Tranexamic acid in bleeding trauma patients: an exploration of benefits and harms

Background: The CRASH-2 trial showed that tranexamic acid (TXA) administration reduces mortality in bleeding trauma patients. However, the effect appeared to depend on how soon after injury TXA treatment was started. Treatment within 3 h reduced bleeding deaths whereas treatment after 3 h increased the risk. We examine how patient characteristics vary by time to treatment and explore whether any such variations explain the time-dependent treatment effect. Methods: Exploratory analysis were carried out, including per-protocol analyses, of data from the CRASH-2 trial, a randomised placebo-controlled trial of the effect of TXA on mortality in 20,211 trauma patients with, or at risk of, significant bleeding. We examine how patient characteristics (age, type of injury, presence or absence of head injury, Glasgow coma scale (GCS), systolic blood pressure and capillary refill time) vary with time to treatment and use univariable (restriction) and multivariable methods to examine whether any such variations explain the time-dependent effect of TXA. If not explained by differences in patient characteristics, we planned to conduct separate prespecified subgroup analyses for the early benefit and late harm. Results: There was no substantial variation in age or capillary refill by time to treatment. However, the proportion of patients with blunt trauma, the proportion with head injury and mean systolic blood pressure increased as time to treatment increased. Mean GCS decreased as time to treatment increased. Analyses restricted to patients with blunt trauma, those without head injury and those with a systolic blood pressure <100 mmHg showed that these characteristics did not explain the time-dependent treatment effect. In a multivariable analysis the interaction with time to treatment remained highly significant (p < 0.0001). Separate subgroup analyses that examine how the benefits of early TXA treatment and the harms of late TXA treatment vary by systolic blood pressure (<= 75, 76-89, >89 mmHg); GCS (severe 3-8, moderate 9-12, mild 13-15); and type of injury (penetrating versus blunt) showed no significant heterogeneity. Conclusions: The time-dependent effect of TXA in bleeding trauma patients is not explained by the type of injury, the presence or absence of head injury or systolic blood pressure. When given within 3 h of injury, TXA reduces death due to bleeding regardless of type of injury, GCS or blood pressure.