CRASH-2: Tranexamic Acid in Trauma Hemorrhage

TXA within 3 hr cut all-cause + bleeding mortality without thrombotic excess.

Population

20,211 trauma patients within 8 hr of injury with hemorrhage or hemorrhage risk (SBP <90, HR >110).

Intervention

TXA 1 g IV over 10 min, then 1 g over 8 hr.

Comparison

Matching placebo.

Outcome

All-cause 28-day mortality; secondary: bleeding mortality, thromboembolic events.

Massive multicenter RCT across 274 hospitals in 40 countries. Pragmatic enrollment — clinical suspicion of significant hemorrhage. Time-from-injury subgroups pre-specified.

• All-cause mortality: 14.5% TXA vs 16.0% placebo (RR 0.91, 95% CI 0.85-0.97, P=0.0035).
• Bleeding mortality: 4.9% vs 5.7% (P=0.008).
• TXA <1 hr from injury: 5.3% vs 7.7% (HUGE benefit).
• TXA 1-3 hr: 4.8% vs 6.1%.
• TXA >3 hr: 4.4% vs 3.1% (HARM signal — do not give late).
• No excess thromboembolic events.

Give TXA 1 g IV bolus + 1 g over 8 hr to any trauma patient with significant hemorrhage WITHIN 3 HOURS of injury. Earlier = better. Beyond 3 hr, signal flips toward harm — do not give. Now standard in MTP protocols (military, civilian, prehospital), incorporated into ATLS, REBOA programs. The cleanest, cheapest, easiest blood-saving intervention in trauma.

• Diverse global population — care quality varied across sites; effect size could differ in fully-resourced trauma centers.
• Imprecise time-from-injury reporting in some centers.
• Did not study TXA alongside modern hemostatic resuscitation (1:1:1, viscoelastic-guided) — assume additive.

1. Is your prehospital TXA protocol auditable for the <3-hr window?
2. How do you handle uncertain time-of-injury (e.g. found-down trauma)?
3. Is the >3-hr 'harm' signal real, or noise from differential survival of late-presenting patients?