Tranexamic acid is associated with post-injury mortality in a resource-limited trauma system: Findings from the epidemiology and outcomes of prolonged trauma care cohort study

1 INTRODUCTION

Trauma-related deaths claim 4.4 million lives each year worldwide with unintentional injuries causing 3.2 million deaths and violence-related injuries responsible for another 1.3 million deaths.^{1, 2} Uncontrolled bleeding is the leading preventable cause of death among trauma victims.^{3, 4} Among those with major bleeding, the risk of mortality rises due to the “lethal triad of trauma,” which includes coagulopathy, hypothermia, and acidosis. Treating trauma-induced coagulopathy is particularly challenging and continues to be an area of intense investigation by the global trauma research community.^{5, 6} The lethal triad, along with a host of other pathologic hyperimmune responses during trauma, collectively termed trauma-induced coagulopathy, result in physiologic perturbations that increase post-injury morbidity and mortality.⁷

Hyperfibrinolysis, the pathologic breakdown of the fibrin mesh that stabilizes blood clots, is a major physiologic driver of trauma-induced coagulopathy and death.^{5, 8, 9} Tranexamic acid (TXA), a synthetic derivative of the amino acid lysine, competitively inhibits the enzymatic breakdown of fibrin blood clots and preserves hemostatic clots.^10-12 TXA has been useful in controlling excessive bleeding during various medical conditions and hemorrhage-prone surgical procedures. Additionally, TXA's ability to decrease fibrinolysis helps blunt the inflammatory response associated with hemorrhagic shock, potentially reducing the risk of multiple organ dysfunction.^{13, 14} TXA is also commonly used for minimizing and managing active hemorrhage during operative procedures with anticipated heavy blood loss.^{10, 14, 15}

The existing literature on the impact of TXA for post-injury mortality generally demonstrates that TXA improves survival, but several reports demonstrate mixed results.^16-19 A recent systematic review and bias-adjusted meta-analysis of randomized controlled trials suggests that TXA use for trauma leads to a reduction in 24-h and 30-day mortality.²⁰ A few other studies exist that either did not find or were uncertain about an association between TXA treatment and mortality.^21-26 Some studies reported that, among civilian adults hospitalized with traumatic hemorrhagic shock, those treated with TXA had a significant mortality reduction at 28 days, but not at 24 or 48 h, while others conversely reported an early survival advantage that is not maintained over the ensuing 6 months.^{27, 28} These varied findings may be partially attributable to study design and administration of TXA in patient populations with varying risk for fibrinolysis.²⁹

In line with efficacy studies, several observational studies have investigated the impact of timing and dosing of TXA. While some studies demonstrate that TXA reduces bleeding and mortality from hemorrhage, its effectiveness is closely tied to the timing of treatment.^30-32 Administering a “full” dose of TXA (i.e., completing a 1-g loading bolus and initiating a 1-g infusion) within 3 h of injury significantly lowers the risk of death from bleeding, whereas starting treatment after 3 h may be ineffective or even harmful. Several pharmacologic and clinical investigations suggest that 1 g of TXA may be as effective as a full conventional 2-g TXA dose.^{33, 34} Further evidence regarding the effectiveness of TXA is needed, including studies from diverse patient populations and regions of the world experiencing the highest trauma mortality with limited access to surgical care, such as low- and middle-income countries (LMICs).

The purpose of this study is to assess the association between TXA administration within 3 h of injury and subsequent 24-h mortality among severely injured patients with extra-cranial injuries with, or at risk for, major hemorrhage. This study uses secondary data from a prospective, multicenter cohort study in an LMIC population. Specifically, the study targets individuals with hemorrhage or those with a heightened risk of hemorrhagic shock to determine whether TXA treatment administered within 3 h from the time of injury has a significant effect on mortality.

2 METHODS

3 RESULTS

A total of 4007 eligible patients met the cohort inclusion criteria, but 12 patients (0.3%) had missing values for one or more key covariates (missing at random) and were subsequently excluded from the analysis (Figure 1). The remaining 3995 patients, including 502 (13%) who received TXA within 3 h of injury and 3493 (87%) who did not receive any TXA, were included in the PS model and IPTW computation (see Tables S1 and S2). Following the IPTW trimming approach, 388 untreated patients were excluded.

Table 1 provides the demographic, injury, and clinical characteristics of the weighted cohort (N = 3607) by treatment group. IPTW effectively balanced the distribution of key covariates between the treated and untreated groups (SMDs < 10% for all covariates). Overall, the average age of patients was 32.9 years (SD, 10.3 years), with 83.9% being male. Penetrating injuries comprised 72.3% of the cohort; 15.9% of the cohort had wounds. About 62% of patients had an NISS of <9 (mild), while 12.9% and 25.1% had an NISS of 9–15 (moderate) and >15 (severe), respectively. In terms of blood product transfusions, 3.1% of patients received whole blood, 11% received packed red blood cells or platelets, and 15.2% received plasma, freeze-dried plasma, or cryoprecipitate. Patients in the TXA group received significantly more blood products than individuals in the non-TXA groups. Among patients who received TXA within 3 h (n = 502), the average time from injury to initial TXA administration was 1.64 h (SD: 0.76 h). Regarding TXA dosage, 76.1% received 1 g of TXA, and 23.9% received 2 g (i.e., 1-g initial bolus followed by initiation of a 1-g infusion).

The results from the covariate-adjusted weighted logistic regression models for both primary and secondary outcomes, expressed as marginal effects, are summarized in Table 2. The group treated with TXA had a 38% lower odds of 24-h all-cause mortality compared to the untreated group (mOR, 0.62; 95% CI, 0.49–0.78). This significant reduction in all-cause mortality was similarly observed for secondary outcomes, including 48-h mortality (mOR, 0.57; 95% CI, 0.42–0.77), 7-day mortality (mOR, 0.61; 95% CI, 0.45–0.83), and 30-day mortality (mOR, 0.63; 95% CI, 0.48–0.82). Detailed results from the adjusted logistic regression models for these outcomes are provided in Table S3. Further, recognizing the importance of early time points in hemorrhage control, we analyzed TXA's impact at 6- and 12-h post-injury. While the 6-h reduction in mortality odds (mOR, 0.73; 95% CI, 0.49–1.09) was not significant, the 12-h reduction by 35% (mOR, 0.65; 95% CI, 0.49–0.86) was significant, as presented in Table S4.

The subgroup analyses largely aligned with the primary analysis, showing considerably lower all-cause mortality odds among patients who received TXA, except in two cohorts: those treated within 1 h of injury and those with blunt trauma. As per Table 3A, patients receiving TXA within 2 h of injury showed 43% lower odds of 24-h mortality compared to untreated patients (mOR, 0.57; 95% CI, 0.45–0.73), while those treated within 1 h showed no significant difference (mOR, 0.85; 95% CI, 0.68–1.07), likely due to the small size of the treated group, with less than 5% receiving TXA within this time frame in the weighted cohort. Patients receiving 1 g of TXA within 3 h were associated with lower 24-h mortality odds compared to untreated patients (mOR, 0.73; 95% CI, 0.56–0.94). Among patients with the highest risk of hemorrhage, TXA within 3 h was associated with a 60% lower 24-h mortality (mOR, 0.40; 95% CI, 0.30–0.50). In the no-blood-products subgroup, mortality was rare among TXA-treated patients, with no deaths at 24 h, 48 h, or 7 days, and one death at 30 days, preventing statistical modeling due to complete separation and insufficient events (Table S5). Regarding injury type, patients with penetrating injuries receiving TXA within 3 h showed lower 24-h mortality odds (mOR, 0.58; 95% CI, 0.43–0.78), while those with blunt injuries showed no significant difference (mOR, 0.83; 95% CI, 0.50–1.38), noting that patients with both injury types were excluded due to small sample size (Table 3B). These mortality reduction trends persisted across 48-h, 7-day, and 30-day intervals. Among TXA-treated patients, no statistically significant difference in mortality odds was observed between those receiving 1 and 2 g of TXA, as shown in Table 3B.

Regarding the sensitivity analyses, results were largely consistent with the primary analysis, supporting the robustness of the IPTW-weighted with trimming approach. Patients' baseline characteristics in each of the three sensitivity analyses are provided in Table S6. Table 4 summarizes the marginal treatment effects on primary and secondary outcomes for each of the three sensitivity analyses. In the first analysis, considering the full cohort without covariate balancing, patients treated with TXA within 3 h showed lower, though not statistically significant, 24-h mortality odds compared to untreated patients (mOR, 0.85; 95% CI, 0.64–1.13), likely due to covariate imbalance. The second analysis, using untrimmed IPTW, yielded results closely aligned with the primary analysis (mOR, 0.61; 95% CI, 0.48–0.77 for 24-h mortality), indicating that excluding extreme PS estimates did not impact the weighting process or the treatment effect estimation. Finally, the PSM with optimal full matching produced marginal odds ratio estimates for primary and secondary outcomes closely comparable to the primary analysis, with TXA administration within 3 h associated with 28% lower odds of 24-h mortality (mOR, 0.72; 95% CI, 0.56–0.91). Ratios of raw mortality counts by treatment group in each of these sensitivity analyses can be found in Table S7.

4 DISCUSSION

This multicenter observational study contributes additional evidence in support of TXA, as advocated for by the US National Trauma Research Action Plan and trauma experts worldwide.⁵⁷ We observed an association of a 38% reduction in odds of 24-h all-cause in-hospital mortality among severely injured South African patients with, or at heightened risk for, extra-cranial hemorrhage who received TXA within 3 h of injury versus those who did not. Consistent mortality benefits were also observed across secondary endpoints, subgroup analyses, and sensitivity analyses, underscoring a robust association between TXA administration within 3 h and lower mortality. Notably, while the mortality benefit remained significant for patients with penetrating injuries, it did not reach significance among those with blunt trauma. This study uniquely extends the evidence for TXA to trauma populations experiencing a high proportion of penetrating trauma (72% in our cohort) and within resource-limited settings where access to blood products and timely surgical care are limited. Importantly, the mortality benefit from TXA was independent of blood products or other life-saving interventions. Notably, the median time from injury to TXA administration was 1.6 h, and a 1-g dose was given in the majority (76%) of cases.

We acknowledge that the association with mortality reduction in our study is higher than that reported in other observational and interventional studies. A recent meta-analysis of seven randomized controlled trials found that TXA reduced death risk by 24% at 24 h and 11% at 1 month.²⁰ In the older CRASH-2 randomized placebo-controlled trial, TXA significantly reduced mortality—TXA given within 1 h of injury decreased bleeding death risk by 32%, and by 21% if administered between 1 and 3 h.⁵⁸ Further, the MATTERs Study, a retrospective analysis of wartime injuries, similarly showed lower in-hospital mortality of TXA versus no TXA patients (17.4% vs. 23.9%, respectively), a 27% decrease in all-cause mortality.⁵⁹ These and other findings indicate that TXA significantly decreases mortality in trauma patients, particularly when administered within 3 h of injury. The large mortality reduction in our study likely reflects the high proportion of penetrating injuries (72%), with over one-half of all patients experiencing stab wounds predominantly to the torso. This injury profile increases the likelihood of noncompressible bleeding, and/or an inflammatory response from visceral injury, which may make the population more responsive to TXA therapy. However, this mechanism of injury differs from the predominant firearm and collision mechanisms seen in other studies, which may further contribute to the larger difference seen in our study. Additionally, access to facilities is delayed, and access to surgical care and blood products is limited in this population.^{36, 60} The cumulative result may be a selection bias in which patients with major noncompressible hemorrhage die before reaching the facilities, selecting for patients' whose hemorrhage is more amenable to nonsurgical therapies, such as TXA. Those reasons may help explain why the benefit of TXA is more magnified in our study population compared to studies conducted in high-income settings that are not as resource-constrained. Hence, our study has implications for similar resource-constrained health systems, for example, low-income nations, austere and remote settings, and combat casualty care.

Existing literature suggests that hemostatic resuscitation with blood components or whole blood confers a significant survival benefit in hemorrhage. In our study cohort, 19% of patients received blood components or whole blood, raising the question of whether the observed mortality benefit was attributable to blood transfusions rather than TXA. However, in our IPTW-weighted models, patients who received specific blood components or whole blood demonstrated higher odds of mortality compared to those who did not receive transfusions, a finding that contrasts with the observed mortality-reducing effect of TXA but should be interpreted cautiously due to the covariate imbalance of blood products (Table S3). Thus, we can reasonably conclude that the mortality benefit and the substantial effect sizes associated with TXA in our study were not influenced by the administration of blood products.

Among several subgroup analyses in our study, one key finding revealed that administration of 1 g of TXA demonstrated comparable effectiveness to 2 g doses in reducing mortality odds. These findings align with recent literature, including Gunn et al. (2024), who found no significant differences in 28-day mortality across three TXA dosing strategies (1-g bolus vs. 1-g bolus plus 1-g infusion vs. 2-g bolus).³⁴ Similarly, El-Menyar et al.⁶¹ evaluated the safety and efficacy of a second in-hospital TXA dose versus placebo and reported that the additional dose provided no advantages over a single prehospital dose in terms of mortality rates, blood transfusion requirements, thromboembolic complications, organ failure, or hospital length of stay. Further validation in pharmacokinetic studies is needed. However, a 1 g TXA dose becoming the standard could yield pragmatic benefits for early prehospital administration (e.g., EMS transport times may only accommodate a 1-g loading dose) and increased feasibility of intramuscular TXA administration (current TXA concentrations permit a maximum of 1 g per intramuscular injection, not 2 g). A 1 g TXA dose has pragmatic implications.

5 CONCLUSION

This study suggests that early administration of TXA in severely injured trauma patients is significantly associated with reduced mortality risk across multiple time points. The protective effect of TXA was particularly pronounced in patients with penetrating injuries and those at the highest risk of hemorrhage, when administered within 1 to 3 h post-injury. Notably, a 1-g dose was effective in reducing mortality. The mortality reduction benefits were consistent across different analytical approaches, strengthening our findings. While TXA showed remarkable effectiveness in most subgroups, its impact varied by injury type. These findings support incorporating TXA into early trauma care protocols in LMIC environments, though further research is needed to optimize its use.

FUNDING INFORMATION

This work was supported by United States Department of Defense award numbers/grant log numbers: W81XWH-20-2-0042/BA190049, W81XWH-22-1-0883/RC210170, HT9425-23-1-1025/PR220453.

CONFLICT OF INTEREST STATEMENT

The authors have disclosed no conflicts of interest.