Community Salt Substitution for Reducing Cardiovascular Risk

Summary

Published: February 2021

What is the problem?

Cardiovascular disease is the most common cause of death worldwide,1 and occurs most commonly among people who are middle-aged or elderly.2 The World Health Organization (WHO) and other public health organizations state that high blood pressure (hypertension)3 is a major contributor to cardiovascular risk, including the risk of heart attack, stroke, and heart failure, and to kidney failure.4 We discuss the evidence for the link between blood pressure and cardiovascular risk below.

Global Burden of Disease (GBD) estimates suggest that in 2017, high blood pressure was responsible for:

• 10.4 million deaths globally, representing 19% of all deaths
• 10%, 16%, and 22% of all deaths in countries with low, low-middle, and middle socio-demographic indices, respectively5

We have not reviewed the methodology underlying these estimates, and we view them with some uncertainty.

Systolic blood pressure (SBP) is the pressure in a person’s arteries as the heart is contracting, while diastolic blood pressure (DBP) is the pressure between contractions.6 For the sake of simplicity, we focus on SBP in this report, as SBP and DBP usually change in the same direction.7

Dietary salt (sodium chloride) intake is a determinant of blood pressure. The most recent Cochrane meta-analysis of RCTs reports that reducing salt intake modestly reduces blood pressure, but the effect is larger among people with hypertension, in which the average reduction in SBP is 5.5 millimeters of mercury (mm Hg, the unit in which blood pressure is typically measured).8

Dietary potassium is also a determinant of blood pressure; two recent meta-analyses of RCTs suggest that increasing potassium intake tends to reduce blood pressure among people with hypertension. The average reduction in SBP in these meta-analyses, Poorolajal et al. 2017 and Filippini et al. 2017, is 4.3 and 4.5 mm Hg, respectively.9

What is the program?

This report focuses on replacement of normal table salt with a salt substitute, in which potassium chloride replaces a portion of the sodium chloride, for all members of a community. The intention of salt substitution is to reduce the risk of diseases related to high blood pressure by reducing sodium intake and increasing potassium intake.

Community salt substitution 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 two trials.10

Trials have used different methods for promoting the adoption of a salt substitute within a community:

• In Zhou et al. 2013, research staff delivered salt substitute to participating households.11
• In Li et al. 2016, villages received access to salt substitute in conjunction with a community education program. In some villages, the cost of the salt substitute was subsidized to make it equivalent to normal salt.12
• In Bernabe-Ortiz et al. 2020, research staff retrieved and replaced table salt in community households, free of charge.13

Other trials specifically target people at high risk of hypertension-related diseases, but we will not discuss them in depth here.14

Does the program have strong evidence of effectiveness?

Overall, we believe there is fairly strong evidence that community salt substitution reduces cardiovascular morbidity and mortality. This evidence comes from two sources: direct evidence that salt substitution reduces cardiovascular mortality in people at high cardiovascular risk, and indirect evidence that salt substitution reduces blood pressure in whole communities, which other evidence suggests should reduce cardiovascular mortality and morbidity. We put roughly equal weight on both.

We did not identify direct evidence that salt substitute distribution to entire communities reduces cardiovascular morbidity or mortality. However, one non-community salt substitution RCT in elderly men reports a 41% reduction in cardiovascular mortality.

The indirect evidence comes from three community salt substitution RCTs, all of which report that potassium-enriched salt substitute modestly reduces SBP by between 1.1 and 6 mm Hg. We combine this with evidence from a meta-analysis of RCTs demonstrating that blood pressure-lowering drugs reduce the risk of high-burden cardiovascular diseases by between 17% and 28%. Based on the totality of this evidence and other considerations, our rough best guess is that community salt substitution reduces deaths from cardiovascular disease by 2.5% or 21%, depending on which source of evidence we focus on.15

Potassium-enriched salt substitute may harm people with severe kidney disease. Estimates suggest that this effect is greatly outweighed by the benefit of potassium-enriched salt substitute, but we remain uncertain about these estimates.

Direct evidence of effectiveness

We identified one trial (Chang et al. 2006) that measures the effect of salt substitution on cardiovascular mortality, though not in a community setting. It finds a 41% reduction (95% confidence interval, 5% to 63%) in cardiovascular mortality in elderly men. However, we view this evidence as fairly uncertain. After adjustments, our best guess based on this intervention is that community salt substitution reduces cardiovascular mortality by 21%.16

To identify RCTs of community salt substitution interventions that report direct impacts on cardiovascular disease outcomes, such as heart attacks and stroke, we performed a medium-depth scientific literature search.17 We also spoke with Graham MacGregor, a researcher who studies salt and health and is chairman of the nonprofit organization World Action on Salt & Health (WASH).18

We did not identify RCTs of community salt substitution interventions that report cardiovascular disease outcomes. However, we did identify one cluster-RCT, Chang et al. 2006, that reports the impact of 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.19 The salt substitute was 49% sodium chloride, 49% potassium chloride, and 2% “other additives.”20 At baseline, the men had an average age of 75 and average SBP of 131 mm Hg.21

The study 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.22 In the experimental group, urinary sodium excretion decreased by 17%, and urinary potassium excretion increased by 76%.23
• Cardiovascular mortality was 41% lower in the experimental group than the control group, with a 95% confidence interval of 5% to 63%.24
• Mortality from heart failure and stroke were 70% and 50% lower, respectively, in the experimental group relative to the control group.25
• Non-cardiovascular mortality did not differ significantly between groups.26

The interpretation that salt substitution reduced cardiovascular mortality is strengthened by the observation that the reduction in mortality primarily occurred via diseases that are strongly linked to hypertension.

Although we have not evaluated the trial in detail, it has several obvious features that we believe limit its value for evaluating the effectiveness of community salt substitution:

• The intervention targeted elderly male veterans rather than all members of a community, so its relevance to community-wide salt substitution is uncertain. Differences in age range, sex, and possibly other characteristics could result in different outcomes in the two intervention types.
• The trial was a cluster-RCT with only five clusters, which is too few to expect randomization to even out baseline differences between groups.27 This weakens our ability to draw strong cause-and-effect conclusions from the findings.
• The effect size of the most important finding, that cardiovascular mortality was 41% lower in the experimental group, is imprecisely estimated. The 95% confidence interval includes the possibilities that the intervention reduced cardiovascular mortality by as little as 5% or as much as 63%.28
• The trial was not preregistered, meaning that the authors had the freedom to select analyses that potentially presented the findings in a more favorable light.

Overall, we believe this trial provides some degree of support to the hypothesis that community salt substitution reduces cardiovascular mortality, but the study’s limitations and the fact that it was not a community trial leave us with substantial uncertainty. After applying adjustments that reflect our skepticism, our best guess based on this intervention is that community salt substitution reduces cardiovascular mortality by 21%.29

We are aware of one large, ongoing salt substitution RCT that was preregistered and will report stroke as its primary outcome, but it also targets high-risk individuals rather than entire communities.30

Indirect evidence of effectiveness

We identified three community salt substitution RCTs (Zhou et al. 2013, Li et al. 2016, and Bernabe-Ortiz et al. 2020), all of which report that potassium-enriched salt substitute modestly reduces blood pressure, although the finding was not statistically significant in Li et al. 2016. The effect appears to be primarily attributable to an increase in potassium intake rather than a reduction in sodium intake.

These trials do not report cardiovascular mortality. A meta-analysis of RCTs (Ettehad et al. 2016) demonstrates that blood pressure-lowering drugs reduce the risk of major cardiovascular disease events by 20% per 10 mm Hg decline in SBP. Combining these estimates suggests that community salt substitution lowers cardiovascular disease risk by 2% to 3%.

Together, we believe this constitutes relatively strong evidence that the intervention reduces cardiovascular morbidity and mortality.

Effect of community salt substitution on blood pressure

To identify RCTs of community salt substitution interventions that report indirect markers of cardiovascular disease risk, such as blood pressure, we performed a medium-depth scientific literature search.31

This search identified three trials:

• Zhou et al. 2013 was a cluster-RCT conducted in rural China that included 462 people and lasted two years.32 Families were included in the study if they had at least one member with hypertension and a high salt intake.33 This trial is therefore not perfectly representative of community salt substitution, but we included it because the outcomes reflect all adult family members rather than only those with hypertension.34 A salt substitute composed of 65% sodium chloride, 25% potassium chloride, and 10% magnesium sulfate was delivered to participating families.35 Average SBP was 152 mm Hg at baseline.36 We have not found evidence that the study was preregistered.
• Li et al. 2016 was a cluster-RCT conducted in rural China that included approximately 2,566 people and lasted 18 months.37 A salt substitute composed of 65% sodium chloride, 25% potassium chloride, and 10% magnesium sulfate was made available in intervention villages, either at full price or subsidized to match the price of normal salt, and an education campaign promoted its use.38 Although the paper does not report baseline blood pressure, average SBP in the control group during the intervention was 142 mm Hg.39 The study was preregistered with blood pressure and urinary sodium excretion as its primary outcomes.40
• Bernabe-Ortiz et al. 2020 was a stepped-wedge cluster-RCT conducted in rural Peru that included 2,376 people and lasted 5 to 30 months, depending on the cluster.41 In intervention households, the research team replaced normal salt with salt substitute composed of 75% sodium chloride and 25% potassium chloride, free of charge.42 Average SBP was 113 mm Hg at baseline, which is within the normal range and considerably lower than the other two trials.43 The study was preregistered with blood pressure as its primary outcome.44

We believe Li et al. 2016 and Bernabe-Ortiz et al. 2020 may be more informative than Zhou et al. 2013, due to their larger size and preregistration. Zhou et al. 2013 also selected families based on higher-than-average risk,45 which may be less relevant to pure community salt substitution interventions. However, we have not evaluated the three trials in depth.

The trials report the following effects on urinary sodium and potassium excretion, which are measures of sodium and potassium intake:

• Zhou et al. 2013 does not report a significant effect of salt substitution on urinary sodium or potassium excretion.46
• Li et al. 2016 reports that salt substitution reduced urinary sodium excretion by 6% and increased potassium excretion by 16%.47
• Bernabe-Ortiz et al. 2020 reports that salt substitution did not impact urinary sodium excretion and increased potassium excretion by 32%.48

Together, this suggests that the impact of community salt substitution on sodium intake is likely small or nonexistent, and its impact on potassium intake may range from null to moderate. However, we have remaining uncertainty about this conclusion because we have not investigated the methods used to measure urinary sodium and potassium in these trials, or the implications of methodological differences between trials.

The trials report the following effects on blood pressure:

• Zhou et al. 2013 reports that at endline, SBP was 6 mm Hg lower (95% confidence interval, 10 lower to 2 lower) in the intervention group than in the control group, and DBP was 3 mm Hg lower (95% confidence interval, 6 lower to 1 lower).49
• Li et al. 2016 reports that at endline, SBP was 1.1 mm Hg lower (95% confidence interval, 3.3 lower to 1.1 higher) in the intervention group than in the control group, and DBP was 0.7 mm Hg lower (95% confidence interval, 2.2 lower to 0.8 higher).50 Effects on blood pressure tended to be larger in villages that received a price subsidy for the salt substitute, corresponding with higher self-reported salt substitute use.51 At endline, the prevalence of hypertension was 1.3% lower (95% confidence interval, 5.1 lower to 2.5 higher) in the intervention group than the control group.52 None of these differences were statistically significant.
• Bernabe-Ortiz et al. 2020 reports that SBP was 1.3 mm Hg lower (95% confidence interval, 2.2 lower to 0.4 lower) in the intervention group than in the control group, and DBP was 0.8 mm Hg lower (95% confidence interval, 1.4 lower to 0.1 lower).53 It also reports that among people without hypertension at baseline, the incidence of hypertension was 51% lower (95% confidence interval, 29% lower to 66% lower) in the intervention group than in the control group.54

All three trials report that salt substitution modestly reduced average blood pressure in a community setting, with overlapping 95% confidence intervals, although the finding was not statistically significant in one trial. Notably, in this trial, participants had to pay for salt substitute rather than having it delivered for free, so lower adherence may have contributed to this finding. This is consistent with the observation that the effect size in this trial tended to be larger among villages that received a price subsidy for the salt substitute.55

We are uncertain how to interpret the provocative finding in Bernabe-Ortiz et al. 2020 that salt substitution reduces the incidence of hypertension by 51% in participants without hypertension at baseline. Its effect size is fairly precisely estimated and highly statistically significant,56 suggesting that it is unlikely to be due to chance. However, it could reflect a delaying of hypertension rather than durable prevention,57 in which case it would overestimate the long-term impact of the intervention.

In addition, several meta-analyses of RCTs conducted primarily in people with hypertension confirm that salt substitution reduces SBP and DBP,58 and effect sizes are substantially larger in these meta-analyses than in community trials. These findings are consistent with the hypothesis that potassium-enriched salt substitutes reduce blood pressure, but we have not vetted these meta-analyses or the RCTs that underlie them.

Overall, we believe this research provides relatively strong evidence that community salt substitution reduces average blood pressure.

Impact of blood pressure reduction on cardiovascular disease risk

There is strong evidence from placebo-controlled RCTs that blood pressure-lowering drugs substantially reduce the risk of high-burden cardiovascular events such as stroke and heart attack.59 A meta-analysis of 123 RCTs, Ettehad et al. 2016, reports the following effects for each 10 mm Hg reduction in SBP:60

• 20% reduction in combined major cardiovascular disease event risk (95% confidence interval, 17% to 23%)
• 17% reduction in heart attack risk (95% confidence interval, 12% to 22%)
• 27% reduction in stroke risk (95% confidence interval, 23% to 32%)
• 28% reduction in heart failure risk (95% confidence interval, 22% to 33%)
• Nonsignificant 5% reduction in kidney failure risk (95% confidence interval, 7% increase to 16% decrease)
• 13% reduction in all-cause mortality risk (95% confidence interval, 9% to 16%)61

In addition, the impact of blood pressure on cardiovascular risk is supported by observational studies relating blood pressure to cardiovascular outcomes.62 Although we believe there is strong evidence that reducing blood pressure reduces cardiovascular morbidity and mortality, the reduction in blood pressure reported in salt substitution trials tends to be smaller than in most drug trials,63 so we have some remaining uncertainty about the analogy. Nevertheless, we believe this constitutes fairly strong evidence that community salt substitution reduces cardiovascular disease risk.

For the purposes of our cost-effectiveness analysis, we assume a linear relationship between blood pressure reduction and cardiovascular risk reduction.64 Applying the cardiovascular risk reduction identified in Ettehad et al. 2016 to the degree of blood pressure reduction identified in community salt substitution trials, we estimate that community salt substitution reduces the risk of cardiovascular death by 2.5%.65

Potential offsetting/negative effects

Increasing potassium intake may elevate the risk of death in people who have a reduced ability to excrete potassium, such as those with severe chronic kidney disease.66 A recent modeling study estimates that nationwide potassium-enriched salt substitution in China would cause 11,000 deaths per year among people with severe chronic kidney disease,67 in a population of 1.4 billion people.68 The same study estimates that nationwide potassium-enriched salt substitution would avert 461,000 deaths from cardiovascular disease annually.69 This implies that the harm of salt substitution would offset 2.4% of its mortality benefit.70 The study also notes that despite the risk to people with chronic kidney disease, this group is still predicted to experience a net mortality benefit from salt substitution due to a reduced risk of cardiovascular diseases.71 We remain uncertain about these estimates because we have not vetted them.

We are not aware of other substantial harms of potassium-enriched salt substitutes; however, we have not thoroughly searched for them. Notably, the amount of potassium supplied by salt substitute is within the range of traditional diets containing fruits and vegetables.72

How cost-effective is the program?

A preliminary cost-effectiveness analysis suggests that community salt substitution is below the range of cost-effectiveness of the opportunities that we expect to direct marginal donations to (about 10x cash or higher, as of 2021). This cost-effectiveness analysis is in an early stage and therefore is not directly comparable to the cost-effectiveness analyses of our top charities. As a general rule, our estimates of a given program's cost-effectiveness tend to go down as we gain more information, and we would conduct a more thorough cost-effectiveness analysis before deciding to recommend funding to this program.

Note that our cost-effectiveness analyses are simplified models that do not take into account a number of factors. There are limitations to this kind of cost-effectiveness analysis, and we believe that cost-effectiveness estimates such as these should not be taken literally due to the significant uncertainty around them. We provide these estimates (a) for comparative purposes and (b) because working on them helps us ensure that we are thinking through as many of the relevant issues as possible.

Although salt substitution is inexpensive and likely reduces the risk of cardiovascular events and death, the population average risk reduction is offset by the ongoing cost of the intervention across most reasonable assumptions. However, we have not deeply researched intervention models or cost, and it is possible that there are more cost-effective ways to implement the intervention than those we considered, perhaps by conducting larger-scale education campaigns or replacing salt at the level of food processing.

A sketch of the cost-effectiveness model is below:

• Prevalence and consequences of the problem. Global Burden of Disease (GBD) estimates suggest that cardiovascular disease accounts for 27% of all deaths in countries with a low, low-middle, or middle socio-demographic index (SDI).73
• Effect of the intervention on deaths. Using blood pressure effects to estimate impact on cardiovascular mortality, we estimate that community salt substitution would avert 14.3 deaths per 100,000 people annually in countries with a low, low-middle, or middle SDI.74 Using the effect size estimate from Chang et al. 2006, a targeted salt substitution trial discussed above, we estimate that community salt substitution would avert 120 deaths per 100,000 people annually in countries with a low, low-middle, or middle SDI.75
• Effect of the intervention on morbidity. We assume the intervention reduces cardiovascular morbidity in proportion to its impact on cardiovascular deaths. To estimate this proportion, we used GBD data to calculate the ratio between years lost to disability and mortality in low-, low-middle-, and middle-SDI countries.76
• Cost of the program. We are uncertain about program costs, but we use a range of estimates based on a charity paying for between 50% and 100% of the cost of salt substitute,77 plus operating costs,78 or paying only for a community health education campaign.79 We estimate that community salt substitution costs between approximately US$1.31 and US$4.37 per person, per year, depending on the type of intervention.80
• Cost-effectiveness. Our preliminary estimate is that the intervention costs between approximately $2,400 and $30,600 per death averted. This range largely reflects which effect we use (indirect estimates based on effects on blood pressure or direct effects based on Chang et al. 2006).81 Putting equal weight on these approaches leads to a best guess of roughly $16,500 per death averted.82 After applying GiveWell’s moral weights to compare various types of good outcomes,83 we estimate that community salt substitution is below the range of cost-effectiveness of opportunities that we would consider directing funding to.

We remain uncertain about the following:

• Effect size. We estimate the effect of community salt substitution on cardiovascular morbidity and mortality in two ways: first, using the impact on blood pressure reported by trials of community salt substitution, and second, using the impact on cardiovascular mortality reported by a trial of salt substitution in men living in a retirement home, Chang et al. 2006. Each of these methods has limitations. The first method estimates cardiovascular morbidity and mortality indirectly by using the effect of the intervention on blood pressure, and the indirectness of this method requires uncertain assumptions. The second method is more direct, but relies on evidence from a group of participants that differs substantially from the participants in a community-wide salt substitution intervention, which introduces possible bias and uncertainty. We adjust for key sources of bias in the second method, but we are uncertain how effective these adjustments are.84
• Costs. We have substantial uncertainty about the cost figures we use. After conducting a limited search, we did not identify total cost estimates for community salt substitution programs. We use a range of estimates based on a charity paying for between 50% and 100% of the cost of salt substitute,85 plus operating costs,86 or paying only for a community health education campaign.87 We believe it is plausible that this range includes the true cost of the program, but we would require more information to determine this.
• Value of averting deaths in middle and older age. Most of the programs we recommend funding impact children. However, salt substitution primarily averts death in middle and older age, when cardiovascular disease is most deadly. We believe that averting the death of a middle-aged person with ten years to live is probably less valuable than averting the death of a child with 60 years to live. For this reason, we assign lower value to averting deaths via salt substitution relative to other interventions such as insecticide-treated nets, which affects how salt substitution compares to other programs in cost-effectiveness. However, this choice is highly uncertain and reflects value judgments by us, people demographically similar to participants in programs we recommend funding, and our donors.88
• External validity of RCT data on blood pressure lowering. To estimate the impact of community salt substitution on mortality, we use RCT data on the impact of blood pressure-lowering treatments on hypertension-related outcomes.89 These data have several limitations for this purpose. First, they reflect cardiovascular events rather than mortality.90 Second, they reflect the impact of anti-hypertensive drugs, which may not impact blood pressure via the same mechanism as salt substitution. We have not evaluated this possibility. Third, we are uncertain about how the duration of these interventions compares to the duration of salt substitution interventions. It is possible that interventions of longer duration would yield larger effects, and therefore that these data lead us to underestimate the impact of salt substitution.
• Coverage. Our cost-effectiveness analysis assumes relatively good intervention coverage, except in our model of an education-only intervention, because we use an estimate of blood pressure lowering from Bernabe-Ortiz et al. 2020, in which table salt was retrieved and replaced in each participating household.91 We view this as a relatively high-intensity version of the intervention, and we are uncertain how well communities would adhere under conditions typical of charity implementation. We apply an adjustment for lower coverage in our model of an education-only intervention.92
• Harms. We rely on estimates of the potential harm of salt substitution from Marklund et al. 2020,93 but we have not vetted this study, nor have we searched comprehensively for other possible harms of salt substitution. Note that we do not incorporate this estimate into our preliminary cost-effectiveness analysis due to its small impact.

Is there room for more funding?

Given that cardiovascular disease is the leading cause of mortality globally,94 and blood pressure is one of its primary risk factors,95 there is likely to be extensive room for more funding for this intervention.

After undertaking a very limited investigation, we did not identify obvious opportunities to fund charities directly implementing community salt substitution. We spoke with Graham MacGregor, a researcher who studies salt and health and is chairman of the nonprofit organization World Action on Salt & Health (WASH).96 WASH promotes salt substitution as part of its broader salt reduction efforts, but a focus on salt substitution would require program development, and we believe the cost-effectiveness of such a program may be challenging to evaluate.97

We also spoke with Laura Cobb, Director of Nutrition Policy and Surveillance at Resolve to Save Lives (RTSL), who stated that her organization has conducted substantial work in salt substitute promotion.98 However, it has encountered significant resistance over concerns about possible harm to people with kidney disease.99

Key questions for further investigation

• The Salt Substitute and Stroke Study (SSaSS) is a large, ongoing salt substitution RCT with stroke as a primary outcome.100 We may update our report with information from that trial once the stroke outcome has been published.
• Are there intervention models that are substantially less costly than the range we assumed in our cost-effectiveness analysis?
• Are interventions targeted to people at high cardiovascular risk more cost-effective?
• What proportion of people would use salt substitute in conditions typical of charity implementation?
• Would the implementing charity pay the full cost of the intervention, or would the cost be leveraged by other funders, such as a government?

Our process

• We performed a medium-depth scientific literature search for RCTs of community salt substitution.101
• We performed less structured, shallow searches for background information on blood pressure and its relationship with cardiovascular morbidity and mortality, sodium intake, and potassium intake.
• We consulted the GBD Results Tool for estimates of the mortality burden of hypertension-related diseases.
• We constructed a preliminary cost-effectiveness analysis of community salt substitution.
• We spoke with Graham MacGregor, a researcher who studies salt and health and is chairman of the nonprofit organization WASH, and Laura Cobb, Director of Nutrition Policy and Surveillance at Resolve to Save Lives (RTSL), an organization that has conducted substantial work in salt substitute promotion.102

Sources

Document	Source
American Heart Association, "Health Threats From High Blood Pressure"	Source (archive)
Bernabe-Ortiz et al. 2020	Source
CDC, "High Blood Pressure Symptoms and Causes"	Source (archive)
Chang et al. 2006	Source
Ettehad et al. 2016	Source
Filippini et al. 2017	Source
GiveWell, 2020 update on GiveWell's moral weights	Source
GiveWell, cost-effectiveness analysis of community salt substitution, 2020	Source
GiveWell, estimate of AMF cost per net, 2019 (unpublished)	Unpublished
GiveWell's non-verbatim summary of a conversation with Dr. Graham MacGregor, June 18, 2020	Source
GiveWell's non-verbatim summary of a conversation with Dr. Laura Cobb, September 18, 2020	Source
Graudal, Hubeck-Graudal, and Jurgens 2017	Source
Greer et al. 2019	Source
Hernandez et al. 2019	Source
Huang et al. 2020	Source
Institute for Health Metrics and Evaluation, Global Burden of Disease, GBD Results Tool, High systolic blood pressure risk 2017 (accessed November 12, 2020)	Source (archive)
Jafarnejad et al. 2020	Source (archive)
Kochanek et al. 2019	Source (archive)
Li et al. 2013	Source
Li et al. 2016	Source (archive)
Marklund et al. 2020	Source (archive)
Neal et al. 2017	Source
Peng et al. 2014	Source
Poorolajal et al. 2017	Source (archive)
Prospective Studies Collaboration 2002	Source
Sijbesma and Christoffers 2009	Source
Tanner, Candland, and Odden 2015 (working paper)	Source (archive)
U.S. Department of Agriculture, FoodData Central, Nutrition information for raw banana (accessed December 18, 2020)	Source (archive)
U.S. Department of Agriculture, FoodData Central, Nutrition information for roasted potato (accessed December 18, 2020)	Source (archive)
U.S. National Library of Medicine, ClinicalTrials.gov, "High Cardiovascular Risk Management and Salt Reduction in Rural Villages in China" (accessed November 9, 2020)	Source (archive)
U.S. National Library of Medicine, ClinicalTrials.gov, "Launching a Salt Substitute to Reduce Blood Pressure at the Population Level in Peru" (accessed November 9, 2020)	Source (archive)
WHO, "Cardiovascular diseases"	Source (archive)
WHO, Fact Sheet Detail, "Hypertension," 2019	Source (archive)
WHO, Q&A Detail, "Noncommunicable diseases: Hypertension," 2015	Source (archive)
World Bank, World Bank Open Data, "Population, total - China"	Source (archive)
Zhou et al. 2013	Source

• 1

  “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"

• 2

  Kochanek et al. 2019, table 6, p. 35, row "Major cardiovascular diseases" and subentries, and table 7, p. 38

• 3

  “Normal adult blood pressure is defined as a blood pressure of 120 mm Hg [millimeters of mercury] when the heart beats (systolic) and a blood pressure of 80 mm Hg when the heart relaxes (diastolic). When systolic blood pressure is equal to or above 140 mm Hg and/or a diastolic blood pressure equal to or above 90 mm Hg the blood pressure is considered to be raised or high.” WHO, Q&A Detail, "Noncommunicable diseases: Hypertension," 2015

• 4
  • “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"

• 5

  Institute for Health Metrics and Evaluation, Global Burden of Disease, GBD Results Tool, High systolic blood pressure risk 2017 (accessed November 12, 2020)

• 6

  “Blood pressure is measured using two numbers:

  The first number, called systolic blood pressure, measures the pressure in your arteries when your heart beats.

  The second number, called diastolic blood pressure, measures the pressure in your arteries when your heart rests between beats.” CDC, "High Blood Pressure Symptoms and Causes"

• 7
  • In meta-analyses of salt reduction and potassium supplementation, SBP and DBP typically move in the same direction. For example, there is reasonably strong evidence that low-salt diets reduce both SBP and DBP in white, black, and Asian people.

    “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.”

    Graudal, Hubeck-Graudal, and Jurgens 2017, abstract

  • Potassium supplementation tends to reduce both SBP and DBP. “Overall, potassium supplementation decreased systolic blood pressure of 4.48 mm Hg (95% CI 3.07–5.90) and diastolic blood pressure of 2.96 mm Hg (1.10–4.82).” Filippini et al. 2017, abstract

• 8

  “Sodium reduction from an average high usual sodium intake level (201 mmol [millimoles]/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, Hubeck-Graudal, and Jurgens 2017, abstract

• 9
  • “Compared to placebo, potassium supplementation resulted in modest but significant reductions in both SBP (MD -4.25 mmHg; 95% CI: -5.96 to -2.53; I² = 41%) and DBP (MD -2.53 mmHg; 95% CI: -4.05 to -1.02; I² = 65%).” Poorolajal et al. 2017, abstract
  • “Overall, potassium supplementation decreased systolic blood pressure of 4.48 mm Hg (95% CI 3.07-5.90) and diastolic blood pressure of 2.96 mm Hg (1.10-4.82).” Filippini et al. 2017, abstract

• 10
  • “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

• 11

  “The two kinds of salt were delivered in 1 kg bags that were identical in appearance and were differentiated by a three-digit code.” Zhou et al. 2013, p. 428

• 12

  “Intervention villages—the salt reduction program comprised community-based health education and availability of reduced-sodium, added-potassium salt substitute at village shops. The health education component was delivered by the township health educators with assistance from the village council and the village doctors through public lectures, the display and distribution of promotional materials, and special interactive education sessions targeted towards individuals at elevated risk of vascular diseases [15]. The salt substitute was made available for purchase in all intervention villages and promoted through the health education component of the intervention. Residents in villages randomized to the price subsidy were provided with coupons that enabled the purchase of salt substitute at the same price as usual salt.” Li et al. 2016, p. 3

• 13

  “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

• 14

  “From each village, approximately 35 high-risk adult participants with either (1) a history of stroke or (2) age 60 years or above with inadequately controlled high blood pressure, were enrolled.” Huang et al. 2020, p. 137

• 15

  This is based on effect sizes for the direct and indirect estimation methods, adjusted for internal and external validity concerns. To calculate these figures, we divided the estimated number of cardiovascular deaths averted by the baseline rate of cardiovascular deaths we assumed in our cost-effectiveness analysis, then applied external and internal validity adjustments. The broad range reflects the large differences between the estimates generated by the direct and indirect methods. See here for how we arrived at the 2.5% lower bound and here for how we arrived at the 21% upper bound.

• 16

  This is based on effect sizes for the direct estimation method, adjusted for internal and external validity concerns. To calculate these figures, we divided the estimated number of cardiovascular deaths averted by the baseline rate of cardiovascular deaths we assumed in our cost-effectiveness analysis, then applied external and internal validity adjustments. See calculations here.

• 17

  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.

• 18

  GiveWell's non-verbatim summary of a conversation with Dr. Graham MacGregor, June 18, 2020

• 19
  • “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

• 20

  “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

• 21

  Chang et al. 2006, p. 1292, table 1

• 22

  “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

• 23

  “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

• 24

  “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

• 25

  Chang et al. 2006, p. 1293, table 2, rows "Heart failure" and "Cerebrovascular disease"

• 26

  “The non-CVD deaths were not significantly different between the groups (P = 0.479). The non-CVD incidence was 8030.0 and 7990.0 per 100 000 person-years for the experimental and control groups, respectively.” Chang et al. 2006, p. 1292

• 27

  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)

• 28

  “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

• 29

  This is based on effect size for the direct estimation method, adjusted for internal and external validity concerns. To calculate these figures, we divided the estimated number of cardiovascular deaths averted by the baseline rate of cardiovascular deaths we assumed in our cost-effectiveness analysis, then applied external and internal validity adjustments. See calculations here.

• 30
  • The Salt Substitute and Stroke Study (SSaSS) trial. A detailed description of the study is published in Neal et al. 2017. Interim results, including blood pressure outcomes, are published in Huang et al. 2020.
  • "The SSaSS has been designed to test whether sodium reduction achieved with a salt substitute can reduce the risk of vascular disease. The study is a large-scale, open, cluster-randomized controlled trial done in 600 villages across 5 provinces in China. Participants have either a history of stroke or an elevated risk of stroke based on age and blood pressure level at entry. Villages were randomized in a 1:1 ratio to intervention or continued usual care. Salt substitute is provided free of charge to participants in villages assigned to the intervention group. Follow-up is scheduled every 6 months for 5 years, and all potential endpoints are reviewed by a masked adjudication committee. The primary end point is fatal and nonfatal stroke, and the 2 secondary endpoints are total major cardiovascular events and total mortality." Neal et al. 2017, abstract

• 31

  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.

• 32
  • “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. Of the 462 individuals in the trial, 372 completed the study (81%).” Zhou et al. 2013, abstract
  • "All families were randomized by a computerized randomization scheme into one of the two groups. Everyone in the same family received the same treatment (normal salt or salt substitute)." Zhou et al. 2013, "Subjects and methods" section

• 33

  “The families were potentially eligible for the study, if the family members satisfied all the following conditions: (1) At least one member in the family was a hypertensive patient. We defined hypertension as having systolic blood pressure >140 mm Hg or a diastolic blood pressure >90 mm Hg. (2) The participant had an estimated daily sodium intake of ≥260 mmol [millimoles] per day.” Zhou et al. 2013, p. 427

• 34

  This is not stated explicitly in the paper, but is implied by the fact that only 51% of participants had hypertension. “The subjects included 234 females (51%) and 237 subjects with hypertension (51%).” Zhou et al. 2013, p. 428

• 35
  • “The two kinds of salt were delivered in 1 kg bags that were identical in appearance and were differentiated by a three-digit code.” Zhou et al. 2013, p. 428
  • “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

• 36

  This represents the average of the control and intervention groups from table 1. Zhou et al. 2013, p. 429, table 1

• 37
  • “This study was a cluster-randomized trial done over 18 months in 120 townships (one village from each township) from five provinces.” Li et al. 2016, abstract
  • “A population survey was done at the end of the intervention period amongst an age- and sex-stratified random sample of 20 or more consenting adults drawn from each of 119 villages, resulting in 2,566 survey participants.” Li et al. 2016, p. 3

• 38
  • Li et al. 2016 does not describe the composition of the salt substitute, but it is included in a previous publication. “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
  • “The salt substitute was made available for purchase in all intervention villages and promoted through the health education component of the intervention. Residents in villages randomized to the price subsidy were provided with coupons that enabled the purchase of salt substitute at the same price as usual salt.” Li et al. 2016, p. 3

• 39

  Li et al. 2016, p. 7, table 2

• 40

  "Primary Outcome Measures:

  1. Mean systematic blood pressure level [Time Frame: October 2010 - December 2012]
  2. 24 hour urinary sodium [Time Frame: October - December 2012]"

  U.S. National Library of Medicine, ClinicalTrials.gov, "High Cardiovascular Risk Management and Salt Reduction in Rural Villages in China" (accessed November 9, 2020)

• 41
  • “We report on a population-wide implementation of this strategy in a stepped-wedge cluster randomized trial (NCT01960972). The regular salt in enrolled households was retrieved and replaced, free of charge, with a combination of 75% NaCl and 25% KCl. A total of 2,376 participants were enrolled in 6 villages in Tumbes, Peru.” Bernabe-Ortiz et al. 2020, abstract
  • Figure 3 of Bernabe-Ortiz et al. 2020 suggests that the maximum length of the intervention was 30 months, and its minimum length was five months. However, since it was a stepped-wedge cluster-randomized trial with six clusters, only one-sixth of participants received the intervention for the full 30 months. Bernabe-Ortiz et al. 2020, p. 379, figure 3

• 42

  “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

• 43

  Bernabe-Ortiz et al. 2020, p. 376, table 1

• 44

  "Primary Outcome Measures:

  1. Systolic/diastolic blood pressure (mmHg) [Time Frame: Three years]"

  U.S. National Library of Medicine, ClinicalTrials.gov, "Launching a Salt Substitute to Reduce Blood Pressure at the Population Level in Peru" (accessed November 9, 2020)

• 45

  “The families were potentially eligible for the study, if the family members satisfied all the following conditions: (1) At least one member in the family was a hypertensive patient. We defined hypertension as having systolic blood pressure >140 mm Hg or a diastolic blood pressure >90 mm Hg. (2) The participant had an estimated daily sodium intake of ≥260 mmol per day.” Zhou et al. 2013, p. 427

• 46

  “At the 24-month follow-up visit, first morning urine sodium concentrations were 177 mmol l^-1 (interquartile range 137–261) for the salt substitution group and 184 mmol l^-1 (interquartile range 108–228) for the normal salt group. The corresponding potassium concentrations were 29 mmol l^-1 (interquartile range 18–45) for the salt substitution group and 29 mmol l^-1 (interquartile range 18–51) for the normal salt group. There were also no significant differences between the two groups for median urine sodium and potassium (P=0.36 and P=0.66, respectively).” Zhou et al. 2013, p. 428

• 47

  “Among 1,903 people with valid 24-hour urine collections, mean urinary sodium excretion in intervention compared with control villages was reduced by 5.5% (-14mmol/day, 95% confidence interval -26 to -1; p = 0.03), potassium excretion was increased by 16% (+7mmol/day, +4 to +10; p<0.001), and sodium to potassium ratio declined by 15% (-0.9, -1.2 to -0.5; p<0.001).” Li et al. 2016, abstract

• 48

  “The levels of sodium in 24-h urine samples (see Supplementary Table 4) at the end of the study and baseline were 3.95 ± 1.83 (s.d.) g and 3.94 ± 1.86 g, respectively, and mean difference 0.01 (95% CI (0.25, −0.23)). These results were similar across all the study villages. In contrast, the levels of potassium were higher at the end of the study (2.60 ± 1.20 g) than at baseline (1.97 ± 1.20 g), and mean difference 0.63 (95% CI (0.78, 0.47)).” Bernabe-Ortiz et al. 2020, p. 375

• 49

  Zhou et al. 2013, p. 430, table 2

• 50

  “Mean blood pressure differences were -1.1 mm Hg systolic (-3.3 to +1.1; p = 0.33), -0.7 mm Hg diastolic (-2.2 to +0.8 p = 0.35).” Li et al. 2016, p. 6

• 51

  Li et al. 2016, p. 8, table 3

• 52

  Li et al. 2016, p. 7, table 2

• 53

  “The fully adjusted intention-to-treat analysis showed an average reduction of 1.29 mm Hg (95% confidence interval (95% CI) (−2.17, −0.41)) in systolic and 0.76 mm Hg (95% CI (−1.39, −0.13)) in diastolic blood pressure.” Bernabe-Ortiz et al. 2020, abstract

• 54

  “Among participants without hypertension at baseline, in the time- and cluster-adjusted model, the use of the salt substitute was associated with a 51% (95% CI (29%, 66%)) reduced risk of developing hypertension compared with the control group.” Bernabe-Ortiz et al. 2020, abstract

• 55

  Li et al. 2016, p. 8, table 3

• 56

  “Among participants without hypertension at baseline, in the time- and cluster-adjusted model, the use of the salt substitute was associated with a 51% (95% CI (29%, 66%)) reduced risk of developing hypertension compared with the control group.” Bernabe-Ortiz et al. 2020, abstract

• 57
  • “Our results also point to a lower incidence of hypertension in the participants receiving the intervention, a key clinical and public health finding. Whether this is a short-term effect (that is, the intervention did not prevent hypertension onset but rather delayed it) remains to be studied further. Because the endocrine system in charge of salt regulation, the renin–angiotensin–aldosterone system, continues to receive larger amounts of sodium or lower amounts of potassium, it is probable that blood pressure will start to increase until it reaches hypertensive thresholds. Longer follow-ups, with and without intervention, are required to assess whether the endocrine system develops salt resistance. However, our findings show no evidence of an interaction between time and intervention.” Bernabe-Ortiz et al. 2020, p. 376
  • "Because of the potential problems with hypertension incidence, blood pressure change is a more reliable efficacy metric to use in salt substitution trials. However, one complicating factor of measuring efficacy this way is that people's blood pressure tends to rise very gradually with age. This means that if the benefit of salt substitution is not necessarily to decrease blood pressure, but to prevent a gradual rise in blood pressure over the course of a lifetime, it may not be detectable in trials that only last a few years. The INTERSALT study found that the rate at which blood pressure increases with age is associated with the amount of salt consumed in a society, which lends plausibility to the idea that salt substitution interventions could function by preventing or slowing this increase." GiveWell's non-verbatim summary of a conversation with Dr. Graham MacGregor, June 18, 2020

• 58
  • “Pooled results showed that salt substitutes had a significant effect on SBP (mean difference: -4.9 mm Hg; 95% CI: -7.3, -2.5 mm Hg; P < 0.001) and DBP (mean difference: -1.5 mm Hg; 95% CI: -2.7, -0.3 mm Hg; P = 0.013).” Peng et al. 2014, abstract
  • “[Low-sodium salt substitutes] decreased SBP (MD -7.81 mm Hg, 95% CI -9.47 to -6.15, p<0.00001) and DBP (MD -3.96 mm Hg, 95% CI -5.17 to -2.74, p<0.00001) compared with control.” Hernandez et al. 2019, abstract
  • “Pooled weighted mean differences showed significant reductions of SBP (WMD - 8.87 mmHg; 95% CI - 11.19, - 6.55, p < 0.001) and DBP (WMD - 4.04 mmHg; 95% CI - 5.70, - 2.39) with no statistically significant heterogeneity between the 11 included comparisons of SBPs and DBPs.” Jafarnejad et al. 2020, abstract

• 59

  “Meta-regression analyses showed relative risk reductions proportional to the magnitude of the blood pressure reductions achieved. 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). However, the effect on renal failure was not significant (0·95, 0·84–1·07).” Ettehad et al. 2016, abstract

• 60

  For more detail on this categorization of events, see the following:

  "Data were also extracted for major cardiovascular disease events (defined as fatal and non-fatal myocardial infarction, sudden cardiac death, revascularisation, fatal and non-fatal stroke, and fatal and non-fatal heart failure), coronary heart disease (fatal and non-fatal myocardial infarction and sudden cardiac death, excluding silent myocardial infarction), stroke (fatal and non-fatal, excluding transient ischaemic attacks), heart failure (new diagnosis of heart failure, hospital admission, or death), renal failure (end-stage renal disease resulting in dialysis, transplantation, or death), and all-cause mortality." Ettehad et al. 2016, Pg. 959.

• 61

  “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). However, the effect on renal failure was not significant (0·95, 0·84–1·07).” Ettehad et al. 2016, abstract

• 62

  “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.” Prospective Studies Collaboration 2002, abstract

• 63

  Most blood pressure-lowering drug trials represented in Ettehad et al. 2016, a meta-analysis of the impact of blood pressure-lowering drugs on cardiovascular event risk, reduced SBP by 5 or more mm Hg. Ettehad et al. 2016, p. 959, figure 2

• 64

  This is consistent with the model applied by Ettehad et al. 2016; the approximately linear function can be observed in figure 2. However, we note that other functions may fit the data in figure 2, and we have not investigated this deeply. Ettehad et al. 2016, p. 959, figure 2

• 65

  This is based on effect sizes for the indirect estimation method, adjusted for internal and external validity. To calculate these figures, we divided the estimated number of cardiovascular deaths averted by the baseline rate of cardiovascular deaths we assumed in our cost-effectiveness analysis, then applied external and internal validity adjustments. See calculations here.

• 66
  • “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. Individuals with chronic kidney disease are advised to limit dietary potassium and avoid potassium enriched salt substitutes.” Marklund et al. 2020, p. 2
  • “The risks of potassium-enriched salt substitutes include a possible increased risk of hyperkalemia and its principal adverse consequences: arrhythmias and sudden cardiac death, especially in people with conditions that impair potassium excretion such as chronic kidney disease. There is insufficient evidence regarding the effects of potassium-enriched salt substitutes on the occurrence of hyperkalemia.” Greer et al. 2019, abstract

• 67

  “The intervention could potentially produce an estimated 11000 (6422 to 16562) additional deaths related to hyperkalaemia in individuals with chronic kidney disease.” Marklund et al. 2020, abstract

• 68

  Total population as of 2019. See World Bank, World Bank Open Data, "Population, total - China".

• 69

  “Nationwide implementation of potassium enriched salt substitution could prevent about 461000 (95% uncertainty interval 196339 to 704438) deaths annually from cardiovascular disease, corresponding to 11.0% (4.7% to 16.8%) of annual deaths from cardiovascular disease in China; 743000 (305803 to 1273098) non-fatal cardiovascular events annually; and 7.9 (3.3 to 12.9) million disability adjusted life years related to cardiovascular disease annually.” Marklund et al. 2020, abstract

• 70

  11,000/461,000 = 0.0239

• 71

  “The net effect would be about 450000 (183699 to 697084) fewer deaths annually from cardiovascular disease in the overall population and 21000 (1928 to 42926) fewer deaths in individuals with chronic kidney disease.” Marklund et al. 2020, abstract

• 72

  In Bernabe-Ortiz et al. 2020, 24-hour potassium excretion in the intervention group was 2.6 grams per day, which reflects total daily intake of potassium from salt substitute and other foods: "In contrast, the levels of potassium were higher at the end of the study (2.60 ± 1.20 g) than at baseline (1.97 ± 1.20 g), and mean difference 0.63 (95% CI (0.78, 0.47))." This is approximately the amount of potassium found in four medium potatoes or five large bananas (see U.S. Department of Agriculture, FoodData Central, Nutrition information for raw banana (accessed December 18, 2020) and U.S. Department of Agriculture, FoodData Central, Nutrition information for roasted potato (accessed December 18, 2020)).

• 73

  Data are available in our cost-effectiveness analysis.

• 74

  See our calculations here.

• 75

  See our calculations here. The figure of 120 deaths is obtained after applying internal and external validity adjustments to the raw number of deaths averted.

• 76

  See our calculations here.

• 77
  • For the cost of salt, we used the lowest cost cited in a community salt substitution RCT, which is US$0.65 per kg in Li et al. 2016: “Salt substitute costs about twice as much as usual salt (4CNY [US$0.65] vs. 2CNY [US $0.33] per kg).” Note that another RCT, discussed below, cited a much higher cost.
  • In their community salt substitution program, Bernabe-Ortiz et al. 2020 note that they were able to obtain a negotiated price of US$4 per kg: "During the present study, before the intervention the cost of 1 kg of the salt substitute to the general public was 35 PEN (~$US10), and through the project we were able to achieve a price reduction to 14 PEN (~$US4), further indicating opportunities for scaling up implementation efforts."

• 78

  Estimated using the operating cost associated with the distribution of a single insecticide-treated net ($2.59) from our cost-effectiveness analysis of the Against Malaria Foundation (AMF). The data come from this spreadsheet, which is restricted to GiveWell employees. See here and here for how we use this operating cost in our cost-effectiveness analysis for this intervention.

• 79

  Community RCTs of salt substitution do not provide estimates of this aspect of intervention cost. We used the median cost cited in an analysis of hygiene promotion interventions in low-income countries: “Irrespective of their results, their costs ranged from US$1.05 to 1.74/pp/yr, with an average of US$1.31” (Sijbesma and Christoffers 2009, p. 426). In our preliminary cost-effectiveness analysis, these figures are not adjusted for inflation.

• 80

  See this section of our cost-effectiveness analysis for how we calculated these estimates.

• 81

  See our calculations here and here.

• 82

  (2,400+30,600)/2 = 16,500

• 83

  See this document, "Headline results compared to 2019" and "Graph of age-weighted results."

• 84

  See our cost-effectiveness analysis here.

• 85
  • For the cost of salt, we used the lowest cost cited in a community salt substitution RCT, which is US$0.65 per kg in Li et al. 2016: “Salt substitute costs about twice as much as usual salt (4CNY [US$0.65] vs. 2CNY [US $0.33] per kg).” Note that another RCT, discussed below, cited a much higher cost.
  • In their community salt substitution program, Bernabe-Ortiz et al. 2020 note that they were able to obtain a negotiated price of US$4 per kg. "During the present study, before the intervention the cost of 1 kg of the salt substitute to the general public was 35 PEN (~$US10), and through the project we were able to achieve a price reduction to 14 PEN (~$US4), further indicating opportunities for scaling up implementation efforts."

• 86

  Estimated using the operating cost associated with the distribution of a single insecticide-treated net ($2.59) from our cost-effectiveness analysis of AMF. The data come from this spreadsheet, which is restricted to GiveWell employees. See here and here for how we use this operating cost in our cost-effectiveness analysis for this intervention.

• 87

  Community RCTs of salt substitution do not provide estimates of this aspect of intervention cost. We used the median cost cited in an analysis of hygiene promotion interventions in low-income countries: “Irrespective of their results, their costs ranged from US$1.05 to 1.74/pp/yr, with an average of US$1.31” (Sijbesma and Christoffers 2009, p. 426). In our preliminary cost-effectiveness analysis, these figures are not adjusted for inflation.

• 88

  See this document for more information on the values we assign to different outcomes achieved by programs we evaluate.

• 89

  Ettehad et al. 2016

• 90

  "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

• 91

  “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

• 92

  See the “External validity adjustments” tab, "Coverage" section, of our cost-effectiveness analysis.

• 93

  “The intervention could potentially produce an estimated 11000 (6422 to 16562) additional deaths related to hyperkalaemia in individuals with chronic kidney disease.” Marklund et al. 2020, abstract

• 94

  “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"

• 95

  “Meta-regression analyses showed relative risk reductions proportional to the magnitude of the blood pressure reductions achieved. 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). However, the effect on renal failure was not significant (0·95, 0·84–1·07).” Ettehad et al. 2016, abstract

• 96

  GiveWell's non-verbatim summary of a conversation with Dr. Graham MacGregor, June 18, 2020

• 97
  • "Several organizations are actively working to reduce salt consumption in various contexts around the world, but there may not be any that are primarily or exclusively focused on implementing salt substitution interventions at scale. Many salt reduction programs are designed to reduce salt intake by all means possible, such that salt substitution is only one part of a broader program."
  • "If it received additional funding, WASH would also be interested in working on salt substitution implementation. WASH is currently waiting on the results of a salt substitution trial that is currently in progress. If the salt substitution trial shows promising results, WASH plans to use the results to convince the Chinese government to gradually increase the potassium content of the country's salt supply over a matter of years. However, the Chinese government may be hesitant to take such steps, preferring to keep both ordinary salt and salt substitute available on the market. With additional funding, WASH may be able to develop salt substitution implementation programs in other contexts as well."

  GiveWell's non-verbatim summary of a conversation with Dr. Graham MacGregor, June 18, 2020

• 98

  "RTSL has been working at both the country and global levels to promote the use of salts that are higher in potassium and lower in sodium. There is a sizable body of research on low-sodium salts, and RTSL seeks to translate that research into interventions that could be implemented at scale." GiveWell's non-verbatim summary of a conversation with Dr. Laura Cobb, September 18, 2020

• 99

  "Implementing policies to increase the use of low-sodium salts has been challenging, particularly because of concerns from governments and physicians about hyperkalemia, a potentially fatal increase of blood potassium levels in people with severe kidney disease."
  GiveWell's non-verbatim summary of a conversation with Dr. Laura Cobb, September 18, 2020

• 100

  "The SSaSS has been designed to test whether sodium reduction achieved with a salt substitute can reduce the risk of vascular disease. The study is a large-scale, open, cluster-randomized controlled trial done in 600 villages across 5 provinces in China. Participants have either a history of stroke or an elevated risk of stroke based on age and blood pressure level at entry. Villages were randomized in a 1:1 ratio to intervention or continued usual care. Salt substitute is provided free of charge to participants in villages assigned to the intervention group. Follow-up is scheduled every 6 months for 5 years, and all potential endpoints are reviewed by a masked adjudication committee. The primary end point is fatal and nonfatal stroke, and the 2 secondary endpoints are total major cardiovascular events and total mortality." Neal et al. 2017, abstract

• 101

  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.

• 102

  See GiveWell's non-verbatim summary of a conversation with Dr. Graham MacGregor, June 18, 2020 and GiveWell's non-verbatim summary of a conversation with Dr. Laura Cobb, September 18, 2020.