The Rhythm | Issue 21
Issue 21 | Week of 24 August 2026
No major cardiology congress fell this week. The European Society of Cardiology (ESC) Congress 2026 begins next week in Munich, running 28 to 31 August.
Cardiology Intelligence Briefing: how far the evidence travels
Once a therapy, test, or clinical pathway has been shown to work, the next question is how far that evidence can be trusted beyond the setting in which it was first established.
This week, several long-standing assumptions held up surprisingly well. Finerenone's benefits were consistent in patients with a history of cancer, suggesting that prior malignancy is a marker of higher risk rather than a reason to withhold treatment.2 Cold-stored platelets remained noninferior for haemostatic efficacy after storage for up to 21 days, challenging a logistical constraint that has shaped platelet supply for decades. The higher rate of re-exploration for bleeding, however, prevents the result from being read as an uncomplicated win.1 In donor-heart assessment, estimated cardiovascular risk appeared more informative for transplant outcome than whether coronary angiography had been performed, questioning how much the invasive test adds once donor risk is already known.3
Other evidence travelled less well. An AI-ECG model that performed strongly in hospital cohorts lost substantial discrimination and positive predictive value when applied to a lower-prevalence community population.5 A model's performance belongs partly to the population in which it is used, not just to the algorithm itself.
The week also offered a different warning about overreach. Repeat ablation was associated with less than half the five-year mortality of antiarrhythmic therapy after recurrent atrial fibrillation (3.3% vs 8.2%), but registry data cannot exclude the possibility that differences between the patients selected for each strategy explain part of the observed effect.4
Knowing that something works is no longer enough. As cardiology's major therapies and tools mature, progress will increasingly depend on defining where their benefits persist, where they weaken, and which assumptions still deserve to survive.
