Potassium supplements took about 5 points off blood pressure at the clinic, and about 3 over a full day, in 803 people not on medication. The reviewers rate that evidence low to very low

Eat more potassium, the advice goes, and your blood pressure will thank you. Plenty of earlier reviews back that up, but they often mixed people on medication with people not on it, and foods with pills. This one, published in the Journal of the American Heart Association on 30 October by a team at Universidad Peruana Cayetano Heredia in Lima, asks something narrower. In adults with high blood pressure who aren't yet being treated for it, what does potassium on its own, as a supplement, actually do?

They found 19 randomised trials with 803 people, run between 1982 and 2021 in 11 countries. Every trial had to confirm exposure by measuring potassium in 24-hour urine, and trials that added other minerals or relied on salt substitutes or diet changes were left out. Most used potassium chloride, at a typical dose of 64 mmol a day. Sixteen of the 19 compared it with a placebo.

At the clinic, potassium lowered systolic pressure, the top number, by 5.3 mm Hg and diastolic, the bottom number, by 4.5. That's the figure in the title, and it's a respectable one. But only 6 of the trials also strapped on a monitor to track pressure for 24 hours, and there the falls were 2.8 and 3.2 mm Hg. On the top number, that's about half the clinic result. The authors point out that round-the-clock readings are more closely tied to heart disease and organ damage than clinic ones, so the smaller figure may be the one that matters more.

Those numbers come from the review's main statistical model, which gives a precision-weighted average across the trials. The authors are clear that this isn't a claim that everyone responds alike. The trials disagreed sharply, with an I² of 92.3% for clinic systolic pressure, where anything above about 75% is usually read as high. When they reran the sums allowing for that disagreement, the clinic average stayed clearly below zero, at 7.2 mm Hg. The 24-hour results did not: 2.4 mm Hg systolic, with a confidence interval from a 5.2 fall to a 0.4 rise.

The most telling number is one few reviews report. A prediction interval estimates what a new trial in a comparable setting might find, and for clinic systolic pressure it ran from a 24.5 mm Hg fall to a 10.1 mm Hg rise. So the average points to a benefit, but it doesn't promise one in any given setting. Part of that spread traces to one 48-person Kenyan trial. Dropping it took the I² down to 66.3%, and the prediction interval to between a 12.1 fall and a 1.7 rise, which still includes no effect at all.

Why the spread? The hints are interesting but exploratory, based on comparing whole trials rather than individual people. Pressure fell more in trials where it started higher, by about 4 mm Hg more systolic for every extra 10 mm Hg at the start, and in longer trials, by 1.6 mm Hg a week. Dose explained little, and there was no clear threshold. The trials were also short, with a median of 4 weeks and none past 16, so nobody can say from this whether the effect lasts. And 14 of the 19 were crossover trials, where people take both treatments in turn. Only 5 of those reported a washout gap between them, so one phase could have bled into the next.

Side effects were more common on potassium: 51 of 398 people reported at least one, against 12 of 374 on the control, or 12.8% against 3.2%. They were mostly mild and in the gut. Few people quit because of them, 4 of 504 against none of 480. What's less reassuring is what the safety data can't show. Reporting was patchy, and the review's safety outcomes were withdrawals and side effects, not blood potassium levels, which is the obvious worry with high doses. The trials also largely left out people aged 80 and over. The US National Institutes of Health notes that with kidney disease or drugs such as ACE inhibitors and potassium-sparing diuretics, even intakes below the recommended level can push blood potassium too high.

Scale matters here too. By our arithmetic, the typical trial dose is about 2,500 mg of potassium a day. Most potassium-only supplements on sale contain up to 99 mg, so that's about 25 pills, or the potassium in about six medium bananas. And this review says nothing direct about the bananas: the authors say supplements and foods aren't equivalent and their results shouldn't be stretched to diet. Their own bottom line is cautious. Potassium supplements might have a conditional add-on role in selected people with healthy kidneys and blood tests in place, but they shouldn't delay or replace blood pressure drugs when those are called for. What's needed now is what this evidence lacks: larger parallel trials of at least 8 to 12 weeks, with 24-hour monitoring built in.