What the Trials Show: Blood Pressure First, Cardiovascular Outcomes Next
The subject in this chapter is the evidence for potassium-containing lower-sodium table salt: what randomized trials measured, what changed compared with regular salt, and how far those results can travel. The central judgment is not simply whether the intervention “works.” It is which outcomes show the clearest signal, how large that signal is, who was studied, and what uncertainty remains between a lower blood pressure reading and fewer cardiovascular events or deaths.
That distinction matters because health claims often compress several steps into one sentence. A substitute can change sodium and potassium exposure. That change can alter urinary or serum electrolyte measurements. Blood pressure can then move, and a long-term intervention might affect strokes, heart attacks or mortality. Each step is important, but evidence at one step does not automatically prove the next. The trials give us a way to separate those claims.
How the randomized evidence was built
The strongest part of the evidence comes from randomized controlled trials. In this design, participants are assigned to use a lower-sodium salt substitute, regular salt or, in some studies, no intervention. Random assignment helps balance many differences between groups, so a later difference is more plausibly related to the assigned salt strategy than to a pre-existing difference between participants. It does not make every limitation disappear. The comparator, the length of follow-up, adherence, outcome measurement and the people who were eligible still determine what the result means.
The World Health Organization's 2025 guideline reviewed 26 randomized trials published from 1986 through 2021. Together they involved 34,961 adults and 92 children. Most trials included some participants with hypertension, and the largest trial enrolled people at elevated risk of stroke. Twenty-three of the 26 trials studied substitutes that replaced sodium with potassium. Most examined discretionary use: salt that a person added during cooking or while eating.
That sounds like a broad body of evidence, but it is a selected one. Every trial excluded people in whom increased potassium could potentially cause harm, including people with kidney disease or impaired renal function and people taking potassium-sparing medicines. Some also excluded people with diabetes. The selection was ethically understandable for a trial, but it narrows the population to which the results directly apply. A randomized design can strengthen the comparison inside an eligible population while leaving an important question unanswered outside it.
A newer systematic review and network meta-analysis searched through October 2025 and included 34 randomized trials with 37,063 participants from 15 countries. It found that half of the studies were conducted in China. Its larger and more current evidence base is useful for comparing substitute formulations, but it does not remove the underlying selection issue. The studies still differ in diet, food preparation, baseline blood pressure, duration and eligibility. The two reviews also used different search dates and eligibility criteria, so their different trial counts describe different evidence sets rather than a simple disagreement.
The comparator is equally important. “Lower-sodium” does not mean lower sodium than every other possible diet; in these trials it generally means lower-sodium substitute versus regular salt. If participants in both groups use similar amounts of salt, the contrast tests the composition change. If one group uses less salt overall, or if participants receive different advice about the rest of their diets, the comparison may include more than the substitute itself. The WHO guideline's recommendation also concerns table salt added at home or at the table, not every source of sodium in manufactured food or restaurant meals.
The evidence is therefore best read as an answer to a defined question: among adults selected for these trials, what happened when a potassium-containing lower-sodium substitute was used in place of regular salt, usually for discretionary use? That is a more precise question than whether a product labeled “low sodium” is healthy for everyone.
The primary blood-pressure result is relatively clear. WHO reports that, compared with regular salt, assignment to a lower-sodium substitute reduced systolic blood pressure by an average of 4.76 mmHg and diastolic blood pressure by 2.43 mmHg. Follow-up ranged from 56 days to five years, and WHO rated the evidence for these average reductions as moderate certainty. The newer network analysis also found blood-pressure reductions across several substitute categories, with its larger formulation groups generally producing reductions of roughly 4–7 mmHg in systolic pressure.
An average effect is not a personal forecast. It describes the difference between groups across the studies, not the change that every participant experienced. Some people may have had a larger response, some a smaller one and some no meaningful change. The estimate also does not show that a person should alter medication or use a particular product without considering their health circumstances. It shows that, in the studied comparisons, changing the salt formulation was associated with a measurable average reduction in blood pressure.
The word “associated” needs care here. Random assignment supports a causal interpretation of the assigned intervention within the trial, but the intervention is not just a chemical ingredient in isolation. It includes the instruction or opportunity to use a substitute, the actual amount used, the foods in which it was used and the time over which participants continued using it. The result is causal for that package of exposure under those study conditions; it is not proof that any lower-sodium packet will produce the same effect in any setting.
The evidence path is not a ladder in which every lower rung guarantees the one above it. It is a set of related questions. Biomarkers can show that the intervention changed exposure. Blood pressure can show a clinically important physiological response. Cardiovascular events and mortality test whether that response corresponds to outcomes people care about most. The farther the claim travels from the measured exposure, the more follow-up, events, comparability and certainty it generally requires.
From blood pressure to deaths: what each result can support
Urinary and serum electrolytes sit close to the intervention. In the WHO review, 24-hour urinary sodium fell by an average of 19.98 mmol, or 459 mg, across 11 randomized trials, while urinary potassium rose by 11.44 mmol, or 450 mg. Serum potassium rose by a mean 0.12 mmol/L. These results are consistent with the intended composition change: less sodium and more potassium were available to the body. But the urinary findings were highly heterogeneous, and WHO did not grade them. They are signals about exposure and physiological response, not direct evidence that a person will avoid a stroke.
Blood pressure is closer to a patient-relevant outcome because sustained high blood pressure is itself harmful and is a major pathway to cardiovascular disease. Still, it remains an intermediate measure when the claim is about heart attacks or deaths. A reduction of a few millimetres of mercury can be meaningful at population scale, but the trials must still show whether events changed, how many events occurred and how confident we should be that the result is not due to chance, bias or a particular study context.
WHO reports favorable findings for some cardiovascular outcomes. Compared with regular salt, lower-sodium substitutes were associated with a rate ratio of 0.70 for non-fatal acute coronary syndrome, with a 95% confidence interval from 0.52 to 0.94. The guideline translated that estimate into about 150 fewer cases per 100,000 person-years, with an interval from 250 fewer to 30 fewer. For cardiovascular death, the rate ratio was 0.77, with a 95% confidence interval from 0.66 to 1.00, corresponding to about 180 fewer deaths per 100,000 person-years, with an interval from 310 fewer to no fewer. These are population-level estimates under the evidence review's baseline assumptions, not a prediction of an individual's personal risk.
The result for non-fatal stroke is more cautious. WHO reports a risk ratio of 0.90, but its confidence interval runs from 0.80 to 1.01. The estimated absolute effect was 20 fewer strokes per 100,000 people, but the interval ranged from 40 fewer to two more. The point estimate leans toward benefit, yet the interval includes little difference and a small possible increase. Saying that the result is compatible with benefit is more accurate than saying the intervention has definitively prevented stroke.
A confidence interval is not a verdict that an intervention either works or does not work. It describes the range of effects reasonably compatible with the data and the model. A narrow interval can still surround a small effect; a wide interval can reflect few events or inconsistent studies. When an interval crosses the point of no difference, the correct conclusion is usually that the study set has not established a clear effect for that outcome—not that the intervention has been proven useless.
The newer review adds another layer by comparing formulation groups. Its most compelling evidence for mortality came from moderate-potassium substitutes containing 25–40% KCl and low-sodium substitutes containing 60–79% NaCl. It estimated fewer deaths over five years at both intermediate and high cardiovascular risk, with moderate to high certainty for those particular analyses. It also found that these groups probably reduced systolic blood pressure and might reduce non-fatal cardiovascular events. But some event findings were low to moderate certainty, and the apparent mortality benefit depended heavily on one large Chinese trial, an issue that must remain attached to the result rather than being buried after the headline.
Safety outcomes illustrate why “no difference” needs a denominator and a study population. The WHO review found little to no difference in hyperkalaemia between lower-sodium substitutes and regular salt, with moderate-certainty evidence. Yet very few studies reported hyperkalaemia, the definitions varied and the review judged the potassium-related information unreliable. A rare outcome can be difficult to estimate even in a large evidence base, especially when people at greatest risk were excluded.
The newer network analysis found no significant increase in intervention-related adverse events, withdrawals due to adverse events or withdrawals overall for the main substitute groups. Acceptance was also comparable to regular salt in the trials that measured it. These findings are reassuring for the selected participants and products studied. They are not evidence that kidney disease, impaired renal function or potassium-sparing medicines make no difference. The absence of an observed excess in a selected population cannot answer a safety question about people who were not adequately represented.
Evidence certainty combines several judgments. Reviewers can lower certainty because studies have a high risk of bias, results are inconsistent, the population or intervention is indirect, the estimate is imprecise or publication bias is plausible. The network review also evaluated whether the different comparisons were sufficiently comparable for indirect conclusions. This matters because a large pooled sample can conceal differences in age, baseline blood pressure, diet, food preparation and formulation. The number of participants is relevant, but it is not a substitute for comparability.
The most defensible interpretation is therefore layered. There is moderate-certainty evidence that, in studied adults, lower-sodium salt substitutes reduce systolic and diastolic blood pressure compared with regular salt. There are favorable signals for some cardiovascular events and deaths, but their certainty and applicability vary. Mortality conclusions in the newer analysis are especially dependent on a large Chinese trial, while non-Chinese trials contributed no mortality data. Safety findings are generally reassuring in selected trial populations, but they do not establish safety for people with impaired potassium handling.
That interpretation also shows what new evidence would change the conclusion. Large, well-conducted trials in populations with different food systems and dietary patterns could test whether the cardiovascular and mortality findings travel beyond the settings that dominate the current evidence. Better reporting in people with kidney disease, diabetes or relevant medicines could clarify the safety boundary. Longer follow-up could show whether adherence and benefits persist. More complete measurement of total dietary sodium and potassium could reveal how much of the effect comes from the substitute rather than other changes in eating.
The trials answer more than a marketing claim, but less than a universal rule. They show a consistent enough blood-pressure signal to make the intervention credible, while also showing why event, mortality and safety claims require more careful wording. To understand whether one lower-sodium product deserves more confidence than another, the evidence must next be compared by formulation rather than grouped under one broad label.