POISE-3: Tranexamic Acid in Non-Cardiac Surgery

TXA reduced bleeding without an offsetting thrombosis penalty.

Population

9,535 adults ≥45 years undergoing noncardiac surgery with at least one cardiovascular risk factor and a moderate-to-high bleeding risk.

Intervention

Tranexamic acid 1 g IV at start + 1 g at end of surgery.

Comparison

Placebo.

Outcome

Composite primary efficacy: life-threatening, major, or critical-organ bleeding within 30 days. Composite primary safety: MI, non-hemorrhagic stroke, peripheral arterial thrombosis, symptomatic VTE within 30 days.

International, randomized, double-blind, placebo-controlled trial. Factorial design with second arm randomizing hypotension-avoidance strategy (separate analysis).

• Bleeding composite: 9.1% TXA vs 11.7% placebo (HR 0.76, 95% CI 0.67-0.87, p<0.001). Significant reduction.
• Cardiovascular safety composite: 14.2% TXA vs 13.9% placebo (HR 1.02, 95% CI 0.92-1.14). NOT statistically inferior to placebo, but did not meet pre-specified non-inferiority margin (so technically: failed to prove non-inferiority).
• Subgroup: TXA benefit consistent across surgery types.
• Mortality: similar.

TXA has graduated from 'use in trauma + OB' to 'reasonable in many noncardiac surgery patients.' POISE-3 showed clear bleeding reduction; the safety story is 'not worse but not formally proven equivalent.' Most experts have integrated peri-induction TXA 1 g (+1 g at end) into many high-bleeding-risk cases (orthopedic, spine, vascular) when patients don't have absolute contraindications (active intravascular thrombosis, history of seizure on prior TXA).

• Composite endpoints can mask divergent component effects.
• Failed to formally prove cardiovascular non-inferiority — a real but small concern, especially in vascular patients.
• Open-label dose 1g IV × 2 may not be optimal for all body weights (10-20 mg/kg may be more standard in some protocols).

1. When do you give TXA peri-operatively now? Has this trial expanded those indications?
2. How do you weigh the bleeding reduction against the small unproven thrombosis signal in patients with prior CVA?
3. Does dose matter? 1 g flat vs 10-20 mg/kg vs 30 mg/kg loading — what's your institutional protocol?