Tranexamic acid for trauma: optimal timing of administration based on the CRASH-2 and CRASH-3 trials.

This article investigates the optimal timing and patient criteria for administering tranexamic acid (TXA) in trauma care, using machine learning–based causal forest models applied to data from the CRASH-2 and CRASH-3 randomized trials involving 28,448 trauma patients. The study finds that TXA is most effective in reducing 24-hour mortality when given within 2 hours of injury, with benefits extending beyond 2 hours in patients with severe traumatic brain injury (Glasgow Coma Scale [GCS] score < 9). The optimal treatment criteria identified are administration within 2 hours of injury or a GCS score below 9, encompassing 60% of the cohort, and these criteria were associated with sustained mortality benefits at 28 days. The findings support earlier TXA administration than current guidelines recommend and highlight the utility of machine learning methods to refine treatment strategies based on patient heterogeneity.