Potassium and blood-pressure treatment: the result that can change the plan

In clinic, two weeks apart

A patient in her early sixties with type 2 diabetes and an estimated glomerular filtration rate of 40 mL/min per 1.73 m², has blood pressure above target on three drugs. Spironolactone is added. At the next visit the home readings have fallen by more than 10 mm Hg and everyone is pleased. Then the laboratory report arrives: serum potassium 5.7 mmol/L, up from 4.8. The drug that worked is now the drug under review, and the next decision is no longer about blood pressure at all.

That sequence is common enough to deserve a note of its own. The existing Veritas note on angiotensinogen silencing, aldosterone synthase inhibition and endothelin blockade covers the new drug classes. This one stays with a single laboratory value and asks three questions: why potassium moves, what the best-known trial of managing it actually measured, and how to read trial results when eligibility and monitoring were tighter than ordinary practice.

Why the value moves

Serum potassium is held within a narrow range by renal excretion, and the distal nephron does most of the work. Aldosterone drives potassium secretion in the collecting duct in exchange for sodium reabsorption. Anything that raises distal sodium delivery and aldosterone activity, such as a loop or thiazide-type diuretic, tends to lower potassium. Anything that blocks aldosterone synthesis or action, or reduces the angiotensin II signal upstream of it, tends to raise potassium.

The two directions have different consequences for treatment. Hypokalaemia with potassium-wasting diuretics is usually gradual and manageable with dose adjustment, potassium-sparing combinations or supplementation, and it rarely forces withdrawal of an effective regimen. Hyperkalaemia with angiotensin-converting enzyme inhibitors, angiotensin receptor blockers and especially mineralocorticoid receptor antagonists is the more frequent reason an effective drug is stopped, because the margin above 5.5 mmol/L is where arrhythmic risk begins to be taken seriously. The risk is concentrated in people with reduced glomerular filtration, diabetes, older age, or concurrent potassium supplements and salt substitutes.

AMBER: what twelve weeks of persistence establishes

The AMBER trial set out to test whether a potassium binder could keep people on spironolactone (Agarwal et al., Lancet 2019). It enrolled 295 adults with uncontrolled resistant hypertension and chronic kidney disease, defined as an estimated glomerular filtration rate of 25 to 45 mL/min per 1.73 m², at 62 centres in ten countries. Everyone received open-label spironolactone, starting at 25 mg daily, and was randomised to patiromer 8.4 g daily or placebo. Entry potassium was between 4.3 and 5.1 mmol/L, and randomisation was stratified by that value and by diabetes.

The primary endpoint was not blood pressure. It was the proportion of participants still taking spironolactone at week 12: 86 percent with patiromer against 66 percent with placebo, a difference of 19.5 percentage points (95 percent confidence interval 10.0 to 29.0). Hyperkalaemia was the mechanism behind the gap. In a later indirect comparison using AMBER data, serum potassium of 5.5 mmol/L or above occurred in 35 percent of the patiromer group and 64 percent of the placebo group, and discontinuation attributed to hyperkalaemia in 7 percent and 23 percent (Agarwal et al., Clinical Kidney Journal 2023). Systolic blood pressure fell by a similar amount in both arms over the 12 weeks, roughly 11 to 12 mm Hg, which is what one would expect when most people in both arms were still taking the active drug for most of the period.

So AMBER established something specific: in this population, with this binder, more people could stay on spironolactone for three months. It did not establish that staying on spironolactone for three months improves cardiovascular or kidney outcomes, because no such endpoint was measured and the trial was far too short to measure one. Whether long-term binder use is safe, affordable and acceptable to patients outside a 12-week protocol remains an open question, and a cardiovascular outcome trial of the strategy has not been published.

Eligibility and monitoring change what a result means

Three design features of AMBER matter when reading it against a clinic list.

First, kidney function. AMBER required an eGFR of 25 to 45. People with better kidney function were not studied because their hyperkalaemia risk on spironolactone is lower, and people below 25 were excluded because the risk was judged too high. The result therefore applies to a band of kidney function, not to resistant hypertension in general.

Second, entry potassium. Participants began between 4.3 and 5.1 mmol/L. Someone presenting at 5.3 was not eligible, and the trial says nothing about starting spironolactone in such a patient with or without a binder.

Third, monitoring. Potassium was measured at frequent scheduled visits throughout the 12 weeks, with pre-specified thresholds triggering dose reduction or withdrawal. Frequent testing catches rises early, which both protects participants and makes an intervention look safer than it would under the sparser monitoring that is realistic in primary care. When a service cannot reproduce the trial's monitoring schedule, it cannot assume the trial's safety profile.

The table below sets AMBER beside the largest potassium-related trial in a different setting, discussed in the final section. It is a description of design, not a ranking. The two trials studied different populations with different interventions and endpoints, and comparing their event rates directly would be meaningless.

Dietary potassium is a different measurement

Clinicians are often asked whether a potassium-enriched salt substitute is a good idea for someone whose serum potassium already worries them. The two quantities should be kept apart. Dietary potassium intake is a flux, measured in grams per day; serum potassium is a concentration, held steady by excretion as long as the kidney and the hormonal axis can respond. In people with normal kidney function, substantial dietary potassium raises serum levels very little. In people with impaired excretion, or on drugs that impair it, the same intake can matter.

The Salt Substitute and Stroke Study tested a substitute of 75 percent sodium chloride and 25 percent potassium chloride against regular salt in 20,995 people from 600 villages in rural China (Neal et al., NEJM 2021). Participants had a history of stroke or were aged 60 or over with high blood pressure; 72.6 percent had had a stroke. People using potassium-sparing diuretics or potassium supplements, and people with serious kidney disease, were excluded. Over a mean of 4.74 years the stroke rate was 29.14 against 33.65 per 1000 person-years (rate ratio 0.86, 95 percent CI 0.77 to 0.96), with lower rates of major cardiovascular events and death. Serious adverse events attributed to hyperkalaemia were 3.35 against 3.30 per 1000 person-years (rate ratio 1.04, 95 percent CI 0.80 to 1.37). A prespecified analysis of the 15,249 participants with prior stroke found recurrent stroke and death both lower with the substitute, again without a detectable hyperkalaemia signal (Ding et al., JAMA Cardiology 2025).

Serum potassium was not routinely measured, so the safety outcome depended on hyperkalaemia being clinically recognised and recorded, and the people at highest risk had been excluded at the outset. A 2026 network meta-analysis of 34 trials and 37,063 participants reached the same practical conclusion: the mortality and event benefits are driven largely by SSaSS, generalisability beyond populations with high discretionary salt use is limited, and kidney disease and hyperkalaemia risk should be ruled out before a substitute is recommended (Lai et al., BMC Medicine 2026).

What to take from this

The potassium result in the opening case does not have a single correct response, and this note offers none; the 2024 European Society of Cardiology guideline sets out the monitoring schedule and thresholds for renin-angiotensin blockers and mineralocorticoid receptor antagonists (McEvoy et al., European Heart Journal 2024). What the evidence does support is a way of reading it. A rise in potassium on an effective drug is an expected consequence of the mechanism, not a surprise. The best trial of keeping people on that drug shows persistence over three months in a defined kidney-function band under close monitoring, and is silent on events. Dietary interventions that add potassium were tested in people whose kidneys could handle it, with serum levels unmeasured. Each result is useful exactly as far as its population, monitoring and endpoint allow.

Teams designing trials in which potassium management affects eligibility or persistence can discuss study design confidentially.