Pushing potassium to the top of normal cut atrial fibrillation events

Design

prespecified analysis of a randomised controlled trial, time-to-first-event with death as competing risk (POTCAST)

Population

1200 adults with an implantable cardioverter-defibrillator or CRT-defibrillator and plasma potassium 4.3 mmol/L or below; median age 64, 80.3% male, 32.5% with prior atrial fibrillation

Primary outcome

composite of atrial fibrillation hospitalisation with treatment change or inappropriate shock, hospitalisation for cardioversion or ablation, or amiodarone loading

Effect

8.0% (48/600) vs 12.2% (73/600), 2.80 vs 4.30 events per 100 person-years, subdistribution HR 0.65 (95% CI 0.45 to 0.93), P=0.02

Low-normal potassium has long been associated with atrial fibrillation, and this prespecified analysis of POTCAST tested whether doing something about it helps. Adults with an implantable cardioverter-defibrillator or a resynchronisation defibrillator and a plasma potassium of 4.3 mmol/L or below were randomised to potassium-raising treatment plus standard care, or standard care alone. Median age was 64, 80.3% were men, and 32.5% already had atrial fibrillation.

Uptitration raised plasma potassium by 0.3 mmol/L against controls. Over a median 3.3 person-years, the composite endpoint - acute hospitalisation for atrial fibrillation leading to a treatment change or for inappropriate shock, hospitalisation for cardioversion or ablation, or amiodarone loading - occurred in 48 of 600 in the high-normal group (8.0%, 2.80 per 100 person-years) and 73 of 600 controls (12.2%, 4.30 per 100 person-years). The subdistribution hazard ratio, with death treated as a competing risk, was 0.65 (95% CI 0.45 to 0.93, P=0.02). The effect looked similar with and without prior atrial fibrillation.

This is an unusually cheap intervention aimed at a modifiable number that most clinics already measure and then ignore once it is inside the reference range. The population is specific - device patients, so a group with structural heart disease and close follow-up - and the endpoint is a composite dominated by healthcare utilisation rather than by stroke or death. What it does justify is treating a potassium of 3.8 in a device patient as a target rather than a normal result, with the monitoring that implies.

• In device patients, treat a plasma potassium in the low-normal range as a target to raise, not a normal result.
• Expect the achievable difference to be modest - this trial moved potassium by 0.3 mmol/L, not by a whole unit.
• Recheck potassium and renal function after any uptitration, particularly alongside renin-angiotensin blockade or mineralocorticoid receptor antagonists.
• Note the endpoint is atrial fibrillation-related healthcare events, not stroke or mortality.
• The effect held in patients with and without prior atrial fibrillation, so this is prevention as well as suppression.

The statistics, in plain English

A subdistribution hazard ratio treats death as a competing risk rather than as censoring, which matters in a device population where mortality is not rare - it stops the analysis crediting the intervention for events that could not happen because the patient died. The interval, 0.45 to 0.93, excludes 1.0 but only just at the upper end, and it rests on 121 events in total. In absolute terms the difference is 1.5 events per 100 person-years, or roughly one event avoided for every 67 patients treated for a year - meaningful for an intervention this cheap, modest as a clinical effect. The subgroup consistency is reassuring but subgroups in a 1200-patient trial are underpowered by construction.

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.