Oxford University Press's
Academic Insights for the Thinking World

New sodium intake research and the response of health organizations

The American Journal of Hypertension (AJH) recently published the findings of a comprehensive meta-analysis monitoring health outcomes for individuals based on their daily sodium intake. The results were controversial, seemingly confirming what many notable hypertension experts have begun to suspect in recent years: that levels of daily sodium intake recommended by governmental agencies like the CDC are far too low, perhaps dangerously so.

Media outlets were quick to broadcast the findings, and the response from the CDC and organizations like the American Heart Association were much the same as in the past, dismissing the analysis, without pointing to specifics, as relying on “faulty methodology” and “flawed data.”

We recently spoke to Dr. Niels Graudal, lead author of the meta-analysis published in AJH, to understand, among other things, the details of his research and his opinion on the reaction of governmental health agencies to new findings on sodium intake.

Could you start by talking to us about the nature of a meta-analysis? What about your meta-analysis makes its findings more valid than, say, a single, localized study?

Population studies are accepted in health science as a means to define associations between health-factors. For instance, the associations between blood pressure, cholesterol, and mortality have been defined by such studies. Meta-analyses integrate the results from many individual studies to provide an average of the association of the “risk factor” to outcome. Such analyses help to reach a consensus, and constitute the core of the Cochrane Collaboration, which systematically organizes medical research information on the basis of scientific evidence.

Was there a common methodology among the studies included in your meta-analysis?

In population studies on sodium intake, the individually-measured sodium intake is used to categorize the participants in groups of low, intermediate, and high sodium intake. The groups are followed for years, while mortality (death rate) and morbidity (disease rate) in the different groups are recorded. Successively, the association between sodium intake and mortality/morbidity is calculated.

heart rate

What are some possible obstacles encountered in population studies like this?

There are factors which could bias the result in a wrong direction, so-called “confounders.” For instance, sodium intake typically increases with energy intake. Sick participants with a low energy intake may therefore eat less sodium than healthy people, and overweight participants predisposed to diabetes and cardiovascular disease may eat more sodium than healthy people. Therefore, the energy intake is a confounder, which could explain a potential increased mortality in participants with a low and a high sodium intake. However, there are statistical methods that allow us to correct for such confounders in order to ensure for accurate findings; such methods are used in almost all such studies, and have been for many years.

What were the specific findings of your meta-analysis on sodium intake?

Perhaps most importantly, the implications of these findings are that the present recommendation from the CDC that individuals should reduce sodium intake to below 2300 mg/day is too restrictive, and that the majority (about 95%) of the global population presently eat sodium within the safest range (2,645-4,945 mg/day) and therefore have no need to alter their intake.

Our present analysis showed that both high sodium intake and low sodium intake were associated with increased mortality when compared with the present usual sodium intake of most individuals worldwide, which is between 2,645 and 4,945 mg per day. In spite of the fact that sodium intake is somewhat difficult to measure precisely, the signal from the nearly 275,000 participants we looked at was abundantly clear.

Perhaps most importantly, the implications of these findings are that the present recommendation from the CDC that individuals should reduce sodium intake to below 2300 mg/day is too restrictive, and that the majority (about 95%) of the global population presently eat sodium within the safest range (2,645-4,945 mg/day) and therefore have no need to alter their intake.

How did you account for those participants in your analysis that were already suffering from, say, hypertension or obesity? Might they have affected the findings in some way?

When we excluded groups of populations with diseases from our analysis and only included healthy populations, which were random samples of the general population and within which multiple statistical adjustments for confounders had been performed, the results concerning low sodium intake were even more significant, indicating that confounders could not have affected the outcome of our analysis.

Is your meta-analysis the first scientific research to suggest that extremely low levels of sodium intake like those promoted by the CDC may actually be associated with negative health outcomes?

Actually, a 1984 paper published in the journal Science questioned the wisdom of population-wide sodium intake reduction on the basis of an investigation of about 10,000 participants. The FDA immediately published a high-profile response in The New York Times, claiming that the findings were likely the result of a statistical fluke or “something wrong with the analysis.” This immediate move to quell any dissenting evidence seems to have governed the debate ever since.

More recently, though, a population study published while our meta-analysis was under review showed results very similar to ours. Two of the individual studies (1, 2) included in our analysis also concluded that there was a “U” shaped correlation between sodium intake and mortality (increased risks at very low and high doses). For the record, excluding these two studies from our analysis did not change our results. In the past year, then, four recent studies have independently confirmed increased risks associated with both high and low sodium intakes, and suggest that the present recommendation of less than 2,300 mg/day is in conflict with available science.

What are the arguments against research like this?

Often, health organizations will attempt to call into question researchers’ objectivity by labeling them biased agents of the food industry. In a recent response to a paper showing that the majority of the world’s populations had a salt intake significantly above the recommended 2,300 mg/day, representatives of the World Health Organization (WHO) and World Action on Salt and Health (WASH) accusatorily asked, “why has the food and beverage industry mounted yet another campaign to try to resist beneficial changes, either directly or indirectly through their academic voices?”

Sometimes agencies will willfully misinterpret findings. In a short commentary to the recent Institute of Medicine (IOM) report on sodium intake in populations, nine CDC employees quoted the IOM report as follows: “When it comes to sodium intake levels <2,300mg per day… the committee found insufficient and inconsistent evidence regarding the benefit or harm in certain population subgroups (e.g., individuals with diabetes, chronic kidney disease, or preexisting cardiovascular disease)”. However, the actual quote from the study was this: “science was insufficient and inadequate to establish whether reducing sodium intake below 2,300 mg/d either decreases or increases CVD risk in the general population.” Often, they claim that our data and methods are somehow flawed, though they rarely cite specific instances. In a response to our present meta-analysis, the American Heart Association (AHA) stated that the analysis relied on “flawed data and should not change the way anyone looks at sodium.” They went on to say that:

“…those studies were poorly designed to examine the relationship between sodium intake and mortality, and the findings fail to take into account well-established evidence about sodium intake. Other problems with the new study included unreliable measurements of sodium intake and an overemphasis on studying sick people rather than the general population.”

This is a small selection of the arguments which have been raised for years by representatives of public institutions (WHO, FDA, NIH, CDC, AHA) in response to scientific investigations that don’t agree with their population-wide sodium reduction agenda.

So that we can avoid generalizations, what are the specific studies these organizations usually cite in support of population-wide sodium reduction, and what do you think are their flaws?

The rebuttals of health organizations are almost invariably propped up by vague references to an “immense” – usually unspecified – body of research which “proves” the beneficial effects of sodium reduction for the general population. If they do cite specific studies, it’s usually either

  • A group of randomized trials in which the baseline blood pressure of participants was actually much higher than the average 71 mmHg of the general population: 80-89 mmHg, 83-89 mmHg, borderline hypertensives and hypertensives, and hypertensives (>90 mmHg). Though these studies are obviously not useful for general population policymaking, they are, nonetheless, used to bolster the population-wide sodium reduction agenda of health organizations.
  • A meta-analysis (which, ironically, would have the same hypothetical methodological weaknesses the AHA and CDC supposedly see in ours) that finds increased risk of stroke in individuals consuming more than 4,945 mg/day. The results don’t conflict with our findings (our healthy range is 2,645 – 4,945 mg/day), but also don’t examine negative outcomes for low sodium intake, so are irrelevant to debate around determining an intake range.
  • Follow-ups (1, 2) of two older studies, pooling and analyzing their data with cardiovascular disease (CVD) mortality and all-cause mortality (ACM) as outcomes. These showed no significant difference between the low sodium group (ACM = 2.3%) and the normal sodium group (ACM = 2.6 %) (p = 0.58), thus confirming that sodium reduction may have no effect. The authors did, however, on several occasions, dissect the results by means of multiple adjustments, and did succeed in finding a few marginal or borderline significant results in favor of sodium reduction. The analyses behind these results, though, were not predefined in a protocol and should therefore be considered as having an extremely high risk of bias.


These flawed or irrelevant studies tell us that, concerning the general population, the blood pressure surrogate link between sodium intake and mortality is unreliable. As a matter of fact, the blood pressure surrogate link has been opposed by a meta-analysis that accounted for the full range of global population blood pressure and showed negative side effects for sodium reduction.

Where does this leave us?

As blood pressure is obviously not a reliable link between sodium intake and mortality, the conclusion of the aforementioned 2013 IOM report based on evidence from population studies is the best we have. This report was conducted by independent researchers and sponsored by the CDC, who, for less-than-transparent reasons, chose to follow the example of the AHA by rejecting their own report. As previously mentioned, this report found no evidence to establish whether reducing sodium intake below 2,300 mg/day either decreases or increases CVD risk in the general population. It also found no evidence in support of recommending different sodium intakes to diseased and normal groups, and did find evidence for potential harm in a sodium intake below 1,500 mg. However, the report failed to specify the dimensions of a safe sodium intake zone, but, by implication, indicated that such a safe zone does exist, consistent with the experience of all other essential nutrients.

In your opinion, what should organizations like the CDC and AHA, who control the development and implementation of public health policy, be doing now, in light of this new research?

Any policy like the current one that would aim to have 95% of the world’s population drastically alter their diet ought to be based upon strong, irrefutable scientific evidence.

I think that, instead of immediately moving to accuse dissenting scientists of economic and intellectual corruption, it may be more appropriate for powerful health organizations to ask what scientific mind would buy a theory as simplistic as the one currently governing sodium intake policy (sodium intake leads to high blood pressure, which leads to death) without a modicum of skepticism? Would it be so unreasonable for these groups to at least take our skepticism seriously, instead of reflexively attempting to explain away the results?

Our study provides evidence that a U/J shaped curve exists for the association between sodium intake and health outcome, as it does with all other nutrients. I will be the first to admit that this evidence is based on observational population studies, which are inevitably subject to flaws caused by imprecise measurements and confounders. These flaws, though, are greatly mitigated by the inclusion of a large number of participants, by statistical adjustments, by sensitivity analyses of subgroups, and by consistency in results between several independent studies.

There can never be any scientific guarantee that these safeguards eliminate all flaws; on the other hand, though, in the absence of a conflicting body of data, the IOM report and our analysis should be included in the determination of public policy, not ignored. Any policy like the current one that would aim to have 95% of the world’s population drastically alter their diet ought to be based upon strong, irrefutable scientific evidence.

Image credit: heart health fitness concept. © cosmin4000 via iStockphoto.

Recent Comments

  1. […] For Disease Control) recommends no more than 2,300 mg of sodium per day. That level, say the study’s researchers, is too low for good […]

Comments are closed.