In November 2021, we finished our series A Chemical Hunger, where we argue that the obesity epidemic is the result of environmental contaminants, and that one of those contaminants might be lithium. We hadn’t really expected anyone to read it. But we were wrong — tens of thousands of people have now read the series, and to date the twitter thread giving an overview of the series has more than 2 million views.
In April 2022, we announced the Potato Diet Community Trial. We expected that the potato diet would be really hard to stick to and people would only lose a little weight, if any. But we were wrong — people said the potato diet was easy, enjoyable, and on average, people lost 10.6 lbs over 4 weeks.
Potatoes are really high in potassium, so we wondered if potassium could be the active ingredient causing the weight loss in the potato diet. We decided to try a self-experiment where we took small amounts of potassium salt every day, but it seemed unlikely that such tiny doses could have any effect. But we were wrong — we each lost about 5 pounds over four weeks. One of us kept going and lost 12 lbs over 60 days.
In October 2022, we announced the low-dose potassium community trial (twitter thread here). Even with the results of our self-experiment, it still seemed unlikely that such tiny doses of potassium would do anything for people on average.
Now, you are reading the post with analysis and results from that study.
TABLE OF CONTENTS
- Weight Loss
- Effects Other than Weight Loss
- Future Studies
The goal of this study is to see if the large doses of potassium found in potatoes could plausibly be the reason why people lose weight on the potato diet.
The doses of potassium in this study are small in comparison to the potato diet, only a few thousand milligrams per day. This is much less potassium than people got on the potato diet, so we don’t expect the effect to be large in any practical sense. In fact, we expect that if there is an effect at these doses, it will be quite small, probably a loss of only a few pounds on average. We are just looking to try to see if there is any effect at all.
We are studying potassium because it is a major variable from the potato diet that we can easily look at in isolation, not because we think potassium will be a great or a practical treatment for obesity on its own.
We don’t expect everyone to lose weight on this protocol, or for it to be sustainable in the long term. We just want to know if potassium could be the reason why people lose weight on the potato diet, something that we currently have almost no information about. If it looks plausible, that tells us something about why the potato diet works; and then we can consider, ok wait a minute, why would potassium cause weight loss at all? But more speculation on these points after we look at the results.
Raw data, the analysis script, and study materials are available on the OSF. The dataset is very rich and there’s a good chance that we haven’t found everything there is to find. So if you are statistically inclined, after you’ve finished reading this post we encourage you to download the data and have a look for yourself. If you find anything interesting, or even if you’re just able to confirm our findings, you should write up your analysis on your own blog and let us know about it! Science is a game, please play!
If you recreate these analyses at home, your results may be slightly different than ours because three participants asked that their data not be shared publicly.
Whether or not you like what we’ve done here, we encourage you not to take our word for it. Download the data and materials, perform your own analysis, share your criticisms, run your own study. If you think you can do a better job, maybe you are right! Show us how it’s done.
We collected variables at three points.
First, we collected demographic variables at signup. The variables we collected at this point were:
- chromosomal sex
- reported hormone profile (so we can distinguish trans participants with less ambiguity)
- age in years
- race/ethnicity (from a limited number of options)
- local postal code
- current country of residence
- whether they had done any sort of potato diet in the last year
In response to this last question, the majority told us they had not done any potato diet in the last year, but 40 told us they had done some kind of potato diet on their own, and 7 said they took part in our Potato Diet Community Trial.
After signup, we had people track a number of variables about their health and their diet (and how much potassium they were taking) over the course of the study, on a spreadsheet we provided. You can view a version of that spreadsheet here.
The main variables collected on this sheet were:
- weight (in the morning)
- potassium doses (up to four doses a day)
- variables for whether or not participants consumed meat, eggs, dairy, leafy greens, and tomato products each day (just a 1 for “ate it today” and a 0 for “didn’t eat it today”), because we suspect these foods may be high in lithium (though we’re not sure)
We also included fields for several bonus variables, which were optional but encouraged. These variables were:
- calorie intake
- waist circumference (which a couple people asked for after the potato diet)
- sodium intake
- energy, mood, and ease of study (all on 7-point scales)
- systolic and diastolic blood pressure
- total Cholesterol, as well as LDL and HDL cholesterol
- resting heart rate
- fasting blood glucose
- body temperature
- estimated hours of sleep the night prior
- sleep quality the night prior
- fidgeting (on a 1-7 scale)
- estimated minutes of exercise
- (and several fields for notes)
After we took a look at the data, we realized we had a few questions about aspects of the study that we hadn’t really measured. For example, some people mentioned that they hated the potassium while other people mentioned finding it delicious. But most people didn’t mention this aspect at all, so it would be hard to conduct any analysis related to how much people enjoyed the potassium.
So finally, on December 3rd, we sent a followup survey asking about some of these remaining questions. Five days later, there were 105 responses. We downloaded these responses and added them to the dataset.
The variables we collected at this point were:
- what potassium compound they had primarily consumed
- what form they had taken it in (e.g. salt vs. capsule vs. tablet)
- what brand of potassium they had primarily consumed
- what delivery methods they had used (e.g. in food vs. in a drink)
- change in their appetite
- how much they enjoyed the potassium at the beginning of the trial
- how much they enjoyed the potassium at the end of the trial
- whether they felt leaner or chubbier subjectively
- whether they were intentionally exercising or eating more or less during the trial
- whether they were on some other diet or routine when they started the potassium trial
- and a free-response question asking if there was anything else we should know
For more detail, see the copies of the materials available on the OSF.
As a reminder, the main study protocol was:
- Start with two doses of 330 mg potassium (1/8 tsp Nu-Salt) on the first day.
- If you feel fine, try three or four doses of 330 mg potassium (1/8 tsp Nu-Salt) on subsequent days.
- If you’re feeling fine after 4-7 days, try one dose of 660 mg potassium (1/4 tsp Nu-Salt).
- If you still feel good, keep increasing your dose by small increments. For example, if you are on two doses of 660 mg (1/4 tsp Nu-Salt) a day, you might increase that to three doses of 660 mg, or one dose of 660 mg and one dose of 1300 mg (1/2 tsp Nu-Salt). If a higher dose makes you feel bad, try returning to the dose you were on before and maintain that.
- Try slowly increasing to two doses of 1300 mg (1/2 tsp Nu-Salt) a day. Only go beyond that if you are feeling totally fine.
- You should calibrate based on your own experience — different people will have different needs and different limits. For example, we’d expect someone who weighs 300 pounds would be able to tolerate higher doses than someone who weighs 150 pounds.
- If you feel weird / bad / tired / brainfog and you can’t tell why, try:
- eating something;
- drinking some water;
- getting some sodium;
- and see if any of those help. It may be easy to end up needing food / water / salt and not notice.
- If you still feel weird, try dropping to a lower dose or taking 1-2 days off.
- If at any point you feel sick or have symptoms of hyperkalemia, stop immediately and seek medical attention.
Participants were asked to record their weight every morning, and they were asked to record data up to the weight measurement on the morning of day 29 regardless of whether they stuck to the protocol. That way even if someone found the potassium intolerable, we could still use their data.
A total of 305 people submitted the initial form.
Of those, 15 people filled out the signup form incorrectly in such a way that we couldn’t sign them up (they didn’t enter an email, didn’t enter critical data such as height, etc.). We enrolled the remaining 290 people in the study.
Of the 290 people who were enrolled, 57 never entered any data on their spreadsheet, leaving 233 people who entered at least one day of weight data.
The most common outcome in this group was to make it the full 29 days, but the majority of the 233 people who entered data on day 1 stopped entering weight data before day 29. Here’s the distribution of days completed (as measured by last weight entry) from that group:
As shown above, 104 people entered weights on both the first day and on day 29. This was the criteria we specified in advance for the group we would focus on for the main analysis. Specifically, we said:
Anyone who records data for 29 days is clearly taking the study seriously, even if they weren’t able to stick to the potassium supplements the whole time. … Based on this, our main analysis will focus on participants who provide 4 weeks of data. If you provide a weight measurement for the morning of day 1 and the morning of day 29, so we can calculate your weight before and after, and you took at least one dose of potassium, we will do our best to include you in the analysis.
5. Weight Loss
The main outcome of interest is weight change by the morning of day 29. Here’s the histogram of that variable, with a black vertical line at 0 lbs (i.e. no weight change over 29 days) and a red dashed vertical line at the mean weight change:
On average, people lost weight. The mean is -0.89 lbs, or an average loss of 0.89 pounds over 29 days. With a sample size of 104, this is significantly different from zero in a one-sample t-test, p = .014, and the 95% confidence interval for average weight change is [-1.59, -0.19] pounds.
However, this obscures the data of several people who made it to the end of the study, but who mistakenly didn’t report a measurement on day 29. If we look at the data of everyone who reported a weight on day 28, this is the histogram:
This has a mean of -0.85 lbs and a larger sample size, and is also significant, p = .016.
The same thing is true if we look at everyone’s weight at day 27 — the average weight loss is 0.86 lbs and this is significant, p = .016. The exact cutoff doesn’t matter, which indicates that the result is robust.
People who dropped out before reaching the end of the four weeks also seem to have lost weight on average. You can see that the majority of people who stopped before day 21 are below zero (the horizontal line), indicating they lost some weight over the time they spent on the trial:
In fact, if you look at the weight change from EVERYONE who reported at least two weight measurements (i.e. not including those people who only reported weight for day 1, who literally could not have seen weight change), people still lost 0.79 lbs on average. Here’s the histogram:
Because of the much larger sample size, this is still significant. In fact the p-value is quite a bit lower (p = .0002) and the 95% CI is noticeably narrower, [-1.20, -0.38] pounds.
The average weight loss here is smaller, but remember that about half of these people did not make it the full four weeks! In fact, this analysis includes 26 people who didn’t even make it 7 days.
Looking over the course of the study as a whole, it appears that people slowly lose weight over time, with no apparent changes in the trend:
Of interest here is that the 95% CI excludes zero for the first time on day 7, and that day 25 is the point of greatest average weight change.
Looking at individual trajectories is a right mess, but here’s the plot anyways:
On average it looks like people lose about 0.8 lbs over four weeks on this protocol. This isn’t much weight loss, but it’s statistically distinguishable from nothing.
But obviously some people do lose more weight, sometimes a lot more. Three people lost more than 10 lbs. It’s clear that there is a lot of variation around the small average weight loss. Can we figure out what caused any of this variation?
Well for one thing, some people did not have much weight to lose to begin with. Here’s weight change on day 29 compared to starting BMI:
As you can see, people who started with higher BMIs lost more weight. This correlation is significant, r = -0.269, p = .006, and is exactly what we would expect. People who have a BMI of 22 don’t have much weight to lose, so we should expect to see very little weight loss from them, perhaps no weight loss at all. Meanwhile people with higher starting BMIs have more to lose. It’s interesting to see that the person with the highest starting BMI also lost the most weight.
Many lean people participated in this study, and most seem to have signed up because they wanted to contribute to the research even if they were unlikely to lose weight. This isn’t an experiment, but some of them do provide a sort of baseline response. “I am happy with both weights,” said one participant, “and wasn’t expecting or hoping for a big weight loss number. I thought of myself as somewhat of a ‘control group.’”
If this were a “normal” study, and we were “normal” researchers, we probably would have restricted signups so that only people with a starting BMI of 30 or higher (technically obese) could sign up for the study.
If we had done that, here’s what the analysis would look like. Unsurprisingly, this group lost more weight on average:
The average weight loss for participants who started the trial with a BMI of 30 or above was 1.83 lbs, and again this is significant, p = .031.
Another thing that might matter is what country people are from. This is especially interesting from the perspective of the contamination hypothesis, because we suspect some countries have more contaminants than others. We tried doing a “USA vs. all other countries” analysis, but that was not significant, p = .341. There also doesn’t seem to be a clear effect of what continent people are on, but we can still plot these data:
Nothing groundbreaking here, but we do want to note that we see much less variation in Europe than in North America.
But of course, the main thing we should expect to make a difference in the results of the potassium trial is the amount of potassium!
In this study, everyone was on the same protocol, but some people took much more potassium than others. People were asked to start with two doses of 330 mg on the first day and slowly work up to two doses of 1300 mg a day, but they were asked to drop to a lower dose if a higher dose made them feel bad, and to only go beyond two doses of 1300 mg per day if they were feeling totally fine. We also asked people never to go above 1300 mg in a single dose or 5200 mg in a day.
Given this protocol, it’s natural that some people ended up on higher doses than others. Here’s the distribution of average daily doses for people who made it the full four weeks:
As you can see, there is considerable variation.
With this information, we can compare the amount of potassium people were taking to the amount of weight they lost. When we do, we see a clear relationship, where people who took more potassium lost more weight on average:
This relationship is statistically significant, r = -0.276, p = .005. This is not an experimental result, since we didn’t assign people to different doses, so we shouldn’t assume it’s causal. There are certainly alternative explanations. For example, there may be weird selection issues. People who chose to take more potassium could have been the people who were like “I feel fine, I’ll take more” or people who were like “It’s not working, I’ll take more” or people who were like “I’m losing a little bit of weight, so I’ll take more and lose more”. But this result is also consistent with what we would expect if potassium supplementation was causing the weight loss.
Let’s stop a minute and take a closer look. The regression line here is y = -0.0011x + 1.3110. Essentially what this means is that the model says that on average you would gain 1.3110 lbs if you supplemented no potassium at all for 29 days, but you lose 0.0011 lbs for every mg per day you supplement above that baseline.
For example, someone consuming 2000 mg per day would lose 2.2 lbs more than baseline; since baseline is 1.3 lbs gained, we would expect them to lose about 0.9 lbs on average over 29 days.
The potato diet gives exceptionally high doses of potassium. Sources differ on exact numbers, but the USDA says that a medium potato has about 900 mg of potassium and about 160 calories, so 2000 calories of potatoes a day would give a daily dose of about 11,000 mg potassium.
Plugging that dose into the linear equation above, the predicted weight loss on the potato diet (i.e. on a dose of 11,000 mg/day) would be:
> (-0.0011 * 11000) + 1.3110
It’s hard to get any closer than that — the observed weight loss on the potato diet was 10.6 lbs on average. That’s why we titled the report, LOSE 10.6 POUNDS in FOUR WEEKS with this ONE WEIRD TRICK Discovered by Local Slime Hive Mind! Doctors GRUDGINGLY RESPECT Them, Hope to Become Friends.
Realistically, the fact that the linear equation in this case lines up with the potato diet so well is just an amusing coincidence. The 95% confidence interval on the slope is [-0.0019 to -0.0003], so model fits for 11,000 mg/day include anything from 19.6 lbs to 2.0 lbs lost.
But you have to agree, it is amusing.
This is in fact moderate support for the idea that potassium is the only active ingredient in the potato diet. We say moderate because it’s certainly not conclusive, but it would be hard for the data to be any more consistent with that interpretation.
Another interesting comparison can be found in the relationship between weight loss and total potassium taken over the course of 29 days:
This relationship is also significant (r = -0.209, p = .033), though it’s somewhat smaller than the relationship between weight loss and daily average potassium. This may mean that taking a consistent dose is more important than the amount of potassium you take overall, though the confidence intervals of the two correlations clearly overlap, so don’t conclude too much from this difference.
Other than starting BMI and potassium dosage, we can’t really tell why some people lost more weight than others. Sex, reported hormone profile, age, ethnicity, previous experience with the potato diet — none of them seem to matter.
We asked people to report how often they ate meat, eggs, dairy, leafy greens, and tomato products, and while there are sometimes vague trends, none of these variables are ever significantly associated with weight loss. On the other hand, we should note that these were measured in a very rough fashion (just “did you eat it or didn’t you” for each day), so the variables aren’t sensitive enough to detect anything less than a very strong effect.
We also tried looking at all these variables while controlling for starting BMI and daily average dose, but there still don’t seem to be any associations with these variables and weight loss (though it’s possible we’re missing something.)
Similarly, we looked at the variables from the followup survey, but with the exception of one appetite result we will report below, we didn’t find any associations with these variables and weight loss. Even if there were relationships, we probably wouldn’t find them in these data, because there wasn’t much variation in these variables — most participants took potassium in about the same ways and (per our request) didn’t change their diet or exercise during the trial.
Ease of Weight Loss
So much for absolute weight loss. But what about relative weight loss? Were there signs that the potassium made it easier to lose weight?
Indeed there were, at least in the self-report data. Some people mentioned being surprised at how easy it was to lose weight, and some people mentioned that they were surprised they didn’t gain weight given how poorly they were eating:
(77174810) First of all – holy shit! It’s amazing how well this worked and it’s also surprising that it’s never really been studied before! Thank you for the analysis and thought that you put into this. For this trial, I basically just ate whatever I felt like, went to a football tailgate party nearly every weekend with lots of beer and foods you would not associate with dieting… and still lost nearly 10 lbs! I plan to continue on for at least another couple months so feel free to follow up later if you want to.
I have tried every diet/exercise and variation of CICO, atkins, keto, IF, etc., etc., etc. to try and lose weight. To no one’s surprise, nothing really worked for long and the weight always came back. At the end of 2020 I was over 275. It took me three months of busting my ass to lose 20 pounds and as soon as I started eating “normally” again, I slowly started putting weight back on.
(23881640) I started a quick calorie-restricted diet before the holidays (got to fit into those festive pants!), and I’m combining counting macros, counting calories, AND adding 1 tsp of potassium chloride a day to my water, and the weight is coming off. It’s making the calorie restriction much more bearable. I can tell I’m technically hungry, but adherence is so much easier doing it this way. (I lost 20 pounds before by counting macros, and that was hard.)
(60114890) Trial was very easy. Lost 5.5 lbs.
I definitely attempted to run a calorie deficit. So, this was a deliberate weight loss attempt. I’ve lost the same 5-15 lbs. maybe six times over the last 30 years. This was the first time it wasn’t really painful and didn’t require a lot of discipline. It’s also the fastest rate of weight loss I’ve experienced (1.5-2 lbs/week as opposed to 0.75-1.0 lbs/week). Very very easy. Why? Mostly appetite suppression. Historically I have been able to run 500 kcal deficits with a lot of effort. I was able to run 750-1000 kcal deficits with almost no effort. Real appetite suppression kicked in after second week, at levels of about 1800mg additional potassium. It was ridiculous—yesterday I ate 1300 kcal and burned 2600 kcal and wasn’t really hungry.
…for my purposes, I don’t really care if its placebo. My appetite was substantially suppressed. It was easy to run a 750kcal deficit. I’m going to stay on the diet until I’m at target weight of 185lbs, which would be total loss of 13.6 lbs. Feels very doable.
This wasn’t a universal experience, but we think these reports were interesting.
It seems possible that for most people, small doses of potassium aren’t enough to cause weight loss by themselves, even if they affect your appetite (see below). But they might still be helpful because they enhance other weight-loss approaches.
At this point we would like to draw your attention to the beverages known as “ketoade” and “snake juice”.
Ketoade is a term for home-made electrolyte drinks people sometimes take as part of the ketogenic diet. These almost always include potassium, usually in the form of Morton® Lite Salt™, a half-and-half mixture of potassium chloride and table salt. Since it’s all homebrew, recipes differ widely, but some people are clearly getting several thousand milligrams of potassium a day from their ketoade.
It’s possible that the keto diet works but is hard to stick to, and that ketoade has become popular because it makes weight loss on a restrictive diet much faster and easier. It’s also possible that the keto diet doesn’t cause weight loss at all, and that most successes on the keto diet actually come from people who are taking large amounts of potassium “on the side” as ketoade.
Snake juice is a term for (you guessed it) home-made electrolyte drinks people sometimes take as part of various weight loss strategies, including intermittent fasting, keto, and something called the… snake diet. As far as we can tell, no snakes were harmed in the making of this diet — it appears to refer to how snakes go a long time between meals, since it’s a weight loss strategy about going a long time between meals.
Anyways, snake juice involves drinking a concoction that gives you several thousand milligrams of potassium every day. See this helpful instructional video to learn more. It opens with a man yelling “hey FATTY, behold!” at you, so you just know it is a trustworthy and authoritative source.
In any case, most participants in the potassium trial were essentially drinking ketoade / snake juice / whatever you want to call it: potassium salt and sodium salt mixed in some beverage, often with a little bit of flavoring. And while the effect size was small, on average it seemed to cause weight loss, even without keto or fasting or anything else.
The results of this study suggest that the ketogenic diet community, and this community of “snake people”, have correctly developed a folk wisdom tradition of taking large doses of potassium to amplify their weight loss routines. If so, that is pretty wild, and it speaks well of the value of folk wisdom in solving people’s real problems.
It’s especially interesting that their theories of obesity don’t seem to point at potassium at all. These people don’t think that potassium is the active ingredient here, and they don’t have any idea why potassium might help them lose weight, but they have figured out that they should take it. That’s pretty impressive.
The inverse is true as well. The fact that internet people have settled on potassium salt as part of their folk weight loss routines supports our finding that straight potassium causes weight loss.
6. Effects Other Than Weight Loss
People mentioned a wide variety of effects, but most effects were only mentioned once or twice. One person said that the potassium made their tinnitus worse, but there doesn’t seem to be any sign of this generalizing to other participants.
We did let people report some bonus variables, but most of these variables didn’t get many responses, so we often didn’t end up with a big enough sample size to analyze. For example, only one person reported their total cholesterol on day 29, and no one reported HDL cholesterol, LDL cholesterol, or triglycerides on day 29. So we won’t be taking a closer look at any of those.
Even so, a few things did come through. Here are the effects that people mentioned more than a couple times in the self-report data, or where there were enough measurements to make taking a look worthwhile:
The most commonly mentioned effect of potassium was reduced appetite.
(36100230) I found that my appetite was dulled a bit — My mind focused on food a bit less, I snacked less between meals, and ate slightly smaller servings. I found this started to wane a little bit towards the end of the month — not entirely, but I found myself more likely to feel hungry between meals.
(58007117) Taking the potassium was very easy (with the exception of the few times I put nu-salt into pill casings and took it that way – this caused stomach pain, which I did not experience when just taking it dissolved in liquid). My overall impression is that potassium acts as a mild appetite suppressant.
(11538897) I didn’t think of food while doing the trial. At the lower doses, my hunger was affected but my appetite was not. At the higher doses, both were affected. … There was a huge difference in my general desire for food if I took the supplement in the morning before eating. If I took my first dose with food, I would be thinking about food sooner (though I wouldn’t say it was even hunger, just craving). When I took only the supplement and then went to work, it was almost always that I wouldn’t think of food until after work.
(77174810) I settled on 3 doses of ~990mg (3/8 teaspoon) a day at 0730, 1130, 1600. I felt like this kept hunger at the lowest level overall and was easy to stick with. I found that if I took the supplement when I was already hungry, I’d eat more overall. So I take the dose an hour or so before I’d normally eat a meal.
(19620767) Finished the trial. It was weird, I lost a pound the first day, then nothing for a week, then 4 more pounds, then nothing. My appetite was pretty suppressed the whole time, but due to injury and illness I wasn’t really able to exercise beyond going on walks and doing my PT, I also ate an unusually large amount of junk food for life reasons (depression, birthday cake, etc) without gaining any weight.
(18556224) The potassium didn’t magically decrease the calories I took in — I had to consciously restrict them, or have circumstances dictate that — but it did suppress my hunger, i.e. four weeks I was as hungry during the day (mostly not at all) no matter how much food I had eaten.
I haven’t decided whether the weird feeling that potassium gives me is better or worse than the hunger I’d otherwise experience, since I’ve gotten fairly good at handling that.
I haven’t noticed any cravings during the trial, which is good because that is often a problem for me — not craving things carby things, but craving certain foods I eat anyway (butter, cheese) so that I eat more calories than needed, even though I’m not really hungry for anything, just seeking pleasure.
(49045265) I did notice an reduced appetite. There was only one day during the study I was hungry.
(60114890) I definitely attempted to run a calorie deficit. So, this was a deliberate weight loss attempt. I’ve lost the same 5-15 lbs. maybe six times over the last 30 years. This was the first time it wasn’t really painful and didn’t require a lot of discipline. It’s also the fastest rate of weight loss I’ve experienced (1.5-2 lbs/week as opposed to 0.75-1.0 lbs/week). Very very easy. Why? Mostly appetite suppression. Historically I have been able to run 500 kcal deficits with a lot of effort. I was able to run 750-1000 kcal deficits with almost no effort. Real appetite suppression kicked in after second week, at levels of about 1800mg additional potassium. It was ridiculous—yesterday I ate 1300 kcal and burned 2600 kcal and wasn’t really hungry. …for my purposes, I don’t really care if its placebo. My appetite was substantially suppressed. It was easy to run a 750kcal deficit.
(06769604) My appetite was clearly suppressed, especially in the morning. The issue seemed to be that it would come roaring back in the afternoon and I’d be quite hungry.
This was true even for many people who didn’t lose weight, or who lost only negligible amounts. But it wasn’t universal, and some people explicitly mentioned that there was no change in their appetite.
We found this interesting, so we included a question about appetite changes in the followup survey. In these data, the majority of people reported no change to their appetite, but about a third reported decreased appetite, and six people reported greatly decreased appetite. Only one person reported any amount of increased appetite.
And you probably won’t be surprised to see that reduction in appetite was associated with weight loss:
When we treated this self-report measure as a continuous variable on a 1-5 scale, the relationship was significant, r = 0.295, p = .011. But you’ll also notice that many people who did not lose any weight still reported a reduced appetite, suggesting the potassium had some effect for them, just not enough to cause weight loss.
You might think that potassium caused weight loss because it reduced appetite, which caused people to eat less, which caused weight loss. That may be the case, and several people did mention that they were running a calorie deficit. But we also included a field for people to track their calories if they wanted to, and while only 22 people provided complete data, the correlation in that data is nonsignificant and pretty flat, r = -0.100, p = .659.
You’ll also notice that it’s trending in the “wrong direction”, where people who reported eating more also lost more weight.
We don’t think it’s helpful to conclude that potassium is “just an appetite suppressant”. Clearly it is an appetite suppressant, but like, um, why? Why would it do this? Everything has a mechanism. What is the mechanism for this?
We think potassium reduces appetite because it turns down your lipostat. As we said with the potato diet,
[Reduced appetite] is NOT an explanation any more than “the bullet” is a good explanation for “who killed the mayor?” Something about the potato diet lowered people’s lipostat set point, which reduced their appetite, which yes made them eat fewer calories, which was part of what led them to lose weight. Yes, “fewer kcal/day” is somewhere in the causal chain. No, it is not an explanation.
But even if we accept that potassium turns down your lipostat, you still have to ask, why does it do THAT? What is the mechanism that makes potassium turn down your lipostat’s set point? Well, more discussion in a minute.
Some people mentioned noticeable improvements to their sleep.
(24646801) Regarding sleep, in the month or two prior to the study, I had started to wake semi-regularly (5-6 nights/week & 1-2 times per night) to use the toilet. This tapered off rather quickly during the trial and with few exceptions has not returned. I don’t know enough medically to explain why this would be, but it’s definitely an improvement to my sleep, and I would continue the trial indefinitely to retain this result.
(81847724) Sleep is highly subjective but overall I think my sleep quality improved during the experiment, generally sleeping longer without waking up in the middle of the night.
(87352273) Sleep was the most pleasant surprise. I have issues with insomnia, so I tend to stay awake until 2-3 am when I get really sleepy so I don’t end up just lying awake in bed getting frustrated. With ~2000 mg of potassium as well as magnesium before bed, I found myself naturally getting sleepy and falling asleep around midnight every night without much effort or thinking about it.
We included some bonus variables about sleep in the spreadsheet, but the results are inconclusive.
Sleep quality did go up by 0.2 points, but that was not significant (p = .480).
Hours slept went down somewhat, which is interesting, but that change also was not significant (p = .296).
We should note that most people did not report either sleep variable, so the sample size in both of these cases is less than 40. It looks like potassium may improve your sleep a little and/or may help you sleep less, but this isn’t well-supported and even if there is an effect, the effect is probably small.
This is interesting given that Gwern, who is notorious for his attention to detail, did a self-experiment with potassium citrate and “confirmed large negative effects on my sleep”, with a large apparent effect (d = 1.1). Possible differences may come from the fact that Gwern was originally taking potassium in the evening rather than in the morning, and when he tested this he found a difference; was taking about 4000 mg a day, much higher than most people in this trial; and that he was taking potassium citrate, while most people on this trial were taking potassium chloride. (Also Gwern may just be built different.)
We didn’t find any effect of fidgeting (if anything, people fidgeted less over time), but there were a few self-reports of intense or manic energy.
(87352273) I had really noticeably elevated energy at first, and pretty regularly had the urge to walk or exercise just to burn off some of the nervous energy. The intensity leveled off after the first week or so, but energy overall stayed higher than usual.
(84130320) I had a huge rush of energy, like borderline hypomanic, and I ended up pulling a chest muscle doing pushups because I felt like I was 10 years younger (note to others: you are not actually 10 years younger, do not suddenly do a bunch of pushups). So that sucked.
(93059017) I had so much energy after work that I just needed to walk and I walked an extra mile home.
The participant who lost the most weight (81847724) was also notable for this report:
My mood and energy have been nothing short of fantastic. On a normal day pre-trial, I’d rate my average mood and energy levels in the 4/5 area on the 1-7 scale. Somewhere during week 2 of the trial, I really noticed how elevated I felt in my mood all day long and generally my energy levels were high regardless of the amount of sleep.
However, this increased energy did not seem to be widespread, and some people specifically mentioned not feeling any more energetic.
Looking at the self-report question we included about energy (though FWIW, a sample size of only 29), people’s energy improved by 0.54 points on a 7-point scale, but this was not significant (p = .126).
A couple people noted stimulant-like effects, and strangely, some also mentioned a kind of stimulant reduction or substitution effect.
(36100230) I felt a little more focused after taking the potassium. A few times I wanted to get some caffeine, and took potassium instead, and no longer needed the caffeine.
(72706884) My caffeine intake decreased substantially during the early part of the diet. I typically intake 100-250mg of caffeine daily. This was reduced to 30-60mg every other day during the first 2 weeks. I found supplementing with a 200mg caffeine pill helpful and used one daily during weeks 3 and 4.
(64983306) While taking potassium, I also experienced heightened concentration abilities, as if I was taking ritalin/adderall. This feeling would last for 2-3 hours after taking a dose of potassium.
We can corroborate this with our own experience. Caffeine seemed to have less of an effect for us while on the potassium, and weirdly, seems to have less of an effect still! Not sure what’s up with that.
Only seven people reported their blood pressure readings on day 29, so there wasn’t enough data to do a proper analysis.
However, most of them saw their blood pressure go down, so we figured we should go into some detail anyways.
In the seven cases that reported their BP on both day 1 and day 29, people saw their blood pressure go from:
- 120/81 to 113/77
- 114/64 to 116/63
- 121/91 to 114/78.5
- 123/90 to 123/80
- 131/78 to 130/85
- 111/75 to 99/82
- 121/78 to 126/81
On average, systolic BP went down by 2.9 points, with a maximum of 12 points down; and diastolic BP went down by 1.5 points, with a maximum of 12.5 points down.
Again, these differences are not significant. But with the very small number of people reporting BP, the sample size isn’t large enough to reach statistical significance. Most of these people also had relatively low blood pressure to begin with, so it’s not clear what kind of change you might see if you had hypertension.
People were split on the potassium. Many people found it distasteful, and some people hated it.
(50612600) this is way too disgusting to drink
unbelievable it’s sold as a food product
(79606462) it truly does taste horrible, even dissolved in 12 oz water
Unsurprisingly, many of these people chose to end the trial early, and we can’t blame them.
On the other hand…
(02689028) does liking kcl salt too much count as anything important
(84130320) My experience overall was actually very pleasant. I didn’t think the taste of the KCl was nearly as bad as advertised. To me, it tasted like salt, if salt were perishable and had spent a little bit too long in the refrigerator. Putting it in sparkling water was fairly good, I could tell it was weirdly salty (especially once I got up to 1300mg/dose) but if I just chugged a little, like half a glass, and then topped it back up it was legitimately delicious. If I did a schorle (fruit juice mixed with sparkling water) instead I could barely taste it. … when I felt really bad and backed off of the potassium per the instructions, I craved potassium. Like I really wanted to eat bananas and was like “boy I could really go for some sparkling water with KCl in it.” It was super strange.
(23578149) I went from finding Nu-Sal revolting (even mixed 2:1 with salt) to finding it pleasant.
But one thing is for sure: it really makes you pee.
(7619655) Have you ever eaten a really salty meal, like pizza or Chinese food, and then felt really thirsty afterwards? That’s how the potassium made me feel a lot of the time. It was drink, pee, drink, pee, drink, pee all day. If I didn’t keep up on the drinking, I would get parched lips and a headache. It was hard to keep this up, so I skipped a bunch of days towards the end.
(74537321) I found I had to pee a lot more often depending on how much water I was drinking. I tried to drink a lot of water throughout the day so I could get the most out of my bowel movements, but one issue was I just had to go pee a lot more. It felt like I would drink a cup of water, and then 20 mins later have to pee like I hadn’t gone all day. 🙂 I would say I had 1 to 2 liters of water per day in addition to meal time drinks (milk, juice, diet soda).
We found these self-reports interesting (also hilarious) so in the followup survey, we explicitly asked people how much they enjoyed the potassium. Because some people mentioned that their opinion of the potassium changed over time, we asked them how they felt about it at both the beginning and at the end of the trial:
In the beginning, most people found it unpleasant or disgusting (though you will notice there is still one “very delicious” rating!), but:
By the end, a majority found it either neutral or pleasant, though many people still found it super gross.
You might expect that potassium enjoyment would be related to weight loss, but we didn’t find much evidence for that. We didn’t notice any statistically significant relationships with weight loss, though looking at the plots does seem suggestive:
So it’s possible that people who enjoy potassium salts are more likely to lose weight by eating them, but if so, the effect is probably too small to detect in this study.
The lithium hypothesis is the only theory of obesity that predicts that straight potassium might help people lose weight. It’s not a very strong prediction, we simply noticed that lithium and potassium are both monovalent cations, and that they appear to have some interaction in the brain, where the lipostat is located. But other theories wouldn’t predict a relationship between potassium and weight loss at all.
We first introduced the lithium hypothesis in Part VII of our series A Chemical Hunger, expanding on the idea in Interlude G, Interlude H, and Interlude I. In Part X, the conclusion to the series, we speculated that if obesity is caused by lithium exposure, potassium might be an effective treatment:
Lithium … is an alkali metal ion that appears to affect the brain. Other alkali metal ions like sodium and potassium also play an important role in the brain, and there’s evidence that these ions may compete with each other, or at least interact, in interesting ways (see also here, here, and here). If lithium causes obesity, it may do so by messing with sodium or potassium signaling (or maybe calcium) in the brain, so changing the amount of these ions you consume, or their ratios, might help stop it.
However, the results of this study are not conclusive evidence in favor of the lithium hypothesis, and it benefits us to explore some alternative explanations.
Prosaic explanations like “potassium caused people to lose water weight” would seem to be ruled out by the fact that many people’s appetites got noticeably weaker, and the fact that some people mentioned that they had never lost weight so quickly or easily before. Same thing for placebo.
So the two classes of likely alternatives are that either it’s something confounded with the potassium dose (i.e. when you take more potassium, you also do more X), or that potassium causes weight loss for some other reason than its relationship to lithium.
A natural starting point is to consider whether obesity could be just another disease of deficiency, one you develop if you don’t get enough potassium. Scurvy is the disease that happens when you don’t get enough vitamin C, beriberi is the disease that happens when you don’t get enough vitamin B1, could obesity be as simple as a potassium deficiency?
Unfortunately we think that is not the case. Diseases of deficiency are easy to identify because they regularly crop up in situations where people eat a limited diet for a long time. Both beriberi and scurvy, for example, were common among sailors on long voyages.
Obesity does not really fit this profile. People today may not be getting enough potassium, but if obesity were a disease of deficiency, you would expect to see it showing up in historical records of cities under long sieges, sailors on long voyages, explorers in the Antarctic, and so on.
We see two distant ways to reconcile this idea, however. The first would be if potassium deficiency causes obesity, but only over the very long term. For example, maybe you only develop obesity if you eat a low-potassium diet for 10 years. This would be unusual and we think it is unlikely, but it’s consistent with the data.
The other is if obesity occurs in the rare cases when people both have a potassium deficiency AND have lots of access to calories. Sailors, explorers, and other people who tend to get diseases of deficiency usually are not eating that well in general. Maybe obesity is only triggered when you’re not getting enough potassium, but you can otherwise eat as much as you want. We think this also seems unlikely, but again, we can’t rule it out.
Hydration / Clearance
People drank a lot more water on the potassium trial because the potassium salt made them thirsty, and they had to pee a lot. People also drank a lot of water on the potato diet, for similar reasons. Is it possible that both diets cause weight loss because they encourage you to drink huge amounts of water, and that water flushes your system (or something)?
This seems pretty unlikely to us, though it is consistent with all the evidence. If someone wants to try the super-hydration community trial, where you try to drink 5 liters a day or something (don’t use that number we made it up, figure out what is actually safe), that would be fairly interesting. We don’t expect it would cause comparable weight loss, in part because we think someone would have noticed by now if staying hydrated was enough to cure obesity. But it sure would be interesting if it did!
Potassium and sodium balance each other in biological functions. To regulate the increased amount of potassium they were consuming, we encouraged people to consume more sodium as well, and they may also have naturally craved more sodium as they ate more potassium.
As a result, people on the potassium trial may have been getting more sodium than normal. For similar reasons, people on the potato diet may have been getting more sodium than normal. So one kind of weird possibility is that sodium is what’s causing the weight loss here, not potassium.
We did have this in mind from the start, so one of the bonus variables for the study was estimated daily sodium intake.
Unfortunately, out of the 233 people who entered data, only 20 people tracked their sodium, so we don’t have much evidence. But what evidence we do have doesn’t support this interpretation. People who consumed more sodium actually ended up with higher weights at the end (r = 0.101), though the relationship is not significant (p = .670).
In general we do not expect that sodium is responsible for the weight loss observed in this study, nor would we encourage anyone to try a high-sodium diet. But again, we can’t really rule it out.
Is it possible that potassium increases the clearance of something other than lithium? Just making more urine will increase the clearance of some things! Or could it treat obesity in some other way?
It seems likely, but we can’t really be much more specific than that. Potassium has approximately one zillion roles in biology, so for example if obesity is caused by anything to do with “hormone secretion and action”, which seems like a pretty broad category, potassium could potentially be a treatment. This seems like a question for someone who knows more about biochemistry than we do.
8. Future Studies
There are a number of studies that could be run to get more information. We might run some of them ourselves in the future. For now, here they are as brief sketches.
We know that one of the biggest criticisms we’re going to get on this study is about the lack of blinding and lack of a control group. Everyone in this study took potassium, and everyone knew exactly what they were taking.
Let’s imagine what a control group might look like. It’s well-established that people get heavier as they get older, so over the course of 29 days, people who do nothing should on average end up weighing slightly more by the end. We’re pretty sure that a straight control group would have lost about 0 lbs and maybe would have gained some small fraction of a pound over the course of the study — if you gain 2 lbs a year, that’s about 0.17 lbs a month.
But it’s true that people in this study were paying more attention to their weight and to their diet, and it’s possible if they were taking packets of some other white powder that wasn’t potassium, they would lose weight for some other reason. It’s possible that there’s some level of placebo.
That’s fine, because this study was never intended to be the final word. It’s the first study, not the last.
While the hierarchy of evidence is very important, a meta-analysis of multiple randomized controlled trials doesn’t just happen overnight. With this study, we’ve shown that it’s plausible that potassium by itself could lead to weight loss. There wasn’t evidence for that before.
But now that we have this evidence, it might be worth investing more time and energy in a more controlled or more complex study.
We wouldn’t want to do a straight control group where people do nothing, because that would reduce our effective sample size and it would be boring for participants. Fortunately, there are designs that can help with both problems. Here are two ideas:
First of all, we could run a crossover trial. In this case, the study would run for two months. One half of the participants would be assigned to take potassium for the first month and then take no potassium for the second month. The other half of the participants would be assigned to take no potassium for the first month, and yes potassium for the second month. This allows both groups to serve as controls without reducing our sample size.
Another idea would be to run a dose-dependent experiment. The design might look something like this: one half of the participants would be assigned to a protocol that involves them working up to a dose of 2000 mg of potassium a day. The other half of the participants would be assigned to a protocol that involves them working up to a dose of 4000 mg of potassium a day. (You could also do a dose-dependent experiment with more conditions — some people assigned to 1000 mg a day, some to 2000 mg/day, some to 3000 mg/day, etc.) If potassium is the active ingredient, you should see more weight loss in the group(s) assigned to the higher dose(s).
Comparing different doses allows us to have a control group without having to have a “no treatment” group that spends the month doing nothing. Both groups are providing valuable data, and we still control for the effect of the intervention. It isn’t blinded, but this design guards against placebo effects because it would be hard for the people in the 4000 mg/day group to arrange to lose more weight than the 2000 mg/day group.
The main issue in both cases is statistical power. You might need very large sample sizes to detect these differences, and no one should run one of these studies without conducting a very careful power analysis. But, the designs are theoretically sound.
Other Diet(s) High in Potassium
Potatoes are very high in potassium, but they are not the only food that is very high in potassium. Other foods that are very high in potassium include lima beans, swiss chard, spinach, bamboo shoots, butternut squash, kohlrabi, portabella mushrooms, white beans, bok choy, and many others (though avocado and banana are maybe overrated as sources of potassium!).
If the potato diet causes weight loss because it’s high in potassium, a non-potato diet that is high in potassium might also cause weight loss. So one thing you could do is arrange a trial of some other high-potassium diet and see if that also caused weight loss.
This isn’t a sure thing, however. Other foods do contain potassium, but it’s possible that the potassium is different in these other foods — less bioavailable, released more slowly, part of a different compound, etc. So we don’t think this would be a very strong test of the theory, because it introduces so many new variables.
In addition, we want to note that many of the items on the list of high-potassium foods are foods that we suspect might be high in lithium. In particular, there’s evidence that lithium accumulates in leafy greens, sprouts, and maybe in gourds, which matches most of the foods on the list above. If the potato diet works because it’s high in potassium AND low in lithium, these other high-potassium foods might not have any effect at all.
If we had to pick just one high-K food to test, we would probably pick coconut water. It’s a liquid, so the potassium is probably more available than average. It’s relatively high in potassium, with about 600 mg per cup. It’s easy to find and requires no preparation. And (as far as we know at least) coconut water isn’t swimming with lithium. So if people wanted to try getting 2000+ mg per day of potassium from coconut water, that would be pretty interesting.
Low-Potassium Potato Diet
In the course of designing this study, we came across a set of practices used to remove potassium from potatoes. Some people with serious kidney disease have to avoid consuming too much potassium, and these techniques were developed so they could enjoy potatoes safely. Potassium removal is usually accomplished by slicing or dicing the potatoes in small pieces to increase surface area, and then soaking (before and/or after cooking) or boiling them in water to leach out the potassium (e.g.: link, link). Some sources claim that this can remove more than 50% or even up to 70% of the potassium in potatoes.
We could test these techniques by preparing some potatoes with these methods and sending the potatoes (and the water they were soaked/boiled in, which should contain the removed potassium) to a lab for analysis. If the sliced/boiled/soaked potatoes had much less potassium than potatoes that were baked or roasted or something, that would suggest that these techniques remove potassium as advertised.
We could then use this information to do another test of the weight-loss powers of potassium, by running an experiment with a modified form of the potato diet.
One group would be assigned to eat a potato diet with potatoes prepared in a way that preserves as much potassium as possible (probably baked), and the other group would be assigned to eat a potato diet with potatoes prepared in a way that removes as much potassium as possible (probably boiled and then soaked and then fried). If the preserves-potassium group lost a lot more weight on their potato diet than the removes-potassium group, that would be further strong evidence that potassium is the main active ingredient in the potato diet.
This prediction matches the following tidbit from M’s experience with the potatoes-by-default diet, which makes it seem somewhat more more plausible: “I seemed to be able to eat much more when the potatoes were sliced/grated (e.g. Swiss rosti, Chinese tudousi) than when they were closer to whole potatoes (i.e. diced, potato wedges, etc.). I’m not sure why.”
Some people think that the potato diet works because it is a mono diet. It cuts out most other foods, so there’s very little variety, and some people (e.g. here) think that food variety is part of what makes people gain weight. But if soaking all the potassium out of potatoes made for a much smaller effect, that would mean there was a big difference in weight loss between two otherwise-identical mono diets, which would be hard for food variety to explain.
Potato Diet with Urine Test
One plausible hypothesis is that potassium helps clear lithium from your brain, and this is why it causes weight loss.
If this were the case, most of the lithium that is cleared from the brain should end up in your urine (urinary lithium seems to be a good proxy for levels in the body in general). It should be possible to test people’s urine for a while to establish a baseline, and then start them on the potato diet and see what happens. The level of potassium in their urine should increase dramatically, since there is so much potassium in potatoes. It would be interesting to see if the level of lithium in their urine increased as well.
If urinary potassium levels were correlated with weight loss, that would be more evidence that potassium is the active ingredient (though they might not be correlated, since urinary potassium levels are part of a control system). If urinary potassium levels were correlated with urinary lithium levels, that would be more evidence that potassium is forcing lithium out of your brain (or some other reservoir). And if urinary lithium levels were correlated with weight loss (or frankly, even if they just went up when you started the potato diet), that would be strong evidence in favor of the idea that lithium is the cause of the obesity epidemic.
This could be the smoking gun for the lithium hypothesis, which makes it a pretty attractive idea. The problem is that we don’t have any experience running studies with urine samples, so we’re not sure how to design this study or how to run it. We’re also not sure whether it’s possible to run it over the internet, or if we would have to get a bunch of people together in person. If you do have experience in running studies with urine samples, and you’re interested in helping, please contact us.
However, even this study might not be conclusive. It’s possible that potassium counteracts the effects of lithium but doesn’t increase the rate of clearance. For example, potassium might compete with lithium in the brain without forcing it out. It might reduce lithium absorption in the small intestine. It might keep lithium from leaching out of your bones. It might do something else. (Lithium pharmacodynamics remain poorly understood.) So while it’s plausible that potassium increases lithium clearance, we aren’t confident that’s how things work.
We ran this study because we suspected that potassium might be the active ingredient in the potato diet, that the high levels of potassium found in potatoes might be why a diet high in potatoes causes weight loss. These results support that interpretation.
The weight loss observed in this trial was small on average, but the doses of potassium were intentionally very low. There’s evidence that the relationship between weight loss and potassium consumption is dose-dependent, such that people who took larger doses lost more weight on average. Regression modeling suggests that someone who was consuming a dose of potassium equal to the amount provided by the potato diet would lose a similar amount of weight as people lost on the potato diet.
These results are not decisive. Indeed, no results ever are. However, given the small doses involved, the results could not be more strongly consistent with the potassium hypothesis.
Potassium supplementation is scientifically valuable because it’s relatively controlled. But it’s not very practical, because it’s not clear if large doses of straight potassium salt are safe for most people, and because many people find potassium salt really gross. We strongly recommend that anyone who wants to lose weight should do the potato diet instead. The potato diet gives a much higher effective dose of potassium while probably being a lot safer, and may have other benefits.
The all-potato diet is a relatively big commitment (though much easier than most people expect), so you may prefer to try the half-tato diet instead. This involves getting about 50% of your calories from potatoes and, based on the available case studies, seems to be more than 50% as effective and much less annoying. We plan to study it more soon.
If for some reason the potato diet doesn’t work, we would recommend you try to find some other way to eat a diet that’s exceptionally high in potassium.
If none of these things work for you, then you can try direct potassium supplementation, though you should consult with your doctor, definitely not do it if you have diabetes or kidney disease of any kind, and limit yourself to no more than 5000 mg a day.
We probably will not follow up on this study at 6 months and 1 year, since the average weight loss was so small. It seems unlikely that 0.89 lbs of weight loss will be statistically detectable several months later.
However, several people reported that they are planning to stay on the potassium longer-term, so we may have more results soon from the people who reach 60 days on low-dose potassium.
If you would like to be notified of future stupid studies, or if you want to keep up with our work in general, you can subscribe to the blog by email (below), or follow us on twitter.
And if you feel like reading this post has added a couple of dollars’ worth of value to your life, or if you have lost weight as the result of our research and you think it improves the quality of your life by more than one dollar a month, consider donating $1 a month on Patreon.
Thanks for going on this journey with us.
Your friendly neighborhood mad scientists,
SLIME MOLD TIME MOLD
APPENDIX A: Delivery
People overwhelmingly took potassium chloride (93.3%), overwhelmingly as a salt (92.3%), and mostly as the brand Nu-Salt (62.9%). The most popular method of delivery was to take it dissolved in water, juice, a sports drink, or some other beverage.
We didn’t detect any differences in weight loss for any of these variables, but given that almost everyone took the same kind of potassium in roughly the same way, we wouldn’t have the statistical power to detect any differences unless they were really huge. So there may be differences, but we wouldn’t expect to see any evidence for them in this data, and indeed we do not.
However, the delivery method does seem to make a difference in terms of enjoyment. Here is a sample of people’s recommendations:
(45454797) The metallic taste went away after just a few days and I found the salt to actually taste good with a little apple cider vinegar and water. Gatorade without the sugar! (and easier than pressing lemons all the time)
(40941749) I highly recommend orange fanta if you’re gonna drink your magic potion, and hash browns if you wanna eat it.
(77174810) Yes, KCl tastes gross/weird/bad. I tried a few different concentrations and mixtures with food (don’t mix with a bite of guacamole – yuck!). What I discovered was that mixing it with Simply Strawberry Lemonade makes it very palatable! I dissolved the KCl and a little sea-salt in about 1 oz of water. Then added about 4-6 oz strawberry lemonade. You could damn near sip it this way! Apple cider was the second best mixer.
(94352426) Higher concentrations were only drinkable to me in carbonated drinks made it okay to drink. For me this was the biggest limiting factor, always having to have carbonated water in home, buying it every time I went grocery shopping those bottles are a lot of extra weight.
Though there was considerable variation:
(52533228) By far, the easiest way for me to integrate it into my routine was to add it as a salt substitute in my cooking or meal prep. I could not stand adding it to drinks – the taste was usually awful and harsh. When it was added to food, the flavors mixed well in general and it was much much less noticeable.
(79332762) In terms of taking the potassium, I really disliked it. I would happily take a pill 1-2x per day, but I really dislike the taste of KCL. I tried two approaches to taking it, mustard & lemonade. With mustard it worked ok for low doses (1/8 tsp) but for larger doses it felt like too much salt hitting my stomach at once. With lemonade I don’t want to routinely drink enough lemonade that fully masks the flavor. I also really like lemonade as a treat so making it a daily routine (& making it taste bad) felt weird. I don’t really want to chug powerade/gatorade either.
APPENDIX B: Regulated Success
The body puts in a lot of effort to make sure you don’t get too much potassium. So one thing you might expect to see on this trial is that people start losing weight at first, but as their body acclimates to the extra potassium and their kidneys start filtering it out more aggressively, they stop losing weight and they maybe even gain back the weight that they lost.
Some people did mention something along these lines. For example, participant 98856740 (who submitted after Dec 1 and whose comments are therefore not in the main dataset):
I lost 6 pounds in the first week and then didn’t lose any more. In fact I bounced between that low number and about three pounds higher. During that weight loss period, I felt hot, enough to wake me up at night. I’ve heard people describe hot flashes during menopause that way. Once I got to the plateau stage, I no longer felt hot, just normal. I speculate that my metabolism was using heat to lose weight. I have no idea why it stopped. I don’t think there was anything materially different about the early days.
From the data, we’re not sure what to think. On the one hand, there are very few clear reversals. For example, the number of people who dropped 5 lbs at some point but ended up losing no more than one pound by day 29 is two, specifically these two participants:
But on the other hand, most people hit their minimum weight well before the last week of the study, suggesting that many people hit a plateau early on. Here’s the plot where we highlight each person’s day of minimum weight:
You can see that some people did hit a relatively low minimum weight early on and then never go down further from there. This may be evidence that some people hit a plateau.
APPENDIX C: Accounts of Greatest Weight Loss
Well, my time in the experiment has been shocking, to say the least of it. So obviously I’m morbidly obese so I should probably address that right away considering I’ve lost over 12 lbs during this experiment.
In January 2022, I started working with a doctor that specializes in weight loss. I was put on a low-carb, ketogenic diet 6 days per week with 1 day of free eating anything I wanted, and an exercise routine of moderate walking every other day. My starting weight was ~485 lbs. My compliance with the diet and exercise routine was 100% from January until the start of the potassium trial. My starting weight at the beginning of the trial was 476.2 lbs, so I lost approximately 9 lbs during that 9-month time frame.
I DID NOT change my diet or exercise habits during the trial to any appreciable measure. There were a couple of times I mixed up my exercise routine but mostly I stuck to the same 60 minutes on a treadmill every other day. Any changes to the exercise were noted in the sheet.
Overall I think it’s incredible that the simple change of adding potassium seems to be responsible for a sudden change in the rate at which I was able to lose weight. I will be continuing supplementing potassium going forward, this is the single most amount of progress I’ve made in weight loss in a month.
I’m going to try to think of anything I can disclose here to give context to the data.
Potassium was consumed from Nu-salt and mixed with a Gatorade zero powder that also had some potassium (both details recorded on the sheet). I didn’t have any set schedule for the potassium, I simply added it whenever I felt thirsty and acquired water (up to the dose limit for the day)
My diet was a strict ketogenic diet (under 20 grams total carbohydrates per day, gross carbs, not net) for 6 days per week and one day a week of eating anything I wanted. I do not track calories. I don’t track macros other than the number of carbohydrates consumed to stay under 20. The 20-carb limit includes the 2g carb per serving of the Gatorade zero powder I used to mix the nu-salt.
I weighed myself completely naked on an “Ideaworks JB5824 Extra Wide Talking Scale” between 8:30 and 9:00 AM every day, preferably after having a morning bowel movement. If I didn’t have one, I would still record my weight. I made a note on the sheet whether or not I had a bowel movement for that particular day.
My heart rate was tracked using an AmazFit band with the pulse check feature, typically immediately before or immediately after weighing myself in the mornings.
Sleep is highly subjective but overall I think my sleep quality improved during the experiment, generally sleeping longer without waking up in the middle of the night.
My mood and energy have been nothing short of fantastic. On a normal day pre-trial, I’d rate my average mood and energy levels in the 4/5 area on the 1-7 scale. Somewhere during week 2 of the trial, I really noticed how elevated I felt in my mood all day long and generally my energy levels were high regardless of the amount of sleep.
During the first week of the experiment I remembered to measure my waist circumference as per the CDC method but frankly, I forgot to do that, but I have included a final measurement.
A final note about compounding factors: lithium reduction
I first discovered Slime Mold Time Mold through the “A Chemical Hunger” series of blog posts, but in particular, the section covering lithium is what caught my attention for potential causes of obesity. The reason it caught my attention is I was put on lithium to treat a neurological condition that I was diagnosed with (tourette’s syndrome) when I was 7 years old, and I can positively say that was the time when I began to put on weight steadily over years and decades regardless of my diet and exercise habits. I am 36 years old and have been off lithium for over 10 years now, but the lithium article really resonated with me as a potential cause. So I’ve installed activated carbon and reverse osmosis water filtration systems on all of the water taps in my house since the first lithium post in 2021. The filters I’m using the claim to remove “over 90%” of lithium from water. (City of Cincinnati water, Cincinnati, OH)
So I don’t know how entirely relevant all that could be to the data, but all of the water that I was mixing the potassium in was also water being treated for the removal of lithium specifically (although its been approximately a year of running filtered taps and only the addition of the potassium has resulted in dramatic weight loss)
I did not participate in the potato diet trial.
Anyone that wants to supplement potassium with Nu-salt should try mixing it with the Gatorade zero powder, it almost completely covers the taste and made the trial a breeze.
One last thing, I chose to limit the amount of Nu-salt I was consuming at the 1300mg per serving mark just because I didn’t want to go through my supply of Nu-salt and Gatorade zero powder too quickly. I felt entirely fine with the amount I was consuming and believe I could have easily continued in either increasing to higher doses or adding more 1300mg doses throughout the day.
Well, I feel like I’m rambling at this point but if there are any questions please feel free to ask, in the meantime I’m going to continue supplementing with potassium.
First I just wanted to clarify that I have been following a Time Restricted Eating, or Intermittent Fasting plan since Sep 30th, prior to learning about this study. I was excited to join the study since I found your posts on Twitter talking about the potato diet that people have raved about. I’ve been eating my meals between 12pm and 6pm every day and I’m sure it has contributed significantly to my weight loss. I hope this doesn’t skew the study results too much as a result of my eating schedule.
I did focus on keeping my calories under 3000 per day with a target of 2500. I also made an effort to exercise 2 to 3 times per week of 30 mins or more. That being said, I do think the potassium helped me manage my hunger, and specifically I felt like I didn’t need to eat that much during the day to feel full.
I found the study relatively easy to do. I set reminders for each dose through out the day, as well as a reminder for recording my weight and waist measurements and used an app to track those using my smart scale and smart measuring tape, both from Renpho. I discovered that drinking each dose with straight water was the easiest and fastest way to get it down. I tried with other drinks and things, but I just knew going in that it would taste funny, and got it over with quickly each time.
Starting out I didn’t have an 1/8th teaspoon measure, so I just started with 1/4 teaspoon. Being 6’4″ and 300 lbs, I figured I could handle a larger dose to begin with. Then as a result of not paying attention to the instructions very well, I ended up going up pretty quickly in dosage the first two weeks. For side effects, it was noticeable the first few days where I felt some stomach discomfort, and general unease, but it went away after the first week. The only other side effect that I think was related to the potassium, is that I found I had to pee a lot more often depending on how much water I was drinking. I tried to drink a lot of water throughout the day so I could get the most out of my bowel movements, but one issue was I just had to go pee a lot more. It felt like I would drink a cup of water, and then 20 mins later have to pee like I hadn’t gone all day. 🙂 I would say I had 1 to 2 liters of water per day in addition to meal time drinks (milk, juice, diet soda). I’m going to continue my eating and exercise schedule, but will stop taking potassium and just record my stats each day for the next month. I’d like to really see how the weight loss was impacted by the potassium. I’ll keep updating the spreadsheet and see how things go. I’m happy to talk more about my experience or answer any questions as part of any follow-up.
My 4 weeks are done, although I intend to keep taking potassium given the moderate success I experienced. Taking the potassium was very easy (with the exception of the few times I put nu-salt into pill casings and took it that way – this caused stomach pain, which I did not experience when just taking it dissolved in liquid). My overall impression is that potassium acts as a mild appetite suppressant. Thanks for running this trial, I’m looking forward to reading about the compiled results.
Sorry for the delay- I couldn’t load the sheets properly on my phone, but I was keeping track and am just now getting the chance to fill out the last week. Please excuse the order of the train-of-thought below.
I took my last weight the morning of Thanksgiving and proceeded to eat my weight in food. I haven’t been eating fast food lately but the cravings hit me hard (probably from a combination of eating way too much, alcohol, and not supplementing for a couple of days). My plan for now is to finish up leftovers today, grab some fast food over the next couple of days, and probably restart a 30 day period on Wednesday having gained about 5 pounds in a week.
All of my supplementary data (heart rate, sleep, exercise) was from my fitbit.
It was very true that I didn’t think of food while doing the trial. At the lower doses, my hunger was affected but my appetite was not. At the higher doses, both were affected.
The biggest struggle for me was trying to keep track of my calories. I feel like it negatively impacted my trial because it did affect what I ate, even though I was supposed to eat whatever I wanted. I would eat what I wanted and feel shame/guilt for eating over X amount of calories (arbitrary number from back in my restriction days). The perhaps more interesting way it affected the trial was, once my appetite started being affected by supplementing, I would finish meals that I wouldn’t have because “I had already tracked the calories for it, I should get it,” “how can I track 1/3 of a meal,” etc. For my second attempt at the trial I will not be tracking calories, and hopefully not have the pressure of numbers to affect my eating habits. I understand that it was an optional variable anyways, but hypothetically the change in weight would reflect the appx input anyways.
I did not look into the lithium correlation at all, but if it is important- for meat markers, I only eat white meat. For egg markers, I only eat egg whites. The only thing I noticed that seemed to give me actual hunger pangs was if I drank a significant (about or more than 24 oz in a sitting) diet soda. Of course you can see in my data that alcohol also ruined a couple of days, but that didn’t actually make me feel any more hungry, just more crave-y and less likely to resist eating an entire pizza (apparently).
My work schedule is Fri/Sat nights, Sun-Tues mornings, and a random overtime on either Wed or Thur. Although my Fri shift is the same every week, there is a huge difference between that 3pm-11pm shift and my Tues 530-130 shift in terms of when and what I typically eat (and my sleep schedule).
There was a huge difference in my general desire for food if I took the supplement in the morning before eating. If I took my first dose with food, I would be thinking about food sooner (though I wouldn’t say it was every hunger, just craving). When I took only the supplement and then went to work, it was almost always that I wouldn’t think of food until after work. If I took a dose without food and then went on my walk, even if I had already eaten that day, I would feel very light-headed.
I’m happy I found out about this trial. I am generally pleased with the outcome, if not the methods I specifically used, and am more excited about starting next week with a little less restriction. I’ll still track in case the data is useful for you, but probably only the weight and doses.
[SMTM’s note: despite the comment below, this participant reported losing 8.6 lbs.]
Thanks for running this trial, it was interesting. My subjective feeling is that the potassium supplementation had no discernable effect on my brain function, hunger/diet, or weight. I’m planning to continue supplementing potassium though because my food diary shows my intake of it was very low and I’m curious whether it might have any longer term effects past just the first 4 weeks.
First of all – holy shit! It’s amazing how well this worked and it’s also surprising that it’s never really been studied before! Thank you for the analysis and thought that you put into this. For this trial, I basically just ate whatever I felt like, went to a football tailgate party nearly every weekend with lots of beer and foods you would not associate with dieting… and still lost nearly 10 lbs! I plan to continue on for at least another couple months so feel free to follow up later if you want to.
Interestingly, I was born and raised in Colorado. I lived there for my first 30 years until 2003 when we moved to the East coast and although I am a bigger person (6’5″/225 in 2003), I was never really “heavy” until maybe 2010 or so. I kept putting on weight as I aged into and past my 30s and I just followed conventional “wisdom” that it was due to getting older. Each year I would have a few extra pounds.
I have tried every diet/exercise and variation of CICO, atkins, keto, IF, etc., etc., etc. to try and lose weight. To no one’s surprise, nothing really worked for long and the weight always came back. At the end of 2020 I was over 275. It took me three months of busting my ass to lose 20 pounds and as soon as I started eating “normally” again, I slowly started putting weight back on.
Anyway, you may have just solved obesity. I hope you enjoy being billionaires. Don’t forget us little guys that did nothing but participate in your study when you are trying to decide on the color for your private jet (I think dark blue would be nice).
Notes and observations:
Yes, KCl tastes gross/weird/bad. I tried a few different concentrations and mixtures with food (don’t mix with a bite of guacamole – yuck!). What I discovered was that mixing it with Simply Strawberry Lemonade makes it very palatable! I dissolved the KCl and a little sea-salt in about 1 oz of water. Then added about 4-6 oz strawberry lemonade. You could damn near sip it this way! Apple cider was the second best mixer.
I felt thirsty a LOT of the time, especially in the first week or so. I increased my water consumption by over a quart/day for the duration of the study (still ongoing)
On the weekends, I ate poorly (nutrition wise) but still overall was eating way less than I usually did.
I only tried a 1320mg dose once. I didn’t feel great but cannot say for sure that it was that higher dose. I plan to try two higher doses/day for the second month
I settled on 3 doses of ~990mg (3/8 teaspoon) a day at 0730, 1130, 1600. I felt like this kept hunger at the lowest level overall and was easy to stick with.
I found that if I took the supplement when I was already hungry, I’d eat more overall. So I take the dose an hour or so before I’d normally eat a meal.
I’m very curious about this mechanism for weight loss. Does K+ just act as an appetite suppressant? Or is it more that the lipostat is turned down and that makes you less hungry? If lithium passes through the body fairly rapidly, how long does the effect last on the brain (if that is what is happening)? When I have cut calories in the past, it was an uphill battle to fight hunger. Presumably my lipostat was set too high so I’d be hungry and also not lose weight effectively because my body was not trying to lose weight. Hmm, might make sense… I plan to do this for at least another month if not two. It will be interesting to find out:
Could there be any detrimental long-term effects of taking this much extra K?
If I stopped the supplemental K, will I start to trend back up in weight? How hard will it be to keep the weight off?
How long does the effect last? Will I be (normal) hungry tomorrow if I stop supplements today?
I intend to experiment with the following after I hit my target:
Could I take the supplement every other day or once a week as a “maintenance” dose and keep the weight off? Or maybe just a smaller daily dose?
Looking forward to your further analysis and trial results.