This study is among the first to specifically address VTE as a primary outcome after administration of TXA in a civilian population. TXA was found to be an independent risk factor for VTE when comparing treated and untreated cohorts, which were matched based on readily available clinical characteristics that determine propensity to receive TXA therapy. Treatment with TXA was more frequently associated with mortality, transfusion requirement, increased ICU and hospital length of stay, and higher number of operative interventions during admission in univariate analyses. After adjusting for clinically relevant confounders of thromboembolic complications, which were prespecified based on existing literature, patient weight, presence of hip, pelvis, or leg fracture, and red blood cell transfusion were significantly associated with VTE. Mortality was not found to be significantly different between treated and untreated cohorts in an adjusted model.
Our findings diverge from the CRASH-2 trial, which differed in setting, population, and availability of interventions. Specifically, it has been argued that the limited prehospital care in middle- to low-income countries bestows uncertainty with regard to generalizability to our trauma systems.18 Patients eligible for the CRASH-2 study were those for whom the treating physician had uncertainty with regard to the benefit of TXA, further limiting applicability of this subpopulation to our cohort. The patients in our study, though less severely injured than those represented by the MATTERs study, may have sustained more significant injury than those in the CRASH-2 study as 89% of treated patients and 64% of untreated patients required transfusion in our investigation as compared to half of patients in CRASH-2. These differences represent concerns with regard to the applicability and predictability of previous studies that may affect the accuracy of both the mortality benefit and risk of VTE associated with TXA.19
Combat casualty studies have advanced our understanding of the effects of TXA. While others have demonstrated an association between treatment with TXA and VTE,10 critics have expressed concern with regard to the generalizability of these findings to the civilian trauma population.20 Differences in injury mechanism and severity may contribute to the higher incidence of both MT and VTE in military patients. Even in military populations, VTE occurrence varies widely with studies reporting rates as low as 1.8%8 and 4.0%,9 to as high as 15.6%.10 The MATTERs study demonstrated a crude association between TXA and VTE that was no longer significant after a multivariate logistic regression analysis. In our study, while 11.4% of all patients experienced VTE, only 7.4% of individuals who had not received TXA had DVT or PE. Though clinical guidelines aim to direct appropriate use of TXA, there is concern that increased use of TXA in patients who do not require MT may unnecessarily place patients at an increased risk of thrombotic complications. Similarly, in our study, the majority of the patients who were treated with TXA and suffered VTE did not receive MT. Although 76% of these patients would have been predicted to require MT by their Assessment of Blood Consumption score, it is not possible to determine from available data whether TXA administration prevented need for MT. Further investigations are needed to determine whether VTE rates are higher in the subset of patients who receive TXA preemptively but do not require MT.
Prior propensity matched cohort studies have attempted to address benefits and risks associated with TXA administration, though none have specifically investigated the effect of TXA on thromboembolic complications as a primary outcome. One such single-center study investigating a more severely injured subset of trauma patients than the CRASH-2 trial demonstrated that TXA increased transfusion requirement as well as mortality relative to propensity-matched controls who had not received treatment with TXA.21 Subgroup analysis indicated that TXA may confer survival benefit in patients who require less than eight units of blood transfusion or who do not require surgery. Others have suggested that this survival benefit from TXA may be inversely related to transfusion requirement.22–24 Fibrinolysis shutdown25 has been cited as a potential negative effect of the use of antifibrinolytics in severely injured patients. In our study, increased transfusion requirement and unadjusted mortality among those who received TXA compared with those who did not may be related to these processes. Thus, these studies and our data suggest more selective criteria are needed for the administration of TXA to maximize survival and reduce morbidity, especially in light of our findings that TXA increases risk of VTE. Along these lines, further investigations are necessary to elucidate whether the optimal time to administer TXA may be in the prehospital setting12,26 prior to initiation of transfusion or operative intervention. Additionally, risk of VTE may not be increased when limiting TXA therapy to the prehospital setting. In their propensity-matched cohort analysis, Wafaisade et al. found no increased risk of thromboembolism in civilian patients who had received TXA prior to hospital arrival.22
To date, previous investigations have not considered time-to-VTE in trauma populations. Data from nontrauma literature is insufficient and has used arbitrary cutoffs to designate early and late events.27,28 This study adds insight into differences in posttraumatic VTE distribution over time, which might inform criteria for early and late events. Based on the divergence pattern of cumulative incidence in VTE, the curves largely diverge in the second and third week postinjury, suggesting that risk of late, but not early VTE, may be increased in patients who receive TXA compared with those who had not. Further studies are needed to define chronological thresholds as well as clinical markers associated with temporal trends in VTE distribution. Once identified, a more individualized approach to thromboprophylaxis29,30 can be adopted for patients who initially required TXA and in whom hemostasis has been achieved.
This study has limitations worth noting. Propensity matching was performed using prospectively collected data from our institution’s trauma database. This represents a tradeoff of limiting the number of included subjects in favor of reducing selection bias. These data were restricted and may not include other important clinical criteria that are factored into the decision to administer TXA. Although previous literature cites the benefits of administering TXA within a 3-hour window from time of injury,15 available data required that time of hospital presentation be used as a proxy for time of injury. As our institution is a tertiary referral center with a wide catchment area, time of injury and presentation may have differed markedly. During the last 18 months of the study period, a randomized trial protocol was underway that included prehospital and variable in-hospital TXA dosing when transported by a large regional air medical service, and these patients were excluded since treatment was blinded and patients could receive variable doses of TXA from none (placebo) to 3 g. Additional data regarding prehospital medication and blood product use that may have guided TXA administration were incomplete. We restricted our propensity model to variables available prior to TXA administration, thus there remained important differences between groups among in-hospital variables after matching that required further adjustment in multivariable models. Thromboelastography data, which may have more accurately assessed coagulopathy, hyperfibrinolysis, and TXA administration31 was limited and therefore not incorporated into our analysis. While our multivariate analysis was intended to adjust for confounding, biases may remain. The analysis cohort is small, limiting our power to detect a significant association between TXA and certain factors, including survival; this is particularly notable if the absolute survival advantage is only 1.5% as seen in CRASH-2. Finally, while it is a possibility that providers who were aware that a patient had received TXA may have had a higher suspicion of thromboembolic events, we do not believe that this was a significant source of surveillance bias. Previous studies on surveillance bias refer to screening practices in asymptomatic patients32,33 rather than symptomatic patients, such as those in our study. Although data are varied, since the existing CRASH2 and MATTERs studies (the TXA investigations par excellence) indicate no significant increase in VTE as a result of TXA administration, it is unlikely that a provider would increase their use of imaging studies for VTE diagnosis based solely on whether a patient received TXA.
Previous studies vary both in population of interest and findings with regard to the mortality benefit of TXA and risk of VTE. Our data demonstrates that TXA may be an independent risk factor for VTE development, but was not associated with a survival benefit in this single-center cohort study. Further investigation is needed to identify which injured patients have a survival advantage from TXA, especially given the risks of this intervention. Additionally, the utility of screening for VTE and measures to ensure that prophylaxis is administered rapidly and efficaciously in patients who are treated with TXA should be explored. Such studies will allow a more individualized treatment approach that maximizes benefits and mitigates potential harms.
Significant contributions were made by all listed authors. Specifically, M.D.N. and J.B.B. were responsible for conceptualization of this study. Data acquisition, analysis, and drafting of this article were performed by S.P.M. and J.B.B., M.D.N., M.R.R., J.L.S., and M.E.K. provided assistance in data interpretation, analysis, and critical review of this article.
M.D.N. is an external scientific advisor for Janssen Pharmaceuticals. All other authors have no disclosures.
There are no sources of funding for this study.
M.D.N. and J.B.B. contributed equally as senior author to this publication.
Conflict of interest: M.D.N. is an external scientific advisor to Janssen Pharmaceuticals. Remaining authors have no disclosures.
2. Kauvar DS, Lefering R, Wade CE. Impact of hemorrhage on trauma outcome: an overview of epidemiology, clinical presentations, and therapeutic considerations. J Trauma
3. Gruen RL, Jurkovich GJ, McIntyre LK, Foy HM, Maier RV. Patterns of errors contributing to trauma mortality: lessons learned from 2,594 deaths. Ann Surg
4. Madurska MJ, Sachse KA, Jansen JO, Rasmussen TE, Morrison JJ. Fibrinolysis in trauma: a review. Eur J Trauma Emerg Surg
5. Geerts WH, Code KI, Jay RM, Chen E, Szalai JP. A prospective study of venous thromboembolism
after major trauma. N Engl J Med
6. Brill JB, Badiee J, Zander AL, Wallace JD, Lewis PR, Sise MJ, Bansal V, Shackford SR. The rate of deep vein thrombosis
doubles in trauma patients with hypercoagulable thromboelastography. J Trauma Acute Care Surg
7. Levi M, Levy JH, Andersen HF, Truloff D. Safety of recombinant activated factor VII in randomized clinical trials. N Engl J Med
8. Ng W, Jerath A, Wąsowicz M. Tranexamic acid
: a clinical review. Anaesthesiol Intensive Ther
9. CRASH-2 trial collaborators, Shakur H, Roberts I, Bautista R, Caballero J, Coats T, Dewan Y, El-Sayed H, Gogichaishvili T, Gupta S, Herrera J, et al. Effects of tranexamic acid
on death, vascular occlusive events, and blood transfusion in trauma patients with significant haemorrhage (CRASH-2): a randomised, placebo-controlled trial. Lancet
10. Morrison JJ, Dubose JJ, Rasmussen TE, Midwinter MJ. Military application of tranexamic acid
in trauma emergency resuscitation (MATTERs) study. Arch Surg
11. Johnston LR, Rodriguez CJ, Elster EA, Bradley MJ. Evaluation of military use of tranexamic acid
and associated thromboembolic events. JAMA Surg
12. Nishida T, Kinoshita T, Yamakawa K. Tranexamic acid
and trauma-induced coagulopathy. J Intensive Care
13. Brown JB, Neal MD, Guyette FX, Peitzman AB, Billiar TR, Zuckerbraun BS, Sperry JL. Design of the study of tranexamic acid
during air medical prehospital transport (STAAMP) trial: addressing the knowledge gaps. Prehosp Emerg Care
14. Sainani KL. Propensity scores: uses and limitations. PM R
15. Rosenbaum PR, Rubin D. Constructing a control group using multivariate matching sampling methods that incorporate the propensity score. Am Stat
16. Ramirez RJ, Spinella PC, Bochicchio GV. Tranexamic acid
update in trauma. Crit Care Clin
17. Nunez TC, Voskresensky IV, Dossett LA, Shinall R, Dutton WD, Cotton BA. Early prediction of massive transfusion in trauma: simple as ABC (Assessment of Blood Consumption)? J Trauma
18. Cornelius BG, McCarty K, Hylan K, Cornelius A, Carter K, Smith KWG, Ristic S, Vining D, Cvek U, Trutschl M. Tranexamic acid
: promise or panacea: the impact of air medical Administration of Tranexamic Acid
on morbidity, mortality, and length of stay. Adv Emerg Nurs J
19. Mitra B, Mazur S, Cameron PA, Bernard S, Burns B, Smith A, Rashford S, Fitzgerald M, Smith K, Gruen RL; PATCH-Trauma Study Investigators. Tranexamic acid
for trauma: filling the ‘GAP’ in evidence. Emerg Med Australas
20. Etchill EW, Fang R, Haut ER. Does tranexamic acid
cause venous thromboembolism
after trauma? JAMA Surg
21. Valle EJ, Allen CJ, Van Haren RM, Jouria JM, Li H, Livingstone AS, Namias N, Schulman CI, Proctor KG. Do all trauma patients benefit from tranexamic acid
? J Trauma Acute Care Surg
22. Shiraishi A, Kushimoto S, Oromo Y, Matsui H, Hagiwara A, Murata K. On behalf of the Japanese observational study for coagulation and thombolysis in early trauma (J-OCTET). Effectiveness of early administration of tranexamic acid
in patients with severe trauma. BJS
23. Cole E, Davenport R, Willett K, Brohi K. Tranexamic acid
use in severely injured civilian patients and the effects on outcomes: a prospective cohort study. Ann Surg
24. Harvin JA, Peirce CA, Mims MM, Hudson JA, Podbielski JM, Wade CE, Holcomb JB, Cotton BA. The impact of tranexamic acid
on mortality in injured patients with hyperfibrinolysis. J Trauma Acute Care Surg
25. Moore HB, Moore EE, Gonzalez E, Chapman MP, Chin TL, Silliman CC, Banerjee A, Sauaia A. Hyperfibrinolysis, physiologic fibrinolysis, and fibrinolysis shutdown: the spectrum of post-injury fibrinolysis and relevance to antifibrinolytic therapy. J Trauma Acute Care Surg
26. Wafaisade A, Lefering R, Bouillon B, Böhmer AB, Gäßler M, Ruppert M; Trauma Register DGU. Prehospital administration of tranexamic acid
in trauma patients. Crit Care
27. McClendon J Jr, Smith TR, O'Shaughnessy BA, Sugrue PA, Thompson SE, Koski TR. Time to event analysis for the development of venous thromboembolism
after spinal fusion ≥ 5 levels. World Neurosurg
28. Tzeng CW, Curley SA, Vauthey J, Aloia TA. Distinct predictors of pre- versus post-discharge venous thromboembolism
after hepatectomy: analysis of 7621 NSQIP patients. HPB (Oxford)
29. Costantini TW, Min E, Box K, Tran V, Winfield RD, Fortlage D, Doucet J, Bansal V, Coimbra R. Dose adjusting enoxaparin is necessary to achieve adequate venous thromboembolism
prophylaxis in trauma patients. J Trauma Acute Care Surg
30. Ko A, Harada MY, Barmparas G, Chung K, Mason R, Yim DA, Dhillon N, Margulies DR, Gewertz BL, Ley EJ. Association between enoxaparin dosage adjusted by anti-factor Xa trough level and clinically evident venous thromboembolism
after trauma. JAMA Surg
31. Chapman MP, Moore EE, Ramos CR, Ghasabyan A, Harr JN, Chin TL, Stringham JR, Sauaia A, Silliman CC, Banerjee A. Fibrinolysis greater than 3% is the critical value for initiation of antifibrinolytic therapy. J Trauma Acute Care Surg
32. Haut ER, Schneider EB, Patel A, Streiff MB, Haider AH, Stevens KA, Chang DC, Neal ML, Hoeft C, Nathens AB, et al. Duplex ultrasound screening for deep vein thrombosis
in asymptomatic trauma patients: a survey of individual trauma surgeon opinions and current trauma center practices. J Trauma
33. Haut ER, Pronovost PJ. Surveillance bias in outcomes reporting. JAMA
Dr. Adrian A. Maung (New Haven, Connecticut): Good morning. I want to thank the Association for the opportunity to discuss this very interesting paper and congratulate Dr. Myers and her colleagues on a very well-presented study, as well as the acceptance of their paper into the Journal of Trauma.
Since the publication of the CRASH-2 trial in 2010, despite the relatively small absolute decrease in mortality, TXA has been viewed as a magical potion for patients in hemorrhagic shock. Yet, subsequent studies have raised concerns that there are potential side effects, including this paper.
Dr. Myers reports increased rates of VTE in patients who received TXA compared to patients identified by propensity score matching who did not receive TXA.
But this study does have several limitations, some of which have been already acknowledged in the presentation that I would like to discuss and point out further.
1) The propensity matching used very reasonably criteria that were known to the clinicians at the beginning when the patient arrived yet this created some important differences between the two groups, including higher rates of penetrating injury in the TXA group as well as higher level of patients who did not receive VTE prophylaxis at all in the TXA group.
So do you think these results would have been different if you had matched your groups based on injury severity, use of VTE prophylaxis, et cetera, or perhaps that’s a topic for another paper.
2) Pittsburgh is a very well-experienced, very well run trauma center but the rates of VTE prophylaxis were quite low and their rates of VTE were much higher compared to CRASH-2 or MATTER trials. How do you think this affected your results?
3) Did you examine the need for surgery, especially surgery required for bleeding control, between the two groups and how it differed?
And my last question is what is the take-home message? What do we do when we get home from San Diego and see that patient in hemorrhagic shock? Do we still give TXA? Should we be more vigilant about VTE prophylaxis? About VTE screening? What should we do?
Again, congratulations on a great study. Thank you.
Dr. Matthew Martin (San Diego, California): Nicely presented paper. I do have significant concerns, along with Adrian, about the data. One of the reasons you do propensity matching is to pseudo-randomize but your groups are so different on multiple factors that impact the VTE rate so I’m just a little unclear about how you then make the leap to TXA is the associated or causal factor?
Another important factor that we would need to know is what was the use of routine screening for DVT, and for PE? With a retrospective study it’s always a challenge to get at the true incidence of VTE unless there has been uniform and routine screening, particularly for DVT. And then trying to nail down any causal factors becomes highly unreliable and biased, such as jumping from TXA as causing the VTEs.
We also know that DVT and PE now are two different processes with actually different risk factors so did you look at them independently as opposed to the combined group of lumping them together? Thanks.
Dr. Krista L. Kaups (Fresno, California): Good morning. My question really relates very much to Dr. Martin’s question and this is the difficulty of a retrospective study is how were PE and – well, how was VTE diagnosed. Was this by ultrasound? By CT scan? Do you know how was this determined retrospectively? Thank you.
Dr. Nasim Ahmed (Neptune, New Jersey): My question to you about the propensity score analysis, when you matched that two groups and you said that was a good match, instead of doing a relative risk reduction or absolute risk reduction why did you do the regression analysis? Thank you.
Dr. Martin Schreiber (Portland, Oregon): Dr. Myers, Dr. Martin asked you how you will change management based on your findings. You showed no benefit and only negative outcomes. Will you stop giving TXA?
Dr. Sara P. Myers (Pittsburgh, Pennsylvania): I would like to begin by thanking Dr. Maung for his comments and questions. I will first address his question regarding the propensity matching methodology itself. We acknowledge that there are significant differences between treated and untreated cohorts. Other studies that have used propensity score methodology to address use of TXA have similar limitations. There have been, to my knowledge, four propensity score matched studies that address similar questions with regard to TXA administration. These investigations match on very different variables, ranging from 4 variables up to 20. It is certainly possible that, had the groups been matched on other parameters, the results may have been different. That having been said, I do not believe that matching on injury severity of VTE prophylaxis would be appropriate. Matching is meant to be done on variables that predict the propensity that a patient would receive TXA. In this sense, we would not have matched on injury severity score, which is retrospectively assessed, or VTE prophylaxis, which occurs after the fact. ISS and VTE prophylaxis may be confounders, which is why we adjusted for these in our regression model. As such, I would like to emphasize that the cohorts are different based on the variables that we were adjusting for in the regression but they were not different in the predictive factors that we used to match. Ultimately, prospective randomized studies are needed to determine which patients benefit most from TXA, especially given the potential risks of this intervention.
Dr. Maung’s second question relates to how our VTE prophylaxis and rates compared to the CRASH-2 and MATTERs trials, and how this may have affected our results. As I mentioned, there are concerns regarding the generalizability and applicability of these trials to the patients we treat as Trauma Surgeons in the United States. I concede that the rate of VTE prophylaxis was low. This may reflect changes in VTE prophylaxis practices over time, but when we revisited patient charts to abstract the rationale for withholding prophylaxis, we found that the majority of patients had either died or been discharged prior to VTE prophylaxis being initiated. Additionally, in the treated group, VTE occurred in 12% of the patients who didn’t receive thromboprophylaxis prior to thromboprophylaxis being started. In this subset of patients, thromboprophylaxis was delayed given concern for bleeding risk. The rates of VTE quoted in the literature vary depending on injury severity and patient population. Though our results differ from those in the CRASH-2 and MATTERs trial, this may, again, stem from the fact that our patient population was fundamentally different than in these two studies.
We did examine the need for surgery. Patients treated with TXA required significantly more operative intervention(s) for bleeding control than the untreated cohort.
Turning to Dr. Maung’s final question- what is the take-home message? The take-home message is that TXA may increase the risk for serious adverse events such as VTE without a concomitant survival benefit. While this may not be true for all patients, it is likely true for some. I believe that we need to identify the patients who maximally benefit from TXA. Individuals who do not benefit from the agent may be exposed to unnecessary risks associated with its administration. Either way, it is important to acknowledge that as a hemostatic adjunct, TXA may increase the risk of thrombosis. We should administer TXA when we believe it will be a life-saving intervention, as that is our priority, and remain vigilant that VTE may occur taking care to institute appropriate prophylaxis, and perhaps, screening as well.
Finally, I will address the question regarding our screening process. We did not have routine screening of either DVT or PE. All of the VTE were diagnosed based on a clinical suspicion. PE was diagnosed based on CTA primarily; individuals suspected of having a DVT received ultrasound.
I would like to end by thanking the audience and Dr. Maung for their comments and questions, and appreciate the opportunity to present our data.