Abstract Tranexamic acid (TXA) reduces mortality in patients with trauma; however, optimal patient selection remains unclear. This study aimed to identify trauma subgroups most likely to benefit from TXA administration by integrating systematic evidence mapping with trauma phenotype analysis derived from the Japan Trauma Data Bank. We conducted a two-phase study: first, a systematic search of MEDLINE, Web of Science, and the Cochrane Library databases (inception to June 28, 2024) and identified randomized controlled trials (RCTs) evaluating TXA in trauma; second, we assessed TXA's association with mortality across phenotypes derived through machine learning–based clustering using 14 variables available during initial trauma care. Among eligible studies, control group mortality and number needed to treat (NNT) were calculated and visualized via bubble plots (size = sample size). Five RCTs (n = 894–20,127; published 2010–2023) and one phenotype study (n = 24,058; four phenotypes) were included, all reporting mortality as an outcome. At approximately 1 month post-injury, control group mortality in RCTs ranged from 10 to 21.8%, whereas in-hospital mortality across phenotypes ranged from 3.9 to 51.4%. NNT varied from 22 to 68 (RCTs) and from 10 to 98 (phenotype study), with an inverse relationship between baseline mortality and NNT, indicating greater TXA benefit in higher-risk groups. This study suggests that TXA is more effective in trauma subgroups with higher baseline mortality. Phenotype-driven stratification using initial clinical data may support more targeted TXA administration and improve patient outcomes. Further research is needed to validate these phenotypes for clinical implementation.
Tachino et al. (2026) studied this question.