Blood Pressure Lowering Treatment Trialists' Collaboration — blood pressure-lowering treatment across the spectrum of cardiovascular risk: an individual participant data meta-analysis

BPLTTC IPD meta-analysis

Paciente / Población Intervención / Exposición Comparación Desenlace

In 357,440 participants in 51 randomised blood-pressure trials, blood-pressure-lowering treatment reduced major cardiovascular events across the full spectrum of predicted risk compared with placebo or less intensive treatment.

N

357.440pacientes

Diseño

One-stage individual participant data meta-analysis of 51 randomised trials of blood-pressure-lowering drugs (20 placebo-controlled, 23 active-controlled, 8 intensity comparisons; at least 1,000 patient-years per arm) in 357,440 participants; mean age 65 years, 41.6% women, mean baseline blood pressure 152/87 mmHg, median follow-up 4.2 years; Cox models stratified by trial, effects expressed per 5 mmHg systolic reduction and reported by decile of predicted 5-year cardiovascular risk on both relative and absolute scales; safety analysed in the 19 trials with an inactive comparator; funded by the British Heart Foundation.

Desenlace

First major cardiovascular event — stroke, ischaemic heart disease or heart failure causing death or hospitalisation — across deciles of predicted 5-year cardiovascular risk

Relevancia

1Define la práctica — el ensayo en el que se apoya la guía.

ResultadoPer 5 mmHg systolic reduction the hazard ratio was 0.90 (95% CI 0.88–0.92) overall and ran from 0.85 to 0.92 across the ten deciles, with no interaction (P > 0.99) and no trend (P = 0.20) — including the lowest decile, 35,744 people at a mean predicted risk of 3.0%. Absolute benefit tracked baseline risk instead: the 5-year risk difference rose from 0.4 to 2.6 percentage points, a number needed to treat of 255 falling to 38, while baseline blood pressure itself added little. Adverse events were more frequent with treatment and did not vary by risk decile: haemodynamic HR 1.66 (1.53–1.80), renal 1.49 (1.26–1.75), neurovestibular and fall-associated 1.09 (1.03–1.15), with no excess of falls or fractures.

Bidel Z, Nazarzadeh M, Mihaylova B, et al. N Engl J Med. 2026; published online 30 August. Presented at ESC Congress 2026, Hot Line 9.

Discusión y crítica

Three hundred fifty-seven thousand participants settle the question the trials were never designed to answer: the proportional benefit of lowering blood pressure is the same in someone with a 3% five-year risk as in someone with 42%, and the 71,488 people in the two lowest deciles were the stratum where randomised evidence had effectively been absent. What changes across the spectrum is arithmetic, not biology — the same 10% relative reduction is one event per 255 treated at the bottom and one per 38 at the top. Since the harms are relative too and equally flat, the net gain thins where the risk does, which is an argument for risk as the gate rather than blood pressure, not for treating everyone. Predicted risk is model-dependent and the safety data come from only 19 trials with an inactive comparator, so where exactly the line falls remains a judgement.