Bayesian Statistics to Reanalyze Data From the STAAMP Trial

Tranexamic acid (TXA) is associated with improved survival following trauma in prior prehospital trials, shaping clinical practice. The Study of Tranexamic Acid During Air and Ground Medical Prehospital Transport (STAAMP) trial did not find a difference in mortality between TXA and placebo. A bayesian approach that incorporates prior clinical evidence may better characterize the impact of TXA.

To evaluate the probability of mortality benefit associated with prehospital TXA in trauma.

This quality improvement study used a post hoc bayesian analysis of data from the STAAMP trial, a prospective, multicenter, double-masked, placebo-controlled, phase 3 randomized clinical trial conducted at level I trauma centers in the US from May 1, 2015, to October 31, 2019. Patients at risk for hemorrhage within approximately 2 hours of injury were randomized to receive prehospital TXA or placebo. The data analysis was performed between January 1, 2024, and December 31, 2025.

The primary outcome was 30-day mortality assessed using frequentist statistics. Bayesian hierarchical logistic regression models were built to estimate the posterior probability of mortality associated with TXA. The prior distributions were informed by Clinical Randomization of an Antifibrinolytic in Significant Hemorrhage 2 (CRASH-2) and Prehospital Antifibrinolytics for Traumatic Coagulopathy and Hemorrhage (PATCH-Trauma) trial data, and the influence of prior selection, sample size, and varying risk thresholds was also evaluated.

Of 903 STAAMP patients analyzed (median [IQR] age, 39 [26-55] years; 668 male [74.0%]), 447 received TXA and 456 received placebo. The majority of patients (n = 760 [84.2%]) sustained blunt injuries, with a median injury severity score of 12 (IQR, 5-22). In the frequentist analysis, the 30-day mortality rates were 8.1% and 9.9% for the TXA and placebo groups, respectively (hazard ratio, 0.81; 95% CI, 0.59-1.11). Using bayesian models, the estimated posterior risk ratios were 0.91 (95% credible interval [CrI], 0.85-0.97) with a CRASH-2 prior, 0.80 (95% CrI, 0.65-0.97) with a PATCH-Trauma prior, and 0.82 (95% CrI, 0.52-1.23) under a noninformative prior. The results were robust across confounders, site clustering, and alternative priors. The posterior probability that TXA reduced mortality ranged from 84% to 99%.

This quality improvement study using a post hoc bayesian reanalysis of the STAAMP trial suggested a high probability that prehospital TXA would improve survival. Bayesian methods may offer refined inference and support clinical decision-making in prehospital trauma care.