From the World Hypertension League: Understanding the Science that Supports Population-Wide Salt Reduction Programs

3 May 2017

Jacqui Webster, Temo Wasqanivalu, JoAnn Arcand, Kathy Trieu, Francesco Cappuccio, Lawrence Appel, Mark Woodward, Norm Campbell and Rachael McLean

Full paper available here.

Independent systematic reviews of the totality of evidence have led governments and international agencies around the world to agree that population-wide interventions to reduce salt are beneficial for health. However some studies (which are largely criticised for their design and methodology and, therefore, invalid results) produce findings that conflict with the evidence base. As a result, these studies gain a lot of publicity which confuse the general public, clinicians and program leaders, and can therefore impede program implementation.

To address this, the Standards for Salt Research group - composed of national and international health and scientific organisations - will be developing a set of processes and criteria to which new research concerning salt should adhere to. This will ensure that only robust scientific studies are used when national and international recommendations relating to salt reduction are reviewed.

In their paper, the World Hypertension League provided an overview of the main arguments used against reducing salt, and the counter arguments of evidence that supports population-wide salt reduction.

1. Can we make recommendations about salt without consensus on the science? Salt is an unsettled and hot topic for investigation and so it would be premature to make recommendations in the absence of trial data.

Response: There is consensus about the need to reduce population salt intake from the organizations that have systematically and independently examined the totality of the evidence. Public health policy recommendations need to be based on critical appraisals of the totality of the evidence by expert scientific groups overseen by governmental or nongovernmental health and scientific organizations. As in most areas of public health, the evidence base is incomplete, as it is lacking in definitive randomized controlled trials (RCTs) for cardiovascular disease (CVD) outcomes, due to practical and ethical considerations.
Nevertheless, multiple independent review processes in different countries have all concluded that typical dietary salt intake is too high, that it creates serious health problems, and that consequently salt intake should be reduced. The current WHO recommendation is that salt (sodium chloride) intake should be <5 g/d for adults, with lower levels in children based on their lower caloric needs. The evidence supporting the need to reduce salt and the impact of reducing salt on health is increasing. However, every year there are a few controversial studies that get most of the media coverage and cause some people to question current recommendations. These studies are usually linked to a small group of individuals, several with ties to commercial interests. Such studies are not appropriately designed to assess the association between salt intake and disease as outlined below.

2. What are the estimated potential benefits of salt reduction on the burden of disease and death? Some estimates of the potential benefits of salt reduction in terms of burden of disease and deaths are much lower than initially thought.

Response: Reducing salt intake would save millions of lives a year globally. High dietary salt is associated with wide-ranging health problems including hypertension, stroke, CVD, bone demineralization, kidney stones, gastric cancer, and kidney disease. The greatest health risk is the increase in blood pressure (BP) caused by excess salt intake. Although the magnitude of change in BP from modest changes in dietary salt are relatively small in individuals, on a population scale, a reduction in salt explains between one fifth and one third of cases of hypertension. The Global Burden of Disease Study estimated that increased dietary salt is the second leading behavioural risk for death and the third leading behavioural risk for disability globally, such that 4.129 million deaths and more than 83 million disability-adjusted life years lost could be attributed to high dietary sodium consumption in 2015. Health economic analyses have found that reducing dietary salt is cost saving or highly cost effective in different investigations. Such evidence has provided the rationale for the WHO and the World Economic Forum to identify population-wide sodium reduction as a “best-buy” intervention for public health.

3. Does reducing salt reduce BP long term? Most evidence shows that reducing salt reduces BP only in the short term (<6 months). Should we be cautious in extrapolating from short-term studies of a few months or weeks to what might be expected in the long term?

Response: A large body of evidence supports the fact that reducing dietary salt has a long-term effect on BP. In a wide variety of animal studies, BP continues to increase if salt intake is increased, and this is not attenuated over time. In human epidemiological studies, habitually high dietary salt intake is associated with high BP and increasing BP with age. BP would not continue to rise with age if changes in BP were only for a short time after exposure. In long-term population intervention studies that reduced dietary salt and lasted for decades, decreases in BP have matched or exceeded reductions predicted from results of shorter-term clinical trials. In clinical trials of reducing dietary salt, BP reduces in proportion to reductions in dietary salt. Many trials lasting 4 weeks or longer have shown reductions in BP. Several high-quality randomized clinical trials that reduced dietary salt have demonstrated reductions in BP up to 18 months with no attenuation in reduction of BP. Most longer-term trials have not been able to sustain reductions in dietary salt among participants and have been challenged regarding confounding with other co-interventions. Nevertheless, in the sodium reduction arm of the Trials of Hypertension Prevention (TOHP) 2 study, systolic BP was still reduced in the intervention arm compared with control after 3 years. In conclusion, well-designed meta-analyses of salt reduction trials clearly demonstrate that salt reduction will have a long-term impact on BP.

4. What is the evidence that lowering BP in non-hypertensive persons will reduce CVD mortality? Some studies suggest that lowering BP increases mortality and cardiovascular events (other than stroke) in people without hypertension.

Response: Lowering BP will reduce CVD mortality, including in persons without hypertension. The extensive evidence from epidemiology and higher-quality analyses of randomized clinical trials shows a direct relationship between BP and CVD. This relationship between BP and CVD is evident at SBP values beginning at ≈115 mm Hg, a finding that suggests that non-hypertensive persons will also benefit from reductions in BP. The long-term follow-up from the TOHP studies, which showed that reduced sodium was associated with reduced CVD, were all performed in pre-hypertensive patients. Reductions in CVD are related to the extent to which BP is reduced as well as to the absolute cardiovascular risk of the study population. The controversy arises from retrospective nonrandomized analyses of studies that include individuals who are already sick. Some such studies have found a U-or J-shaped curve that can be explained by reverse causality (whereby BP may be low because of preclinical or prevalent diseases such as myocardial infarction or heart failure, which then lead to increased risk of premature death). Studies where aggressive BP-lowering therapy is used in populations susceptible to clinical hypotension may also cause cardiovascular and renal disease. A recent study that suggested a relationship between reduced BP and increased CVD events included patients with diabetes, and therefore treatment and other factors were likely to have influenced the results. Such studies do not provide evidence to counter support for population-wide salt reduction strategies. An overview of some of the main weaknesses in these studies is provided below.

5. Why are there no randomized clinical trials proving that salt reduction reduces CVD? These must be performed to provide robust data proving that salt reduction reduces CVD before we implement interventions.

Response: Pooled data from RCTs with long-term follow-up for CVD suggest that salt reduction reduces CVD. The reasons few RCTs have been designed and conducted specifically to examine whether salt reduction reduces CVD is because adequately designed and appropriately powered trials would be too expensive, impractical, or potentially unethical, given the high quality of evidence documenting the benefits of salt reduction. This is the same in many other fields of public health (eg, tobacco control, obesity reduction, actions to reduce climate change, air pollution, physical inactivity, and excessive alcohol consumption). Because of the high content of salt in the food supply, long-term RCTs of salt reduction have found it hard to sustain reduced dietary salt for more than 3 years in free-living patients. The alternative approach of asking patients to increase dietary salt would be unethical due to the evidence that eating too much salt harms health. In most clinical trials, the relatively small number of people who have a CVD event precludes the establishment of any relationship between salt intake and CVD. However, studies that have pooled the data from RCTs of salt reduction, or salt replacers, using meta-analyses, have demonstrated a reduction in cardiovascular events with reduced dietary sodium. The existing evidence is strong enough to support the implementation of strategies to reduce salt intake. Meta-analyses of cohort studies of salt intake in healthy populations, or where filters were used to exclude low-quality research, have also shown that salt intake is associated with CVD. Long-term experience with salt reduction in Finland, Japan, and the United Kingdom has shown an association between salt reduction and reduced BP and CVD at the population level. Long-term follow-up of individuals in the TOHP trials, which carefully assessed usual salt intake through repeat 24-hour urine collections, found that salt intake <2300 mg/d was associated with reduced CVD.

6. Aren’t low levels of sodium consumption associated with increased CVD events and mortality? Some studies show that low levels of sodium (<3000 mg sodium or 7.5 g salt/d) are associated with increased mortality, especially in non-hypertensive persons. The most recent study quoted is the Prospective Urban Rural Epidemiology (PURE) study.

Response: It is important to emphasize that consumption of <3 of sodium a day is not considered low sodium consumption and that such levels are not associated with increased CVD events and mortality. Population salt intake for communities with diets that do not have added salt are nearly all below 1 g of sodium (2.5 g of salt) per day. So, although sodium is required for proper bodily function, the minimum physiological amount required is <1 g of sodium per day. Very few populations are currently consuming such low amounts, but in the few remaining hunter-gatherer populations where they do, BP does not rise with age and hypertension and CVD are uncommon. Reliable vital statistics are unavailable in these isolated populations, but mortality appears to be related to infectious diseases and other problems of economically developing regions that are unrelated to salt intake.
While some of the studies that suggest lower amounts of dietary salt are associated with increased CVD events and mortality are based on very large numbers of participants, the methodologies are seriously flawed. Hence, their findings should be interpreted with caution and are not appropriate to guide policy. Systematic reviews of cohort studies that have criteria to exclude lower-quality studies, as well as systematic reviews of RCTs, have found that CVD is reduced (not increased) with lower intakes of dietary sodium. Some of the common methodological problems have been addressed extensively in previous reviews and are summarized as follows:

Measurement error: The most accurate technique to estimate usual dietary salt intake is the collection of multiple, high-quality 24-hour urine samples. A single urine sample is often used to estimate usual intake; however, this is also less accurate because of large day-to-day variation in dietary intake and because excretion varies widely even with a fixed salt intake. Several of the studies showing a J-shaped relationship between salt intake and health outcomes rely on a single spot urine assessment to estimate each person’s long-term usual sodium intake. Single spot urine sodium samples cannot accurately assess an individual’s usual salt intake, because sodium intake varies meal to meal and day to day and is also impacted by seasonal food availability. Other studies have had a large proportion of 24-hour urine collections that were incomplete, leading to inaccurate estimates of salt intake that would impact the results. Incomplete 24-hour urine collections have lower sodium levels and do not reflect true intake. Other studies have used food surveys that may also not reliably estimate salt intake. In contrast, a follow-up study of participants in the TOHP study used multiple 24-hour sodium measurements and documented that people with lower sodium intakes have a lower risk of CVD and total mortality. More recent analysis has shown that this relationship is still present after a median of 24 years of follow-up, with no evidence of a J-shaped curve.
Reverse causality: In observational studies, persons consuming the least salt may be more likely to have the highest risk of cardiovascular events and death because they were already ill when they entered the study–the problem known as “reverse causation.” These people are likely to consume fewer calories and therefore eat less salt because they are ill, rather than being ill as a result of eating less salt. For example, in the PURE study, older age, having diabetes, and having a history of CVD were more common among patients with lower estimated sodium intakes than those with higher intakes.
Confounding factors: Residual confounding cannot be excluded in observational studies, even when multiple factors are controlled for in the analysis. Specifically, studies often do not assess or control for factors that may address the health outcome of interest. Such factors may include chronic kidney disease, family history of CVD, and levels of or changes in nutrient or calorie intake (which might be related to age, physical activity, or chronic disease status). RCTs, such as TOHP, control for confounding factors through randomization at baseline and therefore provide a higher standard of evidence than observational studies.

7. Why can’t untimed collections (eg, spot samples) such as those used in the PURE study and others, which show a J-shaped curve, accurately reflect the usual long-term salt intake of individuals?

Response: Spot urine samples are an unreliable measure of individual salt intake and should not be used as the basis for correlating salt intake with health outcomes, such as BP or CVD. There is ongoing research to examine the potential for use of spot urine samples from a large sample of the population to estimate mean population salt intake. However, spot urine samples do not provide an accurate assessment of an individual’s intake. The reasons are as follows:

Salt intake varies from day to day and from meal to meal. Therefore, the sodium content of a spot urine sample reflects what was just eaten rather than usual salt intake over an extended period in an individual.
Other factors affect spot urine sodium excretion concentration and include state of hydration, posture, renal function, diurnal variation, and other regulatory functions.
The equations used to calculate 24-hour salt intake from spot urine samples include several variables strongly associated with disease outcomes (age, sex, and urine creatinine concentration), which means that the estimate is not independent of other potential confounding factors.
The correlation between spot urine samples and 24-hour urine estimates of salt intake varies from one population to another and so the equations cannot be applied without validation studies. For example, in the main PURE study, the correlation was relatively high at around 0.7; however, the correlation in the PURE China population was <0.3. Also, validation studies show differences of 8000 to 9000 mg sodium per day in spot and 24-hour urine samples from the same individuals even when both are from the same collection day.
Bland-Altman plots of spot urine vs 24-hour urine plot show bias at high and low intakes. Spot urine overestimates at low intake and underestimates at high intake, which means that risk is exaggerated at low intake and underestimated at high intake. Therefore, spot urine samples do not reflect individual intakes well across their range.