BRIGHT-4: bivalirudin's advantage holds in the patients you assumed did not need it

Design

prespecified subgroup analysis of a randomised, open-label trial

Population

6,016 STEMI patients undergoing radial-access primary PCI; 4,581 (76.1%) at low bleeding risk by CRUSADE <30

Primary outcome

30-day composite of all-cause death or BARC types 3–5 bleeding

Effect

low bleeding risk 1.4% vs 2.9% (HR 0.49, 95% CI 0.32–0.75); higher risk 8.1% vs 9.2% (HR 0.88, 95% CI 0.62–1.26)

The standard reasoning about bivalirudin is that it earns its place by reducing bleeding, so its benefit should concentrate in patients already at high bleeding risk. This prespecified analysis of BRIGHT-4 — 6,016 STEMI patients treated by radial primary PCI, randomised to bivalirudin with a 2–4 hour high-dose post-procedural infusion or to heparin monotherapy — tested that directly, splitting the cohort at a CRUSADE score of 30.

The higher-risk group had four times the event rate of the low-risk group (8.6% vs 2.2%), as expected. But the treatment effect did not follow. In low bleeding risk patients the 30-day composite of death or BARC 3–5 bleeding fell from 2.9% to 1.4% (HR 0.49, 95% CI 0.32–0.75). In the higher-risk group it moved from 9.2% to 8.1% (HR 0.88, 95% CI 0.62–1.26), which does not exclude no effect.

Read the absolute numbers rather than the ratios. The reduction was 1.5 percentage points in the low-risk group and 1.1 in the higher-risk group — the absolute interaction test was flatly non-significant at p=0.81. What differs is the baseline, not the benefit. The practical conclusion is the opposite of the intuitive one: a low CRUSADE score is not a reason to skip bivalirudin.

• Do not use a low bleeding risk score to justify heparin monotherapy in radial primary PCI.
• The regimen tested was bivalirudin plus a high-dose infusion for 2–4 hours after PCI, not a procedural bolus alone — the post-PCI infusion is part of the intervention.
• All of this is radial access; it does not transfer to femoral primary PCI.
• Bivalirudin costs substantially more than heparin and is not stocked in every Indian cath lab — the case for it here is mortality as well as bleeding, which changes the cost conversation.
• CRUSADE remains useful for predicting who will bleed; it is not useful for choosing the anticoagulant.

Why it matters

It removes the rationale many operators use to reserve bivalirudin for the patients they judge most likely to bleed.

Don't overread it

This is a prespecified subgroup analysis, not a separately powered trial in each stratum.

The statistics, in plain English

This is why absolute and relative interaction tests are reported separately. The relative interaction was significant at p=0.04 because a hazard ratio of 0.49 and one of 0.88 look very different — but when the underlying risk differs fourfold, identical absolute benefits produce very different ratios. The absolute interaction, at p=0.81, says the two groups gained about the same number of percentage points. Note also that the higher-risk confidence interval runs from 0.62 to 1.26: that arm was smaller (1,435 patients) and simply lacked the power to resolve an effect this size.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.