Sodium Mortality Debate

Within the medical community there continues to be a debate about whether lower sodium consumption reduces cardiovascular mortality. The debate involves two schools of thought. The first is that the lower the sodium consumption, the lower the cardiovascular mortality. The other group feels that there is an association of sodium consumption with cardiovascular mortality that is U-shaped or J-shaped. At both the higher and lower ends of sodium consumption there is an increase in mortality. In the center portion, which approximately corresponds to the average present American consumption, there is the lowest mortality.

This debate has prompted better and better studies with better and better data. As the better data is reported the weight of evidence favors more and more the direct correlation. In other words, the lower the consumption of sodium, the less the cardiovascular mortality. This is probably because it improves the sodium potassium ratio, but potassium is rarely considered.

To support the U-Curve, one after another meta-analysis is repeated. There are many problems with the U-Curve argument. A recent meta-analysis (1) touting the U-curve prompted a commentary (2) that discussed the problems with the U-Curve argument.

Recent Meta-analysis

Karl Pearson
Karl Pearson, A Founder Of Modern Statistics
The recent meta-analysis that prompted the commentary included 25 prior studies. 23 of the studies were prospective cohort studies and 2 were randomized controlled trials. However there are several problems whenever a secondary examination is made of studies that are not designed for the purpose of the secondary study. In this case the studies were not designed to study the sodium cardiovascular disease interaction. This resulted in some standard methodologic challenges.

In the case of the 23 cohort studies, every study was a secondary examination of data collected for a purpose that differed from the purpose of the secondary study. The original studies were not designed to find the relationship of sodium intake and cardiovascular disease (CVD). They were designed to find answers to different questions. This means the quality of the original studies may have been excellent for the question they were designed to answer, but the quality was poor for answering the sodium-CVD question.

There are 3 main problems with the methods used for every secondary study of cohorts to analyze the relationship of sodium to CVD. For this secondary study, the meta-analysis, the 3 main problems are: 1. systematic bias in measuring sodium, 2. reverse causation, and 3. imprecise estimation of urinary sodium.

Systematic Bias

Whenever an original study is designed to study something besides the purpose of the secondary examination, the original study may be slanted so that the area being examined in the secondary study is not evenly distributed. This would result in systematic bias. A review of 31 independent sodium-CVD cohort studies found 3 to 4 issues with methods per study. These issues had the potential to alter the direction of the association in 96% of the studies. That means the result could be the opposite of the true result 96% of the time.

Reverse Causation

In the case of sodium and cardiovascular disease, reverse causation is likely in secondary reviews of data. Participants in the original studies often have cardiovascular disease. They are often on a low sodium diet, either self-determined or determined by their doctor. The participants also are often on diuretics or other medications that change the amount of sodium in their urine, leading to imprecise estimates of sodium intake (discussed below). So rather than low sodium intake leading to cardiovascular disease, cardiovascular disease leads to low sodium intake, i.e. reverse causation.

Imprecise Estimates

Limited urinary sodium collections are used in many studies. Previous studies had shown some correlation of short urinary collections of sodium with 24 hour collections. Many of the studies included in this particular meta-analysis had only spot collection or short period collections. Of the 25 studies that were included in the meta-analysis only 2 had multiple 24 hour urine collections.

But it is now known that limited urinary sodium collections are an imprecise estimate of sodium intake. Only recently sodium excretion has been discovered to have a cyclical variation lasting longer than a day. A recent study (3) showed the imprecision of spot, and even single 24 hour, urinary sodium collections. The findings come out of a study done for the Mars flight program.

In a highly controlled environment, the researchers showed that there will be a high degree of inaccuracy in 24 hour urinary sodium collections as an estimate of sodium consumption. In a single 24 hour urine collection, only one half of the collections will be accurate within 3 grams of the amount of sodium taken in during the prior 24 hours. If 3 sequential samples are examined, the collection will achieve 75% accuracy. And if 7 samples are used, it will have a 92% accuracy.

The Problems With This Study

In addition to the problems of all secondary studies that use re-purposed data, there also were problems specific to this meta-analysis. The meta-analysis authors chose cutpoints for their groups that were entirely different than any other studies. The cutpoints were markedly different from those chosen by the Institute of Medicine (IOM) and American Heart Association (AHA) for their recommendations. The authors chose 2645 mg/day and 4945 mg/day instead of the 1500 mg/day and 2300 mg/day used by the IOM. The difference makes comparison of the meta-analysis data with other studies very difficult. It also raises the question of why the researchers chose different cutpoints.

The authors of the meta-analysis also did multiple testing for comparisons of their outcomes across 4 outcomes and 5 levels of sodium intake. This multiple testing increased the probability of their results only being due to chance.

And finally, it could not be determined how consistent their findings were because of inadequate data in the article and in the supplementary material. Two of the cohorts that they used were the NHANES I and NHANES III. When alternative analyses of these cohorts were done, entirely different results were obtained showing no increased risk from the lower salt intake. Using the alternative analyses resulted in the hazard ratio from lower salt intake being 0.99. This ratio meant there was no increase in risk of cardiovascular death.

Studies With Higher Quality Data

The commentary then cited 2 recent cohort studies, not included in the meta-analysis, that were of higher quality than those included in the meta-analysis. Both of these studies showed an association of higher sodium intake and CVD. One of these studies was the TOHP study that showed a linear trend for the relationship between sodium and cardiovascular disease even at the lower end of sodium intake. This study used only healthy participants and collected multiple 24 hour collections. This is far more relevant to answering a question about prevention than one including patients who already have the disease.

Another high quality study was a randomized controlled study designed specifically to examine the relationship of sodium and potassium to cardiovascular disease. It was the Taiwanese study that we discussed in our post here. In this particular study the cardiovascular mortality hazard ratio was reduced to 0.59 when the sodium was reduced. However in this Taiwanese study some of the sodium was replaced with potassium so that an even higher ratio of potassium to sodium was obtained in the diet than would have been obtained simply by reducing sodium. This study was not included in the meta-analysis.

As the commentary relates, the highest quality studies show that less sodium intake results in less cardiovascular disease in healthy people, even at the lower levels recommended by the IOM and AHA. Secondary analysis of studies designed for another purpose have not given results of adequate quality. But such secondary studies will continue to provide enough confusion that little will be done to change national food practices. It will be up to individuals to reduce sodium intake and increase potassium intake on their own.

If you want to reduce your chance of cardiovascular disease, hypertension and strokes to the lowest possible chance, do not wait for the nation to change the food supply. Do it on your own. Reduce your sodium consumption to less than 1500 mg per day and increase your potassium consumption to more than 4700 mg per day. Go on the high potassium foods diet.
1. Compared with usual sodium intake, low- and excessive-sodium diets are associated with increased mortality: a meta-analysis. Graudal N, Jürgens G, Baslund B, Alderman MH. Am J Hypertens. 2014 Sep;27(9):1129-37. doi: 10.1093/ajh/hpu028. Epub 2014 Mar 20.

2. Sodium and cardiovascular disease: what the data show. Whelton PK, Appel LJ. Am J Hypertens. 2014 Sep;27(9):1143-5. doi: 10.1093/ajh/hpu138.

3. Agreement Between 24-Hour Salt Ingestion and Sodium Excretion in a Controlled Environment. Lerchl K, Rakova N, Dahlmann A, Rauh M, Goller U, Basner M, Dinges DF, Beck L, Agureev A, Larina I, Baranov V, Morukov B, Eckardt KU, Vassilieva G, Wabel P, Vienken J, Kirsch K, Johannes B, Krannich A, Luft FC, Titze J. Hypertension. 2015 Oct;66(4):850-7. doi: 10.1161/HYPERTENSIONAHA.115.05851. Epub 2015 Aug 10.