In a nutshell
Dietary salt modification aims to reduce cardiovascular risk through two approaches: (1) reducing overall salt intake, and (2) replacing regular salt with salt substitute in which part of the sodium is replaced with potassium. Our model evaluates a potential grant that would fund a charity to implement both interventions across general populations in low- and middle-income countries.
We estimate that it costs about $855 to avert a death via dietary salt modification. We think the intervention can be cost-effective because:
- It targets the leading cause of death globally.
- It probably reduces population cardiovascular risk by a small but meaningful amount (~3%).
- It is very cheap per person covered (~$0.08 USD/person/year).
We have a number of uncertainties that make our current cost-effectiveness estimate very uncertain, including:
- Our estimate of the impact of dietary salt modification on the risk of cardiovascular death relies heavily on one randomized controlled trial that has a wide uncertainty interval.
- We are uncertain about the impact of dietary salt modification on mortality risk in the general population, due to the limited generalizability of the evidence on potassium-enriched salt substitute, and weaker evidence on dietary salt reduction.
- The intervention requires behavior change and we are uncertain how effective this would be in contexts other than those that have been studied.
- We are uncertain about important implementation details that could impact cost-effectiveness.
We should be able to reduce our uncertainty in some of these areas as we learn more about where and how the intervention would be conducted.
Published: July 2025 (February 2021 version)
Summary
What we think this program will do
Dietary salt modification programs are a public health intervention that promotes the use of potassium-enriched salt substitute and reduces overall salt intake in the general population of low- and middle-income countries. The resulting decrease in sodium intake and increase in potassium intake would reduce blood pressure at a population level, likely reducing the risk of cardiovascular events like heart attacks (more).
While we are not yet certain about implementation details, this intervention may involve a combination of support for government policies and other activities: advocating for less severe warning labels on potassium-enriched salt substitute, public education campaigns, advocacy to medical professionals, support for the salt industry and retailers to increase availability of potassium-enriched salt substitute, and market shaping and subsidy to lower the price of salt substitute (more).
Why we think this program is cost-effective
Our initial work suggests that this program is probably cost-effective because:
- We think the intervention is probably effective for increasing the use of potassium-enriched salt substitute and reducing normal salt intake. A similar intervention has been successfully implemented at scale in the past, and it forms a basis for our assumptions about intervention coverage (more). Coupled with broad support for sodium reduction from WHO member states, we believe the intervention is likely to be feasible in additional locations. We estimate that the intervention would increase the use of potassium-enriched salt substitute by about 18 percentage points.
- It targets cardiovascular disease, the most common cause of death globally and in many low- and middle-income countries. IHME estimates suggest that each year, about 0.2-0.3% of people die of cardiovascular disease in China and India, the locations we focus on in our cost-effectiveness analysis (more).
- It probably reduces the population-wide risk of cardiovascular events like heart attacks and strokes by a small but meaningful amount (~3% after adjustments). We use a successful dietary salt modification intervention, the SMASH trial, as a demonstration of feasibility at scale and a model for our cost-effectiveness analysis (more). Data from the SMASH trial are used to estimate the impact of the intervention on uptake of potassium-enriched salt substitute and reduction of sodium intake. In turn, uptake of potassium-enriched salt substitute is converted into deaths averted using the results of a large randomized controlled trial in China (the SSaSS trial), which reports that potassium-enriched salt substitute reduces the risk of dying from cardiovascular disease by 13% (95% confidence interval, 4% to 21%) in people at high risk (more). We add to this the estimated benefit of reducing overall salt intake, which was not part of the SSaSS trial’s intervention and has a more uncertain evidence basis (more). We scale the effect size downward to reflect lower intervention impact in an implementation setting (18 pp increase in salt substitute vs. 72 pp in the SSaSS trial) and the possibility that it may be harder to modify salt intake from processed foods vs. home use (70% adjustment). We roughly estimate that most of the benefit (~77%) comes from increasing the use of potassium-enriched salt substitute, and the rest (~23%) comes from reducing overall salt intake, though this would depend on implementation details (more).
- It is very cheap per person covered. Based on preliminary information from an implementing organization, we estimate that population-wide dietary salt modification would cost about 8 cents per person, per year the program is active, though this would depend on implementation details (more).
- We think this intervention would probably not happen without GiveWell support. This is our initial impression from speaking with an implementing organization, though we plan to investigate further.
Our best guess is that the version of dietary salt modification we have modeled is about 14 times as cost-effective as unconditional cash transfers1 in China and India, which is above our current funding bar (10x cash). This cost-effectiveness estimate is at an early stage. As a result, we have more uncertainty about this estimate than for some of our other grants. A sketch of our cost-effectiveness analysis and level of uncertainty is in the table below, based on preliminary information from a potential implementing organization. Key parameters are highlighted in bold.
What we are estimating | Best guess | Confidence intervals (25th - 75th percentile) | Implied cost-effectiveness |
---|---|---|---|
Grant amount | $9.7M | ||
Number of people in the intervention area | 40 million | 25-55 million | 10-19x |
Intervention impact on behavior, proxied by increase in salt substitute market share | 18 pp | 10-26 pp | 8-21x |
Duration of intervention impact on dietary intake, years | 4.25 | 4-6 | 14-20x |
Baseline annual cardiovascular death rate in intervention area, per 100,000 | 254 | 200-300 | 12-17x |
Relative risk of cardiovascular death after adjustments | 0.97 | 0.96-0.98 | 13-17x |
Deaths averted | 11,348 | ||
Initial cost-effectiveness estimate | |||
Cost per death averted | $855 | ||
Moral weight of averting one cardiovascular death, units of value | 30 | 20-40 | 10-19x |
Initial cost-effectiveness estimate, x cash | 12 | ||
Summary of primary benefits, % of total | |||
Reduced mortality | 87% | ||
Reduced morbidity | 13% | ||
Additional adjustments | |||
Medical costs averted adjustment | 20% | 10-30% | 13-16x |
Final cost-effectiveness estimate, multiples of cash transfers2 | 14 | ||
See our full cost-effectiveness analysis for this intervention here.
We have also considered other perspectives that are not captured explicitly in these cost-effectiveness estimates.
- Expert opinion on the effectiveness of dietary sodium reduction is mixed. Most experts support dietary sodium reduction, but some are opposed to it, and the issue is controversial (more). Coupled with the more uncertain evidence base for dietary sodium reduction, this reduces our enthusiasm for the intervention somewhat. However, we note that as modeled, the majority of the benefit of this intervention (~77%) is estimated to come from potassium-enriched salt substitution rather than dietary sodium reduction. The effectiveness of the former has a stronger evidence basis and is less controversial.
Main reservations
- We are uncertain about the impact of dietary salt modification on mortality risk in the general population. The randomized controlled trial of potassium-enriched salt substitution that is the primary basis for our cost-effectiveness analysis was conducted in high-risk people in a specific setting, and although the findings of a second trial with a less selected population are consistent with it, we still have some uncertainty about how well the findings would generalize to average-risk people in other settings (more). Adding to this uncertainty, we rely heavily on a secondary analysis of the SSaSS trial that has a wide uncertainty range. In addition, we are highly uncertain about the impact of dietary salt reduction on the overall risk of death, and we do not think this uncertainty is likely to be resolved in the near future (more). This is because there are no randomized trials of dietary salt reduction that were designed and powered to detect changes in the risk of death, such trials are currently thought to be infeasible, and the observational evidence is debated (more).
- The intervention depends on individual behavior change, and we are very uncertain about population adherence outside the context in which it has been studied. We make very uncertain assumptions about how well people will adhere to the intervention. Assumptions are based on unpublished information from a potential implementing charity and data from the SMASH trial, but adherence could differ greatly due to intervention details and location (more).
- We are not yet certain about important implementation details. This creates uncertainty about how well our current cost-effectiveness analysis represents the intervention we would be funding. For example, the intervention is less cost-effective if the implementing charity bears the cost of potassium-enriched salt substitute.
- We are highly uncertain about the moral weight we use for averting cardiovascular deaths. We currently assume that averting a typical death from cardiovascular disease is about one-fourth as valuable as averting the death of a child under 5. The difference is explained by age at death– cardiovascular disease typically kills adults who are middle-aged or older and have fewer remaining years to live than children. Moral weights are subjective and people may have substantially different intuitions about them, particularly when comparing scenarios that are very different. In addition, the intervention probably tends to avert the deaths of people who are at high risk of cardiovascular death, and therefore who have lower-than-average life expectancy even after a particular event is averted. We do not adjust for this because it would raise ethical concerns and we think it is counterbalanced by population-wide reductions in latent cardiovascular risk (more).
- Potassium-enriched salt substitute may cause health risks in people with chronic kidney disease. The evidence on this remains very uncertain, but we estimate that it offsets about 5% of the deaths averted by the intervention. Importantly, the intervention is thought to be net-beneficial even in people with chronic kidney disease because its cardiovascular benefits probably outweigh its risks in that population. Nevertheless, the real or perceived risk of the intervention in people with chronic kidney disease could impact its perception by governments and the public, and ultimately its feasibility (more).
Basics of the program
What problem does this program address?
Cardiovascular disease is the most common cause of death worldwide,3 including in many low- and middle-income countries,4 and occurs most frequently among people who are middle-aged or elderly.5 Three quarters of these deaths occur in low- and middle-income countries.6 High blood pressure (hypertension)7 is a major contributor to cardiovascular risk, including the risk of heart attack, stroke, heart failure, and kidney failure.8
Dietary salt (sodium chloride) intake is a determinant of blood pressure, and reducing it lowers blood pressure in randomized controlled trials, particularly among people with hypertension (-5.5 mm Hg systolic).9 Since high blood pressure is an important risk factor for cardiovascular events like heart attacks and strokes,10 most experts and public health bodies like the World Health Organization believe lowering salt intake reduces cardiovascular risk on a population level.11 Potassium intake also impacts blood pressure, with effects opposite those of sodium.12
Global Burden of Disease (GBD) modeling suggests that high sodium intake is the single largest dietary risk factor for premature death.13 GBD estimates suggest that in 2019, high sodium intake was responsible for:
- 1.9 million deaths globally, representing 3.3% of all deaths.
- 1.1%, 2.4%, and 4.8% of all deaths in countries with low, low-middle, and middle socio-demographic indices, respectively.14
GBD does not publish estimates of the mortality burden of low potassium intake. We have not reviewed the methodology underlying these estimates, and we view them with some uncertainty.
As of 2023, all WHO countries have committed to aggressive salt reduction targets (-30% by 2030), but few have taken recommended actions to achieve this.15
What is the program?
This program intends to reduce deaths and illness from cardiovascular disease by promoting the use of potassium-enriched salt substitute and reducing overall salt intake in the general population of low- and middle-income countries.
Potassium-enriched salt substitute is a replacement for normal table salt (sodium chloride) in which potassium replaces a portion of the sodium. Large-scale trials have used different formulations of salt substitute, with sodium chloride levels ranging from 65% to 75%, potassium chloride levels ranging from 25% to 30%, and 10% magnesium sulfate in some trials.16
While we are not yet certain about implementation details, preliminary information from a potential implementing organization suggests that this intervention may involve a combination of the following:
- Government advocacy and support, such as advocating for less severe warning labels on potassium-enriched salt substitute.17
- Public education campaigns to increase public awareness of the benefits of dietary salt modification.
- Advocacy to medical professionals to promote dietary salt modification.
- Support for the salt industry and retailers to promote and increase the availability of potassium-enriched salt substitute.
- Market shaping and subsidy to lower the price of salt substitute.
We use the Shandong Ministry of Health Action on Salt and Hypertension (SMASH) intervention as a demonstration of feasibility and a model for our cost-effectiveness analysis because it resembles the intervention described above and was successfully implemented on a large scale.
SMASH was a government-led 5-year public health campaign in Shandong province, China, that focused on salt reduction and replacement.18 Shandong contained 96 million people at the time of its 2010 census,19 one year before the SMASH intervention began,20 and its salt intake was above the national average.21
Program activities engaged multiple stakeholders in government, industry, and the public, and included:
- An extensive public media campaign targeting sodium reduction, including messaging in newspapers, broadcasts, pamphlets, posters, signs, and billboards.
- Development of voluntary local food standards and regulations for caterers, supermarkets, and food manufacturers to reduce the salt content of foods.
- Low-sodium food displays in supermarkets.
- Distribution of over 13 million scaled salt spoons.22
- Promotion of salt substitute that was 70% sodium chloride, 30% potassium chloride.23
How many people does dietary salt modification reach?
Intervention area and cost per person
Based on preliminary information from a potential implementing organization, we assume the intervention would cover an area containing about 40 million people, 20 million in China and 20 million in India. The preliminary budget for the intervention would be $9.7 million USD for a three-year program, which implies an annual cost of $0.08 USD per person covered.24
Intervention adherence
We assume the impact of the intervention on behavior would be 70% of what occurred in the SMASH trial, as proxied by market share of potassium-enriched salt substitute (intervention impact of 18 pp on market share).
The precise number of people whose diet is affected by the intervention cannot be estimated since it will result in widely varying levels of adherence between households. However, we use the proportion of small retail sales of table salt that are potassium-enriched salt substitute as a proxy for population adherence to the intervention, relative to the SMASH trial. In the SMASH trial, reported market share of potassium-enriched salt substitute was 0.4% at baseline, and 26.1% at endline, implying an intervention impact of 25.7 pp.25
Based on a preliminary estimate from a potential implementing organization, we assume the endline market share of potassium-enriched salt substitute would be 23% in an implementation setting. Roughly assuming a 5% baseline market share, this implies an intervention impact of 18 pp in an implementation setting.26 This figure implies that adherence in an implementation setting would be 70% of what occurred in the SMASH trial. Although this figure is a guess and we are very uncertain about it, we view it as reasonable since it is lower than what was achieved in the SMASH trial.
Main reservations and uncertainties
Our main reservations about coverage are the following:
- The intervention depends on individual behavior change, and we are very uncertain about population adherence outside the context in which it has been studied. The only example of this program at scale we are aware of is the SMASH trial, and it is not clear that the intervention would get similar adherence in other contexts.
- Our estimate of the population contained in the intervention area is very preliminary, and may change as implementation details are refined.
What impact does dietary salt modification have?
Summary
Our cost-effectiveness analysis models two main benefits from dietary salt modification:
- Reduced cardiovascular mortality
- Reduced cardiovascular morbidity
A summary of the contributions of each type of benefit to our estimate of the modeled value of the program is below.
What we are estimating | % of modeled benefits |
---|---|
Reduced cardiovascular mortality | 87% |
Reduced cardiovascular morbidity | 13% |
A brief summary of our modeling process:
- We estimate the impact of the SMASH intervention on coverage of potassium-enriched salt substitute and dietary salt reduction based on differences between baseline and endline data, and adjustment for secular trends.
- We estimate the impact of potassium-enriched salt substitution and dietary salt reduction on cardiovascular mortality using data from a RCT (the SSaSS trial) and a modeling study (Zhang et al. 2018), and scale them for coverage to estimate the impact of SMASH intervention on cardiovascular mortality.
- We adjust for the difference in coverage between the SMASH intervention and the implementation scenario, based on information from the charity.
We also include a supplemental adjustment to account for medical costs averted as a result of fewer cardiovascular events. Rather than explicitly modeling this, we have applied a 20% adjustment based on previous modeling work for other interventions.27
Overall, we estimate that it costs about $855 to avert a death via dietary salt modification in China and India, and the intervention is about 14 times as cost-effective as spending on unconditional cash transfers. However, this is a preliminary estimate that we plan to refine with additional work.
Impact of the SMASH intervention on coverage of potassium-enriched salt substitute and sodium intake
Our cost-effectiveness analysis models the cardiovascular benefit of the SMASH trial using the intervention’s reported impact on the use of potassium-enriched salt substitute and its reported impact on sodium intake, then adjusts the estimated cardiovascular benefit to fit the current implementation scenario. The SMASH trial reports that the use of potassium-enriched salt substitute increased by 26 pp from baseline to endline, sodium intake declined by 25%, and systolic blood pressure declined by 1.8 mmHg. Since the study did not have a control group,28 we believe these findings are probably confounded by secular trends. The changes in nutrient intake that can be attributed to the intervention were probably smaller than reported, and the change in systolic blood pressure was probably larger than reported.
See our description of the SMASH population and intervention here. The impact of the SMASH trial is published in Xu et al. 2020. There is no control group, so the study compares data collected at baseline vs. in the last year of the intervention, which spans a 5-year period.29
The study does not report the impact of the intervention on cardiovascular events or mortality. Key findings from a representative sample of adults:
- Potassium-enriched salt substitute “accounted for more than a quarter of sales of small packaged retail salt by the end of the study period.”30 Industry data suggest that market share of potassium-enriched salt substitute was 0.4% at baseline and 26.1% at endline, implying an intervention impact of 25.7 pp.31 This is the most important outcome of the trial for our cost-effectiveness model, since potassium-enriched salt substitution accounts for 77% of the estimated benefit of the intervention.
- Mean sodium intake was 25% lower, and mean potassium intake was 15% higher, at endline vs. baseline. These were measured objectively using 24-hour urine collections from a representative sample of the population. Although confidence intervals for the difference are not provided, p-values imply that these differences are precisely estimated (p < .001).32 The reduction in sodium intake is the second-most important outcome for our cost-effectiveness model, since it accounts for 23% of the estimated benefit of the intervention.
- Mean systolic blood pressure (SBP) was 1.8 mmHg lower and diastolic blood pressure (DBP) was 3.1 mmHg lower in a representative sample of the population. Although the trial does not report confidence intervals for the difference, the p-value for the change in SBP implies substantial uncertainty about the size of the effect (p = 0.04). The difference in DBP is more precisely estimated (p < .001).33 We use changes in blood pressure to sense-check our main cost-effectiveness estimates, after adjusting for nationwide secular trends in blood pressure (see below).
- “Knowledge, attitudes, and behaviors associated with dietary sodium reduction and hypertension improved significantly.”34 We do not directly use this estimate in our cost-effectiveness analysis, but it supports the other outcomes.
Because the study did not have a control group, these differences could be confounded by secular trends unrelated to the intervention. We believe this is likely.
Secular trends suggest that sodium intake has been decreasing and potassium intake has been increasing nationwide in China, implying that the intervention may not have had as large of an impact on intake as implied by the Xu et al. 2020. Between 1997 and 2011, average sodium intake in China declined by ~17%35 and fruit and vegetable intake (the main source of dietary potassium) increased.36
This implies that about one quarter of the decline in sodium intake that occurred during the SMASH intervention,37 and an unknown fraction of the increase in potassium intake, may have been due to secular trends rather than the intervention. This further implies that the true reduction in sodium intake caused by the SMASH intervention may have been closer to 19% rather than 25%.38 We use the former figure in our cost-effectiveness analysis.
Nationally representative data suggest that in China as a whole between 1991 and 2011, SBP increased by 6 mmHg in men and 5 mmHg in women.39 Assuming this rate of increase applies to the 5-year period of the SMASH study, we would counterfactually expect a 1.4 mmHg increase in SBP.40 This implies that the intervention may have actually reduced SBP by about 3.2 mmHg,41 which is a larger effect size than reported (1.8 mm Hg). However, this adjustment for secular trends is very uncertain because it assumes nationwide trends apply to Shandong province and were linear over the period of time in question.
Given the size of the impact of potassium-enriched salt substitute and dietary sodium reduction on blood pressure in randomized controlled trials, and the lower intervention intensity achieved in SMASH relative to trial settings, we are skeptical that the true impact of the intervention on SBP was 3.2 mmHg. For the purposes of sense-checking our main estimate of cardiovascular mortality reduction in our cost-effectiveness analysis, we cut the secular trend adjustment in half to 0.7 mmHg, yielding an estimate that the SMASH trial reduced population-wide SBP by 2.5 mmHg. We use this figure to sense-check our main cost-effectiveness analysis.
Together, this suggests that the intervention was successfully implemented and substantially increased the use of potassium-enriched salt substitute, decreased overall sodium intake, and reduced blood pressure on a population level. However, the degree of reduction is uncertain, mainly due to the absence of a control group.
Impact of potassium-enriched salt substitution and dietary salt reduction on cardiovascular mortality
We estimate that the version of dietary salt modification we model would reduce the population-wide risk of cardiovascular death by about 3% in China and India, after adjustments. The intervention has two elements, whose benefits we estimate separately and then sum and adjust to arrive at this 3% figure: potassium-enriched salt substitution and dietary salt reduction. To estimate deaths averted, we multiply this 3% reduction by the rate of cardiovascular death in China and India estimated by the Institute for Health Metrics and Evaluation (IHME), which was 322 and 185 per 100,000 annually in 2019. We estimate that 77% of the modeled benefit of the intervention comes from potassium-enriched salt substitution, which has a fairly strong evidence base, and 23% from dietary salt reduction, which has a weaker and more contested evidence base.
Our cost-effectiveness analysis first models the impact of the SMASH dietary salt modification trial on cardiovascular mortality, then uses local cardiovascular mortality rates and external validity adjustments to adapt this estimate to the settings our funding would support.
Our estimate of the impact of potassium-enriched salt substitution on cardiovascular mortality comes from the SSaSS trial, a large and high-quality cluster-randomized trial conducted in China that reports that potassium-enriched salt substitution reduces cardiovascular mortality by 13% in people at high risk of stroke (95% confidence interval, 4% to 21%).42 We scale its effect size sharply downward to reflect much lower intervention impact in the SMASH trial vs. the SSaSS trial (26 pp increase in coverage of salt substitute vs. 72 pp in the trial).
Our estimate of the impact of dietary salt reduction on cardiovascular mortality is more indirect and much more uncertain, since there are not informative randomized controlled trials of salt reduction with mortality outcomes. We use an estimate from Zhang et al. 2018, a modeling study that estimates the impact of dietary salt reduction in the SMASH population on cardiovascular mortality via the impact of sodium reduction on blood pressure.
Summing these two benefits implies that dietary salt modification reduced the population-wide risk of cardiovascular death by about 6% in the SMASH trial.
We then adjust this estimate downward for internal validity (0.93) and external validity (0.49) concerns. Our main internal validity concern is the indirect estimation method used in Zhang et al. 2018. Our main external validity concerns are lower expected adherence in the current implementation setting vs. the SMASH trial (18 pp vs. 26 pp, proxied by intervention impact on market share of salt substitute) and the possibility that people in the current implementation setting may get more of their salt from processed food vs. the SSaSS trial, and this may be harder to modify (70% adjustment). After applying adjustments, this yields a reduction in cardiovascular mortality risk of about 3%.
Two sense-checks based on the estimated impact of the SMASH trial on blood pressure return estimates that are roughly consistent with the main estimate (83% and 95% of the main estimate).
We have four main uncertainties about this evidence. First, we are unsure how well the findings of the SSaSS trial apply to the general population in other settings. Second, we rely on a secondary analysis of the SSaSS trial with a wide uncertainty interval, increasing our uncertainty about this estimate. Third, we are very uncertain how much dietary salt reduction impacts mortality, due to its weaker and more contested evidence base. Fourth, this intervention requires substantial behavior change, and we are unsure what level of adherence would be achieved in novel settings.
Impact of potassium-enriched salt substitution and dietary salt reduction on blood pressure
Reducing blood pressure is the main mechanism by which dietary salt modification is thought to reduce cardiovascular events and deaths.43 Therefore, as part of our evidence evaluation process we considered the impact of potassium-enriched salt substitution and dietary salt reduction on blood pressure. Meta-analyses of randomized controlled trials provide strong evidence that reducing sodium intake, increasing potassium intake, and potassium-enriched salt substitution lower blood pressure, though the effect occurs primarily in people with hypertension.
A 2020 Cochrane meta-analysis of randomized controlled trials reports that reducing dietary salt intake by about 8 grams per day on average (a large reduction; ~1.3 tsp)44 reduces blood pressure, particularly among people with hypertension.
- In white people without hypertension, salt reduction reduced systolic blood pressure by an average of 1.1 mmHg (95% confidence interval, 0.6 to 1.6; 89 studies; GRADE high-quality evidence). The reduction tended to be larger in black people, and was similar but not statistically significant in Asian people.
- In white people with hypertension, salt reduction reduced systolic blood pressure by an average of 5.5 mmHg (95% confidence interval, 4.6 to 6.5; 85 studies; GRADE high-quality evidence). The reduction was similar in black people, and tended to be larger in Asian people.45
The paper reports that publication bias did not significantly impact these findings.46
A 2013 Cochrane meta-analysis that only considered longer-term trials with moderate dietary salt restriction, averaging a reduction of 4.4 grams per day of salt (~0.7 tsp), reports similar findings.47 However, even this more modest reduction in salt intake is larger than would be expected from population-wide interventions similar to SMASH. Together, this provides strong evidence that dietary salt reduction lowers blood pressure, though the effect is small in people without hypertension.
Potassium intake also impacts blood pressure, with effects opposite those of sodium.48 Part of the explanation for this effect is that increasing potassium intake increases sodium excretion by the kidneys.49 Therefore, the impacts of the two minerals on blood pressure are linked.
The evidence of the impact of potassium intake on blood pressure is summarized in a 2012 meta-analysis of randomized controlled trials conducted by the World Health Organization. The intervention was predominantly potassium supplementation, although two of 22 trials used dietary advice to increase potassium intake instead.50 Potassium and sodium intake were highly variable across trials.51
- In people with hypertension, increasing potassium intake from typical levels reduced systolic blood pressure by an average of 4.7 mmHg (95% CI 2.4 to 7.0; 17 trials).52
- In people without hypertension, increasing potassium intake from typical levels did not impact systolic blood pressure, but this is based on only three trials (0.1 mmHg systolic; 95% CI –0.77 to 0.95; 3 trials).53
- In populations with mixed hypertension status, increasing potassium intake reduced systolic blood pressure by an average of 3.0 mmHg, but this is based on only two trials (95% CI 0.3 to 5.7; 2 trials).54
Neither baseline sodium intake nor between-group difference in potassium intake were significant predictors of the change in blood pressure.55 Funnel plots do not raise concerns about publication bias.56 This provides strong evidence that potassium supplementation lowers blood pressure, though the effect appears to be limited to people with hypertension.
Potassium-enriched salt substitute combines the effects of decreasing sodium and increasing potassium intake, although outside of tightly controlled research settings it typically has a larger proportional effect on potassium intake than on sodium intake.57 A 2022 meta-analysis of 21 randomized controlled trials, some of which were conducted in people with hypertension and some of which were conducted in mixed populations,58 reports that it decreases systolic blood pressure by 4.6 mm Hg (95% confidence interval 3.1 to 6.1; 19 trials).59 A test of publication bias finds some evidence of it, suggesting that the effect size may be overestimated.60
Together, this provides strong evidence that dietary sodium reduction and potassium-enriched salt substitution reduce blood pressure, though the effect occurs primarily in people with hypertension.
Impact of potassium-enriched salt substitution on cardiovascular mortality
One high-quality RCT, the SSaSS trial, reports that replacing 72% of typical salt intake with potassium-enriched salt substitute reduces cardiovascular deaths by about 13% (95% confidence interval, 4% to 21%) in people at high risk of stroke. This estimate is what we use in our cost-effectiveness analysis. A second less informative RCT reports a larger effect size of 36% (95% confidence interval, 8% to 56%), and a third RCT at high risk of bias is consistent with the other two (41%; 95% confidence interval, 5% to 63%). We believe this strongly supports the effectiveness of salt substitution. However, the effect size reported in the SSaSS trial has a wide uncertainty interval, leading us to be quite uncertain about the true size of the effect.
We performed a medium-depth scientific literature search for trials of potassium-enriched salt substitute with cardiovascular and/or mortality outcomes.61 This returned six trials, three of which had too few deaths and/or cardiovascular events to return informative estimates.62 The three trials that were large enough to be informative are:
- The Salt Substitute and Stroke (SSaSS) trial. This large cluster-randomized trial is the primary basis for the estimate of the impact of potassium-enriched salt substitute on cardiovascular mortality we use in our cost-effectiveness analysis because it is the largest, highest-quality trial we identified (more).
- Yuan et al. 2023. This large cluster-randomized trial impacts how we interpret and adjust the findings of the SSaSS trial for our cost-effectiveness analysis (more).
- Chang et al. 2006. This cluster-randomized trial has serious limitations so it does not directly impact our cost-effectiveness analysis, but we view it as directionally consistent with the other trials (more).
We also identified a meta-analysis that pools the cardiovascular mortality findings from three trials, Yin et al. 2022. We do not use this estimate in our cost-effectiveness analysis because 95% of study weight comes from the SSaSS trial, and most of the remaining weight comes from Chang et al. 2006, which we believe is at high risk of bias (more).
The SSaSS trial
The SSaSS trial was a five-year cluster-randomized trial in 600 villages across five provinces in China.63 About 35 people were selected from each village,64 for a total of 20,995 people.65 Participants were at high risk of stroke, due to a history of stroke or being older than 60 with high blood pressure.66 88% of participants had a hypertension diagnosis at baseline, 79% were taking blood pressure-lowering drugs, and average blood pressure was high (154 mmHg systolic).67
In this setting, the majority of dietary salt intake came from salt added during home cooking.68 The intervention provided salt substitute containing about 70% sodium chloride and 30% potassium chloride, in sufficient amounts for the cooking needs of the household.69 It also included advice to use salt substitute instead of regular table salt, and to use the salt substitute more sparingly than they are accustomed to using salt.70
The researchers measured urinary sodium and potassium excretion over 24-hour periods once per year in representative samples to estimate the impact of the intervention on sodium and potassium intake. Compared to the control group, the average reduction in sodium intake was 0.35 grams per day, representing an 8% decrease from baseline.71 Compared to the control group, the average increase in potassium intake was 0.80 grams per day, representing a 57% increase from baseline.72 Therefore, the intervention primarily increased potassium intake and only modestly reduced sodium intake. These figures imply that on average, participants replaced 72% of regular salt with salt substitute.73 This was associated with an average decline in SBP of 3.3 mmHg relative to the control group.74
The primary outcome of the trial was stroke incidence. Compared with the control group, after 4.7 years the intervention group experienced 14% fewer strokes (95% confidence interval, 4% to 23%), and this was statistically significant (p = 0.006).75 In our cost-effectiveness analysis, we use the impact of the intervention on cardiovascular death, which was a secondary outcome. The intervention reduced cardiovascular deaths by 13% (95% confidence interval, 4% to 21%), and this was statistically significant after correction for multiple comparisons.76
The main concern with salt substitution is hyperkalemia, an abnormal elevation of blood potassium level in people who have kidney disease that can lead to death.77 The intervention group did not experience significantly more serious adverse events related to hyperkalemia,78 although participants at high risk of hyperkalemia were excluded at baseline so it is not clear how relevant this is to population-wide interventions.79 Lawrence Appel, professor of medicine at Johns Hopkins University who studies sodium and cardiovascular health, argued in feedback to GiveWell that the procedures to exclude people with chronic kidney disease in the SSaSS trial were probably not very effective because many people with the condition do not know they have it and self-report can be unreliable. Therefore, the lack of a significant effect on hyperkalemia in the trial was probably meaningful.
We believe the SSaSS trial is convincing for the following reasons:
- The findings of the trial are consistent with priors, which increases our confidence that they are valid. Based on the preregistered power analysis, the researchers expected a reduction of stroke risk of about 13%, and the trial reports a 14% reduction in stroke risk.80
- The research team used best practices in trial design that reduce the risk of bias. The study design and analysis were publicly registered before the study began.81 Sample size required to detect an effect was determined in advance by a power calculation.82 The registry page declared a single primary outcome,83 and it remained the primary outcome in the paper.84 The analysis was conducted on an intention-to-treat basis.85 This means that participants were included in the final figures whether or not they adhered to the intervention. This gives a better picture of real-world effectiveness and reduces bias.
- The p-values for the outcomes are fairly small, suggesting a low risk that the findings are due to chance.86
- P-values of secondary outcomes are corrected for multiple comparisons,87 reducing the risk of false positives.
The SSaSS trial has two main limitations for our purposes. First, although we are confident that the intervention was effective, we are uncertain about its precise effect size. For example, although the intervention group experienced 13% fewer cardiovascular deaths than the control group, the 95% confidence interval ranges from a 4% reduction to a 21% reduction.88
Second, since the intervention was conducted in people at high risk of stroke, it is less directly relevant to population-wide salt substitution interventions in which beneficiaries would have a range of risk profiles. In our CEA, we assume the relative risk of cardiovascular death in the SSaSS trial would apply to the general population, even though absolute risk would be lower. This is an uncertain assumption, although we believe the findings of the trial Yuan et al. 2023 suggest that the results of the SSaSS trial are probably applicable to a more general population (more).
Yuan et al. 2023
Yuan et al. 2023 was a two-year cluster-randomized trial conducted in 48 residential elderly care facilities with 1,612 men and women in four regions of Northern China with high salt intake and a high prevalence of hypertension.89 The baseline prevalence of hypertension in trial participants was about 62%, 39% were taking blood pressure-lowering drugs, and average blood pressure was moderately elevated (~138 mmHg systolic).90 The trial involved potassium-enriched salt substitute and progressively reduced salt intake, implemented separately and together in different arms, and compared against a control group with no intervention.91 The salt reduction intervention had little impact on salt intake so we will not consider it further.92
Facility kitchens were provided with either normal table salt or potassium-enriched salt substitute that was 62.5% sodium chloride, 25% potassium chloride, and 12.5% dried flavorings.93
Compared with the control group at endline, sodium intake declined by 0.2 grams per day (nonsignificant) and potassium intake increased by 0.5 grams per day in the potassium-enriched salt substitute group compared with the control group.94 These changes are somewhat smaller than observed in the SSaSS trial (0.4 and 0.8 grams per day, respectively).95 As in the SSaSS trial, the intervention primarily increased potassium intake rather than reducing sodium intake.
The primary outcome of the trial was differences in systolic blood pressure between groups.96 Compared with the control group at three time points, the potassium-enriched salt substitute group had an average systolic blood pressure 7.1 mmHg lower (95% confidence interval, 3.8 to 10.5 mmHg).97
Cardiovascular and mortality outcomes were secondary outcomes of the trial.98 Deaths from cardiovascular disease were 36% lower (95% confidence interval, 8% to 56%) in the potassium-enriched salt substitute group, and all-cause mortality was nonsignificantly 16% lower (95% confidence interval, 13% higher to 37% lower), relative to the control group.99 These secondary outcomes are not corrected for multiple comparisons, implying that they underestimate uncertainty.100
Despite lower compliance with salt substitute relative to the SSaSS trial, the impact on blood pressure and deaths from cardiovascular disease was apparently larger in Yuan et al. 2023. However, the 95% confidence intervals overlap, so we cannot confidently conclude that the results are different.
We think this trial has some of the same strengths as the SSaSS trial, including preregistration101 and sufficient size to identify statistically significant impacts on cardiovascular outcomes. However, Yuan et al. 2023 also has several limitations that collectively reduce its evidence value for measuring the impact of the intervention on cardiovascular mortality:
- It was not designed or powered to detect intervention effects on cardiovascular events or mortality.102
- The uncertainty range around the reduction in cardiovascular mortality is large, encompassing the possibility that potassium-enriched salt substitute reduced cardiovascular mortality by 8% to 56%.103
- Secondary outcomes including cardiovascular mortality were not corrected for multiple comparisons, meaning that the reported uncertainty ranges are underestimated.104 Cardiovascular mortality outcomes would probably not be statistically significant after correcting for multiple comparisons.
- Participants were people living in residential elderly care facilities in which food was provided, so it has less relevance to a community-wide intervention like the SMASH trial.
For these reasons, we think the findings of Yuan et al. 2023 are not as informative as those of the SSaSS trial, and we do not directly use them in our cost-effectiveness analysis (we may do so in the future if we combine them via meta-analysis). However, they do indirectly impact our cost-effectiveness analysis by increasing our confidence in how the SSaSS findings can be generalized to the general population of low- and middle-income countries:
- One limitation of the SSaSS trial for estimating impacts on the general population is that participants were selected on the basis of high stroke risk, including hypertension (88% were diagnosed with the condition at baseline).105 In contrast, while study regions in Yuan et al. 2024 were selected on the basis of high hypertension rates,106 individual participants were not selected on the basis of risk factors other than age. Although hypertension at baseline was common (62%),107 it was not as common as in the SSaSS trial, and average blood pressure at baseline was also substantially lower (138 vs. 154 mmHg systolic).108 For comparison, in 2018 the average prevalence of hypertension among Chinese men age 50-59 was 47%, in men 60-69 it was 55%, and in women of the same ages it was 41% and 55%, respectively.109 The findings of Yuan et al. 2023 suggest that the results of the SSaSS trial are also applicable in populations that are not stringently selected for high blood pressure.
- The findings of Yuan et al. 2023 provide suggestive evidence that the true impact of the SSaSS trial’s intervention on cardiovascular mortality may have been somewhat underestimated, given its wide uncertainty interval (13% reduction in cardiovascular deaths; 95% confidence interval, 4% to 21%).110 In the future we may formally quantify this by pooling the results using meta-analysis.
Given the moderately high prevalence of hypertension in Yuan et al. 2024, we still think there is some risk that the cardiovascular mortality benefit reported in these trials would be smaller in the general population. However, this may be roughly counterbalanced by the possibility that the SSaSS trial underestimated the benefit of the intervention, so we do not adjust for it.
Chang et al. 2006
Chang et al. 2006 tested the impact of potassium-enriched salt substitution on cardiovascular mortality in elderly male Taiwanese military veterans living in a retirement home in which food was provided.
Five kitchens serving approximately 2,200 men were randomized to either continue serving typical food (control group) or prepare food with salt substitute (experimental group) for approximately 31 months.111 The salt substitute was 49% sodium chloride, 49% potassium chloride, and 2% “other additives.”112 At baseline, the men had an average age of 75 and average systolic blood pressure of 131 mmHg.113
The trial reports the following key outcomes:
- Estimated average sodium intake during the trial was 5.2 grams per day in the control group and 3.8 grams per day in the experimental group.114 In the experimental group, urinary sodium excretion decreased by 17%, and urinary potassium excretion increased by 76%.115
- Cardiovascular mortality was 41% lower in the experimental group than the control group (95% confidence interval, 5% to 63%).116
Although we have not evaluated the trial in detail, it has features that we believe limit its value for evaluating the effectiveness of community-wide dietary salt modification interventions resembling SMASH:
- It was a cluster-RCT with only five clusters, which is too few to expect randomization to even out baseline differences between groups.117 This increases the risk of bias.
- It was not preregistered, which increases the risk of bias.
- It targeted elderly male veterans living in a retirement home in which all food was provided, so it has less relevance to a community-wide intervention like the SMASH intervention.
Nevertheless, its findings are directionally consistent with those of the SSaSS trial and Yuan et al. 2023.
Impact of dietary salt reduction on cardiovascular mortality
Our estimate of the impact of dietary salt reduction on cardiovascular mortality in the SMASH trial comes from Zhang et al. 2018, a modeling study. It indirectly estimates that dietary salt reduction in the SMASH population averted 6,700 cardiovascular deaths, by modeling the impact of sodium reduction on blood pressure, and the impact of changes in blood pressure on cardiovascular mortality risk. It does not model the added benefit of increasing potassium intake from potassium-enriched salt substitute. After adjustment for nationwide secular trends in sodium intake, we use this figure to estimate that dietary salt reduction lowered cardiovascular mortality risk by 1.7% in the SMASH trial, though some of this overlaps with the effect of potassium-enriched salt substitution on sodium intake, which we adjust for to avoid double-counting. We are highly uncertain about this estimate because there is little direct evidence on the effect of sodium reduction on cardiovascular mortality, and the issue is contentious among experts.
We are not aware of compelling direct evidence testing the hypothesis that reducing salt intake reduces the risk of cardiovascular events or mortality, for example high-quality RCTs reporting these outcomes.118 We do not believe it is likely that such evidence will become available in the near future, because high-quality RCTs do not appear to be feasible.119
Available estimates of the impact of dietary salt reduction on cardiovascular risk mainly rely on observational studies and indirect estimates that use the impact of the intervention on blood pressure as an intermediate variable (sodium intake -> blood pressure -> cardiovascular risk). Findings from these types of studies are more uncertain than high-quality RCTs because they are more susceptible to confounding and they may miss benefits or harms that occur outside the specific mechanism that is modeled.
We rely on the indirect method in our CEA, due to the availability of a study that directly models the SMASH population at the time of the intervention, Zhang et al. 2018. To estimate the impact of dietary sodium reduction on cardiovascular deaths, the study used baseline blood pressure and sodium intake measures form the SMASH trial, cardiovascular mortality from the Shandong Death Registration System, an estimate of the impact of sodium intake on blood pressure from a meta-analysis of RCTs, and an estimate of the impact of blood pressure on cardiovascular mortality risk from IHME (based on observational evidence).
Zhang et al. 2018 reports that reducing sodium intake by the amount observed in the SMASH trial would be expected to prevent 6,700 deaths per year in Shandong province.120 Since we believe the reported impact of the SMASH trial on sodium intake is likely overestimated due to a nationwide decline in sodium intake, we adjust this downward by 24%.121 Using the population of Shandong province at the time, and an IHME estimate of the cardiovascular mortality rate in that province, we estimate that the sodium reduction component of the intervention reduced the cardiovascular mortality rate in Shandong province by 1.7%.122 We further adjust this downward by 15% to account for the fact that a small degree of sodium reduction is assumed in the mortality reduction estimate we use from the SSaSS trial of potassium-enriched salt substitute.123
We are very uncertain about this estimate, for several reasons:
- The estimation method is indirect, and high-quality direct evidence is not available to sense-check it.
- The impact of dietary salt reduction on cardiovascular risk (and all-cause mortality) is uncertain due to a lack of strong evidence (more). It is also controversial among experts,124 although most believe it is beneficial (more).
- We are uncertain about our adjustment for secular trends in sodium intake because it is based on nationwide trends that we do not know are relevant to Shandong province specifically.
To more broadly evaluate the hypothesis that dietary salt reduction lowers cardiovascular risk and all-cause mortality, we have conducted a review of the evidence on this subject.125 We believe the evidence as a whole supports the hypothesis that dietary salt reduction reduces cardiovascular mortality risk without substantial offsetting effects, but we are very uncertain about it because the evidence is complex, it has important limitations, and experts disagree on how to interpret it. Our reasoning is as follows:
- The main way dietary salt reduction is suggested to lower cardiovascular risk is via its impact on blood pressure, a well-established causal risk factor for cardiovascular disease.126 RCTs provide strong evidence that dietary salt reduction lowers systolic blood pressure, with larger effects in people with hypertension than without hypertension.127
- Because of the causal impact of blood pressure on cardiovascular risk, the impact of salt reduction on blood pressure would be expected to reduce cardiovascular risk and all-cause mortality unless it is counterbalanced by substantial harmful effects. We have not identified compelling evidence that such effects exist, although to our knowledge, the evidence also does not rule them out. The main potential mechanism of harm that has been highlighted by sodium reduction critics is the change in blood-pressure-regulating hormones that occurs with salt restriction. We do not think the evidence is compelling that these hormonal effects cause dietary salt restriction to be net-harmful, though the lack of direct evidence prevents us from reaching strong conclusions. See this section of the report for a more thorough discussion of this issue.
- The most rigorous meta-analysis of observational studies we identified, Ma et al. 2022, reports a positive linear relationship between sodium intake and cardiovascular event risk, consistent with the impact of salt reduction on blood pressure in RCTs.128 Additionally, it reports a nonsignificant U-shaped relationship between sodium intake and all-cause mortality, which is statistically significant in some other meta-analyses, suggesting the possibility of higher call-cause mortality at both high and low intakes.129 At face value, this suggests that the apparent benefit of low salt intake on cardiovascular death risk may be offset by higher risk of death from non-cardiovascular causes. We believe the trend toward higher all-cause mortality at low salt intake is likely due to reverse causality bias and thus does not reflect offsetting harms, but we are uncertain about this.130
- Relatively weak evidence from studies in rodents and primates, and human observational studies, mostly suggests that the long-term impact of salt intake on blood pressure may be larger than reflected in these shorter-term RCTs.131
Relative contributions of potassium-enriched salt substitution vs. dietary salt reduction to the benefits of dietary salt modification
As modeled, we estimate that 77%132 of the benefit of dietary salt modification comes from potassium-enriched salt substitution, and 23% comes from dietary salt reduction. As in our main estimate of the cardiovascular benefit of dietary salt modification, this is based on results from the SSaSS trial (salt substitution) and Zhang et al. 2018 (salt reduction), scaled to match the adherence level we expect in an implementation scenario.
To estimate this figure, we divide the estimated cardiovascular mortality benefit of potassium-enriched salt substitution in the SMASH trial by the estimated total benefit.133 This estimate depends heavily on assumptions about relative adherence to salt substitution and salt reduction in the beneficiary population. Our CEA assumes that relative adherence in the beneficiary population would be the same as the SMASH trial, but this can be adjusted based on intervention and population characteristics.
Adjustments
We apply downward adjustments to our estimate of the impact of dietary salt modification on cardiovascular mortality and morbidity. An internal validity adjustment of 0.93 is applied for evidence quality due to the limitations of Zhang et al. 2018. An external validity adjustment of 0.49 is applied for differences between study and implementation contexts. This adjustment accounts for differences in adherence (0.70) and the possibility that sodium intake may be harder to modify in an implementation context because more of it comes from processed food rather than salt added during cooking (0.70). The implied relative risk of cardiovascular death after all adjustments is 0.97, or a ~3% decrease.
Internal validity adjustment
We apply an internal validity adjustment of 0.93. This is a weighted average of two adjustments, applied to the two studies that provide the effect size estimate in our CEA:
- An adjustment of 1.0 for the SSaSS trial of potassium-enriched salt substitution. We do not apply a downward internal validity adjustment to its estimate because it has high-quality methods, its findings are consistent with priors, and it is directionally consistent with a lower-quality trial (more). We give this adjustment 77% weight in the final adjustment because we estimate that potassium-enriched salt substitution accounts for 77% of the effect of the dietary salt modification intervention we model.
- An adjustment of 0.7 for Zhang et al. 2018. This study uses an indirect modeling method to estimate the impact of dietary salt reduction on cardiovascular mortality. Our downward adjustment reflects our uncertainty about this estimate, which could be susceptible to bias and may miss offsetting harms (more). We give this adjustment 23% weight in the final adjustment because we estimate that dietary sodium reduction accounts for 23% of the effect of the dietary salt modification intervention we model (independent of the small impact of potassium-enriched salt substitution on sodium intake).
Taking a weighted average of these two adjustments yields a final internal validity adjustment of 0.93.
External validity adjustment
We apply an external validity adjustment of 0.49, intended to adjust our estimated effect size for differences between study vs. implementation settings. This reflects two components:
- An adjustment of 0.70 for lower intervention impact on behavior change in the implementation setting vs. the SMASH trial. This reflects 18 pp estimated intervention impact in the implementation setting vs. 26 pp in the SMASH trial, proxied by retail sales of salt substitute (more).
- An adjustment of 0.70 for the possibility that sodium intake may be harder to modify in an implementation context because more of it comes from processed food rather than salt added during cooking. The SSaSS trial was conducted in a rural setting where most of ingested salt was added during cooking in the household,134 and this may not be the case for an implementation setting, though this remains to be determined.
Multiplying these two adjustments together yields a final external validity adjustment of 0.49.
We do not adjust for baseline sodium or potassium intake. Our preferred meta-analysis of observational studies, Ma et al. 2022, reports that the relationship between sodium intake and cardiovascular risk is compatible with a linear function, and the same is true for potassium intake.135 This implies that increasing potassium intake by 500 mg per day would have the same impact on the absolute risk of cardiovascular disease whether a person started off with a low or high intake. In turn, this implies that baseline sodium and potassium intake should not be adjusted for. However, we are very uncertain about this because the linearity of the relationships becomes questionable at more extreme values,136 and a linear relationship is counterintuitive since one would expect the share of cardiovascular events attributable to low potassium intake (for example) to increase as potassium intake decreases.
IHME estimate of cardiovascular death rates
To estimate deaths averted, we multiply the relative reduction in cardiovascular mortality by the rate of cardiovascular death in China and India estimated by the Institute for Health Metrics and Evaluation (IHME), which was 322 and 185 per 100,000 annually in 2019,137 or around 0.2-0.3%.138 We plan to refine these figures when more specific locations have been identified.
We have not investigated the accuracy of IHME estimates of cardiovascular mortality in China and India, and we regard them with some uncertainty. However, cardiovascular disease is a common cause of death and it has distinctive signs and symptoms, so this is not a major concern.
Sense-checks
Two sense-checks using the estimated impact of the SMASH intervention on blood pressure are roughly consistent with the main estimate (83% and 95%), however both have major limitations.
As a sense-check on our estimate of the impact of dietary salt modification on cardiovascular mortality risk, we model the effect size in two other ways:
- Beginning with the estimated impact of the SMASH intervention on population blood pressure, we translate this into a relative risk of cardiovascular mortality using a meta-analysis of RCTs of blood pressure-lowering drugs reporting the relationship between blood pressure reduction and cardiovascular mortality (Ettehad et al. 2016). This returns an estimate that the intervention reduced cardiovascular mortality risk by 5%, implying an effect size 83% as large as the main estimate.139
- Beginning with the estimated impact of the SMASH intervention on population blood pressure, we translate this into a relative risk of cardiovascular mortality using a meta-analysis of observational studies of the relationship between blood pressure and cardiovascular mortality (Prospective Studies Collaboration 2002). This returns an estimate that the intervention reduced cardiovascular mortality risk by 6%, implying an effect size 95% as large as the main estimate.140
We regard these estimates as roughly consistent with the main estimate. However, both sense-checks have major limitations:
- Both sense-checks rely on our estimate of the population-wide change in SBP in the SMASH trial, which we very roughly adjust for secular trends, since the trial did not have a control group and the Chinese population as a whole experienced significant increases in SBP over that period. We assume that the true reduction in blood pressure resulting from the intervention was 2.5 mmHg, as opposed to 1.8 mmHg reported in the paper.141 We are very uncertain about how much to adjust for secular trends in SBP, if at all.
- Both sense-checks are indirect estimates, increasing our uncertainty about their outputs. For example, they do not reflect the possibility of additional benefits or harms that are not mediated via blood pressure.
- Both sense-checks are not fully independent of our main estimate, since the paper we use to estimate the impact of dietary sodium reduction on cardiovascular mortality, Zhang et al. 2018, also uses an indirect estimation method with blood pressure as an intermediate variable. However, Zhang et al. 2018 only accounts for 23% of the main estimate in our CEA.
- The meta-analysis underlying the first sense-check does not report on cardiovascular mortality, so we use the impact of blood pressure-lowering drugs on major cardiovascular event risk as a proxy for it. However, the meta-analysis does report a significant reduction in all-cause mortality.142
- The meta-analysis underlying the second sense-check also does not report on overall cardiovascular mortality, so we use the association between blood pressure and the risk of death from ischemic heart disease (heart attack) in people 60-69 years old as a proxy for it.143
- The first sense-check is based on trials of blood pressure-lowering drugs lasting 0.5 to 6.5 years. This may underestimate the possible long-term benefits of sodium reduction on blood pressure and cardiovascular risk (more).
- The second sense-check is based on observational evidence and could include residual confounding, potentially overestimating benefits of sodium reduction on blood pressure and cardiovascular risk.
Impact of dietary salt modification on cardiovascular morbidity
We assume that the relative impact of dietary salt modification on cardiovascular morbidity is the same as its relative impact on cardiovascular mortality. We apply this relative risk to annual years lost to disability (YLDs) from cardiovascular disease in China and India estimated by IHME.144 Our views about the strengths and weaknesses of the evidence on morbidity averted by this intervention are similar to those we have about mortality averted.
We have not investigated the accuracy of IHME estimates of cardiovascular morbidity in China and India, and we regard them with some uncertainty.
Additional benefits and downsides
Additional benefits
We believe this intervention probably has benefits beyond averting deaths and illness from cardiovascular disease. By averting cardiovascular events, it probably reduces expenses and economic opportunity costs associated with them, including lost income due to illness or having to care for ill family members. In our early-stage CEA, we apply a standard 20% adjustment to account for this,145 but we are uncertain about this figure and have not looked into it further.
Potential downsides
We are aware of four ways in which dietary salt modification could cause harm:
- A potential increase in all-cause mortality. Meta-analyses of observational studies report that low salt intake is correlated with higher all-cause mortality than average salt intake.146 Experts disagree about how to interpret this correlation. We think it is probably not causal, i.e. reducing salt intake probably does not increase all-cause mortality, but we have some remaining uncertainty due to a lack of strong evidence either way. We note that the degree of salt reduction we would expect from the specific program we model is too small for this to be even a theoretical concern, though that would not necessarily be true for programs in populations with lower baseline sodium intake. We expand on this below.
- In people with chronic kidney disease, potassium supplementation may increase the risk of hyperkalemia, a potentially fatal disorder.147 Two modeling studies based on hypothetical potassium-enriched salt substitution interventions in China and India report that hyperkalemia may cause a significant number of deaths, offsetting 2-10% of the deaths averted due to the intervention.148 However, salt substitution is estimated to confer a net reduction in mortality even among people with chronic kidney disease.149 In comments on this report, Laurence Appel pointed out to us that this concern remains fairly hypothetical. He shared a 2019 paper concluding that there is not enough evidence to come to strong conclusions about the impact of potassium-enriched salt substitute on hyperkalemia risk.150
- Potassium-enriched salt is more expensive than regular salt.151 The annual marginal cost of salt substitute per person is small152 but it would have to be borne by beneficiaries, the charity, or the government. Our cost-effectiveness analysis suggests that if the cost is borne by beneficiaries, it has little impact on cost-effectiveness.
- Reducing intake of added salt may also reduce the intake of iodine and other essential nutrients that are added to salt as fortificants. The WHO argues in a 2022 brief that salt reduction and salt iodization are compatible, but it may be necessary to adjust the concentration of fortificant to compensate for lower salt intake.
We incorporate the second and third items into our cost-effectiveness analysis. We estimate that excess deaths caused by hyperkalemia offset about 5% of the deaths averted by the intervention.153
Impact of dietary salt intake on all-cause mortality
Meta-analyses of observational studies report that low salt intake is correlated with higher all-cause mortality than average salt intake.154 This finding is not statistically significant in our preferred meta-analysis,155 Ma et al. 2022, but the trend is still toward higher risk.
We are not confident about how to interpret these findings as they are controversial in the scientific community (more), but we think reverse causality is the most likely explanation for higher mortality risk at low salt intakes. In other words, we think the correlation between lower salt intake and higher mortality risk in observational studies probably does not reflect a causal impact of low salt intake on mortality. See our discussion of reverse causality here.
In addition, we do not believe this is even a theoretical concern for the specific intervention we model because sodium intake is not expected to be reduced to a degree where all-cause mortality would be elevated.156 However, it could be a theoretical concern for locations where baseline sodium intake is below 4 grams per day.
If dietary salt reduction lowers blood pressure and cardiovascular mortality, it would have to cause substantial offsetting harm to result in a net increase in all-cause mortality. The most commonly cited mechanism we identified is that sodium restriction causes a compensatory increase in the circulating levels of the blood-pressure-regulating hormones renin, aldosterone, and noradrenaline157 . Researchers disagree about whether these changes are harmful158 .
Our understanding is that the main mechanism by which these hormonal changes could be harmful is via an increase in cardiovascular and kidney disease risk.159 However, we think the evidence suggests that dietary salt restriction probably reduces cardiovascular disease risk (more). In addition, a class of blood-pressure-lowering drugs called thiazide diuretics causes similar hormonal changes,160 and these drugs reduce cardiovascular events in randomized controlled trials.161 Regarding kidney disease, in limited searches we did not find direct evidence that dietary salt restriction or thiazide diuretics impact the risk of kidney disease in either direction, but we note that the most common causes of chronic kidney disease are high blood pressure and diabetes,162 and dietary salt restriction lowers blood pressure. Overall, we think it is unlikely that the hormonal changes caused by dietary salt restriction cause net harm via cardiovascular or kidney disease.
A second possible mechanism is that, in rodents, the blood-pressure-regulating renin-angiotensin system can lose some of its ability to respond to acute physiological stress if it is chronically activated by very low salt intake.163 Its ability to compensate for changes in blood pressure is weakened, leaving animals more vulnerable to death from blood loss and potentially other conditions that deplete blood volume.164 As an explanation for possible offsetting harms of dietary salt restriction, we view this as speculative for two reasons. First, the evidence supporting it as a cause of death is from rodents subjected to major blood loss, and it is not clear that this is relevant to the most common causes of death in humans. Second, the effects were observed at very low levels of salt intake that are not clearly relevant to humans eating a moderately salt-restricted diet.165
An argument that reduces our level of concern about dietary salt restriction is that humans appear to have mostly evolved with a sodium intake less than half of the intake that is typical in industrialized countries,166 suggesting that the physiology associated with lower sodium intake is unlikely to be harmful.
Overall, we think moderate dietary salt restriction probably does not increase all-cause mortality, but we have some remaining uncertainty about it due to the absence of direct evidence that it does not. To the extent one believes dietary salt restriction is harmful, targeting it to people who have particularly high salt intake is one way to reduce the risk.
Main reservations and uncertainties
- We are uncertain about the impact of potassium-enriched salt substitution on mortality risk in the general population. The SSaSS was conducted in high-risk people in a specific setting,167 and although a second trial in a less selected population is consistent with it (see above), we still have some uncertainty about how well this would generalize to average-risk people in other settings. We currently assume that the relative (not absolute) risk of cardiovascular mortality reported in the SSaSS trial is applicable to the general population across different settings, but this is uncertain. In addition, we rely on a secondary analysis of the trial and it has a wide confidence interval (13% reduction in cardiovascular mortality, 95% confidence interval 4% to 21%).168
- We are highly uncertain about the impact of dietary salt reduction on mortality risk. The evidence supporting dietary salt reduction is weaker than the evidence supporting potassium-enriched salt substitution, and its benefits are contested by some experts (more).169 We do not think this uncertainty is likely to be resolved in the near future.170
- The intervention depends on individual behavior change, and we are very uncertain about population adherence outside the context in which it has been studied. We make very uncertain assumptions about how well people will adhere to the intervention. Assumptions are based on unpublished information from a potential implementing charity and data from the SMASH trial, but adherence could differ greatly due to intervention details and location.
- We are not yet certain about important implementation details. This creates uncertainty about how well our current cost-effectiveness analysis represents the intervention we would be funding. For example, the intervention is less cost-effective if the implementing charity bears the cost of potassium-enriched salt substitute.
- We are very uncertain about the duration of impact of the intervention. We currently assume that the intervention takes two years to fully ramp up, losing 0.75 years of full impact equivalents during this time, and that the benefit of the intervention persists for 2 years of full impact equivalents after it ends. These figures are currently guesses. We may be able to refine our estimate of the duration of impact, but we believe it will remain quite uncertain.
- We are highly uncertain about the moral weight we use for averting cardiovascular deaths. We currently assume that averting a typical death from cardiovascular disease is about one-fourth as valuable as averting the death of a child under 5.171 The difference is explained by age at death– cardiovascular disease typically kills adults who are middle-aged or older and have fewer remaining years to live than children.172 It is also the case that, since the intervention averts deaths among people who are at high cardiovascular risk, they probably do not go on to live as long as average-risk people of the same age. However, we do not incorporate this into our CEA, for two reasons. First, we think it raises difficult moral issues because it undervalues saving the lives of people who need help the most.173 Second, since cardiovascular disease is a cumulative process that accrues slowly over many years due to exposure to risk factors including high blood pressure,174 we think this potential downside is counterbalanced by population-wide reductions in latent cardiovascular risk. In other words, the intervention benefits people who would not have a cardiovascular event averted in a particular year by reducing their risk in subsequent years. We do not know whether the net result of both factors causes us to underestimate or overestimate benefits overall.
- We are uncertain about the potential for leverage and funging to impact the cost-effectiveness of the intervention. We have not yet considered the impact of leverage and funging due to limited details about the intervention, and we think it is likely to reduce cost-effectiveness when taken into account. This is because our model does not currently account for costs incurred by governments, or the possibility that we might funge another donor. We should be able to model these variables once we have additional details about the intervention, though they will likely remain uncertain.
- We are uncertain about how to model medical costs averted. We apply our standard 20% adjustment for medical costs averted,175 but it is based on data from malaria prevention interventions and we have not yet looked into the possibility that it may be inaccurate in this scenario.
Additional perspectives beyond our cost-effectiveness model
We have considered other perspectives that are not captured explicitly in these cost-effectiveness estimates. Mixed expert opinion on the effectiveness of dietary salt reduction reduces our enthusiasm for the intervention somewhat, although our model estimates that most of the benefit of the intervention comes from potassium-enriched salt substitution.
Expert opinion on the effectiveness of dietary salt reduction is mixed
Public health bodies including the WHO generally believe typical dietary salt intake is harmful and reducing it is beneficial.176 WHO advocates for dietary salt reduction, and all WHO member states have committed to aggressive sodium reduction targets (-30% by 2030),177 implying broad support for the intervention.
However, some researchers argue that reducing sodium intake from average intake in most countries is probably not beneficial, and may even be harmful. Therefore, sodium reduction policy is not justified. The main arguments for this position are reviewed in O’Donnell et al. 2020. The arguments revolve primarily around observational studies suggesting that low salt intake is associated with adverse outcomes like higher all-cause mortality, and hormonal changes that the authors regard as potentially harmful. We think it is unlikely that a moderate reduction in sodium intake increases all-cause mortality, but given the uncertainty of the evidence, we cannot completely rule it out (more).
The fact that expert opinion is mixed increases our uncertainty about the effectiveness of the intervention. While the majority (77%) of the benefit of the version of dietary salt modification we have modeled comes from potassium-enriched salt substitution, which is not subject to this concern, interventions focusing more on dietary salt reduction would be more subject to it.
External reviews of our report and cost-effectiveness analysis
We sought external reviews of this report, and our cost-effectiveness analysis, by experts on both sides of the sodium reduction debate, and by Resolve to Save Lives, a nonprofit leading sodium reduction efforts.
- Lawrence Appel and Matti Marklund are researchers who study dietary sodium intake and cardiovascular health at Johns Hopkins University. They generally agreed with the conclusions of an earlier draft of the report but provided many comments on it, a number of which we incorporated. Key points:
- They argued that it is very unlikely (<5%) that sodium reduction increases all-cause mortality.
- Dr. Marklund stated that the simple method we use in our cost-effectiveness analysis (comparative risk assessment) is adequate, but he prefers the more complex Markov modeling method, which allows for evaluating health impact and cost-effectiveness over varying time horizons.
- Dietary salt reduction may have larger benefits over a lifetime because it may attenuate the age-related increase in blood pressure that occurs in people with typical salt intake.
- GBD modeling suggests that high sodium intake is the single largest dietary risk factor for premature death.178
- The idea that potassium-enriched salt substitute causes harm via hyperkalemia in people with kidney disease is currently hypothetical, not supported by direct evidence. They provided a review paper they authored that supports this, Greer et al. 2019.179
- A class of blood-pressure-lowering drugs called diuretics causes changes in blood-pressure-regulating hormones that are similar to salt restriction,180 and these drugs reduce cardiovascular events in randomized controlled trials.181 Therefore, since diuretics reduce cardiovascular events despite causing hormonal changes similar to those of salt restriction, it is unlikely that these hormonal changes would negate the cardiovascular benefits of reducing blood pressure through dietary salt restriction.
- If dietary salt modification were achieved by policy change, it may have longer-lasting effects than what we model.
- Franz Messerli is a cardiologist and coauthor of the paper O’Donnell et al. 2020. He expressed high uncertainty about what the true impact of a dietary salt modification program would be. Other key feedback:
- He agreed that potassium-enriched salt substitute probably reduces cardiovascular event risk, but pointed out that its main impact in the SSaSS trial was due to an increase in potassium intake rather than a decrease in sodium intake. We agree that this is probably true, and it is reflected in this report and our CEA.
- Due to the weakness of the evidence, he does not think a conclusion can be reached about whether modest dietary salt restriction impacts cardiovascular events and all-cause mortality in the general public. He argued that dietary salt restriction is probably beneficial in people who have both hypertension and a high intake of salt.
- He pointed to an observational analysis in O’Donnell et al. 2020 suggesting that the lowest all-cause mortality rate is associated with average sodium intake and high potassium intake.182 We have some uncertainty about how to interpret this; we discuss our skepticism of observational associations between sodium intake and all-cause mortality here.
- Resolve to Save Lives broadly agreed with the conclusions of an earlier draft of the report and stated that the cost-effectiveness model seems generally reasonable. Other key feedback:
- They argued that it is very unlikely (<2%) that sodium reduction increases all-cause mortality.
- They pointed out that another potential downside of dietary salt reduction is that it may reduce iodine intake in places where salt is fortified with iodine, unless efforts are coordinated with iodine fortification programs. We added this to the report.
- They argued that GiveWell’s moral weights undervalue the impact of cardiovascular morbidity on peoples’ lives. The morbidity burden we assume comes from Global Burden of Disease estimates of the morbidity caused by cardiovascular events, multiplied by our moral weight for morbidity. We did not update our moral weight for morbidity in response to this feedback because it is standard across all of our models. However, in future work we may investigate the accuracy of Global Burden of Disease estimates of the morbidity caused by cardiovascular events. Additionally, we account for the direct and indirect financial impacts of averting cardiovascular events with an adjustment that represents averted medical costs.
- They argued that the benefit of the program would probably extend further than the two years of “full impact equivalents” we assume accrue after the end of the program, but did not provide evidence of this. In response, we modeled impact decay over time, using a very rough guess that effect size fades exponentially by ⅓ per year, due to reduced individual adherence, and the possibility that dietary salt policy would have been implemented counterfactually. This yielded an estimate of 1.99 impact-years,183 very close to our initial assumption of 2.0, so we did not revise the CEA. However, we remain highly uncertain about how quickly program benefits fade relative to the counterfactual. Resolve to Save Lives pointed out that if the impacts are primarily via policy, they may be longer-lasting than if they are primarily via individual behavior change. We will update the model if we think a specific funding opportunity is likely to have a longer or shorter impact than what we assume.
Is there room for more funding?
Cardiovascular disease is the leading cause of death globally.184 In order to help address this burden, all WHO countries have committed to aggressive salt reduction targets (-30% by 2030), but few have taken significant action.185 Therefore, this may be an excellent time to provide support for salt substitution and reduction efforts. We believe this implies a large amount of room for more funding, but we are not sure how much.
We have identified one charity that is seeking funding to implement this type of intervention, Resolve to Save Lives. We have not researched other potential implementing organizations. Based on unpublished conversations with Resolve to Save Lives, we believe they may have the capacity to use several times more than the $9.7M modeled in our early-stage cost-effectiveness analysis, and that the specific program we fund would probably not occur without GiveWell support.
Sources
- 1
This estimate of the value per dollar donated to unconditional cash transfers is out of date as of 2024. We are continuing to use this outdated estimate for now to preserve our ability to compare across programs, while we reevaluate the benchmark we want to use to measure and communicate cost-effectiveness.
- 2
This estimate of the value per dollar donated to unconditional cash transfers is out of date as of 2024. We are continuing to use this outdated estimate for now to preserve our ability to compare across programs, while we reevaluate the benchmark we want to use to measure and communicate cost-effectiveness.
- 3
“Cardiovascular diseases (CVDs) are the number 1 cause of death globally, taking an estimated 17.9 million lives each year. CVDs are a group of disorders of the heart and blood vessels and include coronary heart disease, cerebrovascular disease, rheumatic heart disease and other conditions. Four out of 5 CVD deaths are due to heart attacks and strokes, and one third of these deaths occur prematurely in people under 70 years of age.” WHO, "Cardiovascular diseases"
- 4
See the IHME GBD Compare tool for causes of death.
- 5
Kochanek et al. 2019, table 6, p. 35, row "Major cardiovascular diseases" and subentries, and table 7, p. 38
- 6
"Over three quarters of CVD deaths take place in low- and middle-income countries." WHO, Cardiovascular diseases (CVDs)
- 7
"Hypertension (high blood pressure) is when the pressure in your blood vessels is too high (140/90 mmHg or higher)...Hypertension is diagnosed if, when it is measured on two different days, the systolic blood pressure readings on both days is ≥140 mmHg and/or the diastolic blood pressure readings on both days is ≥90 mmHg." WHO, Fact Sheet Detail, "Hypertension," 2019
- 8
- “Among other complications, hypertension can cause serious damage to the heart. Excessive pressure can harden arteries, decreasing the flow of blood and oxygen to the heart. This elevated pressure and reduced blood flow can cause:
- Chest pain, also called angina.
- Heart attack, which occurs when the blood supply to the heart is blocked and heart muscle cells die from lack of oxygen. The longer the blood flow is blocked, the greater the damage to the heart.
- Heart failure, which occurs when the heart cannot pump enough blood and oxygen to other vital body organs.
- Irregular heart beat which can lead to a sudden death.
Hypertension can also burst or block arteries that supply blood and oxygen to the brain, causing a stroke.
In addition, hypertension can cause kidney damage, leading to kidney failure.”
WHO, Fact Sheet Detail, "Hypertension," 2019- “Left undetected (or uncontrolled), high blood pressure can lead to:
- Heart attack — High blood pressure damages arteries that can become blocked and prevent blood flow to the heart muscle.
- Stroke — High blood pressure can cause blood vessels in the brain to clog more easily or even burst.
- Heart failure — The increased workload from high blood pressure can cause the heart to enlarge and fail to supply blood to the body.
- Kidney disease or failure — High blood pressure can damage the arteries around the kidneys and interfere with their ability to filter blood effectively.
- Vision loss — High blood pressure can strain or damage blood vessels in the eyes.
- Sexual dysfunction — High blood pressure can lead to erectile dysfunction in men or lower libido in women.
- Angina — Over time, high blood pressure can lead to heart disease or microvascular disease (MVD). Angina, or chest pain, is a common symptom.
- Peripheral artery disease (PAD) — Atherosclerosis caused by high blood pressure can cause a narrowing of arteries in the legs, arms, stomach and head, causing pain or fatigue.”
American Heart Association, "Health Threats From High Blood Pressure"
- “Among other complications, hypertension can cause serious damage to the heart. Excessive pressure can harden arteries, decreasing the flow of blood and oxygen to the heart. This elevated pressure and reduced blood flow can cause:
- 9
- “High sodium consumption (>2 grams/day, equivalent to 5 g salt/day) and insufficient potassium intake (less than 3.5 grams/day) contribute to high blood pressure and increase the risk of heart disease and stroke… Most people consume too much salt—on average 9–12 grams per day, or around twice the recommended maximum level of intake… The principal benefit of lowering salt intake is a corresponding reduction in high blood pressure.” Salt Reduction, WHO 2020.
- “Sodium reduction from an average high usual sodium intake level (201 mmol/day) to an average level of 66 mmol/day, which is below the recommended upper level of 100 mmol/day (5.8 g salt), resulted in a decrease in SBP/DBP of 1/0 mmHg in white participants with normotension and a decrease in SBP/DBP of 5.5/2.9 mmHg in white participants with hypertension. A few studies showed that these effects in black and Asian populations were greater.” Graudal et al 2017, abstract.
- 10
Ettehad et al. 2016 is a meta-analysis of RCTs of blood pressure-lowering drug interventions. It reports that reducing blood pressure reduces the risk of cardiovascular events. “Every 10 mm Hg reduction in systolic blood pressure significantly reduced the risk of major cardiovascular disease events (relative risk [RR] 0·80, 95% CI 0·77–0·83), coronary heart disease (0·83, 0·78–0·88), stroke (0·73, 0·68–0·77), and heart failure (0·72, 0·67–0·78), which, in the populations studied, led to a significant 13% reduction in all-cause mortality (0·87, 0·84–0·91).” Ettehad et al. 2016, abstract.
https://doi.org/10.1016/S0140-6736(15)01225-8 - 11
“High sodium consumption (>2 grams/day, equivalent to 5 g salt/day) and insufficient potassium intake (less than 3.5 grams/day) contribute to high blood pressure and increase the risk of heart disease and stroke… Most people consume too much salt—on average 9–12 grams per day, or around twice the recommended maximum level of intake… The principal benefit of lowering salt intake is a corresponding reduction in high blood pressure.” Salt Reduction, WHO 2020.
- 12
“High sodium consumption (>2 grams/day, equivalent to 5 g salt/day) and insufficient potassium intake (less than 3.5 grams/day) contribute to high blood pressure and increase the risk of heart disease and stroke.” Salt Reduction, WHO 2020.
- 13
“In 2017, 11 million (95% uncertainty interval [UI] 10–12) deaths and 255 million (234–274) DALYs were attributable to dietary risk factors. High intake of sodium (3 million [1–5] deaths and 70 million [34–118] DALYs), low intake of whole grains (3 million [2–4] deaths and 82 million [59–109] DALYs), and low intake of fruits (2 million [1–4] deaths and 65 million [41–92] DALYs) were the leading dietary risk factors for deaths and DALYs globally and in many countries.”
GBD 2017 Diet Collaborators 2019, abstract. - 14
- 15
“All 194 Member States have committed to reducing population sodium intake by 30% by 2030, demonstrating strong consensus on sodium reduction as a lifesaving strategy, this report, however, shows that only nine countries have fully established the recommended policies to reduce sodium intake. Globally we are off track to achieve the target.”
Launch of the WHO Global report on sodium intake reduction, WHO 2023. - 16
“The villages were randomly assigned in a 1:1 ratio to the intervention group, in which the participants used a salt substitute (75% sodium chloride and 25% potassium chloride by mass), or to the control group, in which the participants continued to use regular salt (100% sodium chloride).”
Neal et al. 2021, abstract.
“We conducted a double-blind, randomized controlled trial among 200 families in rural China to establish the 2-year effects of a reduced-sodium, high-potassium salt substitute (65% sodium chloride, 25% potassium chloride, 10% magnesium sulfate) compared with normal salt (100% sodium chloride) on blood pressure.” Zhou et al. 2013, abstract
“The regular salt in enrolled households was retrieved and replaced, free of charge, with a combination of 75% NaCl and 25% KCl.” Bernabe-Ortiz et al. 2020, abstract
“The salt substitute contains, on average, 65% sodium chloride, 25% potassium chloride and 10% magnesium sulphate and is iodized and commercially available in China, but mostly not used in rural areas.” Li et al. 2013, p. 4 - 17
Increasing potassium intake can pose risks for people with chronic kidney disease. See our discussion of this more.
- 18
“To assess the association of a government-led, multisectoral, and population-based intervention with reduced salt intake and blood pressure in Shandong Province, China… This cross-sectional study used data from the Shandong–Ministry of Health Action on Salt and Hypertension (SMASH) program, a 5-year intervention to reduce sodium consumption in Shandong Province, China.” Xu et al. 2020, abstract.
- 19
- 20
“SMASH was officially launched in March 2011, and implementation began after the baseline survey was conducted from June to July, 2011.” Xu et al. 2020, p. 878
- 21
“The Shandong Province of eastern China is the second most populous province in China, with more than 90 million residents.16 Shandong has been one of China’s largest salt-producing provinces since ancient times.17 In 2002, the mean dietary salt intake was 12.6 g per day in Shandong, which was above the national average.” Xu et al. 2020, p. 878
- 22
“The SMASH program used multisectoral intervention strategies and was implemented by 15 administrative departments across the province (eTable 1 in the Supplement). The program developed local food standards and regulations and promoted salt-reduction actions among caterers, supermarkets, and food-processing enterprises. Low-sodium food displays were established in 1461 supermarkets, promoting customer awareness of sodium labeling. To promote sodium reduction in home cooking, more than 13 million scaled salt spoons were widely distributed to households to facilitate awareness and measurement of salt added to food. The Shandong government launched intensive media campaigns to support the intervention. By the end of 2015, 1777 newspapers and 26 668 public broadcasts were disseminated to deliver relevant messaging to the population province-wide. Millions of posters, pamphlets, and signs were displayed in local schools, communities, restaurants, and cafeterias. More than 74 000 low-sodium diet advertising billboards were set up in communities, and 69 431 trainings were organized in local health agencies for the community members.” Xu et al. 2020, p. 878-9
- 23
“The bureau also promoted use of a low-sodium salt substitute that contains 30% potassium chloride.” Xu et al. 2020, p. 879
- 24
$9.7M / 40M = $0.24. $0.24 / 3 = $0.08
- 25
Xu et al 2020 states that potassium-enriched salt substitute “accounted for more than a quarter of sales of small packaged retail salt by the end of the study period.” The paper does not provide a specific figure.
Xu et al. 2020, p. 884.
Industry data provide specific baseline and endline figures of 0.4% and 26.1%.
Effect evaluation of Shandong - Ministry of Health Action on Salt and Hypertension (SMASH), Shandong Science and Technology Press 2018, table, p. 134-5 (unpublished). - 26
This 5% figure is a placeholder and is not based on direct evidence from implementation locations. It is based on the general information below, plus the assumption that market share is probably higher in China than in India.
Yin et al. 2021 reports that “An interviewee from the Chinese salt manufacturing industry reflected that the market share of reduced-sodium salts was less than 10% of the total salt market, with products mainly available in urban areas.”
Zhang et al. 2023 presents data from a nationally representative survey in China reporting that nationwide, 12.2% of people report using low-sodium salt substitute. We regard this as a likely overestimate, because of self-report bias, and the fact that a positive response does not necessarily imply consistent use. On the other hand, the survey occurred in 2015-16 and market share may have increased somewhat since then.
See our calculations here. - 27
See our write-up here.
- 28
"The primary limitation of this study is that SMASH was a province-wide intervention, and there was no control group." Xu et al. 2020
- 29
“To evaluate the intervention, 2 provincial representative cross-sectional surveys were performed in 2011 (preintervention survey) and 2016 (postintervention survey).” Xu et al. 2020, p. 878
- 30
Xu et al. 2020, p. 884.
- 31
Effect evaluation of Shandong - Ministry of Health Action on Salt and Hypertension (SMASH), Shandong Science and Technology Press 2018, table, p. 134-5 (unpublished).
- 32
“The 24-hour urinary sodium excretion decreased 25% from 5338 mg per day (95% CI, 5065-5612 mg per day) in 2011 to 4013 mg per day (95% CI, 3837-4190 mg per day) in 2016 (P < .001), and potassium excretion increased 15% from 1607 mg per day (95% CI, 1511-1704 mg per day) to 1850 mg per day (95% CI, 1771-1929 mg per day) (P < .001).” Xu et al. 2020, abstract.
- 33
“Adjusted mean systolic blood pressure among all participants decreased from 131.8 mm Hg (95% CI, 129.8-133.8 mm Hg) to 130.0 mm Hg (95% CI, 127.7-132.4 mm Hg) (P = .04), and diastolic blood pressure decreased from 83.9 mm Hg (95% CI, 82.6-85.1 mm Hg) to 80.8 mm Hg (95% CI, 79.4-82.1 mm Hg) (P < .001).” Xu et al. 2020, abstract.
- 34
Xu et al. 2020, abstract.
- 35
Li et al. 2016, table 2, p. 823.
- 36
Li et al. 2016, table 2, p. 823-4.
- 37
17% / 14 y = 1.21%
1.21% x 5 = 6.1%
6.1% / 25% = 0.24 - 38
25% * (1 - 24%) = 19%
- 39
Li et al. 2016, table 2, p. 822.
- 40
(5 + 6) / 2 = 5.5
(5.5 / 20) x 5 = 1.4 - 41
1.8 + 1.4 = 3.2
- 42
"The salt substitute was also shown to be beneficial with respect to death from vascular causes (22.9 events vs. 26.3 events per 1000 person-years; rate ratio, 0.87; 95% CI, 0.79 to 0.96)." Neal et al. 2021.
- 43
“High sodium consumption (>2 grams/day, equivalent to 5 g salt/day) and insufficient potassium intake (less than 3.5 grams/day) contribute to high blood pressure and increase the risk of heart disease and stroke… Most people consume too much salt—on average 9–12 grams per day, or around twice the recommended maximum level of intake… The principal benefit of lowering salt intake is a corresponding reduction in high blood pressure.” Salt Reduction, WHO 2020.
“Sodium reduction from an average high usual sodium intake level (201 mmol/day) to an average level of 66 mmol/day, which is below the recommended upper level of 100 mmol/day (5.8 g salt), resulted in a decrease in SBP/DBP of 1/0 mmHg in white participants with normotension and a decrease in SBP/DBP of 5.5/2.9 mmHg in white participants with hypertension. A few studies showed that these effects in black and Asian populations were greater.” Graudal et al 2017, abstract. - 44
The meta-analysis reports that average sodium intake declined from 201 mmol/d to 66 mmol/d, for a difference of 135 mmol/d. The molar weight of sodium is 22.99 u. 22.99 x 0.135 = 3.1 grams of sodium. Sodium is 40% of the mass of sodium chloride. 3.1 / 0.4 = 7.8 grams of salt. There are about 6 grams of salt in a teaspoon. 7.8 / 6 = 1.3.
“The average sodium intake was reduced from 201 mmol/day (corresponding to high usual level) to 66 mmol/day (corresponding to the recommended level).”
Graudal et al 2017, abstract. - 45
“The effect of sodium reduction on blood pressure (BP) was as follows: white people with normotension: SBP: mean difference (MD) ‐1.09 mmHg (95% confidence interval (CI): ‐1.63 to ‐0.56; P = 0.0001); 89 studies, 8569 participants; DBP: + 0.03 mmHg (MD 95% CI: ‐0.37 to 0.43; P = 0.89); 90 studies, 8833 participants. High‐quality evidence.
Black people with normotension: SBP: MD ‐4.02 mmHg (95% CI:‐7.37 to ‐0.68; P = 0.002); seven studies, 506 participants; DBP: MD ‐2.01 mmHg (95% CI:‐4.37 to 0.35; P = 0.09); seven studies, 506 participants. Moderate‐quality evidence.
Asian people with normotension: SBP: MD ‐0.72 mmHg (95% CI: ‐3.86 to 2.41; P = 0.65); DBP: MD ‐1.63 mmHg (95% CI:‐3.35 to 0.08; P =0.06); three studies, 393 participants. Moderate‐quality evidence.
White people with hypertension: SBP: MD ‐5.51 mmHg (95% CI: ‐6.45 to ‐4.57; P < 0.00001); 84 studies, 5925 participants; DBP: MD ‐2.88 mmHg (95% CI: ‐3.44 to ‐2.32; P < 0.00001); 85 studies, 6001 participants. High‐quality evidence.
Black people with hypertension: SBP MD ‐6.64 mmHg (95% CI:‐9.00 to ‐4.27; P = 0.00001); eight studies, 619 participants; DBP ‐2.91 mmHg (95% CI:‐4.52, ‐1.30; P = 0.0004); eight studies, 619 participants. Moderate‐quality evidence.
Asian people with hypertension: SBP: MD ‐7.75 mmHg (95% CI:‐11,44 to ‐4.07; P < 0.0001) nine studies, 501 participants; DBP: MD ‐2.68 mmHg (95% CI: ‐4.21 to ‐1.15; P = 0.0006). Moderate‐quality evidence.”
Gradual et al. 2020, abstract. - 46
“The funnel plots of all analyses were investigated. For each funnel plot, all studies giving rise to asymmetry were eliminated. The resulting effect was compared with the original analysis. All these analyses showed only marginal effects without significance (not shown).”
Gradual et al. 2020, p. 22. - 47
“Thirty-four trials (3230 participants) were included. Meta-analysis showed that the mean change in urinary sodium (reduced salt vs usual salt) was -75 mmol/24-h (equivalent to a reduction of 4.4 g/d salt), the mean change in BP was -4.18 mmHg (95% CI: -5.18 to -3.18, I2=75%) for systolic and -2.06 mmHg (95% CI: -2.67 to -1.45, I 2=68%) for diastolic BP. Meta-regression showed that age, ethnic group, BPstatus (hypertensive or normotensive) and the change in 24-h urinary sodium were all significantly associated with the fall in systolic BP,explaining 68% of the variance between studies. A 100 mmol reduction in 24 hour urinary sodium (6 g/day salt) was associated with a fall insystolic BP of 5.8 mmHg (95%CI: 2.5 to 9.2, P=0.001) after adjusting for age, ethnic group and BP status. For diastolic BP, age, ethnic group,BP status and the change in 24-h urinary sodium explained 41% of the variance between studies. Meta-analysis by subgroup showed that,in hypertensives, the mean effect was -5.39 mmHg (95% CI: -6.62 to -4.15, I 2=61%) for systolic and -2.82 mmHg (95% CI: -3.54 to -2.11, I2=52%) for diastolic BP. In normotensives, the mean effect was -2.42 mmHg (95% CI: -3.56 to -1.29, I 2=66%) for systolic and -1.00 mmHg(95% CI: -1.85 to -0.15, I 2=66%) for diastolic BP.”
He, Li, and MacGregor 2013, abstract. - 48
“High sodium consumption (>2 grams/day, equivalent to 5 g salt/day) and insufficient potassium intake (less than 3.5 grams/day) contribute to high blood pressure and increase the risk of heart disease and stroke.” Salt Reduction, WHO 2020.
- 49
“Dietary K+ intake induces rapid urinary excretion not only of K+ but also of Na+, a phenomenon called K+-induced natriuresis.”
Gritter, Rotmans, and Hoorn 2018, p. 16. - 50
“All of the included studies intended to compare health outcomes between a group of
participants consuming a normal or usual potassium intake to a group consuming increased
potassium. In one study, the intervention was dietary advice or education, plus a tablet
12 supplement (Berry et al., 2010). In two studies, the intervention was dietary advice or
education (Chalmers et al., 1986; Siani et al., 1991). The remaining 20 studies used a
supplement intervention (Barden et al., 1986; Bulpitt et al., 1985; Forrester & Grell, 1988;
Fotherby & Potter, 1992; Grobbee et al., 1987; Gu et al., 2001; He et al., 2010; Kaplan et al.,
1985; Kawano et al., 1998; MacGregor et al., 1982; Matlou et al., 1986; Obel, 1989; Overlack
et al., 1991; Patki et al., 1990; Richards et al., 1984; Siani et al., 1987; Smith et al., 1985; Trial
Hyp Prv Col, 1992; Valdés et al., 1991; Whelton et al., 1995).”
Effect of increased potassium intake on blood pressure, renal function, blood lipids and other potential adverse effects, WHO 2012, p. 11-12. - 51
“In two studies, the achieved potassium intake in the intervention group was less than that
necessary for a urinary potassium excretion of 70 mmol/day (Forrester & Grell, 1988; Gu et
al., 2001). In five studies, the potassium intake at follow-up in the intervention group was at
least equal to that necessary for a urinary potassium excretion of 70 mmol/day, but less
than 90 mmol/day (Berry et al., 2010; Kaplan et al., 1985; Patki et al., 1990; Siani et al., 1991;
Siani et al., 1987). In 11 studies, the potassium intake at follow-up was at least equal to that
necessary for a urinary potassium excretion of 90 mmol/day, but less than 120 mmol/day for
the intervention group (Barden et al., 1986; Bulpitt et al., 1985; Chalmers et al., 1986;
Fotherby & Potter, 1992; Kawano et al., 1998; MacGregor et al., 1982; Matlou et al., 1986;
Obel, 1989; Smith et al., 1985; Trial Hyp Prv Col, 1992; Whelton et al., 1995). In four studies,
the potassium intake at follow-up was greater than that necessary for a urinary potassium
excretion of 120 mmol/day for the intervention group (Grobbee et al., 1987; He et al., 2010;
Richards et al., 1984; Valdés et al., 1991).
One study reported a baseline sodium intake of less than 2 g/day (Smith et al., 1985). At
baseline, 18 studies reported a sodium intake of 2–4 g/day (Barden et al., 1986; Berry et al.,
2010; Bulpitt et al., 1985; Chalmers et al., 1986; Forrester & Grell, 1988; Fotherby & Potter,
1992; Grobbee et al., 1987; He et al., 2010; Kaplan et al., 1985; Kawano et al., 1998;
MacGregor et al., 1982; Matlou et al., 1986; Obel, 1989; Overlack et al., 1991; Siani et al.,
1987; Trial Hyp Prv Col, 1992; Valdés et al., 1991; Whelton et al., 1995). Four studies
reported a baseline sodium intake of more than 4 g/day (Gu et al., 2001; Kawano et al.,
1998; Patki et al., 1990; Siani et al., 1987).”
Effect of increased potassium intake on blood pressure, renal function, blood lipids and other potential adverse effects, WHO 2012, p. 12. - 52
“The meta-analysis of change in systolic blood pressure is shown in Figure 3.2 and Table 3.53.
Systolic blood pressure was reduced by increased potassium intake relative to normal
potassium intake by 3.06 mmHg (95%CI: 1.42, 4.70). The reduction in systolic blood pressure
in studies specifically targeting individuals with hypertension was 4.68 mmHg (95%CI: 2.40,
6.96), which was statistically significantly greater than the reduction from the trials targeting
individuals with normal blood pressure (0.09 mmHg, 95%CI: –0.77, 0.95). The two studies
reporting on heterogeneous populations including some individuals with hypertension and
some with normal blood pressure showed a reduction in systolic blood pressure of 2.95
mmHg (95%CI: 0.26, 5.65); this reduction was significant, but was not significantly different
from that found in individuals with hypertension (Table 3.54).”
Effect of increased potassium intake on blood pressure, renal function, blood lipids and other potential adverse effects, WHO 2012, p. 14. - 53
“The meta-analysis of change in systolic blood pressure is shown in Figure 3.2 and Table 3.53.
Systolic blood pressure was reduced by increased potassium intake relative to normal
potassium intake by 3.06 mmHg (95%CI: 1.42, 4.70). The reduction in systolic blood pressure
in studies specifically targeting individuals with hypertension was 4.68 mmHg (95%CI: 2.40,
6.96), which was statistically significantly greater than the reduction from the trials targeting
individuals with normal blood pressure (0.09 mmHg, 95%CI: –0.77, 0.95). The two studies
reporting on heterogeneous populations including some individuals with hypertension and
some with normal blood pressure showed a reduction in systolic blood pressure of 2.95
mmHg (95%CI: 0.26, 5.65); this reduction was significant, but was not significantly different
from that found in individuals with hypertension (Table 3.54).”
Effect of increased potassium intake on blood pressure, renal function, blood lipids and other potential adverse effects, WHO 2012, p. 14. - 54
“The two studies reporting on heterogeneous populations including some individuals with hypertension and some with normal blood pressure showed a reduction in systolic blood pressure of 2.95
mmHg (95%CI: 0.26, 5.65); this reduction was significant, but was not significantly different
from that found in individuals with hypertension (Table 3.54).”
Effect of increased potassium intake on blood pressure, renal function, blood lipids and other potential adverse effects, WHO 2012, p. 14. - 55
“The difference in achieved potassium intake between the intervention and control groups
had no significant effect on the reduction in systolic blood pressure (Figure 3.5 and Table
3.53).”
“Baseline sodium intake also had no statistically significant impact on the reduction in systolic
blood pressure with increased potassium intake (Figure 3.11 and Table 3.53).”
Effect of increased potassium intake on blood pressure, renal function, blood lipids and other potential adverse effects, WHO 2012, p. 15 and 16. - 56
Effect of increased potassium intake on blood pressure, renal function, blood lipids and other potential adverse effects, WHO 2012, Annex 3, p. 94-95.
- 57
For example, in the SSaSS cluster-randomized trial, replacement of most regular salt with potassium-enriched salt substitute reduced sodium intake by 8% and increased potassium intake by 57%.
The average reduction in sodium intake was 0.35 grams per day. Table 1 on p. 960 states that baseline urinary sodium excretion was 4.3 grams per day. 0.35 / 4.3 = 0.08.
“By contrast, during the 5-year follow-up period, the mean observed difference in 24-hour urinary sodium excretion between the potassium-enriched salt group and the regular salt group was −0.35 g/d (95% CI,−0.55 to −0.15).” Yin et al. 2023, p. 959
The average increase in potassium intake was 0.80 grams per day. Table 1 on p. 960 states that baseline urinary potassium excretion was 1.4 grams per day. 0.80 / 1.4 = 0.57.
“On average, during the 5-year follow-up period, the mean difference in 24-hour urinary potassium excretion between the intervention and control group was +0.80 g (95%CI, +0.71 to +0.90; Figure 1).” Yin et al. 2023, p. 959 - 58
Yin et al. 2022, table 1.
- 59
“There were 21 trials and 31 949 participants included, with 19 reporting effects on blood pressure and 5 reporting effects on clinical outcomes. Overall reduction of systolic blood pressure (SBP) was −4.61 mm Hg (95% CI −6.07 to −3.14) and of diastolic blood pressure (DBP) was −1.61 mm Hg (95% CI −2.42 to −0.79). Reductions in blood pressure appeared to be consistent across geographical regions and population subgroups defined by age, sex, history of hypertension, body mass index, baseline blood pressure, baseline 24-hour urinary sodium and baseline 24-hour urinary potassium (all p homogeneity >0.05). Metaregression showed that each 10% lower proportion of sodium choloride in the salt substitute was 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. There were clear protective effects of salt substitute on total mortality (risk ratio (RR) 0.89, 95% CI 0.85 to 0.94), cardiovascular mortality (RR 0.87, 95% CI 0. 81 to 0.94) and cardiovascular events (RR 0.89, 95% CI 0.85 to 0.94).”
Yin et al. 2022, abstract. - 60
“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).” Yin et al. 2022, p. 5.
- 61
We used the following search terms in the Cochrane Central Register of Controlled Trials (CENTRAL) on July 10, 2020: (salt OR sodium OR NaCl) AND (potassium OR KCl) AND (cardiovascular OR "blood pressure" OR hypertens*) AND community. This returned 68 trials, all of which we considered for relevance. We supplemented this with less structured searches in PubMed that focused on “similar articles” to those we identified via CENTRAL.
We used the following search terms in the Cochrane Central Register of Controlled Trials (CENTRAL) on May 31, 2024: (salt OR sodium OR NaCl) AND (potassium OR KCl) AND (cardiovascular OR "blood pressure" OR hypertens*) AND (death* OR MACE OR coronary OR myocardial OR stroke OR events). The search was restricted from 2020 to June 2024 and to trials. This returned 228 trials, all of which we considered for relevance. We supplemented this with less structured searches in PubMed that focused on “similar articles” to those we identified via CENTRAL.
We also reviewed the trials included in the meta-analysis Yin et al. 2022. - 62
These are the three trials with few deaths and/or cardiovascular events. They will not be discussed further in this report.
- China Salt Substitute Study Collaborative Group 2007. According to figure 4 of the meta-analysis Yin et al. 2022, only four deaths and 5-8 major adverse cardiovascular events occurred in each arm.
- Zhao et al. 2014. According to figure 4 of the meta-analysis Yin et al. 2022, only 1-2 deaths occurred in each arm. Cardiovascular events/deaths are not reported.
- Sun et al. 2021. According to figure 4 of the meta-analysis Yin et al. 2022, 22-28 deaths and 9-16 cardiovascular deaths occurred in each arm. This yielded wide confidence intervals for the impact of the intervention on all-cause mortality and cardiovascular mortality.
- 63
“The study is a large-scale, open, cluster-randomized controlled trial done in 600 villages across 5 provinces in China… Follow-up is scheduled every 6 months for 5 years, and all potential endpoints are reviewed by a masked adjudication committee.” Neal et al. 2017, abstract.
- 64
“About 35 individuals at elevated risk of stroke were recruited in each village and will be followed up for the duration of the trial.” Neal et al. 2017, p. 110.
- 65
“A total of 20,995 persons were enrolled in the trial.” Neal et al. 2021, abstract
- 66
“Elevated risk was defined on the basis of a history of stroke (regardless of type or knowledge of etiology) and/or age 60 years or greater with uncon-trolled high blood pressure (systolic blood pressure≥ 140 mm Hg at visit if on blood pressure–lowering medication or systolic blood pressure ≥ 160 mm Hg if not on blood pressure–lowering medication).” Neal et al. 2017, p. 110.
- 67
“Among the participants at baseline, the mean age was 65.4 years, 49.5% were female, 72.6% had a history of stroke, and 88.4% reported having received a diagnosis of hypertension. The mean blood pressure was 154.0/89.2 mm Hg, and 79.3% of the participants were using at least one blood-pressure–lowering medication — 41.8% were using a calcium antagonist, 22.8% an angiotensin-converting–enzyme inhibitor or angiotensin-receptor blocker, 11.5% a diuretic, 5.7% a beta-blocker, and 0.9% an alpha-blocker.”
Neal et al. 2021, p. 1072. - 68
“However, in northern rural China, large quantities of dietary sodium are consumed, with the majority deriving from salt added to food prepared at home. 14-17 This practice provides a particular opportunity to make significant changes to dietary sodium consumption by providing salt substitute.” Neal et al. 2017, p. 110.
- 69
“The sodium reduction intervention is based on the provision of reduced-sodium salt substitute as a replacement for regular salt among the households of the 35 high-risk individuals in each intervention village.The salt substitute is manufactured in accordance with the national manufacturing standard (GB 2019-2005:which requires a composition of sodium chloride[NaCl]: 70.00% ± 10.00% and potassium chloride [KCl]:30.00% ± 10.00%) and is purchased from a local provider of salt substitute in each county… Sufficient salt substitute is provided to cover the household cooking and food preservation require-ments at an average of about 20 g per person per day to a maximum of 20 kg per year for a household with an average size of 3 persons. The salt substitute is provided free of charge…” Neal et al. 2017, p. 110.
- 70
“The salt substitute is provided free of charge and is accompanied by advice to use the salt substitute instead of salt for all cooking, seasoning,and food preservation purposes. In addition, participants are advised to try and use the salt substitute more sparingly, and not more frequently, than they previously used salt.”
- 71
Table 1 on p. 960 states that baseline urinary sodium excretion was 4.3 grams per day. 0.35 / 4.3 = 0.08.
“By contrast, during the 5-year follow-up period, the mean observed difference in 24-hour urinary sodium excretion between the potassium-enriched salt group and the regular salt group was −0.35 g/d (95% CI,−0.55 to −0.15).” Yin et al. 2023, p. 959 - 72
Table 1 on p. 960 states that baseline urinary potassium excretion was 1.4 grams per day. 0.80 / 1.4 = 0.57.
“On average, during the 5-year follow-up period, the mean difference in 24-hour urinary potassium excretion between the intervention and control group was +0.80 g (95%CI, +0.71 to +0.90; Figure 1).” Yin et al. 2023, p. 959 - 73
“This compares to an observed difference of −0.35 g/d (95% CI, −0.55 to −0.15) and suggests that 72% of baseline regular salt intake was replaced by potassium-enriched salt.” Yin et al. 2023, abstract
- 74
“The mean difference in systolic blood pressure was −3.34 mm Hg (95% CI, −4.51 to −2.18) (Fig. 2).” Neal et al. 2021, p. 1072
- 75
“The mean duration of follow-up was 4.74 years. The rate of stroke was lower with the salt substitute than with regular salt (29.14 events vs. 33.65 events per 1000 person-years; rate ratio, 0.86; 95% confidence interval [CI], 0.77 to 0.96; P=0.006)...” Neal et al. 2021, abstract
- 76
“As compared with regular salt, the salt substitute was also shown to protect against the secondary outcomes of major adverse cardiovascular events (49.1 events vs. 56.3 events per 1000 person-years; rate ratio, 0.87; 95% CI, 0.80 to 0.94; P<0.001) and death from any cause (39.3 events vs. 44.6 events per 1000 person-years; rate ratio, 0.88; 95% CI, 0.82 to 0.95; P<0.001) (Table 1 and Figure 3). The salt substitute was also shown to be beneficial with respect to death from vascular causes (22.9 events vs. 26.3 events per 1000 person-years; rate ratio, 0.87; 95% CI, 0.79 to 0.96) and nonfatal acute coronary syndrome (3.8 events vs. 5.1 events per 1000 person-years; rate ratio, 0.70; 95% CI, 0.52 to 0.93) but not with respect to nonfatal stroke (22.4 events vs. 24.9 events per 1000 person-years; rate ratio, 0.90; 95% CI, 0.80 to 1.01) (Table 1).” Neal et al. 2021, p. 1072
“Adjustment for multiplicity in the analyses of the primary and secondary outcomes was performed post hoc with the use of the Benjamini–Hochberg method.” Neal et al. 2021, p. 1070 - 77
“In addition, concern about the theoretical risks of hyperkalemia and associated sudden death from the use of salt substitutes in patients with serious kidney disease has adversely influenced perceptions of clinicians and the general population.” Neal et al. 2021, p. 1068
- 78
“The rate of serious adverse events attributed to hyperkalemia was not significantly higher with the salt substitute than with regular salt (3.35 events vs. 3.30 events per 1000 person-years; rate ratio, 1.04; 95% CI, 0.80 to 1.37; P=0.76).” Neal et al. 2021, p. 1068
- 79
“Persons were excluded if they or someone living in their household had a potential contraindication to the salt substitute used in the trial; contraindications included use of a potassium-sparing diuretic, use of a potassium supplement, or known serious kidney disease.”
- 80
“The sample of 21,000 participants in 600 clusters (300 intervention and 300 control, 35 participants in each cluster) followed up for a mean of 5 years is projected to provide more than 90% statistical power (with 2-sidedα = .05) to detect a 13% or greater relative risk reduction for stroke in the intervention villages compared with control villages” Neal et al. 2017, p. 112
- 81
The ClinicalTrials.gov registry page is here. It was created in March of 2014, and study recruitment began in April 2014.
“Participants were enrolled in the current trial from April 2014 through January 2015.” Neal et al. 2021, p. 1068 - 82
“The sample of 21,000 participants in 600 clusters (300 intervention and 300 control, 35 participants in each cluster) followed up for a mean of 5 years is projected to provide more than 90% statistical power (with 2-sidedα = .05) to detect a 13% or greater relative risk reduction for stroke in the intervention villages compared with control villages” Neal et al. 2017, p. 112
- 83
“Primary Outcome Measures: Stroke [ Time Frame: 5 years ]”. China Salt Substitute and Stroke Study (SSaSS), ClinicalTrials.gov
- 84
“The primary outcome was stroke, the secondary outcomes were major adverse cardiovascular events and death from any cause, and the safety outcome was clinical hyperkalemia.” Neal et al. 2021, abstract.
- 85
“Primary analyses were performed according to the intention-to-treat principle.” Neal et al. 2021, p. 1070
- 86
“The mean duration of follow-up was 4.74 years. The rate of stroke was lower with the salt substitute than with regular salt (29.14 events vs. 33.65 events per 1000 person-years; rate ratio, 0.86; 95% confidence interval [CI], 0.77 to 0.96; P=0.006)...” Neal et al. 2021, abstract
“As compared with regular salt, the salt substitute was also shown to protect against the secondary outcomes of major adverse cardiovascular events (49.1 events vs. 56.3 events per 1000 person-years; rate ratio, 0.87; 95% CI, 0.80 to 0.94; P<0.001) and death from any cause (39.3 events vs. 44.6 events per 1000 person-years; rate ratio, 0.88; 95% CI, 0.82 to 0.95; P<0.001) (Table 1 and Figure 3). The salt substitute was also shown to be beneficial with respect to death from vascular causes (22.9 events vs. 26.3 events per 1000 person-years; rate ratio, 0.87; 95% CI, 0.79 to 0.96) and nonfatal acute coronary syndrome (3.8 events vs. 5.1 events per 1000 person-years; rate ratio, 0.70; 95% CI, 0.52 to 0.93) but not with respect to nonfatal stroke (22.4 events vs. 24.9 events per 1000 person-years; rate ratio, 0.90; 95% CI, 0.80 to 1.01) (Table 1).” Neal et al. 2021, p. 1072 - 87
“Adjustment for multiplicity in the analyses of the primary and secondary outcomes was performed post hoc with the use of the Benjamini–Hochberg method.” Neal et al. 2021, p. 1070
- 88
“The mean duration of follow-up was 4.74 years. The rate of stroke was lower with the salt substitute than with regular salt (29.14 events vs. 33.65 events per 1000 person-years; rate ratio, 0.86; 95% confidence interval [CI], 0.77 to 0.96; P=0.006)...” Neal et al. 2021, abstract
- 89
“Here, we conducted a clinical trial in which 48 residential elderly care facilities in China (1,612 participants including 1,230 men and 382 women, 55 years or older) were cluster-randomized using a 2 × 2 factorial design to provision of salt substitute (62.5% NaCl and 25% KCl) versus usual salt and to a progressively restricted versus usual supply of salt or salt substitute for 2 years.”
Yuan et al. 2023, abstract.
“The trial commenced in September 2017 and was carried out in 48 residential elderly care facilities, defined as the clusters in the present study, located in four regions in northern China: Xi’an city in Shaanxi province, Hohhot city in Inner Mongolia Autonomous Region, Changzhi County and Yangcheng County in Shanxi province. All regions were selected for their high sodium intake, high prevalence of HTN and history of research collaboration.”
Yuan et al. 2023, p. 982. - 90
Yuan et al. 2023, table 1, p. 976.
- 91
“Here, we conducted a clinical trial in which 48 residential elderly care facilities in China (1,612 participants including 1,230 men and 382 women, 55 years or older) were cluster-randomized using a 2 × 2 factorial design to provision of salt substitute (62.5% NaCl and 25% KCl) versus usual salt and to a progressively restricted versus usual supply of salt or salt substitute for 2 years.”
Yuan et al. 2023, abstract. - 92
Supplementary table 7 reports that 24-hour urinary sodium excretion declined by 45 mmol/L in the salt restriction group and declined by 49 mmol/L in the control group, implying that the intervention was ineffective.
Yuan et al. 2023, supplementary materials table S7, p. 10.
Salt supply records also imply a modest impact on salt intake.
“Supplementary analysis of the study salt supply records showed the average consumption of study salt per person per day was 11.0 ± 4.7 g (equivalent to 36.8 mmol potassium, 117.0 mmol sodium) in participants using salt substitute and 11.6 ± 3.3 g (equivalent to 0.0 mmol potassium, 199.6 mmol sodium) with usual salt (P for total weight = 0.57, P for composition < 0.001), 10.5 ± 3.7 g (equivalent to 17.3 mmol potassium, 145.9 mmol sodium) in participants with restricted supply and 12.2 ± 4.4 g (equivalent to 19.5 mmol potassium, 168.9 mmol sodium) with usual supply (P for total weight = 0.16, P for composition = 0.249).”
Yuan et al. 2023, p. 975-6. - 93
We suspect the “amino acids” probably include glutamate, which has a meaty “umami” flavor. “Facilities assigned to the salt substitute group received salt substitute to replace usual salt. The salt substitute, manufactured by the China Salt General Company at Yulin in China, comprised 62.5% (mg/mg) sodium chloride, 25% (mg/mg) potassium chloride, 12.5% (mg/mg) dried food ingredient flavorings (mushroom, lemon, seaweed, hawthorn, wild jujube) and traces of amino acids. Facilities in the control group received usual salt, 100% sodium chloride, from the same company.”
Yuan et al. 2023, p. 982. - 94
The molecular weight of potassium is 39.1 u, and of sodium is 23.0 u. 39.1 x 0.0128 = 0.5. 0.0097 x 23.0 = 0.2.
“Comparing participants receiving salt substitute versus usual salt, 24-h urinary potassium excretion was increased by 12.8 mmol (95% CI 5.7–19.8) but there was no detectable effect on 24-h urinary sodium excretion (–9.7 mmol; 95% CI –28.4 to 9.1).”
Yuan et al. 2023, p. 975. - 95
See discussion of SSaSS results here.
- 96
“The primary efficacy outcome was SBP, which was taken three times using an OMRON HEM-7136 device following American Heart Association guidelines”
Yuan et al. 2023, p. 982.
This is consistent with the trial’s preregistration page. - 97
“Salt substitute compared with usual salt lowered mean systolic blood pressure (SBP), the primary outcome, by –7.1 mmHg (95% CI –10.5 to –3.8; Bonferroni corrected P < 0.001) (Extended Data Table 2 and Fig. 2). The estimates from the prespecified sensitivity analyses were –6.9 mmHg (95% CI –10.3 to –3.5) in the per-protocol analysis, –7.0 mmHg (95% CI –10.4 to –3.6) with exclusion of facilities from Xi’an and –6.6 mmHg (95% CI –10.3 to –2.8) with imputation for missing data (Extended Data Table 2). All results were statistically significant after adjustment for multiple comparisons using the Bonferroni method.”
Yuan et al. 2023, p. 974. - 98
See the trial’s preregistration page.
- 99
Yuan et al. 2023, table 2, p. 978.
- 100
The legend for table 2, which reports cardiovascular and mortality outcomes, says “The P value was two-sided and was not adjusted for multiple comparison.”
Yuan et al. 2023, table 2, p. 978. - 101
The preregistration page is here.
- 102
The preregistration page describes a power calculation based on reductions in systolic blood pressure, and the primary outcome is listed as systolic blood pressure.
“On the assumptions of a 20% dropout rate, an intraclass correlation of 0.02, the number of clusters of 48 and at least 20 elderly people in each nursing home, and an α value of 0.05, the study to detect a mean 3.0 mmHg reduction in systolic blood pressure (SD, 18 mm Hg) between the intervention groups would have a power of 0.81. To detect a 4.0 mmHg reduction in systolic blood pressure (SD, 18 mm Hg), the power of the same sample size would be 0.96.” - 103
Yuan et al. 2023, table 2, p. 978.
- 104
The legend for table 2, which reports cardiovascular and mortality outcomes, says “The P value was two-sided and was not adjusted for multiple comparison.”
Yuan et al. 2023, table 2, p. 978. - 105
“Among the participants at baseline, the mean age was 65.4 years, 49.5% were female, 72.6% had a history of stroke, and 88.4% reported having received a diagnosis of hypertension. The mean blood pressure was 154.0/89.2 mm Hg, and 79.3% of the participants were using at least one blood-pressure–lowering medication — 41.8% were using a calcium antagonist, 22.8% an angiotensin-converting–enzyme inhibitor or angiotensin-receptor blocker, 11.5% a diuretic, 5.7% a beta-blocker, and 0.9% an alpha-blocker.”
Neal et al. 2021, p. 1072. - 106
“The trial commenced in September 2017 and was carried out in 48 residential elderly care facilities, defined as the clusters in the present study, located in four regions in northern China: Xi’an city in Shaanxi province, Hohhot city in Inner Mongolia Autonomous Region, Changzhi County and Yangcheng County in Shanxi province. All regions were selected for their high sodium intake, high prevalence of HTN and history of research collaboration.”
Yuan et al. 2023, p. 982. - 107
Yuan et al. 2023, table 1, p. 976.
- 108
“Among the participants at baseline, the mean age was 65.4 years, 49.5% were female, 72.6% had a history of stroke, and 88.4% reported having received a diagnosis of hypertension. The mean blood pressure was 154.0/89.2 mm Hg, and 79.3% of the participants were using at least one blood-pressure–lowering medication — 41.8% were using a calcium antagonist, 22.8% an angiotensin-converting–enzyme inhibitor or angiotensin-receptor blocker, 11.5% a diuretic, 5.7% a beta-blocker, and 0.9% an alpha-blocker.”
Neal et al. 2021, p. 1072.
Yuan et al. 2023, table 1, p. 976. - 109
These ages, and older, are when cardiovascular events are the most common. Figures are based on national survey data.
Zhang et al. 2023, data supplement, supplementary Table 4, p. 23. - 110
“As compared with regular salt, the salt substitute was also shown to protect against the secondary outcomes of major adverse cardiovascular events (49.1 events vs. 56.3 events per 1000 person-years; rate ratio, 0.87; 95% CI, 0.80 to 0.94; P<0.001) and death from any cause (39.3 events vs. 44.6 events per 1000 person-years; rate ratio, 0.88; 95% CI, 0.82 to 0.95; P<0.001) (Table 1 and Figure 3). The salt substitute was also shown to be beneficial with respect to death from vascular causes (22.9 events vs. 26.3 events per 1000 person-years; rate ratio, 0.87; 95% CI, 0.79 to 0.96) and nonfatal acute coronary syndrome (3.8 events vs. 5.1 events per 1000 person-years; rate ratio, 0.70; 95% CI, 0.52 to 0.93) but not with respect to nonfatal stroke (22.4 events vs. 24.9 events per 1000 person-years; rate ratio, 0.90; 95% CI, 0.80 to 1.01) (Table 1).” Neal et al. 2021, p. 1072
“Adjustment for multiplicity in the analyses of the primary and secondary outcomes was performed post hoc with the use of the Benjamini–Hochberg method.” Neal et al. 2021, p. 1070 - 111
- “In the present study, we randomized 5 kitchens of a veteran’s retired home in northern Taiwan into experimental and control groups (Figure 1). When a veteran registered into the retired home, he was assigned into one of 11 squads; each squad was composed of 200 people, and 2 squads shared one kitchen. Of 5 kitchens (corresponding to 10 squads), 2 kitchens (corresponding to 4 squads) were randomly assigned to use potassium-enriched salt. The other 3 kitchens (6 squads) were randomly assigned to use regular salt.” Chang et al. 2006, p. 1290
- “Five kitchens of a veteran retirement home were randomized into 2 groups (experimental or control) and veterans assigned to those kitchens were given either potassium-enriched salt (experimental group) or regular salt (control group) for ≈31 mo.” Chang et al. 2006, abstract
- 112
"The potassium-enriched salt was composed of 49% sodium chloride, 49% potassium chloride, and 2% other additives, whereas regular salt was composed of 99.6% sodium chloride and 0.4% other additives. The potassium-enriched salt replaced regular salt in the selected kitchens in a gradual manner.” Chang et al. 2006, p. 1291
- 113
Chang et al. 2006, p. 1292, table 1
- 114
“The average amount of sodium consumed during the trial was estimated to be ≈3.8 g per person per day for the experimental group and 5.2 g per person per day for the control group.” Chang et al. 2006, p. 1294
- 115
“A 17% reduction was observed for the urinary sodium-to-creatinine ratio in the experimental group, but a 76% increase was observed in the potassium-to-creatinine ratio.” Chang et al. 2006, p. 1294
- 116
“A significant reduction in CVD mortality (age-adjusted hazard ratio: 0.59; 95%CI: 0.37, 0.95) was observed in the experimental group.” Chang et al. 2006, abstract
- 117
The following passage is from a World Bank review paper that discusses a similar case. “Despite these promising results, much of the potential impact of this program remains unknown because of challenges in the initial evaluation design that were not corrected for in subsequent studies. From a pair of large villages and a pair of small villages, the design randomly selected one village from each pair to receive the intervention, resulting in one large and one small village being in the treatment group with the other large and small village being assigned to the control. Although the sample size is large (more than 1,000 children), there are too few units of randomization (the four villages in this case) to rely on the Law of Large Numbers to generate treatment and comparison groups that are statistically equivalent. The randomization process allowed only four possible permutations of groupings.” Tanner, Candland, and Odden 2015 (working paper)
- 118
The most recent Cochrane meta-analysis of salt reduction RCTs, Adler et al. 2014, pools four salt reduction trials that report cardiovascular events, reporting a relative risk of 0.76 (95% CI, 0.57 to 1.01). The finding is not quite statistically significant, and it relies heavily on Chang 2006 (49% of study weight), which is confounded by potassium supplementation. In addition, Chang et al. 2006 is intended to be a cluster-randomized trial but it only has five clusters, which is not enough to satisfy the statistical requirements of the cluster-randomized trial design. For these reasons, we do not believe the result reported in Adler et al. 2014 is very informative.
Adler et al. 2014, analysis 1.1, p. 39
“Five kitchens of a veteran retirement home were randomized into 2 groups (experimental or control) and veterans assigned to those kitchens were given either potassium-enriched salt (experimental group) or regular salt (control group) for approximately 31 mo.” Chang et al. 2006, abstract.
The following passage is from a World Bank review paper that discusses a similar case of insufficient number of clusters in a cluster-randomized trial. “Despite these promising results, much of the potential impact of this program remains unknown because of challenges in the initial evaluation design that were not corrected for in subsequent studies. From a pair of large villages and a pair of small villages, the design randomly selected one village from each pair to receive the intervention, resulting in one large and one small village being in the treatment group with the other large and small village being assigned to the control. Although the sample size is large (more than 1,000 children), there are too few units of randomization (the four villages in this case) to rely on the Law of Large Numbers to generate treatment and comparison groups that are statistically equivalent. The randomization process allowed only four possible permutations of groupings.” Tanner, Candland, and Odden 2015 (working paper) - 119
Foti et al. 2020 model RCT designs that could be used to directly test the hypothesis that dietary sodium reduction reduces the risk of cardiovascular events and conclude that they would be “prohibitively expensive”. “We designed and assessed the feasibility of potential individual- and cluster-randomized trials of sodium reduction on cardiovascular outcomes. Based on our assumptions, a trial using any of the designs considered would require tens of thousands of participants and cost hundreds of millions of dollars, which is prohibitively expensive.” Foti et al. 2020, abstract.
- 120
“The benefit of CVD deaths from sodium reduction is considerable with 8800 (6400–13 600), 6700 (4900–11 600), and 8500 (6000–10 800) averted, respectively, if sodium intake was reduced from the 2011 baseline to 3500 mg/day, 4000 mg/day, or reduced by 30%.” Zhang et al. 2018, abstract.
“A study using SMASH baseline data estimated that 6700 deaths from CVD in Shandong could be prevented annually if urinary sodium excretion was reduced to 4000 mg per day. This target was nearly achieved in 2016 (4013 mg per day).” Xu et al. 2020, p. 884. - 121
See our discussion of secular trends here.
See our calculations in the CEA here. - 122
See our calculations here.
- 123
See the adjustment and underlying calculations here.
- 124
See for example this review paper representing the views of 24 sodium reduction skeptics, O’Donnell et al. 2020.
- 125
See the review here.
- 126
Ettehad et al. 2016 is a meta-analysis of RCTs of blood pressure-lowering drug interventions. “Every 10 mm Hg reduction in systolic blood pressure significantly reduced the risk of major cardiovascular disease events (relative risk [RR] 0·80, 95% CI 0·77–0·83), coronary heart disease (0·83, 0·78–0·88), stroke (0·73, 0·68–0·77), and heart failure (0·72, 0·67–0·78), which, in the populations studied, led to a significant 13% reduction in all-cause mortality (0·87, 0·84–0·91).” Ettehad et al. 2016, abstract.
- 127
See our review of the evidence here.
“Sodium reduction from an average high usual sodium intake level (201 mmol/day) to an average level of 66 mmol/day, which is below the recommended upper level of 100 mmol/day (5.8 g salt), resulted in a decrease in SBP/DBP of 1/0 mmHg in white participants with normotension and a decrease in SBP/DBP of 5.5/2.9 mmHg in white participants with hypertension. A few studies showed that these effects in black and Asian populations were greater.” Graudal et al 2017, abstract. - 128
See our discussion of the observational evidence linking sodium intake and cardiovascular risk here.
- 129
See our discussion of the observational evidence linking sodium intake and all-cause mortality here.
- 130
See our reasoning here.
- 131
See our discussion of this evidence here.
- 132
We estimate that 77% of deaths averted are attributable to salt substitution. Because we use our mortality reduction estimate as an approximation of morbidity reduction, 77% of our morbidity reduction estimate is also attributable to salt substitution, and therefore 77% of overall benefits.
- 133
See our calculations here.
- 134
“However, in northern rural China, large quantities of dietary sodium are consumed, with the majority deriving from salt added to food prepared at home. This practice provides a particular opportunity to make significant changes to dietary sodium consumption by providing salt substitute.” Neal et al. 2017, p. 110.
- 135
“Each additional 1000 mg of daily sodium excretion was associated with an 18% increase in cardiovascular risk (adjusted hazard ratio, 1.18; 95% CI, 1.08 to 1.29) (Fig. 1). The spline analysis with pooled data further supported a linear association over the range of sodium excretion within this population (P<0.001 for linearity) (Fig. 2A)... Each additional 1000 mg of daily potassium excretion was associated with an 18% lower cardiovascular risk (hazard ratio, 0.82; 95% CI, 0.72 to 0.94) (Fig. 1). The spline plot also showed a linear trend (Fig. 2B).”
Ma et al. 2022, p. 255. - 136
See figures 2A and 2B. At more extreme values, the 95% confidence range (shaded in blue) widens, encompassing the possibility of nonlinear trends.
Ma et al. 2022, figure 2A and 2B, p. 260. - 137
See data and calculations here.
- 138
185/100,000 = 0.19%. 322/100,000 = 0.32%.
- 139
See our calculations here.
- 140
See our calculations here.
- 141
See our discussion of this here.
- 142
“Every 10 mm Hg reduction in systolic blood pressure significantly reduced the risk of major cardiovascular disease events (relative risk [RR] 0·80, 95% CI 0·77–0·83), coronary heart disease (0·83, 0·78–0·88), stroke (0·73, 0·68–0·77), and heart failure (0·72, 0·67–0·78), which, in the populations studied, led to a significant 13% reduction in all-cause mortality (0·87, 0·84–0·91).” Ettehad et al. 2016, abstract.
- 143
“At ages 40–69 years, each difference of 20 mm Hg usual SBP (or, approximately equivalently, 10 mm Hg usual DBP) is associated with more than a twofold difference in the stroke death rate, and with twofold differences in the death rates from IHD and from other vascular causes. All of these proportional differences in vascular mortality are about half as extreme at ages 80–89 years as at ages 40–49 years, but the annual absolute differences in risk are greater in old age.” Prospective Studies Collaboration 2002, abstract.
- 144
See our calculations here.
- 145
- 146
In the table in Mente et al. 2016, the multivariate-adjusted hazard ratio for <3 g/d relative to 4.00-4.99 g/d is 1.39 (95% CI, 1.23 to 1.58), and the multivariate-adjusted hazard ratio for >7 g/d relative to 4.00-4.99 g/d is 1.39 (95% CI, 1.22 to 1.59).
On p. 15 of the supplementary appendix in Ma et al. 2022, the extremes of sodium intake correspond to about 1.0 g/d and 6.7 g/d. The corresponding multivariate-adjusted hazard ratios are about 1.3 and 1.4. These are estimates based on eyeballing figure S4. In this analysis, differences in all-cause mortality were not statistically significant. - 147
“Concerns have been raised that potassium enriched salt substitutes might increase the risk of clinically important hyperkalaemia in individuals with advanced chronic kidney disease, increasing the risk of sudden cardiac death.” Marklund et al. 2020, p. 2.
- 148
"In the conservative scenario, a nationwide salt substitution intervention was estimated to result in ≈214 000 (95% uncertainty interval, 92 764–353 054) averted deaths from blood pressure reduction in the total population and ≈52 000 (22 961–80 211) in 28 million individuals with advanced chronic kidney disease, while ≈22 000 (15 221–31 840) hyperkalemia-deaths might be caused by the intervention. The corresponding estimates for the optimistic scenario were ≈351 000 (130 470–546 255), ≈66 000 (24 925–105 851), and ≈9000 (4251–14 599). Net benefits were consistent across sensitivity analyses."
Marklund et al. 2022, abstract.
22,000 / 214,000 = 0.10
9,000 / 351,000 = 0.026
"Nationwide implementation of potassium enriched salt substitution could prevent about 461 000 (95% uncertainty interval 196 339 to 704 438) deaths annually from cardiovascular disease, corresponding to 11.0% (4.7% to 16.8%) of annual deaths from cardiovascular disease in China; 743 000 (305 803 to 1 273 098) non-fatal cardiovascular events annually; and 7.9 (3.3 to 12.9) million disability adjusted life years related to cardiovascular disease annually. The intervention could potentially produce an estimated 11 000 (6422 to 16 562) additional deaths related to hyperkalaemia in individuals with chronic kidney disease. The net effect would be about 450 000 (183 699 to 697 084) fewer deaths annually from cardiovascular disease in the overall population and 21 000 (1928 to 42 926) fewer deaths in individuals with chronic kidney disease."
Marklund et al. 2020, abstract.
11,000 / 461,000 = 0.024 - 149
“The net effect would be about 450 000 (183 699 to 697 084) fewer deaths annually from cardiovascular disease in the overall population and 21 000 (1928 to 42 926) fewer deaths in individuals with chronic kidney disease."
Marklund et al. 2020, abstract. - 150
“There is insufficient evidence regarding the effects of potassium-enriched salt substitutes on the occurrence of hyperkalemia. There is a need for additional empirical research on the effect of increasing dietary potassium and potassium-enriched salt substitutes on serum potassium levels and the risk of hyperkalemia, as well as for robust estimation of the population-wide impact of replacing sodium chloride with potassium-enriched salt substitutes.”
Greer et al. 2019, abstract. - 151
“Among salt manufacturers producing both low-sodium salts and regular salts (N=38), the price of low-sodium salts was 1.7 times the price of the regular salts.” Yin et al. 2021, p. 6.
- 152
Based on unpublished communications from Resolve to Save Lives, we estimate a marginal per-person cost of $1 per year.
- 153
See our calculations here.
- 154
In the table in Mente et al. 2016, the multivariate-adjusted hazard ratio for <3 g/d relative to 4.00-4.99 g/d is 1.39 (95% CI, 1.23 to 1.58), and the multivariate-adjusted hazard ratio for >7 g/d relative to 4.00-4.99 g/d is 1.39 (95% CI, 1.22 to 1.59).
On p. 15 of the supplementary appendix in Ma et al. 2022, the extremes of sodium intake correspond to about 1.0 g/d and 6.7 g/d. The corresponding multivariate-adjusted hazard ratios are about 1.3 and 1.4. These are estimates based on eyeballing figure S4. - 155
“ Sodium excretion was not associated with death from any cause (hazard ratio per 1000-mg increase, 1.02; 95% CI, 0.95 to 1.10), but both higher potassium excretion and a lower sodium-to-potassium ratio were associated with a lower risk of death from any cause (Table S5 and Fig. S3).” Ma et al. 2022, p. 9-10.
Ma et al. 2022, figure S4, supplementary appendix p. 15 - 156
Figure S4 of Ma et al. presents a graph of sodium intake vs. all-cause mortality. Taking the graph at face value and putting aside the lack of statistical significance, relative risk does not begin to increase until below about 3 grams per day of sodium. We estimate that average sodium intake in China and India is currently about 4.3-4.4 grams per day, the SMASH trial reduced sodium intake by about 1 gram, and the current intervention would reduce sodium intake by a bit less than that. This would imply a reduction of sodium intake to about 3.5 grams per day, above the point where the relative risk of mortality begins to meaningfully depart from 1.0.
Ma et al. 2022, figure S4, supplementary appendix p. 15
See our calculations to estimate average sodium intake in China and India here.
See our calculation to estimate the reduction of sodium intake in the SMASH trial here. - 157
“During sodium reduction renin increased 1.56 ng/mL/hour (95%CI:1.39, 1.73) in 2904 participants (82 trials); aldosterone increased 104 pg/ mL (95%CI:88.4,119.7) in 2506 participants (66 trials); noradrenalin increased 62.3 pg/mL: (95%CI: 41.9, 82.8) in 878 participants (35 trials); adrenalin increased 7.55 pg/mL (95%CI: 0.85, 14.26) in 331 participants (15 trials); cholesterol increased 5.19 mg/dL (95%CI:2.1, 8.3) in 917 participants (27 trials); triglyceride increased 7.10 mg/dL (95%CI: 3.1,11.1) in 712 participants (20 trials); LDL tended to increase 2.46 mg/ dl (95%CI: -1, 5.9) in 696 participants (18 trials); HDL was unchanged -0.3 mg/dl (95%CI: -1.66,1.05) in 738 participants (20 trials) (All high- quality evidence except the evidence for adrenalin).” Graudal et al. 2020, abstract.
He, Li, and MacGregor 2013. See our discussion of this meta-analysis here. - 158
“The very small effect of sodium reduction on BP in healthy individuals shown in the present review and other reviews including the WHO review, the risk of significant side effects shown in this review [referring to blood lipid and hormone changes], and the possibility that an intervention to reduce BP may not reduce mortality (Wiysonge 2012), and even may increase mortality in some population groups with a normal BP (Brunström 2016) indicate that the BP-effect is not sufficient as a basis for recommendations in the general population, but should be verified in studies directly relating sodium intake with morbidity and mortality.” Graudal et al. 2020, p. 31.
“With salt reduction, there is a small physiological increase in plasma renin activity, aldosterone and noradrenaline. There is no significant change in lipid levels. These results provide further strong support for a reduction in population salt intake. This will likely lower population BP and, thereby, reduce cardiovascular disease.” He, Le, and MacGregor 2013, abstract. - 159
These are the harms that are mentioned in a short StatPearls review on the physiology of the renin-angiotensin system. StatPearls is an online platform that provides healthcare education and resources.
“Overactivation of the [renin-angiotensin-aldosterone system] has been implicated in the pathogenesis of various cardiovascular and renal diseases.[60][61][62] [renin-angiotensin-aldosterone system] is also implicated in the pathogenesis of primary hypertension.[63][64] This has been proven by using medications that block the [renin-angiotensin-aldosterone system] at different steps. Overactivation of the [renin-angiotensin-aldosterone system] is also implicated in the development of secondary hypertension due to primary hyperaldosteronism.”
Physiology, Renin Angiotensin System, StatPearls 2023. - 160
“The average standardised difference in mean [plasma aldosterone] was similar for all classes of diuretic: thiazide/thiazide-like 0.299 (95% confidence interval [CI] 0.150, 0.447), loop 0.927 (0.37, 1.49), MRA/potassium-sparing 0.265 (0.173, 0.357) and combination 0.466 (0.137, 0.796), Q = 6.33, P = .097… In RCTs of diuretic therapy in hypertension, there is an increase in [plasma aldosterone] with all classes of diuretic and no significant between-class heterogeneity.”
McNally et al. 2021, abstract
“Under acute and chronic conditions, diuretics induce an increase in plasma renin activity (PRA)8 but whether diuretics also increases PA has been debated.”
McNally et al. 2021, abstract - 161
“In the network meta-analysis, compared with placebo, angiotensin-converting enzyme inhibitors, dihydropyridine calcium channel blockers, and thiazide diuretics were reported to be similarly effective in reducing overall cardiovascular events (25%), cardiovascular death (20%), and stroke (35%); angiotensin-converting enzyme inhibitors were reported to be the most effective in reducing the risk of myocardial infarction (28%); and diuretics were reported to be the most effective in reducing revascularization (33%).”
Wei et al. 2020, abstract. - 162
“Diabetes and high blood pressure are the most common causes of chronic kidney disease (CKD).”
Causes of Chronic Kidney Disease, National Institute of Diabetes and Digestive and Kidney Diseases 2016 - 163
Dr. Franz Messerli provided the citation Ely, Folkow, and Paradise 1990, which reports that rats maintained on a very-low-salt diet have an impaired blood pressure response to blood loss and stress.
“However, both LNa strains were abnormally sensitive to blood loss and showed attenuated pressor responses to both acute and chronic stress situations.”
Ely, Folkow, and Paradise 1990, abstract. - 164
“The response to slow bleeding was studied in rats on low sodium intake (0.04%) versus 'ordinary' (0.4%) intake. After 12 days on the diets acute experiments were performed on paired, conscious rats. Initially the salt-depleted rats had slightly lower mean arterial pressure (121 +/- 2 versus 129 +/- 3 mmHg, n = 30 pairs, P less than 0.02). Heart rate (HR), cardiac output, plasma volume, extracellular volume and plasma renin activity (PRA) did not differ significantly, while haematocrit was slightly higher in the salt-depleted rats (46.5 +/- 0.3 versus 45.1 +/- 0.5%, P less than 0.05). The animals were then slowly bled 1 ml every 5 min until mean arterial pressure (MAP) fell and remained below 50 mmHg. The controls tolerated up to a 55 +/- 1% loss of initial blood volume, while the salt-depleted rats turned into irreversible shock already after losing 38 +/- 2% (P less than 0.001).”
Göthberg et al. 1983, abstract. - 165
In Ely, Folkow, and Paradise 1990, the amount of sodium in the low-sodium diets were 4-25% the amount in the standard rodent diet. In Göthberg et al. 1983, the amount of sodium in the low-sodium diet was 10% the amount in the standard rodent diet.
“The focus of the following research was on the cardiovascular and neuroeffector effects of dietary Na reduction primarily in normotensive Wistar-Kyoto (WKY) and spontaneously hypertensive rats (SHR)
raised from 4 to 15 weeks on a control Na diet (CNa: 12 mmol per 100 g food) or various low Na diets (LNa: 0.5 to 3 mmol per 100 g food).”
Ely, Folkow, and Paradise 1990, abstract.
“The response to slow bleeding was studied in rats on low sodium intake (0.04%) versus 'ordinary' (0.4%) intake.”
Göthberg et al. 1983, abstract. - 166
Eaton and Konner 1985 estimate that typical hunter-gatherers ate about 690 mg of sodium per day from plant and animal foods, about one-fifth of current US intake.
“The sodium and potassium content for 14 vegetable foods used by recent hunter-gatherers is known. These foods have an average of 10.1 mg of sodium and 550 mg of potassium, respectively, for each 100-g portion. If these values are representative, the daily average of 1463.8 mg of paleolithic vegetable food would have yielded 147.8 mg of sodium. Data on the sodium and potassium content of wild game are unavailable, but if the average values for beef, lamb, pork, and veal (68.75 mg of sodium and 387.5 mg of potassium per 100 g) are assumed to be comparable to those for meat from wild animals, then the 788.2 g of meat in a 35:65 meat:vegetable Paleolithic diet would have provided 541.9 mg of sodium, for a daily total of 689.7 mg.”
Eaton and Konner 1985, p. 286.
Cordain et al. 2002 state that “hunter-gatherers rarely, if ever added salt to their foods”.
Cordain et al. 2002, p. S48.
Campbell et al. 2015 contains a table of sodium intake in hunter-gatherer societies. Intake estimates range from 46 to 1564 mg per day. The upper end of this range is less than half of average US intake. Notably, this excludes some exceptions that lived in coastal areas or near salt deposits and had higher salt intake. However, the paper argues that “hypertension is prevalent in those areas”, suggesting that their higher salt intake was not physiologically normal.
“Several studies have estimated salt intake in hunter-gatherer societies with a range of results (Table I), with nearly all population mean levels less than 2.5 g of salt (sodium 1000 mg) per day. Some of the higher levels of salt intake that have been reported in hunter-gatherer societies may reflect a nutritional transition within these societies and the addition of salt to foods or higher natural sources of salt (eg, marine sources).33,34 For example, we have not included intakes from coastal dwellers in New Guinea,
Quash’Qai tribes people, and similar peoples in Northern Kashmir with high dietary salt from natural sources. Hypertension is prevalent in those areas and hence salt intakes may represent pathophysiological rather than physiological levels.”
Campbell et al. 2015, table I, p. 247-8. - 167
- “Participants have either a history of stroke or an elevated risk of stroke based on age and blood pressure level at entry.” Neal et al. 2017, abstract.
- "The study is a large-scale, open, cluster-randomized controlled trial done in 600 villages across 5 provinces in China." Neal et al. 2017, abstract.
- 168
“As compared with regular salt, the salt substitute was also shown to protect against the secondary outcomes of major adverse cardiovascular events (49.1 events vs. 56.3 events per 1000 person-years; rate ratio, 0.87; 95% CI, 0.80 to 0.94; P<0.001) and death from any cause (39.3 events vs. 44.6 events per 1000 person-years; rate ratio, 0.88; 95% CI, 0.82 to 0.95; P<0.001) (Table 1 and Figure 3). The salt substitute was also shown to be beneficial with respect to death from vascular causes (22.9 events vs. 26.3 events per 1000 person-years; rate ratio, 0.87; 95% CI, 0.79 to 0.96) and nonfatal acute coronary syndrome (3.8 events vs. 5.1 events per 1000 person-years; rate ratio, 0.70; 95% CI, 0.52 to 0.93) but not with respect to nonfatal stroke (22.4 events vs. 24.9 events per 1000 person-years; rate ratio, 0.90; 95% CI, 0.80 to 1.01) (Table 1).” Neal et al. 2021, p. 1072
- 169
See for example this review paper representing the views of 24 sodium reduction skeptics, O’Donnell et al. 2020.
- 170
Foti et al. 2020 assessed the feasibility of conducting randomized controlled trials of dietary sodium reduction with cardiovascular event outcomes, and concluded that they would be “prohibitively expensive”.
“Based on our assumptions, a trial using any of the designs considered would require tens of thousands of participants and cost hundreds of millions of dollars, which is prohibitively expensive. Our estimates may be conservative given several key challenges, such as the unknown costs of sustaining a long-term difference in sodium intake, the effect of differential cotreatment with antihypertensive medications, and long lag time to clinical outcomes. Thus, it would be extraordinarily difficult to conduct such a trial, and despite the high costs, would still be at substantial risk for a spuriously null result.”
Foti et al. 2020, abstract. - 171
See our moral weight calculations for this intervention here.
See our latest moral weight guidance here. - 172
Kochanek et al. 2019, table 6, p. 35, row "Major cardiovascular diseases" and subentries, and table 7, p. 38
- 173
“Even if we thought this was a bad empirical assumption, we think accounting for life expectancy within age cohorts would force difficult ethical trade-offs to the surface. On the one hand, it feels reasonably intuitive to us that if our SMC program was just delaying malaria deaths from age 1 to age 2, this should ‘count’ for less compared to a program that was returning them to average life expectancy. On the other hand, we’d worry that taking into account life expectancy within age cohorts might lead us to assume that infant deaths averted in Sweden are more valuable than those averted in Burkina Faso, since we might expect children in Sweden to live longer anyway. We find this conclusion uncomfortable, since it stands at odds with other intuitions we hold about not wanting to ‘double-penalize’ children living in the most deprived conditions.
In sum, we think accounting for life expectancy within age cohorts would embed ambiguous moral choices and move our moral weights further away from standard measures used in the global health community (e.g. DALYs). We think this raises the bar that the empirical argument would have to meet for us to seriously consider doing this. ”
Are our Top Charities saving the same lives each year? GiveWell 2024. - 174
Cumulative blood pressure over a decade or more predicts the risk of cardiovascular events beyond simply measuring current blood pressure, consistent with cumulative effects of blood pressure on cardiovascular health.
“Lower 10-year cumulative systolic blood pressure was associated with 4.1yr longer survival and 5.4yr later onset of cardiovascular disease, resulting in living longer life with a shorter period with morbidity. Models adjusted for sociodemographic characteristics, cardiovascular risk factors, and index systolic blood pressure demonstrated associations of 10-year cumulative systolic blood pressure (per 130 mmHg×yr change, the threshold for stage-1 hypertension) with cardiovascular disease (HR:1.28, 95%CI:1.20-1.36), coronary heart disease (HR:1.29, 95%CI:1.19-1.40), stroke (HR:1.33, 95%CI:1.20-1.47), heart failure (HR:1.12, 95%CI: 1.02-1.23), and all-cause mortality (HR:1.21, 95%CI:1.14-1.29). These findings emphasize the importance of 10-year cumulative systolic blood pressure as a risk factor to cardiovascular disease, above and beyond current systolic blood pressure.”
Reges et al. 2020, abstract. - 175
- 176
“High sodium consumption (>2 grams/day, equivalent to 5 g salt/day) and insufficient potassium intake (less than 3.5 grams/day) contribute to high blood pressure and increase the risk of heart disease and stroke… Most people consume too much salt—on average 9–12 grams per day, or around twice the recommended maximum level of intake… The principal benefit of lowering salt intake is a corresponding reduction in high blood pressure.” Salt Reduction, WHO 2020.
“Sodium reduction from an average high usual sodium intake level (201 mmol/day) to an average level of 66 mmol/day, which is below the recommended upper level of 100 mmol/day (5.8 g salt), resulted in a decrease in SBP/DBP of 1/0 mmHg in white participants with normotension and a decrease in SBP/DBP of 5.5/2.9 mmHg in white participants with hypertension. A few studies showed that these effects in black and Asian populations were greater.” Graudal et al 2017, abstract. - 177
“All 194 Member States have committed to reducing population sodium intake by 30% by 2030, demonstrating strong consensus on sodium reduction as a lifesaving strategy, this report, however, shows that only nine countries have fully established the recommended policies to reduce sodium intake. Globally we are off track to achieve the target.”
Launch of the WHO Global report on sodium intake reduction, WHO 2023. - 178
“In 2017, 11 million (95% uncertainty interval [UI] 10–12) deaths and 255 million (234–274) DALYs were attributable to dietary risk factors. High intake of sodium (3 million [1–5] deaths and 70 million [34–118] DALYs), low intake of whole grains (3 million [2–4] deaths and 82 million [59–109] DALYs), and low intake of fruits (2 million [1–4] deaths and 65 million [41–92] DALYs) were the leading dietary risk factors for deaths and DALYs globally and in many countries.”
GBD 2017 Diet Collaborators 2019, abstract. - 179
“There is insufficient evidence regarding the effects of potassium-enriched salt substitutes on the occurrence of hyperkalemia. There is a need for additional empirical research on the effect of increasing dietary potassium and potassium-enriched salt substitutes on serum potassium levels and the risk of hyperkalemia, as well as for robust estimation of the population-wide impact of replacing sodium chloride with potassium-enriched salt substitutes.”
Greer et al. 2019, abstract. - 180
“The average standardised difference in mean [plasma aldosterone] was similar for all classes of diuretic: thiazide/thiazide-like 0.299 (95% confidence interval [CI] 0.150, 0.447), loop 0.927 (0.37, 1.49), MRA/potassium-sparing 0.265 (0.173, 0.357) and combination 0.466 (0.137, 0.796), Q = 6.33, P = .097… In RCTs of diuretic therapy in hypertension, there is an increase in [plasma aldosterone] with all classes of diuretic and no significant between-class heterogeneity.”
McNally et al. 2021, abstract
“Under acute and chronic conditions, diuretics induce an increase in plasma renin activity (PRA)8 but whether diuretics also increases PA has been debated.”
McNally et al. 2021, abstract - 181
“In the network meta-analysis, compared with placebo, angiotensin-converting enzyme inhibitors, dihydropyridine calcium channel blockers, and thiazide diuretics were reported to be similarly effective in reducing overall cardiovascular events (25%), cardiovascular death (20%), and stroke (35%); angiotensin-converting enzyme inhibitors were reported to be the most effective in reducing the risk of myocardial infarction (28%); and diuretics were reported to be the most effective in reducing revascularization (33%).”
Wei et al. 2020, abstract. - 182
O’Donnell et al. 2020, figure 3, p. 3371.
- 183
See our calculations here.
- 184
“Cardiovascular diseases (CVDs) are the leading cause of death globally, taking an estimated 17.9 million lives each year.” Cardiovascular diseases, World Health Organization.
- 185
“All 194 Member States have committed to reducing population sodium intake by 30% by 2030, demonstrating strong consensus on sodium reduction as a lifesaving strategy, this report, however, shows that only nine countries have fully established the recommended policies to reduce sodium intake. Globally we are off track to achieve the target.”
Launch of the WHO Global report on sodium intake reduction, WHO 2023.