The ubiquitous presence and excessive consumption of sodium in modern societies have long been recognized as major contributors to the development of endothelial dysfunction, hypertension, cardiovascular disease, chronic kidney disease, and stroke.¹ Internal derangement in the kidney’s handling of sodium and potassium balance and its interplay with the external environment greatly influence the risk of essential hypertension and its cardiovascular sequelae.² Because of the interdependency of sodium and potassium in regulating blood pressure, a deficit in potassium can be just as detrimental as a surplus in sodium.³ In a meta-analysis of 33 randomized clinical trials, Whelton et al⁴ reported that potassium supplementation lowered systolic blood pressure by 4.4 mm Hg and diastolic blood pressure by 2.5 mm Hg and effects were greater in the trials where participants had high sodium intake. Dietary studies also showed the blood pressure and cardiovascular benefits of diets with low sodium and high potassium content.^5,6 However, in societies that are accustomed to the consumption of processed food, regulating the dietary intake of sodium and potassium has proven to be a daunting task. Despite the long-established benefits of potassium-rich and sodium-poor food and tireless efforts to promote healthier diets, tastes and traditions are often entrenched and resistant to change. Substituting regular salt with potassium-enriched salt, therefore, becomes highly attractive as a simple, inexpensive, and potentially scalable solution to enhance cardiovascular health in society.

See related article, pp 1031-1040

The potential benefits and harms of potassium-enriched salt have not been put to the test in randomized clinical trials until recently. A large cluster-randomized trial conducted in China, the SSaSS (Salt Substitute and Stroke Study), showed that compared with regular salt, potassium-enriched salt significantly reduced the rates of fatal and nonfatal strokes, major cardiovascular events, and death in older adults with a history of stroke. Remarkably, the benefits appeared to be gained without significantly elevating the risk for clinical hyperkalemia.⁷ The well-conducted trial has provided the first evidence of the health benefits of salt substitution for a higher risk population.

In the current issue of Hypertension, the SSaSS investigators present a follow-up paper on the cardiovascular-specific effects of potassium-enriched salt.⁸ The large sample size of the original trial is certainly tempting for a secondary analysis as few other data sources exist, especially from clinical trials, to answer a logical follow-up question: does potassium-enriched salt reduce the risks of specific cardiovascular events such as acute coronary syndrome (ACS), heart failure (HF), arrhythmia, and sudden death? Answers to this question would provide more concrete evidence of whether and how salt substitution can modify the cardiovascular risk in this vulnerable population.

To answer this question, Yu et al⁸ performed a secondary analysis of the SSaSS data. They evaluated the effects of salt substitution on 4 outcomes and directly compared their incidence rates between the participants who received potassium-enriched salt versus regular salt. The authors reported the incidence rate ratios (IRRs) of the outcomes of interest, as well as the associated 95% CIs: ACS (6.32 versus 7.65 per 1000 person-years; IRR, 0.80 [95% CI, 0.65–0.99]), HF (9.14 versus 11.32 per 1000 person-years; IRR, 0.88 [95% CI, 0.60–1.28]), arrhythmia (4.43 versus 6.20 per 1000 person-years; IRR, 0.59 [95% CI, 0.35–0.98]), and sudden death (11.01 versus 11.76 per 1000 person-years; IRR, 0.94 [95% CI, 0.82–1.07]). The authors concluded that the use of potassium-enriched salt was more likely to prevent than cause cardiac disease but cautioned that the post hoc nature of the analyses precluded definitive conclusions.

The authors have good reasons to put a cautionary note on the interpretation of the findings. Although the observed incidence rates of the 4 outcomes were all lower in those who had used potassium-enriched salt than those who had used regular salt, 2 of the 4 outcomes (HF and sudden death) have CIs covering 1.0, indicating that the differences were not statistically significant. The other 2 (ACS and arrhythmia) were significant.

It is relevant to note that the analyses were performed without correcting the inflated type I error rates due to multiple tests. In statistics, a type I error rate is the probability of incorrectly rejecting a null hypothesis. When multiple hypotheses are tested, type I errors accumulate. The experimentwise type I error rate is the probability of incorrectly rejecting any of the null hypotheses within a study. The chance that one incorrectly rejects at least 1 of the hypotheses increases with the number of hypotheses tested. The standard practice of clinical trials is to control for inflated type I error rate (ie, multiplicity adjustment). The US Food and Drug Administration of the US Department of Health and Human Services has a longstanding guidance on multiplicity adjustments in clinical trials.⁹ The US Food and Drug Administration’s most recent iteration of the guidance has concluded that failing to appropriately control the type I error rate may increase the risk of a false positive conclusion.⁹ The US Food and Drug Administration document has provided clear and specific instructions on how to control for multiplicity of testing in clinical trials. After multiple comparison adjustments, the P values for ACS, HF, arrhythmia, and sudden death were 0.12, 0.81, 0.12, and 0.65, respectively.

The treatment effects on the cardiovascular-specific outcomes were estimated after the trial’s reporting on the 4 main outcomes: stroke, major adverse cardiovascular events, death from any cause, and hyperkalemia.⁷ Had all outcomes been considered for multiplicity adjustment, even more uncertainty would have been introduced into the interpretation of the findings. So, are the observed differences in the incidence rates of specific cardiovascular outcomes real, or did they happen by chance? We have good reasons to believe that potassium-enriched salt could provide the hypothesized cardiovascular benefits. The main SSaSS publication has firmly established the general health benefits of salt substitution. However, when it comes to data evidence on specific cardiovascular events, we must agree with the authors and admit that the jury is still out. All things considered, one has to admit that evidence from the trial falls short of supporting the claim that K-substituted salt protects against ACS, HF, arrhythmia, and sudden death.

When analyzing multiple end points in clinical investigations, lower outcome event rates and the requirement for multiplicity adjustment often present a formidable, sometimes insurmountable challenge, especially in secondary analyses, which are rarely adequately powered in the design stage of trials. This must have been a challenge that the SSaSS investigators faced in their quest for a more in-depth depiction of the cardiovascular effects of salt substitution. Examination of the time to first event plots further spawns uncertainty of potassium-enriched salt’s effect on an outcome that is biologically plausible—cardiac arrhythmias. The unadjusted rate ratios suggest protection against arrhythmias, but the Kaplan-Meier plots reveal no such signals, nor did the multiplicity-adjusted P values. Lack of statistical significance, however, should never be taken as evidence for lack of effect.¹⁰ Thoughtful and accurate statements of study findings are as just important as careful data analysis. Yu et al⁸ stated their conclusions with a healthy dose of caveats, which should not be lost in the reading of their paper.

Secondary analyses of trial data can be less prescriptive unless they were prespecified. Analysts have the option of using the most appropriate models as the data suggest. With actual data in hand, analytical methods should be chosen to accommodate the features of the data. For example, rates of cardiovascular events were analyzed using Poisson regression. While the approach may be consistent with the analysis of the primary outcomes, whether it can be justified for the secondary outcomes is less clear. After all, when one examines the occurrence of a specific type of event (eg, arrhythmia instead of all cardiovascular events), the rate tends to be much lower; data may also contain more zeros and, thus, exhibit greater variability. All of these complicate secondary analyses and may lead to lower power. For these reasons, many have thought that the Poisson assumption in count data regression was overly restrictive in its requirement of mean equaling the variance. Extra-Poisson variation is a frequent complication of event counts, so is zero inflation, that is, data containing a large number of zeros exceeding what can be accommodated by the probability mass at zero in a Poisson distribution.¹¹ This may be particularly relevant in the current investigation because when one studies a specific type of cardiovascular outcome, no matter how common it is, there are always vastly more people who do not experience the outcome event; this creates a large number of zeros, which may invalidate the Poisson assumption. Properly handling these data issues enhances confidence in the study findings.

Additionally, the occurrence of cardiovascular events is contingent on patient survival. Because the primary analysis has shown that salt substitution reduces all-cause mortality, the different mortality rates between the 2 treatment groups have created a situation of unequal surveillance where participants in the salt substitution group had more chance of being observed for cardiovascular events than those in the regular salt group. Ignoring the competing risks of death could introduce biases in the analysis of cardiovascular events.¹² In other words, cardiovascular event rates would seem to be greater in the group that had fewer patients die if the competing risk of death was not accounted for.

Numerous animal experiments and small human studies have provided a good understanding of the effects of dietary sodium reduction and potassium supplementation. The SSaSS study is the first large-scale clinical trial that established the health benefits of potassium-enriched salt as conferred by a population-level intervention. The study found that potassium-enriched salt prevented strokes, the primary outcome of the trial, and it also reduced the risks of key secondary outcomes such as sudden death and major adverse cardiac events. These findings are of great public health importance. When it comes to the specific types of cardiovascular risks, however, event rates may be too low to generate statistically significant findings. The requirement of multiplicity adjustment has further increased the burden of proof. Nonetheless, the work of Yu et al has shown the plausibility of benefits. While the findings may be inconclusive, the paper does give us hope that future studies will provide more definite evidence.

In summary, even if the specific cardiovascular effects require confirmation from further investigation, the benefits of salt substitution for overall health are highly promising. More research will be needed to explore the optimal approaches for using salt substitution, especially in defining the benefits and risks at the individual level. For example, questions such as the effect and risk modification of moderate or advanced kidney disease on strokes need to be addressed in future studies.

R. Agarwal was supported by the National Heart Lung and Blood Institute grant R01 HL126903 and the Veterans Administration grant I01CX001753.

Disclosures R. Agarwal served as a member of the executive steering committees of the FIDELIO and FIGARO trials and continues to serve on the steering committee of the FIND-CKD and CONFIDENCE trials and as the principal investigator for CONFIDENCE all funded by Bayer. He serves as a consultant for Bayer, Covance, and Boehringer Ingelheim. He served on the executive steering committees for Akebia for their HIF-PHI trials and as a consultant. He chairs adjudication committees for Alnylam and Intercept Pharma and Data safety monitoring boards for Chinook, Vertex, and Eloxx. He receives royalties from UpToDate and serves as Associate Editor of the American Journal of Nephrology and Nephrology Dialysis and Transplantation. W. Tu has nothing to disclose.

For Sources of Funding and Disclosures, see page 1043.

The opinions expressed in this article are not necessarily those of the editors nor the American Heart Association.