I’ve written previously about how salt (sodium) only marginally raises blood pressure, and how the evidence that there is any benefit to reducing salt intake in the diet is weak at best. Now it’s time to look at that other salt, potassium. While sodium raises blood pressure, potassium supposedly lowers blood pressure, and many people have been given the recommendation to replace their table salt with salt substitutes rich in potassium. So, what does the evidence show?
A Cochrane review was published in 2013 that sought to understand if there was any benefit to increasing intake of potassium. The review was funded by, among others, the WHO, the government of Japan, and the government of South Korea. 22 randomized controlled trials, with a total of 1,606 participants, and 11 prospective cohort studies, with a total of 127,038 participants, were included. If you’re unsure about the difference between a randomized trial and a cohort study, then I recommend you read this.
The reason the review chose to include cohort studies, rather than just looking at the more reliable randomized trials (as Cochrane reviews normally do), was because the randomized trials contained too few participants and followed them for too short a time period to be able to say anything about hard outcomes like heart attacks and strokes. Thus the randomized trials were used to determine changes in blood pressure, while the cohort studies were used to look at effects on hard outcomes.
The minimum follow-up period for inclusion in the review was four weeks, which is a reasonable minimum, because any slight variation up or down that only lasts a day or two, or even a week or two, before reverting to baseline, is not going to have a meaningful health effect. Another requirement for randomized trials to be included in the review was that they measure potassium in the urine of participants. This is also a very reasonable requirement, because it gives a much more accurate measure of how much potassium people are actually consuming than just asking them what they ate (there is a very strong correlation between the amount of potassium consumed and the amount excreted in urine).
Of the 22 randomized trials, 20 provided a potassium containing supplement, while two provided dietary advice aimed at getting people to increase potassium intake. The included trials varied in duration from the minimum of four weeks to twelve months.
The cohort studies didn’t have to fulfill the requirement to measure urine potassium, since it’s not usually possible to do urine measurements on everyone in large observational studies, and having that requirement would likely have resulted in the review not being able to include any observational studies. This does however decrease the reliability of the cohort studies, since they are basing their results on how much potassium people say they eat rather than on an actual measurement.
Let’s get to the results.
We’ll start with the randomized trials. Overall, the systolic blood pressure ended up being 3.5 mmHg lower in the potassium treated group than in the control group, while diastolic blood pressure ended up being 2 mmHg lower (systolic blood pressure is the pressure in the arteries when the heart is contracting and pushing blood forward through them, while diastolic blood pressure is the pressure when the heart is relaxing).
The results were slightly better when people with high blood pressure were analyzed separately, with systolic blood pressure dropping 5.3 mmHg while diastolic dropped 3.1. However, considering that the definition of high blood pressure is a systolic blood pressure of at least 140 or a diastolic blood pressure of at least 90, the results are not very impressive even in the high blood pressure group. For people without high blood pressure to start with, there was literally no benefit to increasing potassium intake whatsoever.
What about if we break the results down by dose? Unfortunately, the results don’t get more impressive. The highest dose potassium group (taking 6 grams per day of potassium or greater) did not differ noticeably in blood pressure from the lower dose potassium groups. In fact, there was no evidence of any linear dose-response relationship whatsoever, which is a little bit weird, and suggests that even the small difference found in the review might be due to publication bias (the fact that positive studies are more likely to get published) rather than due to any real effect of potassium. If potassium really does lower blood pressure, we would expect to see a dose-response relationship, as we do for blood pressure lowering drugs. But we don’t.
This train of thought is strengthened by the fact that people with low potassium intake to start with did not benefit more than people with high intake. If the potassium really was having an effect, then you would expect to see the largest benefit in those with the lowest intake at the beginning of the trials.
People with a high sodium intake at the start of the study (more than 4 grams per day) did however appear to benefit more than those with a lower intake, with the high sodium group experiencing a 6.9 mmHg drop in systolic and 2.9 mmHg drop in diastolic blood pressure, as compared with a 2 mmHg drop in both systolic and diastolic blood pressure in those with a sodium intake below 4 grams per day. So it’s possible that an increase in dietary potassium intake is primarily beneficial in people who have a high sodium intake.
What can we conclude from the randomized trials?
Increasing potassium intake might marginally lower blood pressure in people with high blood pressure, especially if they also have a high sodium intake. However, the lack of a dose-response relationship and lack of evidence of greater benefit in people with low potassium intake to start with suggests that this result is likely spurious. For people with normal blood pressure to start with, increasing potassium intake clearly has no effect on blood pressure.
Of course, potassium might have other effects on the body unrelated to its potential effects on blood pressure. And blood pressure is only a surrogate marker, not a meaningful endpoint in itself. So let’s take a look at the cohort studies, and see if there is any benefit of increasing potassium intake on the hard endpoints that actually matter. Three hard end points were analyzed – stroke, cardiovascular disease, and coronary heart disease. The reviewers also wanted to look at overall mortality, but none of the studies provided sufficient data on this for conclusions to be drawn.
The division in to strokes, coronary heart disease, and cardiovascular disease is a bit odd, because two of the categories are actually sub-categories to the third one. Stroke is a form of cardiovascular disease, as is coronary heart disease, so taking the two sub-categories and placing them on a par with the overarching category to which they belong really makes no logical sense. It’s just confusing. No explanation is offered for this breakdown.
Anyway, here’s what the reviewers found. The relative risk of cardiovascular disease was 12% lower among those with a higher potassium intake. The relative risk of coronary heart disease was 4% lower. And the relative risk of stroke was 27% lower. Only the difference in relation to stroke was statistically significant.
So, eating more potassium results in a decreased risk of stroke, right? Well, no, probably not. Observational data are heavily influenced by confounding factors. As an example, the foods with the highest potassium content are all foods that are generally considered to be “health foods”. People who choose to eat more health foods also generally exercise more, smoke less, and have higher status jobs and better incomes. In other words, the scope for confounding is enormous. Which is why anything less than a halving or doubling of risk should automatically be discounted as the result of confounding in observational studies.
Luckily, we don’t have to rely entirely on observational data for hard outcomes any more. A randomized trial has just been published in the New England Journal of Medicine, in which 20,995 men and women who had previously suffered a stroke, or who were over the age of sixty and had a history of high blood pressure, were followed for five years.
The study was carried out in 600 villages in rural China, and it used cluster randomization. What that means is that villages were randomized to the intervention or control group rather than individuals being randomized. Individuals in the intervention group received a salt substitute containing 25% potassium and 75% sodium, and were told to use it instead of their regular salt. Individuals in the control villages were told to get on with life as they had done previously. On average, 35 individuals were recruited per village, with an average age of 65 years and an average starting blood pressure of 154/89 mmHg.
You may well have noticed already that there are two major problems with this study. First, the study was completely unblinded – both the researchers and the participants knew who was in the intervention group and who was in the control group. This will tend to produce more impressive results than you would see in a blinded study, because the participants in the intervention group will make additional lifestyle changes on top of just the intervention (after they start using the salt substitute instead of regular salt, they will start to think of themselves as people who take care of their health, and are therefore more likely to quite smoking and exercise more, to take two examples). Another reason why unblinded studies produce more impressive results is that researchers usually want a positive result (because it’s better for them from a career perspective), and will often do what they can to tilt the results towards a more favourable outcome (which they can easily do when they know who’s in which group).
It’s actually kind of unforgiveable that this study wasn’t blinded. It would have been so easy to produce two identical containers, one with regular salt, and the other with the salt substitute, and carried out a double-blind study. If you’re going to do a big study in which you follow thousands of people for years, you should do it right. Otherwise whatever results you produce will always be of questionable scientific value.
The second problem with the study is that the participants weren’t just increasing potassium. They were also decreasing sodium. That makes it hard to say whether any benefit seen is due to increasing potassium or decreasing sodium.
Let’s get to the results.
Overall, the salt substitute group ended up increasing potassium intake by about 57%, while sodium intake decreased by about 8%. In other words, potassium intake increased by a lot, while sodium intake only decreased a little, so any health benefit is more likely to be due to changes in potassium intake than changes in sodium intake.
The salt substitute group ended up reducing it’s systolic blood pressure by 3.3 mmHg more than the control group, while diastolic blood pressure decreased by 0.7 mmHg. In other words, the results are largely in line with the results from the Cochrane review, with increasing potassium intake resulting in a small reduction in blood pressure.
In terms of hard outcomes, the salt substitute was associated with a 14% reduction in relative risk of stroke. This isn’t as impressive as the 27% reduction seen in the cohort studies, but also isn’t nothing. In absolute terms, it equates to 4-5 fewer strokes per thousand person-years among people at high risk of a stroke. The salt substitute was also associated with a 12% reduction in relative risk of death. Which is on par with the effect on mortality seen in statin trials. So again, not a huge effect, but not nothing.
Can we believe these results?
This is where it becomes a problem that the study was unblinded. It’s impossible to know how much of the benefit seen is due to lack of blinding and how much is actually due to increasing potassium intake. I think it’s reasonable to think of the results of this study as best case scenarios – the benefits might be this big, but they’re probably smaller, and might be non-existent.
What can we conclude from all the evidence we’ve discussed?
Increasing intake of dietary potassium does appear to result in a small reduction in blood pressure in people who have high blood pressure. It may also decrease the risk of stroke and it might even prolong life, but the studies supporting those claims have major methodological flaws, and it’s therefore far from certain whether that is actually the case. Bananas do taste good though.
Dr. Rushworth, it would be very interesting to see your assessment of how Sweden’s unique approach to handling the pandemic stands as of September, 2021. Most media pieces slam Sweden with headlines like “Swedens death rate 10x that of neighbours” and, as a result, I think most people believe that Sweden’s approach has been a failure. My reading of the data suggests otherwise. Please consider doing an uodate.
Agreed
Looks like Finland’s covid cases are on a straight line going up and Norway’s are spiking.
Sweden’s look to be flat and covid is endemic there.
I still think once the dust settles over all these conflicting statistics, the main correlation will be a) general population health and b) early use of prophylactics. And yes, this thing will become endemic. The idea that we can make it go away now looks ridiculous to me. Unless you want everyone to live like bubble boy.
I have read for many years that there are essentially two groups of people – those sensitive to sodium intake, and those who are not. It seems the studies cited oversimplifies this starting point? Likewise, I believe that excessive sodium intake also leaches calcium from our bodies. So, high sodium intake may only raise blood pressure minimally in some individuals, but it is not necessarily a safe practice.
Respons og strøtanker efter at have læst denne artikel: Det er et kendt fænomen, at ‘alternative’ behandlinger ikke leverer store RR-resultater i controlled trials, cohort studier og andre sammenligningsstudier. Grunden til dette er, så vidt jeg kan se, at disse studier kun undersøger målelige ‘materialistiske’ end points.
Der er imidlertid andre ting her i livet end materialisme: – Godt humør, humor, kærlighed, venskab, selvtillid & selvværd samt tilfredshed med livet generelt, er altsammen ting, der giver livet værdi, og som med sikkerhed nedsætter risikoen for sygdomme, men de er sværere at undersøge i en trial, bl.a fordi de er umulige at måle pga folks forskellighed, som gør det svært at kategorisere svarene.
Tankegangen bag den materialistiske livsopfattelse er enøjet, og det kommer til udtryk i denne artikel i dens forsøg på at fastslå, om potassium har målelig ‘effekt’ på blodtrykket, og i pointen om at resultaterne i Kina er tvivlsomme, fordi studiet ikke er blindet. Men er det nu også sandt? Det er kendt, at kvinder kan synkronisere deres menstruationer imod hinanden, hvis de befinder sig i en gruppe, som de er tilfredse og trygge ved at være i. Denne observation modbeviser den materialistiske tankegang, og den sætter den naive forestilling om, at studier er nødt til at være blindede for at være troværdige, i relief.
Man kan spørge: Hvorfor er det egentlig interessant at vide, om potassium per se har effekt på blodtryk? Kan det ikke være ligegyldigt, hvis den overordnede livskvalitet forøges af potassium, eller hvis den overordnede livskvalitet får folk til at indtage potassium? Er brug af potassium et tegn? En confounder? Disse spørgsmål kan studierne slet ikke tage stilling til pga den enøjede måle- og veje tilgang i dem.
Der er mere mellem himmel og jord, siger man, og det gør man jo af en grund. Folk opfinder ikke langtidsholdbare ordsprog ud af den blå luft. Mennesker har både en bevidsthed og en ubevidsthed, og de kommunikerer også ubevidst, hvilket kan observeres ved, at både gaben og latter smitter fra-person-til-person …
Det er til gengæld aldrig bevist, at sygdom smitter person-til-person, hvilket er temmelig grotesk, set i lyset af de seneste 18 måneders begivenheder.
Hello
I recently read that foods rich in flavonoids can help reduce blood pressure
https://www.everydayhealth.com/hypertension/gut-bacteria-and-flavonoid-rich-foods-linked-to-lower-blood-pressure/
I’m a 40 year old, fit and active man with familial hypertension. Frequently over 140/90
And advice would be appreciated
Going low carb may be an effective strategy to lower blood pressure:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6695889/
“As an example, the foods with the highest potassium content are all foods that are generally considered to be “health foods”.”
Likewise, foods rich in flavonoids are also higher in potassium and magnesium, lower in sodium and calcium, which could easily account the minor differences found in these groups.
A medication to lower blood pressure will do just that, regardless whether the pressure is high, normal or already low.
While the proper level of appropriate nutrients allows the body to control the pressure and bring it to normal.
Also unlike a drug, the amount of intake is less meaningful than the level within the body. If you are not deficient, more intake will not help.
Most important is the ratio between minerals – K/Na in this case.
“That makes it hard to say whether any benefit seen is due to increasing potassium or decreasing sodium.”
Doesn’t matter. Either way you are improving the ratio.
Dear Dr Rushworth,
Exactly why is blood pressure of interest to our health? I know we can measure it, so we need to find some reason to make it relevant, but is it? Admittedly, blood pressures vary and people with various sicknesses may have co-varying blood pressures, but don’t they also have other non- measurable physiological variables that also may be important? Is blood pressure, per se, actually of interest to our health or is it just something that is easy to measure, so we force on it any pathology that may stick?
I’d like to know because I am at the evil ‘high normal’ range that used to be okay (if suss) but is now sick. I’ve twice been convinced by doctors to take blood pressure medication, but the first time I felt so shitty that I stopped after two weeks and the second time I broke out in hives and stopped immediately. I’m sure I am just being non-conformist and annoying, but I actually feel better not taking blood pressure medication that I do when I take it.
I know that feeling good is not a primary determinate of modern medicine, and I could drop dead tomorrow, but is the pressure necessary to pump y0ur blood around a disease or just a measure of how your body is coping with the slings and arrows of outrageous fortune?
Well, it’s been shown that high blood pressure increases risk of a number of different diseases, and that lowering blood pressure lowers that risk. I think lifestyle changes should be the first line intervention, and drugs should only be used if absolutely necessary.
Dr Rushworth
Causation or correlation?
E.g. Cochrane states regarding ARB for heart failure: “ARBs were no better than placebo or ACE inhibitors in reducing the risk of death, disability, or hospital admission for any reason” and “ARBs do not reduce total mortality or all‐cause hospitalisations compared with placebo in the treatment of HF patients irrespective of LVEF.”
Does realy lowering blood preasure lower the risk?
Heart failure is not the same thing as hypertension. People who take ARB’s for heart failure are not taking them because of high blood pressure. Having said that, I am surprised to hear tha ARB’s do not show benefit for heart failure. Can you provide a link to the article?
Dr Rushworth
Here is the link to Cochrane Library regarding ARBs and heart failure.
https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD003040.pub2/full?highlightAbstract=arb%7Cfailur%7Cfailure%7Cheart&cookiesEnabled
Dr Rushworth
So why would you prescribe ARB to a heart failure patient if it isn’t for hypertension? What other mechanism is anticipated?
It lowers blood pressure (not necessarily from hypertensive levels) and thereby relieves strain on the heart, because the heart is pumping against less resistance. People with heart failure often have their blood pressure pushed down to 100 systolic, i.e. far below the level you would push someone with high blood pressure.
But is “high blood pressure” in itself a disease, or is it simply the result of some underlying problem? It seems to me that taking chemicals to artificially lower the blood pressure is simply treating the symtoms instead of the problem, which would be childish.
Dr Rushworth
I noticed that the link I gave you regarding ARB and heart failure did work. Try this one instead:
https://www.cochranelibrary.com/cdsr/doi/10.1002/14651858.CD003040.pub2/full?highlightAbstract=arb%7Cfailur%7Cfailure%7Cheart
“Well, it’s been shown that high blood pressure increases risk of a number of different diseases,”
But is bad blood pressure not the result of a disease, rather than the cause of it? Seems to me that blood pressure drugs are only treating a symptom.
It’s both.
Dave, may I recommend Malcolm Kendrick’s book “Doctoring Data”? If I remember rightly, the medical fear of high BP started when it was found that people with very high BPs were terribly vulnerable to strokes.
Whether 140/90 should be a worry is the sort of thing Kendrick discusses.
I’m a bit suspicious of any threshold that is claimed to be a universal truth irrespective of demographic variables – age, sex, and race, for instance.
What’s all this hub-bubb over vitamin K2 as a prophylactic for CVD?
Sorry off topic Dr Rushworth but this may be so important that I wish to hear your opinion on. What do you think of nitazoxamide as a treatment for covid, from the following study:
https://www.medrxiv.org/content/10.1101/2021.04.19.21255441v2.full-text
Saleem:
Looking at the description of Nitazoxamide, dont you think this drug will run into the same wall of resistance as HCQ, Ivermectin, etc.? It is cheap, generic, off-patent and thus of no interest for the pharma industry.
Forget big pharma, we will need to continue using our own brains to assess the research. Knowing about nitazoxamide is just an extra weapon against the virus for those of us in the know and we can only try to advise other people on this research
Dr Rushworth,
You were one of few sensible voices that were critical of the dominant Covid Paradigm. Now, you haven’t written on the topic since July. How come? Could you direct us to other sources that aren’t tin foil hats and/or Trumpist and offer an alternative but balanced analysis of the current situation?
My guess you think yourself as an non racist open and accepting person, probably left leaning with Liberal tenandancies when the very fact of that comment shows your true colours. If a journalist or article of scientist doesn’t fit into your neat box of reality and acceptance, it does not mean that it isn’t science. Open your mind
“aren’t tin foil hats and/or Trumpist ”
Can you leave the fanatical political partisanship out of this? It spoils everything. Thanks
Very nice way of putting it.
Agreed. Can we please have one little tiny corner of the web that is not political?
Wait what, were all of these comments reacting to my innocent little request? I definitely didn’t mean to upset anyone or make a political standpoint. I might have expressed myself clumsily if it appeared that way… My intent was the opposite of political, it was precisely to express the need for non-biased, non-partisan analyses that challenge the dominant view – without risking to be easily dismissed as fanatical or conspiratorial by the holders of that same dominant view. I hope I managed to make myself better understood.
“Whether 140/90 should be a worry is the sort of thing Kendrick discusses.”
Interesting… my BP has always hovered around 140/90, and one aquaintance (an American nurse) freaked out when hearing that and insisted I NEED medication. Went to a (British) doctor who said “meh, don´t worry if you are otherwise healthy. People are different.”
Different medical schools?
Mange
“My intent was the opposite of political, it was precisely to express the need for non-biased, non-partisan analyses that challenge the dominant view ”
Well, you could have done without the “tin foil hats and/or Trumpist”, which does not exactly sound “non-biased, non-partisan”. Thanks.