“Try to cut down on the salt.”
I’m sure many people have gotten this well-meaning piece of advice from their doctor, especially if their blood pressure is a little bit high. It ranks up there with not smoking or drinking alcohol and avoiding red meat, saturated fat, and sugary drinks as one of those things we’re told to do if we don’t want to die prematurely. It is actively promulgated by government health authorities all over the world, including the NHS in the UK and the FDA in the US.
Like most dietary advice that we get from our governments, this recommendation comes from low quality observational studies performed in the mid-twentieth century. The problem with observational studies is that there is a huge scope for confounding effects. As an example, coffee was long thought to be terrible for your health based on observational studies which found that people who drank more coffee tended to die younger. Many doctors recommended that their patients cut down on coffee. Then better data came along, and it became clear that it wasn’t the coffee that was hurting people; the increased death rate was due entirely to the fact that people who drank coffee were also more likely to smoke. Smoking was the confounder in this case. This is the nature of confounding, and it is a problem that exists in all observational studies, which is why we should never just trust observational data. The only way to be sure that you’re removing all confounders from the equation is to do a randomized controlled trial (RCT), which is why this is considered the highest quality form of evidence.
One thing to understand before we get in to the data is that the general hypothesis behind why salt reduction is good for you is that it results in a reduction in blood pressure. So it is interesting to look not just at what the possible reduction in mortality is, but also at how much blood pressure is decreased. If blood pressure decreases by a lot, then that strengthens the case that a salt reduction is good for you, but if blood pressure decreases by only a little or not at all, then that weakens the case. Following the same logic, people with high blood pressure should be expected to see a bigger benefit from salt reduction than people with normal blood pressure, since they have a more dangerously high blood pressure to begin with.
So, what do the trials show?
A systematic review and meta-analysis produced by the Cochrane foundation in 2014 sought to answer this question. It looked at the eight randomized controlled trials that had been carried out up to that point investigating whether salt reduction could result in increased longevity. In total 7,284 individuals were included, about half of which had normal blood pressure, and half of which had high blood pressure. The individual trials lasted from six months to three years, and seven of the research groups also performed long term follow-ups after the initial trials had ended, with the longest such follow-up being around 13 years.
Note that the review did not look at people with heart failure, so that is a caveat. If you have heart failure, then this information does not apply to you.
The interventions were either in the form of advice and coaching by health professionals, or low sodium salt substitution. Participants were divided up based on whether they had normal blood pressure or high blood pressure, and the results were analyzed separately for each group.
Six of the eight trials looked at urinary salt excretion at the end of the study periods, in order to make sure that people in the intervention group were actually consuming significantly less salt than people in the placebo group. Salt excretion was reduced by an average of 27 mmol per day, which is about 0,63 grams. Considering that the average American consumes 3,4 grams of sodium per day, this represents a reduction of 19% .
Ok, on to the results.
Among people with normal blood pressure, 1,3% died over the course of the trials in the reduced salt group compared with 2,0% in the control group. This was a 0,7% absolute difference, but it was not remotely close to being statistically significant. When the long-term follow-up data was included in the analysis, 2,1% had died in the reduced salt group, compared with 2,4% in the control group. So the long-term absolute difference in mortality was 0,3%, and again, it wasn’t even close to being statistically significant.
What does this mean? Basically, if you have normal blood pressure to start, you probably won’t achieve any gain in life expectancy from cutting down your salt intake.
Next we’ll look at the people with high blood pressure. This is the more important group, since logically, it should be more likely to benefit from salt reduction. In the group that reduced their salt intake, 17,1% died during the trials compared with 19,2% in the control group. This was a 2,2% absolute difference in favour of salt reduction, which isn’t nothing, but again, the difference wasn’t even close to being statistically significant (i.e. there is a high probability that it was caused by chance).
When the long term follow-up data was included, 17,3% had died in the reduced salt group, compared with 19,1% in the control group. This was a 1,8% absolute difference, but again, it wasn’t even close to being statistically significant (If you think it’s odd that a lower percentage died in the long term analysis than the short term analysis, here is the explanation: more participants were included in the long-term follow-up than in the shorter follow up because an additional study was included in the analysis).
So, where does that leave us? Basically, the best quality evidence available today does not show that reducing salt intake will cause you to live longer, even if you have high blood pressure. You could argue that there is some benefit, since around 2% more were alive in the intervention group at the end of follow-up. The problem with this argument is that the statistical significance is weak (the 95% confidence interval of the risk ratio is 0.87-1.14, ie the odds that high salt consumption shortens your life are about as big as the odds that it lengthens your life).
But how can salt not be good for you, if reducing salt intake lowers your blood pressure? Well, let’s get to that, since the review looked at the effect of the reduction in salt intake on blood pressure.
Among the people with normal blood pressure at the beginning of the trials, the systolic blood pressure (the pressure in the blood vessels when the heart is contracting) was 1,2 mmHg (mm of mercury, the metric used for measuring blood pressure) less in the reduced salt group than in the control group. Since a normal systolic blood pressure is about 120, and for an average person probably varies from 105 to 140 over the course of a day, an average reduction of 1,2 is unlikely to have any noticeable health impact. It’s basically a rounding error, and it wasn’t statistically significant. The diastolic blood pressure was 0,8 mmHg less in the reduced salt group. Funnily enough, this difference was statistically significant, but in practical terms it is still a rounding error, certainly not something that has any clinical significance.
Among the people with high blood pressure, the average reduction in systolic blood pressure in the low salt group was 4,1 mmHg compared with the placebo group. This reduction was statistically significant, but again, I would seriously doubt its ability to achieve any noticeable clinical effect. Blood pressure lowering drugs generally lower systolic blood pressure by around 15 mmHg and many people with high blood pressure take two or more; I would never bother prescribing a blood pressure lowering drug that only resulted in a 4 mmHg reduction. Diastolic blood pressure was 3,7 mmHg lower in the low salt group than in the placebo group, but that difference was not statistically significant.
So, what can we say about this? Basically, a reduction in salt intake may result in a small reduction in blood pressure, but the reduction is so small that it is unlikely to achieve any meaningful effect on longevity, which explains why the difference seen in mortality is non-existent among people with normal blood pressure, and so small and uncertain that we can’t even be sure it’s a real effect in people with high blood pressure.
One could argue that the decrease in salt intake in these trials was too small to achieve a meaningful effect on health outcomes, and I think that is a valid point. But at the same time, if you decrease salt intake by almost 20% and there is no meaningful difference in blood pressure or mortality, that suggests to me that even if there is an effect, it is so weak that it’s not worth bothering about. One problem with health advice is that people get told to do too many different things at once – better in my opinion to focus on a few things that we know have a meaningful effect.
Conclusion: The highest quality evidence available at this time does not show that people will prolong their lives by decreasing their salt intake. If you like salt, then you should eat it without worrying about your health, especially if your blood pressure is within normal limits.
You might also be interested in my article about the risks of aggressive blood pressure treatment or my article about whether statins save lives.
I take low sodium advice with a … grain of salt. Pink Himalayan preferably !
So, if there isn’t anything in it for the patient, is there perhaps something for the doctor? Do they need this kind of advice to motivate there existance? My doctor looked very sad when I didn’t take the statins, as if I stripped him of his powers. Perhaps the same would apply for salty advice… 😉
Thank you for this, Dr. Rushworth. The best resource I use for salt is “The Salt Fix,” By Dr. James DiNilcolantonio. Salt reduction is especially dangerous for athletes and others who are very active in hot weather (such as me). I eat freely of it. I also take a potassium bicarbonate (1g/day of K) supplement. Both are good for the kidneys, as I understand it.
One of the problems with looking at a singular action of an element such as Na on one health issue, in this case, hypertension, is we fail to look at the wider consequences. We risk “the treatment was a success, however, the patient died” scenario. Salt deprivation may lead to serious cellular depletion of Na and many other minerals. Na is essential for detoxification because it helps to keep toxins such as heavy metals in solution. Salt restriction may result in an increase in infection, fatigue, loss of balance, demineralisation of bones and confusion which happen to be epidemic among the elderly. I worry for the health and quality of life of all of those elderly living in care who are on blanket salt-restricted diets and most for no good reason.
For about 40 years I do not change salt intake of people with high BP. Rather I adjust medication and salt eating habits to attain normal BP, which is easy but time-consuming for the clinician. One pill of KCl retard or 12.5mg of Spironolactone (5-7 days a week) sometimes do marvels.
I’ve seen left heart failures from low body potassium (for evaluation, an ECG is better than serum K+) because of diuretics, reverting only with replenishing potassium body stores.
In a normal temperature room for observing naked people (20-21oC), I look at the veins on the dorsum of the hands when recumbent. If the are full and the blood flows instantly
[press a vein with a finger, squeeze it empty with other finger, away of a communicating vessel, and release the first finger]
it means that arterioles are wide open. If veins aren’t engorged or fill slowly, arterioles are somewhat constricted.
The skin must be of normal colour (not pale, nor cyanotic) and warm if the arterial supply is good. Otherwise, it means some arterial constriction.
People with high BP and engorged veins benefit from a very cautiously, slight, salt reduction, avoiding salty snakes or processed food.
People with high BP and empty veins benefit from an increase in salt ingestion. Which is an anathema, on these days. I always offer for the coming fortnight several free BP measurements. Then people believe me for the rest of their lives, usually long ones.
Low salt intake causes bad kidney function and it is to avoid on those on haemodialysis. This therapy can easily correct imbalances of Na+ or K+.
After any comment, I’ll resume on this theme.
Of course I always have my attention and eyes open for curable causes of high BP. Most are endocrine (including pheochromocitoma) are easy to suspect. Other ‘ring the bell’ when a patient doesn’t behaves as expected — renal artery stenosis, mild chromic renal disease, Kawasaki & all.
Thanks
Sorry by the error. Takayasu disease, not Kawasaki.
Of course I always have my attention and eyes open for curable causes of high BP.
Most are endocrine (including pheochromocitoma) and easy to suspect.
Other ‘ring the bell’ when a patient doesn’t behaves as expected — mild chronic renal disease, renal artery stenosis, Kawasaki & all.
Thanks
Looks as if my habit of salting my food before I taste it, though aggravating to the cook, is a matter of taste. I don*t have blood pressure problems relative to salt intake. Referring to sodium chloride herein, of course, I thank you or your article and the information in it. Looks as if for once, if it makes food taste better, it doesn*t have to be bad for you. i will enjoy more fritos.potato chips and salty treats now, and that after a decade of lowering taste expectations and salt intake. Thank you for this article. From beginning to end its a salt of the earth presentation.
Good analysis. But coincidentally, I recently read this study. Higher blood pressure is just one potential effect of sodium/salt. For decades, since I have low blood pressure, I’ve never worried about my sodium intake. But this has given me pause.
https://www.nature.com/articles/s41593-017-0059-z#Sec12
Dr. Rushworth,
I find your articles here reasonably well written, and your mastery of the material excellent. So please, do not take this criticism as a personal slight.
It is unfortunate that medical professionals, both physicians and researchers, do not delve into statistics more deeply than the mechanics of the calculations. It is a flaw shared by most chemists and physicists too. I would strongly encourage you to study statistics as its own discipline, specifically from the mathematician’s perspective. Discrete Structures is also a must-have complimentary topic to statistics.
If you lack the time or inclination to study statistics on its own, here’s the TL/DR: The statistical methods typically used in applied sciences are based on statistical methodology developed about 100 years ago, 40 years before the advent of the computer. There are simplifying assumptions baked into the methods in order to make them computable by hand.
To motivate why I know you would benefit from a deeper understanding of statistics: RCT is not “the only way to be sure that you’re removing all confounders from the equation”. It’s often not even the best way. That distinction goes to the “matched-pair trial”. In a matched pair trial, subjects are not randomly assigned to a group, rather they are paired with another subject based on matching age, sex, history, and habits, then one goes in the trial group and the other in the control group (chosen randomly). The underlying assumption of RCT (that important, but unenumerated, factors will be evenly distributed) still holds. The drawbacks of the matched-pair trial should be obvious: it requires substantially more information collection, and commensurately more work; and it requires an even number of similar participants. But when the limiting factor is willing participants, rather than researcher time, it ought to be the default method.
A related topic that may be of interest to you: Analysis of Variance with Repeated Measure. I mention this because the review article mentioned above does not, but ought to have, included such an analysis, as the results are likely to have been useful.
Hi Logan,
Thank you for the feedback!
Logan, I am interested in your comments about RCT vs Matched Pair trials.
Two questions, both for you or Sebastian: is there any evidence you can point to that the latter is better than a good RCT at removing confounders?
And, isn’t the N of 1 trial the best option of all?
Thanks,
https://www.drmcdougall.com/misc/2004nl/040700pucoffee.htm