Skip to main content

Journal Club

Cerebrovascular

Tranexamic Acid in Patients Undergoing Noncardiac Surgery

The New England Journal of Medicine | 2022

Rapid review note

Journal Club is a rapid, AI-assisted appraisal layer. It highlights study design, effect estimates, and practice relevance, but it is still a briefing, not a replacement for the paper.

For education only. Not medical advice.

Paper snapshot

Rapid study overview

Open paper

DOI

10.1056/NEJMoa2201171

PMID

N/A

PICO

Population

9535 adults undergoing noncardiac surgery at 114 hospitals in 22 countries who were at risk for bleeding and cardiovascular events.

Intervention

Tranexamic acid 1 g intravenously at the start and end of surgery.

Comparator

Placebo.

Outcomes

Thirty-day major bleeding and thirty-day major cardiovascular complications.

Design

Type

International randomized placebo-controlled trial

Randomized

Yes

Multicenter

Yes

Blinded

Placebo-controlled

Follow-up

30 days

Primary endpoint

Composite bleeding outcome at 30 days and composite cardiovascular safety outcome at 30 days.

Secondary endpoints

  • Individual bleeding components
  • Individual cardiovascular components
  • Disability at 30 days

Practice impact

What this means

POISE-3 showed that tranexamic acid cuts major bleeding in noncardiac surgery, but the trial stopped short of proving cardiovascular safety noninferiority. The practical read is not 'never use it' or 'use it on everyone' - it is most persuasive when a patient has meaningful bleeding risk and only modest thrombotic concern.

Bottom line

Tranexamic acid reliably reduces major perioperative bleeding in noncardiac surgery, but this trial did not fully rule out a small increase in cardiovascular complications, so it supports selective rather than automatic use.

Strength of evidence

high

Recommendation

consider change

Why it matters

  • Bleeding reduction was clear, statistically strong, and clinically relevant.
  • The cardiovascular difference was small in absolute terms but the noninferiority criterion was not met.
  • Use is easiest to justify when bleeding risk is substantial and thrombotic risk is acceptable.

What would change my mind

  • A similarly large trial or pooled analysis definitively showing cardiovascular noninferiority in comparable surgical patients.
  • Procedure-specific evidence showing no important benefit in surgeries with low baseline bleeding risk.

Critical appraisal

How strong is the paper?

Methods critique

Risk of bias

Randomization and placebo control make internal validity strong, although the trial used a per-protocol analysis for the noninferiority safety question rather than a pure intention-to-treat primary analysis.

Confounding

Randomization should minimize baseline confounding, and the large international sample makes major imbalances unlikely.

Missing data

Outcome denominators were slightly smaller for the cardiovascular analysis than for the bleeding analysis, but the paper also reports a similar intention-to-treat sensitivity analysis.

Multiplicity

The superiority test for bleeding and the noninferiority test for cardiovascular safety answer different questions, and the many secondary outcomes should not be overread.

Notes

  • This was a very large trial across 114 hospitals in 22 countries.
  • The key tension is efficacy for bleeding versus incomplete proof of cardiovascular safety noninferiority.
  • The safety margin was prespecified and was not met.

Stats check

NNT

Approximate NNT 39 to prevent 1 composite bleeding event.

Effect sizes

  • Composite bleeding was lower with tranexamic acid: HR 0.76 (95% CI 0.67-0.87; P<0.001).
  • Composite cardiovascular events were similar numerically: HR 1.02 (95% CI 0.92-1.14).
  • Noninferiority was not established because the upper one-sided 97.5% CI was 1.14, above the 1.125 margin.

Absolute effects

  • Bleeding events occurred in 9.1% with tranexamic acid versus 11.7% with placebo, an absolute difference of -2.6 percentage points.
  • Cardiovascular events occurred in 14.2% versus 13.9%, an absolute difference of 0.3 percentage points.
  • Nonhemorrhagic stroke occurred in 0.5% with tranexamic acid versus 0.3% with placebo.

Concerns

  • The cardiovascular result was close to neutral but did not satisfy the prespecified noninferiority margin.
  • The safety primary analysis was per-protocol rather than pure intention-to-treat.
  • A small thrombotic harm cannot be excluded even though the absolute difference was small.

External validity

Who it applies to

Adults undergoing major noncardiac surgery when clinicians want to reduce perioperative bleeding risk.

Who it does not

Patients with procedure-specific contraindications to tranexamic acid or clinical scenarios that differ substantially from the trial's broad noncardiac population.

Generalizability notes

  • The large international sample supports broad perioperative applicability.
  • The result is more about general surgical hemostasis than neurosurgery-specific practice.
  • Clinicians still need to weigh bleeding benefit against the still-open noninferiority safety question.

Evidence trace

Source trace and metadata

Citations (4)

claim_id

methods_critique.risk_of_bias

locator

p. 2

claim_id

stats_check.effect_sizes

locator

p. 1

claim_id

stats_check.concerns

locator

p. 2

claim_id

practice_impact.bottom_line

locator

p. 1

Metadata

Generated at

2026-03-08T15:40:00-05:00

Version

manual-pdf-repair-v1