Temporal Transitions in Fibrinolysis after Trauma: Adverse Outcome Is Principally Related to Late Hypofibrinolysis

We performed a secondary analysis of previously collected data from patients recruited into the ongoing observational Activation of Coagulation and Inflammation in Trauma II study (UK Clinical Research Network ID 5637; ISRCTN 12962642)¹⁸  from a single major urban trauma center. We conducted a descriptive and inferential analysis by reporting the clinical characteristics of our cohort according to their temporal transitions in fibrinolysis and investigating the relationship between these transitions and clinical outcomes. The Activation of Coagulation and Inflammation in Trauma II study is a long-term observational cohort study that began in 2008, investigating early alterations in coagulation and inflammation after injury. For this subset analysis we used contiguous patient samples and data from January 2008 to August 2018. Adult patients (aged 16 yr or older) admitted via a trauma team activation and who arrived within 2 h of injury were screened for inclusion. Full eligibility criteria were previously reported in an earlier subset analysis from the Activation of Coagulation and Inflammation in Trauma II study.²  All procedures performed were in accordance with the ethical standards of the research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. This study was approved by the London – City and East Research Ethics Committee (07/Q0603/29). In accordance with the study protocol, all patients were initially enrolled with assent from a professional consultee independent of the research team. Written informed consent was obtained from all individual participants included in the study that had capacity. For incapacitated patients, consent was provided by a personal consultee where available. The STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) statement was used to guide the reporting of this study.²¹ 

Subjects recruited between 2008 and 2018 with evidence of major trauma and/or bleeding were defined by the following patient characteristics and included in this subset analysis: injury severity score greater than 15; admission coagulopathy (prothrombin time ratio greater than 1.2 or tissue factor–activated ROTEM amplitude at 5 min less than or equal to 35 mm) or shock (lactate greater than or equal to 2 mEq/l or base deficit greater than or equal to 4 mEq/l) on arrival; or transfusion of greater than or equal to 4 units of erythrocytes within first 12 h. Patients with unavailable ROTEM maximum lysis on admission or at 24 h were excluded. Coenrollment into the CRASH-3 (Effects of Tranexamic Acid on Death, Disability, Vascular Occlusive Events and Other Morbidities in Patients with Acute Traumatic Brain Injury) trial was also an exclusion criteria.²² 

Data on patient demographics, admission physiology, injury characteristics, fluid and blood product requirements in the first 24 h, critical care admission, length of stay, venous thromboembolism (VTE) and in-hospital mortality were collected as part of the Activation of Coagulation and Inflammation in Trauma II study.

Blood samples were obtained on admission (0 h) and at 24 h (±2 h) for research purposes. Clinical samples for complete blood count and conventional coagulation tests were obtained simultaneously and processed in the laboratory using standard protocols. Lactate concentration and base deficit were measured by point-of-care blood gas analysis.

ROTEM was performed within 2 h of sample collection.²³  Blood was drawn into 2.7-ml citrated vacutainers (0.109 M buffered sodium citrate, 3.2%; Becton Dickinson, United Kingdom) and warmed at 37°C before analysis. ROTEM delta analyzers (TEM International GmbH, Germany) were used with an automated pipetting system according to the manufacturer’s guidance. Tissue factor–activated ROTEM is the ROTEM test used in this study. Tissue factor–activated ROTEM maximum lysis was recorded after a 60-min assay.^24,25  Assay reproducibility and interrater agreement have been previously investigated.^23,26 

Categories of ROTEM fibrinolysis have been previously defined in an earlier subset analysis with thresholds determined by associations with increased mortality in either hypo- or hyperfibrinolysis.⁹  ROTEM hypofibrinolysis (low) was therefore retrospectively defined as a tissue factor–activated ROTEM maximum lysis of less than 5%; ROTEM hyperfibrinolysis (high) was defined as a tissue factor–activated ROTEM maximum lysis of more than 15%; ROTEM-normal fibrinolysis (normal) was defined as tissue factor–activated ROTEM maximum lysis of 5 to 15%. These ROTEM categories were implemented as exposure variables in our inferential analysis in the form of fibrinolysis states at a single time point and transitions over time. The use of a continuous scale is not meaningful for a diagnostic purpose, considering that any change between 16 and 100% on the maximum lysis scale is representative of increased fibrinolysis but without a linear increase in adverse clinical outcomes.^5,9  Coagulopathy as a prothrombin time ratio greater than 1.2²⁷ ; a severe injury to a specific body region as an abbreviated injury score greater than or equal to 3; massive transfusion as 10 or more erythrocyte units within the first 24 h²⁸ ; multiorgan dysfunction syndrome as a sequential organ failure assessment score greater than or equal to 6 on at least 1 day during the first 7 days of admission^29,30 ; early mortality as death within 24 h; and late mortality as death beyond 24 h. VTE was diagnosed as either deep vein thrombosis on venous ultrasound/computed tomography scan or pulmonary embolism on computed tomography.

We first aimed to assess the natural history of the ROTEM fibrinolysis states and patterns occurring in the first 24 h after injury and their relation to late outcomes. Consequently, all patients who had received tranexamic acid were excluded in the first part of our analysis. We then focused on the impact of tranexamic acid on the most common ROTEM fibrinolysis patterns and their related outcomes. At our institution tranexamic acid is primarily given to bleeding trauma patients for whom the major hemorrhage protocol is activated in the presence of low systolic blood pressure less than 90 mmHg and suspected active hemorrhage. Therefore, we centered the second part of our analysis on the major hemorrhage protocol cohort.

A data analysis and statistical plan was written after the data were accessed. No statistical power calculation was conducted before the study as the sample size of this exploratory analysis was based on available data. Continuous variables were tested for normality using the quantile–quantile plot and Shapiro–Wilk test. As no variable was consistently normally distributed, all are reported as median and interquartile range, with comparisons assessed using the Mann–Whitney U test. Categorical variables are reported as count and percentage, with comparisons assessed using chi-square or Fisher exact tests. The area under the receiver operating characteristic curve (AUC) was used to compare the ability of the levels of maximum lysis on admission and at 24 h to identify multiorgan dysfunction syndrome and late mortality, with statistical significance tested using the DeLong method for two correlated curves. The area under the curve was used to compare the recovery of organ dysfunction. The initial 5% missing components of the sequential organ failure assessment score were resolved by carrying forward the score of the previous day.²⁹  The remaining 1% were managed using the mean of the other observations of the same day within each transition pattern. A two-sided 0.05 level of significance was used. Bonferroni correction was applied for multiple comparisons (further details reported in the captions of table 1, table 2, table 4; and fig. 2, as well as in Supplemental Digital Content table 3, http://links.lww.com/ALN/C737). A complete-case analysis was performed.

To assess the influence of confounding variables on the dynamic changes in the fibrinolysis states and their effect on clinical outcomes, we performed a multivariable logistic regression analysis in the cohort of patients who did not receive tranexamic acid.³¹  Multiorgan dysfunction syndrome was the only outcome variable analyzed due to the limited number of positive events for late mortality and VTE. As confounder variables, we considered demographic variables (age and sex) and variables previously identified as related to multiorgan dysfunction syndrome (base deficit, injury severity score, severe traumatic brain injury, and 24-h crystalloids).³⁰  No variables were considered as effect modifiers. The impact of tranexamic acid was not investigated in a regression analysis given the small number of events in the outcomes of interest when only major hemorrhage protocol patients were included. We have developed two separate models using the fibrinolysis transition patterns and states on admission and at 24 h as categorical exposure variables. Since the size of the high-maximum lysis category at both time points was small (14 patients each, with 2 patients having high-maximum lysis at both time points), we have excluded these categories from the multivariable regression to avoid any overestimation of the odds ratios.³²  The proportion of missing data for multivariable regression analysis was 12 of 372 (3%) and appeared to occur at random across the four remaining transition patterns. We considered the impact of this missing data negligible and so performed a complete-case analysis only.³³  We reported the odds ratios from univariable logistic regression models and the adjusted odds ratios from the multivariable logistic regression models. The odds ratio of a predictor obtained from a univariable analysis compares the likelihood of experiencing an event in the population, whereas the odds ratio of a predictor obtained from a multivariable analysis compares the likelihood of experiencing an event in specific individuals/groups. Multicollinearity was quantified using the variance inflation factor and Cramér’s V. Sensitivity analyses were performed by adjusting the threshold for ROTEM hypofibrinolysis by ±1% and excluding the first 48 h after admission for the classification of multiorgan dysfunction syndrome.^34,35 

Statistical analyses were performed and graphs were generated using RStudio 1.2.1335 (RStudio, Inc., USA) and the following R packages: ggplot2 (H. Wickham, 2016), ggalluvial (J. C. Brunson, 2020), rms (F. E. Harrell, Jr., 2019), plotROC (M. C. Sachs, 2017), and pROC (X. Robin et al., 2011).

Between January 2008 and August 2018, 1,668 patients were enrolled into the Activation of Coagulation and Inflammation in Trauma II study and 1,334 met the inclusion criteria for this subset secondary analysis (Supplemental Digital Content fig. 1, http://links.lww.com/ALN/C737). Subsequent exclusions were attributable to coenrollment in the CRASH-3 trial (n = 119) and missing ROTEM maximum lysis at either admission or 24 h (n = 523), with 39 patients meeting both exclusion criteria. Missing ROTEM data was attributable to early discharge of the patient; patients who consented to remain in the study but declined follow-up blood sampling; ROTEM trace errors; or nonavailability of research personnel. Seven-hundred thirty-one patients were included in the final analysis.

We first examined fibrinolytic transition patterns in 432 patients who did not receive tranexamic acid. On admission, 253 of 432 (59%) had normal-maximum lysis on ROTEM, 26 of 432 (6%) had overt hyperfibrinolysis (high-maximum lysis), and 153 of 432 (35%) had low-maximum lysis. Of those patients still alive at 24 h, 61 of 398 (15%) had low-maximum lysis (table 1); 42 of 61 (69%) had low-maximum lysis on admission, while 19 of 61 (31%) had transitioned from other admission ROTEM lysis states; and 96 of 153 (63%) of the patients with low-maximum lysis on admission were alive but had transitioned to other states by 24 h (fig. 1). Patients excluded because of missing 24-h maximum lysis (n = 349) were representative of the analyzed cohort based on admission maximum lysis: 84 of 349 (30%) were low, 187 of 349 (66%) were normal, and 10 of 349 (4%) were high.

Patients who had low-maximum lysis at 24 h were more severely injured, more likely to have suffered blunt trauma and a severe traumatic brain injury, and more likely to have required a massive transfusion than patients with other late fibrinolysis states (table 1). Within the low-maximum lysis group at 24 h, patients who were persistently low from admission (n = 43) had the highest incidence of severe traumatic brain injury and a lower Glasgow coma score on arrival to hospital (table 2). In contrast, patients who had normal-maximum lysis on admission but transitioned to low-maximum lysis at 24 h (n = 15) had a greater blood component requirement, with 3 of 15 (20%) receiving a massive transfusion. Characteristics of the most infrequent transition patterns are reported in Supplemental Digital Content table 1 (http://links.lww.com/ALN/C737).

Overall mortality was 17 of 38 (45%) for patients who were hyperfibrinolytic (high-maximum lysis) at admission and/or 24 h, compared to 35 of 172 (20%) for those who had low-maximum lysis at any time point and 9 of 223 (4%) for those who always had normal-maximum lysis. There were large differences in late outcomes depending on the transition patterns, with the highest incidence and duration of multiorgan dysfunction syndrome present in those who arrived with low-maximum lysis and remained low at 24 h (fig. 2, A and B). Late mortality was four times higher in this group compared to those patients who always had normal-maximum lysis (table 2), while patients who presented with low-maximum lysis but normalized by 24 h had similar outcomes to those who always had a normal-maximum lysis (table 2). Overall, late outcomes were more closely related to maximum lysis at 24 h than admission maximum lysis levels (fig. 2, C and D). Maximum lysis at 24 h compared to admission maximum lysis was a better predictor of multiorgan dysfunction syndrome (AUC, 0.73 vs. 0.61; P <0.001) and late mortality (AUC, 0.70 vs. 0.54; P =0.023).

The impact of the fibrinolysis transition patterns on the development of multiorgan dysfunction syndrome was then analyzed in relation to other key clinical variables (table 3). Severe traumatic brain injury, while modulated by the associations with the other covariates, was the most important clinical determinant (adjusted odds ratio, 7.40 [95% CI, 3.87 to 14.17]). Age, overall injury severity, and 24-h crystalloids were also predictors of multiorgan dysfunction syndrome. An incremental impact in the risk of organ dysfunction at the population level was evident from the univariable analyses when maximum lysis transitioned from: low on admission to normal at 24 h; normal on admission to low at 24 h; and persistently low both on admission and at 24 h. Whereas, after conditioning on other variables (i.e., focusing on individuals/groups with specific characteristics), only patients transitioning to or remaining in a low lysis state at both time points were characterized by higher odds of multiorgan dysfunction syndrome (adjusted odds ratio, 5.43 [95% CI, 1.43 to 20.61] and 4.85 [95% CI, 1.83 to 12.83], respectively). When the admission and 24-h fibrinolysis states were considered as separate variables, only the latter retained a significant predictive role in multivariable analysis (adjusted odds ratio, 3.54 [95% CI, 1.55 to 8.06]; Supplemental Digital Content table 2, http://links.lww.com/ALN/C737). A moderate degree of collinearity was present between maximum lysis at the two time points (Cramér’s V = 0.34; variance inflation factor = 1.07 and 1.10, respectively).

In sensitivity analysis with ROTEM hypofibrinolysis defined as maximum lysis less than 4%, only 18 patients categorized as normal on admission transitioned to low at 24 h and 19 patients were categorized as persistently low on admission and at 24 h. In these regression models, the odds of multiorgan dysfunction syndrome were only significantly higher in those that transitioned from low-maximum lysis on admission to normal-maximum lysis at 24 h (adjusted odds ratio, 2.25 [95% CI, 1.06 to 4.76]), with low-maximum lysis at 24 h not a predictive factor (adjusted odds ratio, 2.30 [95% CI, 0.90 to 5.91]). In the sensitivity analysis with ROTEM hypofibrinolysis defined as tissue factor–activated ROTEM maximum lysis less than 6%, patients categorized as normal on admission that transitioned to low at 24 h or who were persistently low on admission and at 24 h were again characterized by higher odds of multiorgan dysfunction syndrome (adjusted odds ratio, 5.63 [95% CI, 1.44 to 22.12] and 4.67 [95% CI, 2.04 to 10.70], respectively), with only 24-h low-maximum lysis being a predictive factor (adjusted odds ratio, 3.38 [95% CI, 1.65 to 6.94]). When multiorgan dysfunction syndrome was scored to exclude the first 48 h after admission, only those with persistently low-maximum lysis on admission and at 24 h were characterized by higher odds of multiorgan dysfunction syndrome (adjusted odds ratio, 6.32 [95% CI, 2.36 to 16.96]), with both low admission and 24-h tissue factor–activated ROTEM maximum lysis having a significant role (adjusted odds ratio, 2.04 [95% CI, 1.07 to 3.87] and 2.98 [95% CI, 1.29 to 6.89], respectively).

We next examined the effect of tranexamic acid on these fibrinolytic transition patterns in a subgroup of 315 patients who activated the major hemorrhage protocol. Patients who did not receive tranexamic acid (79 of 315) had similar admission physiology and injury characteristics compared to those who received it before admission sampling (195 of 315; Supplemental Digital Content table 3, http://links.lww.com/ALN/C737). Of the 79 patients who did not receive tranexamic acid, 61 were injured before it was implemented into the major hemorrhage protocol and 18 patients deviated from the protocol after its introduction. Mortality within 24 h was lower in patients who received tranexamic acid before the admission sample (no tranexamic acid vs. tranexamic acid: 23 of 79 [29%] vs. 28 of 195 [14%], respectively; P =0.015), but overall mortality was not significantly different (no tranexamic acid vs. tranexamic acid: 31 of 79 [39%] vs. 55 of 195 [28%], respectively; P = 0.202); Supplemental Digital Content table 3, http://links.lww.com/ALN/C737).

Fibrinolytic transition patterns were markedly altered by tranexamic acid (fig. 3). Admission hyperfibrinolysis was present in 19 of 79 (24%) of those who did not receive tranexamic acid compared to 1 of 195 (<1%) in the non–tranexamic acid group (P < 0.001). Of the patients who did not receive tranexamic acid, 31 of 79 (39%) had low-maximum lysis on admission compared to 136 of 195 (70%) in those who had already been given tranexamic acid (P < 0.001). The proportion of patients with low-maximum lysis at 24 h remained higher in patients in the tranexamic acid group who survived at 24 h (no tranexamic acid vs. tranexamic acid: 14 of 56 [25%] vs. 83 of 167 [50%]; P = 0.006). Of those who survived to 24 h, a higher proportion of patients transitioned to low-maximum lysis at 24 h from normal-maximum admission lysis (no tranexamic acid vs. tranexamic acid: 3 of 25 [12%] vs. 16 of 57 [28%]), and these patients were more likely to be coagulopathic and severely injured (table 4). There was no significant difference in late mortality between patients who did and did not receive tranexamic acid in any of the different transition patterns. The incidence of multiorgan dysfunction syndrome and VTE was higher in persistently low-maximum lysis patients who received tranexamic acid compared to those who did not, although group size was small in the non–tranexamic acid group and patients who received tranexamic acid were significantly more shocked on admission and received more blood components other than erythrocytes (table 4).

In this secondary analysis of previously collected data from a prospective cohort study, we have shown the clinical significance of early transition patterns in ROTEM fibrinolysis states. Viscoelastic hemostatic assays are the only readily available diagnostic tool for detecting fibrinolysis, and in this study, ROTEM at 24 h was found to be a better discriminator of those at risk for multiorgan dysfunction compared to admission ROTEM. Early fibrinolysis transition patterns have potential to aid prognostication of late clinical outcomes which are not easily predicted from adverse admission physiology or injury characteristics alone.

The fibrinolytic response to injury, surgery, and resuscitation after major trauma that is detected by ROTEM is profoundly dynamic, and both the ROTEM hypofibrinolysis state at 24 h and the pattern of transition are key determinants of organ dysfunction and late mortality. The majority of patients had a persistent ROTEM fibrinolysis pattern, but one in three had a dynamic response. Nearly half either arrived with or developed ROTEM hypofibrinolysis by 24 h. Two different patient cohorts appear to be involved in the development of low-maximum lysis at 24 h: those with severe brain injury and those with shock and hemorrhage on admission. These patients with persistent or progressive ROTEM hypofibrinolysis had increased late mortality as well as a greater incidence and duration of multiorgan dysfunction syndrome, in line with previous evidence.^7,11  One in four patients displayed a recovery pattern of ROTEM hypofibrinolysis (low- that transitioned to normal-maximum lysis at 24 h), with mortality comparable to those who always had normal-maximum lysis. In patients who required activation of the major hemorrhage protocol, empiric tranexamic acid administration abolished hyperfibrinolysis on admission, reduced early mortality, and altered fibrinolysis transition patterns, but did not increase late mortality regardless of the pattern itself.

Patients with both normal-maximum lysis on admission who transitioned to a low fibrinolytic state by 24 h and those with persistent ROTEM hypofibrinolysis had increased risks of developing multiorgan dysfunction syndrome compared with those who remained in a normal lysis state. Utilizing the late ROTEM fibrinolysis state may have a role in stratifying risk groups for prophylactic intervention to ameliorate the hypercoagulable state associated with multiorgan dysfunction syndrome^8,36,37  and VTE. A subgroup of particular interest is the persistent ROTEM hypofibrinolysis group,^7,11  a cohort with severe traumatic brain injury but low levels of coagulopathy and transfusion requirements, who importantly are not readily differentiated on admission from those with low-maximum lysis who normalize by 24 h. We have previously shown that in patients with severe traumatic brain injury, ROTEM hypofibrinolysis on admission may in fact represent occult local hyperfibrinolysis driven by the plasminogen receptor S100A10, which is highly expressed in the brain.³⁸  In the context of the CRASH-III trial³⁹  and the survival benefit of tranexamic acid in this subgroup, the mechanisms of the fibrinolytic response in traumatic brain injuries and the precise effect of tranexamic acid on both clinical outcomes and the ROTEM-detectable response remain clear research priorities.

Several types of assays are available to quantify the degree of fibrinolytic activation, including fibrin–matrix systems, turbidimetric tests, single-protein levels, sonorheometry, and several variants of viscoelastic methods.^40–44  The net balance of fibrinolytic activators and inhibitors changes rapidly over time and in response to blood component transfusions. Biomarkers of fibrinolysis—while accepted as the definitive standard—are limited in that they provide an overall measure of fibrinolytic activity that has occurred up until the point of blood sampling rather than determining current fibrinolytic activity in vivo. Moreover, D-dimer, the primary clinical biomarker, is representative not only of raised fibrinolysis but also of increased activation of the coagulation system.⁴²  High levels of D-dimer do occur in patients with severe injuries leading to a physiologic activation of the coagulation system, in the absence of deranged fibrinolysis and abnormal ROTEM parameters. Point-of-care viscoelastic hemostatic assays may closer reflect—albeit with some degree of insensitivity² —the current state of the fibrinolytic system and are currently the only diagnostic tool we have available to assess fibrinolysis in real time. Further biomarkers analyses and comparison with ROTEM patterns over time are required to elucidate the mechanisms underlying the fibrinolytic patterns, what ROTEM hypofibrinolysis represents in vivo, and whether each of the transition patterns described represent the same changes in the regulators of fibrinolysis.^13,14 

Antifibrinolytics have been postulated to switch off normal levels of fibrinolysis, create a persistent fibrinolytic shutdown, and increase mortality and the incidence of multiorgan dysfunction syndrome and VTE, leading some institutions to limit tranexamic acid use to only those patients with ROTEM-diagnosed hyperfibrinolysis.^45,46  We have confirmed that prompt empiric administration of tranexamic acid is associated with reduced ROTEM hyperfibrinolysis and early deaths but importantly did not observe any increase in late mortality in patients who received tranexamic acid. Early low-maximum lysis is a poor determinant for the use of tranexamic acid since a significant proportion of patients will require a massive transfusion and die early; conversely, many other patients with low-maximum lysis on admission will require few blood products and revert to normal-maximum lysis at 24 h with low mortality. As described previously, early low-maximum lysis may signify an occult form of hyperfibrinolysis⁹  or a moderate form of fibrinolysis and coagulopathy,^13,14  while a late low-maximum lysis state, associated with poor outcomes, may represent a true shutdown of the fibrinolytic system in response to a hypocoagulable state during bleeding. The underlying process of ROTEM hypofibrinolysis or the factors leading to it, rather than tranexamic acid per se, appears to determine the actual risk of developing multiorgan dysfunction syndrome. In view of the lower early mortality associated with tranexamic acid and decreased benefit observed with delayed administration,⁴⁵  waiting for ROTEM results is likely to either reduce the efficacy or preclude the treatment of patients that may derive benefit.⁴⁶  In this study, we found no evidence to argue against empiric administration of tranexamic acid, nor did we find support for withholding its administration until results from viscoelastic hemostatic assays are available.

Assessing the impact of tranexamic acid on the incidence of VTE was compounded by the low event rate reporting in the subgroups of interest particularly before the introduction of tranexamic acid. Almost all thrombotic events occurred in patients who received tranexamic acid with late ROTEM hypofibrinolysis, but no binary comparison reached statistical significance. Other known risk factors for VTE (e.g., shock and severe traumatic brain injury [with potential for delayed administration of VTE prophylaxis]), were more common in patients who received tranexamic acid. While randomized clinical trials have not shown an increased rate of VTE in trauma patients who received tranexamic acid,^4,39,47  smaller observational studies have provided some evidence for an association between fibrinolysis shutdown and ROTEM hypofibrinolysis, tranexamic acid, and VTE.^6,11,17  Based on these findings and our data, early stratification of patients at increased risk of VTE (e.g., tranexamic acid administration and persistent ROTEM hypofibrinolysis) and further evaluation of augmented pharmacologic or mechanical VTE prophylaxis strategies are indicated.

There are other several limitations to this study. We focused on severely injured trauma patients, and therefore, the trends described here may not represent the overall trauma population. This study only examined a single viscoelastic hemostatic assay platform and therefore the results may not be directly transferable to other assays, such as thromboelastography. Approximately 30% of the patients were excluded based on missing ROTEM data at 24 h. Although they were representative of the analyzed cohort based on maximum lysis values on admission, multiple imputation was not performed as missing data was most likely nonrandom (e.g., less severely injured patients with earlier recovery could agree to remain part of the study but specifically refuse further blood sampling).³³  The sample size of the study cohort and the use of a standard methodology for dealing with the repeated measurements of tissue factor–activated ROTEM maximum lysis could have affected the results of the multivariable analyses. Future larger studies could explore methodologies better suited to deal with repeated measures.^31,48  At our institution, tranexamic acid was introduced in February 2011 and its use become consistent after 2014. Comparison between tranexamic acid– and non-tranexamic acid–treated patients is potentially confounded by changes in trauma care that occurred during this time, as well as the low numbers of VTE event reports before introduction of tranexamic acid. The tranexamic acid subgroup analysis is also possibly limited by a lack of power in detecting differences in the admission characteristics between the subgroups. As in many trauma systems, tranexamic acid is now given in the prehospital phase soon after injury; therefore, future studies evaluating the pharmacodynamic effects will require prehospital blood sampling before dosing.

In conclusion, this study demonstrates that both delayed-onset and persistent ROTEM hypofibrinolysis are associated with organ dysfunction and late mortality after injury. ROTEM hypofibrinolysis on admission describes a heterogeneous group, with some patients reverting to normal levels of ROTEM fibrinolysis with good outcomes. The transition patterns occurring early after injury may provide valuable prognostic information on clinical outcome that cannot be determined from admission tissue factor–activated ROTEM maximum lysis or clinical variables alone. Empiric tranexamic acid is associated with reduced early mortality and appears to alter the fibrinolytic patterns but does not increase late mortality regardless of the ROTEM fibrinolysis pattern in the first 24 h after injury.

This study was supported in part by research funding from the Fondazione Cassa Rurale di Trento, Trento, Italy (to Dr. Rossetto). The Center for Trauma Science (Blizard Institute, Queen Mary University of London, London, United Kingdom) currently receives departmental support for rotational thromboelastometry reagents and equipment from TEM International GmbH (Munich, Germany) and Hemosonics International (Asnières-sur-Seine, France) and has previously received thromboelastography reagents and equipment support from Haemonetics Corporation (Boston, Massachusetts). The supporting organizations had no role in the design and conduct of the study, preparation of the manuscript, or decision to submit it for publication.

The authors declare no competing interests.