WHAT IS ALREADY KNOWN ON THIS TOPIC

Prior systematic reviews of salt substitutes have reported clear beneficial effects on blood pressure levels, but effects on clinical outcomes and the risk of premature death have been ill-defined due to sparse data.

The recently reported Salt Substitute and Stroke Study (SSaSS), the largest ever trial of a potassium-enriched salt substitute, showed clear blood pressure-mediated protective effects for stroke, major cardiovascular events and premature death.

WHAT THIS STUDY ADDS

Meta-analysis of blood pressure data from 19 trials that included 29 528 participants showed that salt substitutes lowered blood pressure across diverse population subgroups and geographies.

Meta-analysis of clinial outcome data from five trials that included 24 306 participants, mostly from the SSaSS, showed that salt substitutes lowered the risks of total mortality, cardiovascular mortality and cardiovascular events.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Consistent blood pressure-lowering effects of salt substitute in diverse populations across the SSaSS and other trials indicate that the blood pressure-mediated benefits of salt substitute on clinical outcomes are likely to be generalisable across diverse population groups worldwide.

Potassium-enriched salt, such as that used in the SSaSS, should be considered routinely by clinicians seeking to prevent complications in patients with hypertension and by policy makers seeking to reduce the burden of blood pressure-related disease.

Introduction

Cardiovascular diseases are the leading cause of death worldwide, and high blood pressure is a major risk for premature death.1 Excess dietary sodium and insufficient dietary potassium are both well-established causes of high blood pressure. Randomised trials demonstrate that reduced dietary sodium consumption or potassium supplementation lowers blood pressure.2 3 Sodium-reduced, potassium-enriched salt substitutes, in which a proportion of the sodium chloride (NaCl) in regular salt is replaced with potassium chloride (KCl), combine these blood pressure-lowering effects.4

Most trials of salt substitutes done to date have addressed blood pressure-lowering effects and recorded cardiovascular events and deaths only as adverse events and in small numbers.5–7 One prior cluster randomised trial with mortality as the primary outcome reported a protective effect, but the robustness of the finding was uncertain due to the small number of clusters included.8 The recently reported Salt Substitute and Stroke Study (SSaSS), a 5-year cluster randomised trial conducted in 600 villages in China, has for the first time provided unequivocal evidence that salt substitution can protect against stroke, cardiovascular events and death.9 The trial also addressed concerns about possible adverse effects of dietary potassium supplementation, with no evidence of any harmful effect on hyperkalaemia risk among the 20 995 trial participants.9 The benefits for clinical outcomes seen in SSaSS were attributable to blood pressure-lowering, with the magnitude of effects aligned with that anticipated from prospective observational studies and trials of pharmacological blood pressure lowering.10 11

Since the benefits of salt substitutes on clinical outcomes are mediated primarily by blood pressure reduction, comparing the effects of salt substitution on blood pressure across SSaSS and other trials will provide insight into the likelihood that effects on clinical outcomes will be replicated in other populations. Accordingly, this systematic review summarised the effects of salt substitute on blood pressure and clinical outcomes for all available trials and assessed the constancy of the findings across diverse population groups and geographies.

Methods

The protocol for this systematic review has been designed according to the population, intervention, comparison, outcomes and study design approach (online supplemental file) and registered, and the results are reported in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guideline.12

Search strategy

We searched the MEDLINE, Embase and Cochrane databases for the key terms salt substitute, low-sodium salt, reduced-sodium salt and potassium salt to 31 August 2021 without restriction on publication date or language (online supplemental file). Access to the SSaSS dataset was obtained to enable calculation of the effects of salt substitute on blood pressure in participant subgroups.

Eligibility criteria

All parallel group, step-wedged or cluster randomised controlled trials were eligible when adults or clusters were randomly allocated to a salt substitute or regular salt. Regular salt is about 100% NaCl, and salt substitute was defined on the basis of 5% or more of NaCl being replaced with KCl. Outcomes of interest included blood pressure, mortality, cardiovascular mortality and cardiovascular events (a composite of non-fatal stroke, non-fatal acute coronary syndrome or death from a vascular cause).

Study selection

The titles and abstracts of all records identified from the database searches were screened for potential eligibility by one author (XY), and the full-text versions of all potentially eligible studies were obtained for independent review by two authors (XY and KL). Reasons for inclusion or exclusion were recorded, and discrepancies between reviewers were resolved through discussion or with reference to a third author (MT).

Data extraction

Study characteristics, participant characteristics and outcomes reported of included studies were independently extracted by two authors (XY and AP) using a standardised data collection form. Differences in extracted data were resolved by discussion and rechecking the full-text report. Point estimates and measures of uncertainty were extracted directly wherever possible but, if not reported, were estimated from SD, CIs or exact p values as recommended in the Cochrane Handbook for Systematic Reviews.13 For studies that reported the mean differences (MDs) in outcomes only for subgroups, we estimated overall effects by weighting subgroup estimates by the sample size of the subgroups.13

Risk of bias assessment

Two authors (XY and AP) independently determined the risk of bias for each included study using the revised Cochrane Risk of Bias Tool for randomised trials14 with disagreements resolved by discussion.

Statistical analysis

Effects of salt substitute compared with control for the continuous blood pressure and urinary electrolyte outcomes were estimated as MDs with 95% CIs using a random-effects model15 because there was an a priori expectation that effects might differ across studies by more than chance because of the different compositions of the salt substitutes tested. For the clinical outcomes and mortality, we estimated risk ratios (RRs) and 95% CIs using a fixed effect model because of the known methodological limitation of a random-effects model when most of the data for an outcome are derived from one study that is much larger than the others.16 17

Assessment of the heterogeneity across study results was estimated using the I2 statistic. Subgroups of trials were defined by geography (WHO regions), mean age of the participants (above and below 60 years), per cent of male participants (above and below 50%), history of hypertension (all hypertensive vs non-hypertensive or mixed populations), history of cardiometabolic disease (participants with cardiovascular disease or diabetes vs others), proportion of using antihypertensive medication (above and below 50%), mean baseline systolic blood pressure (SBP) (above and below 140 mm Hg), mean baseline diastolic blood pressure (DBP) (above and below 90 mm Hg), baseline body mass index (above and below 25 kg/m2), mean baseline sodium intake (above and below median), mean baseline potassium intake (above and below median), per cent NaCl in the salt substitute (above and below median), per cent KCl in the salt substitute (above and below median) and mean follow-up time (above and below 12 months). Univariable metaregression was also done to explore the association of trial characteristics with effect sizes where there were at least 10 trials with data available for the analysis. Small-study effects and publication bias have been assessed using the regression-based Egger test and funnel plots.18 Sensitivity analyses were done excluding the large SSaSS trial and with the removal of studies considered at high risk of bias.

In addition, we assessed the effects of salt substitutes on blood pressure in SSaSS for each of the prespecified population subgroups reported for the primary stroke outcome in the main paper.9 Estimates were made using analysis of covariance with adjustment for clustering at the village level for participants with measurements at both baseline and final follow-up, with interaction terms fitted to test the constancy of effects across subgroups. Visual comparison of the effects on blood pressure observed in the SSaSS subgroups was made by plotting data alongside the trial groupings defined by the same characteristic, but excluding SSaSS. Statistical analyses were conducted using SAS V.9.4 and Stata V.16.1.

Patient and public involvement

There was no patient or public involvement in this systematic review and meta-analysis.

ResultsSearch results

The literature search identified 4534 unique records. Eighty-six articles were retrieved for full-text review and 39 of them were interventional studies. Excluding 18 articles for the reasons summarised in figure 1, we included 21 studies in the meta-analysis.

Figure 1

PRISMA (preferred reporting items for systematic reviews and meta-analyses) flow diagram illustrating selection of studies. DBP, diastolic blood pressure; PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses; SBP, systolic blood pressure.

Characteristics of included studies

There were 17 individually-randomised trials,5–7 19–32 3 cluster-randomised trials8 9 33 and 1 step-wedged randomised trial.34 Five of the included studies were done in the WHO European Region,20 21 25 27 28 11 in the WHO Western Pacific Region,5 6 8 9 22 24 29 31–33 35 4 in the WHO Region of the Americas,19 23 26 34 and 1 in the WHO South-East Asian Region.30 The range of intervention duration was between 1 month and 60 months. The proportion of NaCl in the salt substitutes varied from 33% to 75%, and the proportion of KCl varied from 25% to 65%. (online supplemental table S1)

Trial sample sizes ranged from 20 to 20 995 (table 1). The mean age of the participants was 21 years in the trial done in the youngest population and 75 years in the trial done among the oldest individuals. The mean baseline SBP levels in the trials varied between 113 mm Hg and 177 mm Hg, and the mean baseline DBP levels varied between 71 mm Hg and 105 mm Hg. The mean daily baseline 24-hour urine sodium excretion was between 2.9 g and 5.5 g, and the mean daily baseline 24-hour urine potassium excretion was between 0.8 g and 3.6 g.

Table 1

Baseline characteristics of participants in the included studies

Risk of bias assessment

The overall risk of bias was deemed to be low for 11 studies, to be of some concern for 8 studies and to be high for 2 studies (online supplemental figure S1). The most frequently identified risk was an absence of clarity regarding the randomisation process.

Effects of salt substitute on blood pressure

Nineteen studies reported the effects of salt substitutes on SBP and DBP (online supplemental figure S2). There was an overall −4.61 mm Hg (95% CI −6.07 to −3.14, p<0.001, I2=76.8) reduction in SBP and a −1.61 mm Hg (95% CI −2.42 to −0.79, p<0.001, I2=73.0) reduction in DBP with salt substitute compared with control (figure 2). Subgroup analyses done to explore the heterogeneity of effects between trials identified trial duration below 12 months (p<0.001) as associated with larger reductions in SBP and DBP. A higher intake level of potassium at baseline (p=0.04) and a lower proportion of NaCl in the salt substitute (p=0.02) were both associated with larger reductions in DBP (figure 3). Likewise, the metaregression showed each 10% lower proportion of NaCl in the salt substitute to be associated with a −1.53 mm Hg (95% CI −3.02 to −0.03, p=0.045) greater reduction in SBP and a −0.95 mm Hg (95% CI −1.78 to −0.12, p=0.025) greater reduction in DBP (table 2). Effects of salt substitute on blood pressure were otherwise consistent and were favourable across countries from WHO regions, and among population strata defined by age, sex, history of hypertension, body mass index, baseline blood pressure and baseline 24-hour urinary sodium excretion (all p homogeneity >0.05) (figure 3).

Table 2

Associations of trial characteristics with the effects of salt substitute on blood pressure in univariable metaregressions

Figure 2

Summary of impact of salt substitutes on outcomes of interest. DBP, diastolic blood pressure; MACE, major adverse cardiovascular events; MD, mean difference; RR, risk ratio; SBP, systolic blood pressure.

Figure 3

Effects of salt substitute on SBP and DBP (mm Hg) in subsets of trials defined by study characteristics. BP, blood pressure; CV, cardiovascular; DBP, diastolic blood pressure; SBP, systolic blood pressure.

Sensitivity analysis excluding SSaSS showed similar pooled estimates for SBP and DBP, and the same was true when two studies considered at high risk of bias were excluded (online supplemental table S2). Effects of salt substitute on blood pressure in participants subgroups in SSaSS were similar across subsets defined by age, sex, education level, history of stroke, diabetes, antihypertensive use and body mass index (all p homogeneity >0.05) (online supplemental figure S3). Patterns of blood pressure lowering in SSaSS subgroups and trials subsets defined by similar criteria were comparable.

Effects of salt substitute on clinical outcomes, mortality and hyperkalaemia

Five studies reported effects of salt substitute on all-cause mortality; three reported effects on cardiovascular mortality; and two reported effects on major cardiovascular events (figure 4). The majority of the data (88%–99%) for all three of these outcomes were derived from one study, SSaSS. The pooled RR for all-cause mortality was 0.89 (95% CI 0.85 to 0.94, p<0.001); the RR for cardiovascular mortality was 0.87 (95% CI 0.81 to 0.94, p<0.001); and the RR for major adverse cardiovascular events was 0.89 (95% CI 0.85 to 0.94, p<0.001). There were two studies that reported no difference in hyperkalaemia events between randomised groups including SSaSS, which recorded 331 clinical hyperkalaemia events.9 33 Two others reported no effect of salt substitute on serum potassium levels (online supplemental figure S4), and six reported no serious adverse events attributable to hyperkalaemia.5 6 24 27 28 30

Figure 4

Effects of salt substitute on the risk of death, vascular death and major adverse cardiovascular events. SSaSS, Salt Substitute and Stroke Study.

Effect of salt substitute on urinary electrolytes

Among the 13 trials with data, salt substitute reduced urinary sodium excretion by −0.48 g/day (95% CI −0.82 to −0.15, p<0.001, I2=78.8) and increased urinary potassium excretion by 0.45 g/day (95% CI 0.25 to 0.65, p<0.001, I2=91.7) (online supplemental figure S5) compared with control. The subgroup analyses identified a greater proportion of KCl in the salt substitute as associated with a greater reduction in urinary sodium excretion (p=0.04) (online supplemental figure S6), and the metaregression showed each 10 years higher age to be associated with a −0.45 g/day (95% CI −0.78 to −0.12, p=0.008) greater reduction in urinary sodium excretion with salt substitution (online supplemental table S3). Effects on urinary potassium excretion appeared to differ across study subsets defined by geographical region (p=0.02) with the largest effects in the European region and the least effects in the Southeast Asian region (online supplemental figure S7). The subgroup of studies with a greater proportion of men suggested larger rises in urinary potassium excretion (p=0.05), and this was echoed by the metaregressions where each 10% higher proportion of men was associated with a 0.29 g/day (95% CI 0.05 to 0.53, p=0.016) greater increase in urinary potassium excretion with salt substitute (online supplemental table S3).

Publication bias

Egger’s regression test indicated asymmetry of funnel plots for effects on DBP (p=0.001) but not for SBP (p=0.1), urinary sodium excretion (p=0.93) or urinary potassium excretion (p=0.45) (online supplemental figure S8). Data were too few to test for bias in reporting clinical outcomes or mortality.

Discussion

The effects of salt substitutes on blood pressure were largely consistent across trial subsets divided by population characteristics, as well as across the predefined SSaSS participant subgroups. Since blood pressure lowering is the mechanism by which salt substitutes confer their cardiovascular protection, the observed consistent blood pressure reductions make a strong case for generalisability of the cardiovascular protective effect observed in SSaSS both outside of China and beyond the SSaSS population.

The magnitude of the cardiovascular protection afforded is likely to be determined by the magnitude of the fall in blood pressure, and trial duration and composition of the salt substitute were the two factors that modified the size of the blood pressure reduction. Trials of longer duration were typically more pragmatic regarding the intensity of intervention, and less blood pressure lowering was likely a consequence of the lesser replacement of dietary sodium with salt substitute2 as well as a greater proportion discontinuing salt substitute early. Salt substitutes in which greater proportions of NaCl are replaced drive greater blood pressure reductions because dietary sodium is reduced more and dietary potassium is increased more. The reasons for the association of age with sodum excretion and geographical region with potassium excretion are not immediately apparent, but great urinary potassium excretion in men may reflect greater use of salt substitute consequent on greater consumption of food.

The size of the global population with known hypertension is about 1.28 billion individuals, though more than half of these are undiagnosed.36 The protective effect against stroke observed in non-hypertensive individuals in SSaSS indicates clear potential for benefit among individuals without hypertension, as does the protection against incident hypertension observed in a community-wide salt-substitute intervention in Peru.34 Further investigation of the potential for population-wide salt substitution is warranted since the available data indicate that salt substitutes are likely to offer large benefits across diverse communities with a very low likelihood of causing harm. Broader population use of salt substitute is supported by the absence of any detectable adverse effect of salt substitutes on hyperkalaemia in this review.37 38 All trials included in the review took pragmatic steps to exclude participants at elevated risk of hyperkalaemia, seeking to exclude those with chronic kidney disease or using medications that elevate serum potassium.39

This review defines the likely widespread efficacy and safety of salt substitutes, but additional work is required to understand the strategies suited for successful scale-up of use in different settings around the world. This will include, for example, making estimates of cost–benefit, understanding community attitudes and exploring the policy landscape with regard to government-led implementation options. The proportion of the benefits attributable to sodium reduction versus potassium supplementation has also been highlighted as a key outstanding question.

This analysis benefits from the systematic approach to the identification of studies and data extraction including access to subgroup data for the largest trial, SSaSS. We complement and advance existing systematic reviews by providing the clearest evidence yet about the effects of salt substitutes on clinical outcomes. The detailed exploration of effects in participant subsets done jointly with metaregression and subgroup analyses provides a comprehensive insight into the factors influencing the effects of salt substitutes and hence the likely generalisabilty of the protection afforded. The study also has some limitations. Complete data were absent for several studies and there was some evidence of reporting bias identified for DBP, though not for other outcomes. We did not search the grey literature and could have missed studies as a consequence. There were no observational data for salt substitute and clinical outcomes identified against which to compare the findings from the trials. The relatively small number of trials available for the metaregressions and subgroup analyses limited statistical power, and there were too few data points to enable multivariable metaregression. Completion of the three ongoing studies identified in the clinical trial registers will not rectify this issue. The relative paucity of data for non-hypertensive individuals makes it difficult to draw definitive conclusions for that participant subset. There was substantial heterogeneity across the included studies for blood pressure outcomes but almost all suggested blood pressure-lowering effects, so the uncertainty is mostly not in the likelihood of benefit but rather in the likely magnitude of the benefit. The data for clinical outcomes are derived mostly from one trial, SSaSS, though the limited clinical outcome data available from other studies are consistent with the effects observed in SSaSS, as are the blood pressure and urinary electrolyte data.

Conclusions

Salt substitutes produce consistent blood pressure-lowering effects across geographies and diverse participant subsets. Blood pressure-mediated beneficial effects of salt substitute on clinical outcomes appear likely to be accrued across a broad range of populations without adverse effects. These findings are unlikely to reflect the play of chance and support the adoption of salt substitutes in clinical practice and public health policy as a strategy to reduce dietary sodium intake, increase dietary potassium intake, lower blood pressure and prevent major cardiovascular events.

Data availability statement

Data are available upon reasonable request.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

Not applicable.