi-square or Fisher’s exact test. Differences in changes in coagulation parameters over time between the three main group comparisons were analyzed with two-way ANOVA. Additionally, mortality differences between ITACTIC arms by type of admission coagulopathy are presented as odds ratios and 95% CI, and differences were analyzed using chi-square or Fisher’s exact test. Differences in times to administration of goal-directed treatments were analyzed with the Wilcoxon rank sum test. P values less than 0.05 were considered statistically significant. Statistical tests of P = 0.05 or above are reported as not significant. Where no P value is reported, statistical comparison of differences was not performed.

Of the 236 ITACTIC patients recruited in London and Oslo, 133 patients were coenrolled into the ACIT study (representing 34% of the whole ITACTIC trial cohort). Patients had a median age of 36 yr, and 83% were male. Median injury severity score was 27, and 39% had sustained a penetrating injury (table 1). In ITACTIC, 58 of these had been enrolled into the conventional coagulation test arm and 75 into the viscoelastic hemostatic assay arm (fig. 1). Demographics, injury characteristics and admission physiology were similar between the two arms and were representative of the whole ITACTIC cohort (supplemental table 1, https://links.lww.com/ALN/D651). Across both ITACTIC study arms, 54% (72 of 133) of patients received an algorithm-directed intervention, and 46% (61 of 133) received only empiric treatments as directed in the institution’s major hemorrhage protocol (fig. 1).

On admission, 71% (93 of 131) presented with viscoelastic or laboratory identified coagulopathy (68% with EXTEM A5 less than 40mm, 33% with prothrombin time ratio greater than 1.2; table 1; supplemental table S1, https://links.lww.com/ALN/D651). Of those patients not coagulopathic on admission, 16% became coagulopathic by EXTEM A5, and 7% developed a prolonged prothrombin time ratio during resuscitation. Hypofibrinogenemia (fibrinogen less than 2 g/l) was present on admission in 66% (75 of 114) of patients and in 65% as measured by FIBTEM less than 10 mm. During bleeding and resuscitation, a further 8 and 12% subsequently developed hypofibrinogenemia as measured by laboratory (Clauss) fibrinogen and FIBTEM A5, respectively. A low admission platelet count was present in 7% (9 of 128) of patients on admission and only developed later during hemorrhage in 3% of those with a normal admission platelet count. In contrast, low platelet function (PLTTEM A5) was present in 70% of patients on admission and developed during bleeding in a further 25% of the patients who were initially normal. Only 8% of patients met the thresholds for additional plasma and 3% for additional tranexamic acid at baseline. Subsequent triggers for plasma were rare with only a single patient meeting the criteria. No patient developed ROTEM-detected hyperfibrinolysis during transfusion. In total, 111 patients (83%) had a coagulation test result that could have triggered an algorithm-guided intervention.

Overall, 28-day mortality in this cohort of patients was 29% (38 of 133; 28% viscoelastic hemostatic assay vs. 29% conventional coagulation test, 0.938; 95% CI, 0.41 to 2.16). The numbers in some subgroups of admission coagulopathy types were very small, but for all types of admission coagulopathy, 28-day mortality was generally lower in the viscoelastic hemostatic assay arm than in the conventional coagulation test arm (supplemental table 2, https://links.lww.com/ALN/D651). Odds ratios for mortality in the viscoelastic hemostatic assay arm compared to the conventional coagulation test arm varied depending on the baseline coagulation derangement, ranging from 0.68 (95% CI, 0.29 to 1.6) for those with EXTEM A5 less than 40 mm at baseline to 0.87 (95% CI, 0.35 to 2.17) for those with fibrinogen less than 2 g/dl (supplemental table 2, https://links.lww.com/ALN/D651).

For patients who were coagulopathic at baseline, coagulation was corrected during resuscitation in a higher proportion of survivors than in those who died (25% vs. 9%; odds ratio, 3.33; 95% CI, 0.25 to 45.11; P = not significant; fig. 2A). There was no difference between the ITACTIC viscoelastic hemostatic assay and conventional coagulation test arms in the proportion of patients who corrected their EXTEM A5 (14% viscoelastic hemostatic assay vs. 17% conventional coagulation test; odds ratio, 0.94; 95% CI, 0.44 to 2.03; P = not significant; fig. 2B). Irrespective of ITACTIC study arm, only patients who received a goal-directed intervention improved their EXTEM A5 during hemorrhage (20% vs. 0%; odds ratio, ∞; P = not significant; fig. 2C). However, as shown in figure 2, the majority of patients did not correct coagulopathy during resuscitation.

The trajectories of coagulation abnormalities during hemorrhage are shown in figure 3 and supplemental figures S1 and S2 (https://links.lww.com/ALN/D651). Survivors had improved coagulation profiles compared to nonsurvivors across all cohorts with admission coagulation deficits (fig. 3; supplemental figs. S1 and S2, https://links.lww.com/ALN/D651). There was no significant difference in correction profiles between the viscoelastic hemostatic assay and conventional coagulation test arms of the ITACTIC trial. Of the 111 patients who triggered an ITACTIC algorithm, only 72 (65%) received a goal-directed intervention before receiving their 12th unit of erythrocyte transfusion. ITACTIC viscoelastic hemostatic assay group patients triggering the algorithm were more likely to receive a goal-directed therapy (52 of 68) than those in the conventional coagulation test group (20 of 43; 76% vs. 47% conventional coagulation test; odds ratio, 3.73; 95% CI, 1.64 to 8.49; P = 0.002). These goal-directed therapies resulted in statistically significant coagulation improvements from baseline for those who presented with low EXTEM A5 or prolonged prothrombin time ratio on admission (both P < 0.001; fig. 3), but not for other laboratory or viscoelastic parameters (supplemental figs. S1 and S2, https://links.lww.com/ALN/D651).

We examined the time delay between coagulopathy detection and administration of the therapeutic intervention (fig. 4). Median time to diagnosis was 14 (8 to 25) min in the viscoelastic hemostatic assay arm and 64 (55 to 73) min in the conventional coagulation test arm. The benefit of early diagnosis in the viscoelastic hemostatic assay arm was overshadowed by the length of time to deliver the goal-directed intervention. Overall, the median time from diagnosis on baseline sample to the first goal-directed intervention was a further 76 (59 to 102) min. Patients in the viscoelastic hemostatic assay arm were more likely to receive a goal-directed intervention (viscoelastic hemostatic assay vs. conventional coagulation test, 31% vs. 19%; odds ratio, 3.48; 1.33 to 9.51; P = 0.005), and the viscoelastic hemostatic assay-guided algorithm delivered the first goal-directed intervention to coagulopathic patients a median of 42 min faster than the conventional coagulation test algorithm. This was still over an hour after the baseline sample was taken and approached 2 h in the conventional coagulation test arm (viscoelastic hemostatic assay vs. conventional coagulation test, 68 [53 to 88] vs. 110 [77 to 123] min; P = 0.005; fig. 4).

This secondary analysis of the ITACTIC trial provides a detailed description of the effect of contemporary approaches to hemostatic resuscitation on trauma-induced coagulopathy. The cohort and subgroup sizes were small, and the results should be interpreted in this context. There are consistent results that suggest goal-directed treatments improved coagulation profiles more than empiric care alone, across the spectrum of coagulation deficits. However, goal-directed interventions were delivered to only around two thirds of patients who needed them. They took over an hour to deliver regardless of diagnostic approach and did not restore coagulation to normal during bleeding and resuscitation. The over-riding picture is that any improvements in diagnosis provided by point of care viscoelastic testing were negated by a failure of delivery or efficacy of the therapeutic interventions.

Hemostatic resuscitation was introduced to directly target trauma-induced coagulopathy.¹²  Major hemorrhage transfusion algorithms attempt to prevent coagulopathy developing or worsening, principally through the empiric administration of whole blood or balanced blood component transfusions.¹³  Before the introduction of these empiric algorithms, coagulation factor levels would fall precipitously during hemorrhage, leading to severe and irreversible bleeding.^14,15  Transfusion with balanced erythrocyte, plasma, and platelet transfusions has been shown to slow the development of coagulation factor deficits. However, the empiric administration of concentrated coagulation therapeutics has not been shown to improve outcomes.^16,17  In ITACTIC patients, the progression of coagulopathy in patients receiving empiric therapy alone (without receiving targeted treatments), was indistinguishable from that described in studies from a decade ago.⁵ 

Goal-directed treatment algorithms aim to identify established coagulation deficits and correct these using concentrated products.⁷  Traditionally, these have been guided using laboratory-based conventional coagulation tests, which have repeatedly been shown to have long turnaround times inappropriate for rapidly bleeding patients.^10,18  The viscoelastic hemostatic assays such as ROTEM and thromboelastogram have moved diagnosis to point of care, providing rapid results and more information on specific coagulation factor defects.^7,10  Our results show that both conventional coagulation test and viscoelastic hemostatic assay-guided approaches produce improved coagulation profiles compared to empiric therapy alone. There were general trends to improved responses with viscoelastic hemostatic assay-guided algorithms and associated nonsignificant trends to lower mortality, but these were not to the extent that was expected, and neither approach led to correction of trauma-induced coagulopathy during hemorrhage.

These findings call into question both the delivery and efficacy of these interventions. The dose of product administered in ITACTIC may have been insufficient, or it may be that the coagulopathy is not correctable by the chosen therapeutic. Most therapeutics administered by the ITACTIC algorithms were blood components, which had to be specifically requested from the transfusion laboratory, often thawed, picked up, delivered to the treating team, and then administered in a queue of other transfusion therapies. Patients who received goal-directed therapy received more erythrocyte transfusions than the empiric group, possibly reflecting the length of time it takes to deliver such an intervention. Although the viscoelastic hemostatic assay-guided arm of the trial delivered diagnostic results faster than the conventional coagulation test arm and patients were more likely to receive an intervention, it appears this was not enough to offset the long delay to delivering and administering the required blood products across the whole trial population.

This study has several limitations. Not all sites were able to recruit to both ITACTIC and ACIT in parallel, and we were only able to analyze just over a quarter of the ITACTIC study cohort. Additionally, there was a large degree of heterogeneity in the patients, leading to wide variation on coagulation responses. Our findings are therefore underpowered in many of the analyses. A second important limitation is that we were unable to track the triggering of an algorithm to the administration of a specific treatment and then the subsequent response of the coagulation system to that intervention. Analysis at this level of granularity was impossible due to the speed of resuscitation, the variable time between ordering an intervention and the patient receiving it, and multiple nearly simultaneous product administrations. This process and human factors analysis is an important area for future research.

In conclusion, major hemorrhage protocols that include goal-directed hemostatic therapy improve coagulation profiles compared to empiric protocols alone. Patients who correct their coagulopathy had a lower mortality, and viscoelastic assays increase the proportion of patients who receive goal-directed treatment. However, the overall benefits of earlier diagnosis appear to be heavily offset by delayed administration of interventions, which are only partially effective. Future progress in correcting trauma-induced coagulopathy and improving clinical outcome will require focus on the rapid delivery and administration of more effective coagulation therapeutics.

The ITACTIC study was part of the Targeted Action for Curing Trauma Induced Coagulopathy (TACTIC) program, funded by the European Commission under the FP-7 HEALTH Contract F3-2013-602771 (https://cordis.europa.eu/project/id/602771). Werfen (Barcelona, Spain) supported the Activation of Coagulation and Inflammation in Trauma (ACIT) and ITACTIC studies through the provision of ROTEM Sigma analyzers and reagents to participating institutions.

Dr. Curry has received research grants from CSL Behring (King of Prussia, Pennsylvania) and Sobi (Stockholm, Sweden) and has served on advisory boards for LFB Biomedicaments (Les Ulis, France), Octapharma (Lachen, Swittzerland), and Sobi. Dr. Maegele has received honoraria for lectures and speakers’ bureaus, congress travel support, and financial support for research projects from Astra Zeneca (Cambridge, United Kingdom), Bayer (Leverkusen, Germany), CSL Behring, Werfen (Barcelona, Spain), LFB Biomedicaments, Octapharma, and Portola Inc. (South San Francisco, California). Dr. Johansson has received unrestricted research grants from Haemonetics Corp. (Boston, Massachusetts) and Octapharma. Dr. Gaarder has received honoraria for lectures from Octapharma, has received research grant support from Haemonetics (Boston, Massachusetts) and Werfen in the form of device and reagent support, and also has previously served on the advisory board for Nycomed (Zurich, Switzerland). Dr. Brohi and Dr. Davenport have received research grant support from Werfen and Haemosonics (Boston, Massachusetts) in the form of device and reagent support and have served on external advisory panels for Octapharma, Astra Zeneca, and Werfen. Dr. Brohi has previously served on external advisory panels for Haemonetics Corp, Werfen, CSL Behring, Bayer, and Astra Zeneca. The other authors declare no competing interests.

Interventions in ITACTIC – supplemental tables and figures, https://links.lww.com/ALN/D651