Pooling five trials, beta-blockers after a heart attack with a normal EF did nothing

Beta-blockers after a heart attack are one of the oldest reflexes in cardiology. An individual-patient meta-analysis published in the New England Journal of Medicine on 9 November reports that, in patients whose left ventricular ejection fraction is at least 50% and who have no other reason to be on the drugs, the reflex does not produce a measurable benefit [s1].

What was pooled

The analysis combined individual patient-level data from five open-label randomised trials that assigned patients with recent myocardial infarction, an LVEF of at least 50%, and no other indication for beta-blocker therapy, to either beta-blocker treatment or no beta-blocker treatment [s1].

The five were REBOOT (7,459 patients), REDUCE-AMI (4,967), BETAMI (2,441), DANBLOCK (2,277) and CAPITAL-RCT (657) — 17,801 patients in total, of whom 8,831 (49.6%) were assigned a beta-blocker and 8,970 (50.4%) were not [s1]. Event rates were analysed with a one-stage fixed-effects Cox proportional-hazards model [s1]. The primary endpoint was a composite of death from any cause, myocardial infarction, or heart failure [s1].

Individual-patient meta-analysis is a stronger design than pooling published summary estimates: it lets the analysts apply consistent definitions and a single model across every trial rather than stacking effect sizes computed differently by different teams.

The result

Over a median follow-up of 3.6 years (IQR 2.3 to 4.6), a primary endpoint event occurred in 717 patients (8.1%) assigned a beta-blocker and 748 patients (8.3%) assigned none — hazard ratio 0.97 (95% CI 0.87 to 1.07; P = 0.54) [s1].

The components pointed in no consistent direction. Death from any cause: 335 versus 326 events, hazard ratio 1.04 (95% CI 0.89 to 1.21) [s1]. Myocardial infarction: 360 versus 407, hazard ratio 0.89 (95% CI 0.77 to 1.03) [s1]. Heart failure: 75 versus 87, hazard ratio 0.87 (95% CI 0.64 to 1.19) [s1].

The confidence interval on the primary result is what makes this different from an underpowered null. An upper bound of 1.07 and a lower bound of 0.87 rules out both a meaningful harm and the kind of benefit — a 20% or 25% relative reduction — that beta-blockers were historically credited with in this setting.

Why the old evidence said otherwise

The trials that established beta-blockers after myocardial infarction were run in a different medicine. REDUCE-AMI's own authors put it directly: most of those trials enrolled patients with large infarctions and predate biomarker-based diagnosis of myocardial infarction, percutaneous coronary intervention, modern antithrombotics, high-intensity statins and renin-angiotensin-aldosterone system antagonists [s2].

REDUCE-AMI, published in April 2024, was the first large modern test. It randomised 5,020 patients in Sweden, Estonia and New Zealand who had undergone coronary angiography after an acute myocardial infarction and had an LVEF of at least 50%, to long-term metoprolol or bisoprolol, or to no beta-blocker [s2]. Over a median 3.5 years, the primary composite of death from any cause or new myocardial infarction occurred in 199 of 2,508 patients (7.9%) on a beta-blocker and 208 of 2,512 (8.3%) without one — hazard ratio 0.96 (95% CI 0.79 to 1.16; P = 0.64) [s2]. Secondary endpoints, including death from any cause (3.9% versus 4.1%) and hospitalisation for heart failure (0.8% versus 0.9%), showed no advantage either [s2].

The November meta-analysis is that finding, replicated across four further trials and roughly 3.5 times as many patients, with a tighter interval.

What it does not cover

The scope is narrow and deliberately so. Every patient in these trials had an ejection fraction of at least 50% and no other indication for a beta-blocker [s1]. The analysis says nothing about patients with reduced ejection fraction, in whom beta-blockers have well-established mortality benefit; nothing about patients taking them for arrhythmia, angina or hypertension; and nothing about the mildly reduced range between 40% and 50%.

All five trials were open-label [s1]. That is standard for a treat-versus-no-treat design, but it means outcomes like heart failure hospitalisation, which involve clinical judgement, are not blinded. Hard endpoints — death, myocardial infarction — are less exposed to that.

The analysis was funded by Centro Nacional de Investigaciones Cardiovasculares Carlos III and others, and was prospectively registered in PROSPERO [s1].

What follows

Nothing here is guidance, and no reader should stop or start a prescription on the basis of a trial result. Beta-blockers are not drugs to discontinue abruptly, and the decision about any individual patient depends on why they were started.

What the meta-analysis does is move the burden of proof. For a patient with a preserved ejection fraction after myocardial infarction and no separate indication, the evidence for routine long-term beta-blockade is now a precise null across 17,801 randomised patients. Whether guideline committees translate that into a recommendation to stop, or merely to stop starting, is the next thing to watch.

Sources

• [s1] Beta-Blockers after Myocardial Infarction with Normal Ejection Fraction, The New England Journal of Medicine, 9 November 2025. https://doi.org/10.1056/NEJMoa2512686
• [s2] Beta-Blockers after Myocardial Infarction and Preserved Ejection Fraction, The New England Journal of Medicine, 7 April 2024. https://doi.org/10.1056/NEJMoa2401479