Disappointing Outcome With Bitter Melon Protocol

This is an N of 1 study on a specific supplement that reportedly has the outcome of lowered blood glucose levels. In June of 2015 I went on a ketogenic diet. I was enjoying the benefits of weight loss and increased alertness, but I noticed my blood glucose was creeping up in the daily readings. I wanted to see if I could turn that around and bring the readings lower.

I heard on the Ben Greenfield Fitness podcast that Bitter Melon lowers blood glucose levels. I purchased the supplement and decided to test it. The results were surprising, but not in a fun way.

suprised scientist

My Question

Is Bitter Melon effective for in lowering my blood glucose and should I continue to take it?

What I Did

I took Bitter Melon on a randomized schedule for 20 days, then took it every day for another 14 days. As I did so, I recorded my glucose levels after waking, at 10am, and 3pm each day.

How I Did It

I used a Google Spreadsheet to generate a random “one” or “zero” for each day in the 20 day randomized period. This gave me instructions on whether or not to take the Bitter Melon. After 20 days, I took it every day for 13 days.

The daily dose was two tablets, which is 900mg of the supplement. I excluded readings on days where food intake was outside of a normal range, or something extraordinary was happening like a flight or no sleep.

At the end of the period, I looked at the difference on the randomized days using a TTest. I also looked at the effect on my glucose for the period where I was taking the supplement every day.

What I Learned

I had a disappointing outcome with Bitter Melon protocol. There was no change in the continued climb in my blood glucose levels. This graph shows the period from the beginning of the diet to the end of the Bitter Melon test:

Morning Reads

The period of taking Bitter Melon was from 8/9/2015 onward, so for me it appeared to be ineffective in arresting the upward movement of the glucose levels.

The first measure I looked at was the morning readings based on having taken the supplement the day prior. There was no effect: p=.84. Recall that to have demonstrated a significant difference in the two data sets we are looking  for p<.05.

The second measure was a 10am reading after having taken the pills at around 7am. Doing the TTest comparing the days where I had taken the supplement vs. not there was no effect: p=.25.

The final measure I compared was a measure at 3PM, comparing the days where I had ingested Bitter Melon vs. not. The test showed a significant difference (P=.057), but the result was reversed from what I would expect. The average glucose reading on the days I took Bitter Melon was higher than days I did not take the pills. That is an odd outcome.

Finally, looking at the effect of taking the supplement daily there was no significant change in the morning glucose level taking an intermittent dose versus a daily dose (p=.17).

image (7)

As a result of this analysis I will no longer take Bitter Melon and save myself the money. And the data opens up another line of research. Something is driving my glucose levels upward. I am tracking my food intake and my carbohydrates are very low. I am not eating any processed foods. The next step is to eliminate all supplements and see if the source is there as well as up my fiber levels. One thing is for sure. A “one size fits all” ketogenic diet ripped straight from the podcasts is not the cure all for my physiology. Nor is it probably for yours either.


Sign up for the QuantXLaFont Newsletter
Get our lifestyle tips and studies delivered to your inbox.
Thank you! We don't spam :)

13 thoughts on “Disappointing Outcome With Bitter Melon Protocol

  1. You don’t mention when you took the bitter melon. With breakfast and after the first measurement?

    Also, the graph would be much more informative if datapoints were colored by melon status. Given the size of the swings, there could easily be a substantial rise or fall caused by the treatment which is masked by the variability – at n=20, you’re not well-powered to detect anything.

    > The average glucose reading on the days I took Bitter Melon was higher than days I did not take the pills. That is an odd outcome.

    I don’t think this is odd. You say that the last part of your protocol was “After 20 days, I took it every day for 13 days.”; since there is a very clear and steep increase in blood glucose (actually, the increase is *so* steep that I find it suspicious all on its own – if you were really having 70mg/ml readings at the beginning and ended at 130mg/ml, how were you still alive and functional at all timepoints, or still alive now?), and you weighted ‘melon days’ towards the end, then the average melon day will indeed have a higher average blood glucose reading. Because those 13 days were not randomized, they are not balancing confounders across the intervention & control data (such as, say, a temporal trend) and the assumptions of the t-test are clearly violated.

    A quick fix for that is to throw out the non-randomized data and keep using the t-test. A better fix would be a linear model with day-index 1:n as a predictor. (So something like `lm(BloodGlucose ~ Melon + TimeOfDay + Day)`.) Even better would be a multilevel or latent-variable model to deal with the substantial error in each blood glucose reading – they’re generally only rated for ±5 accuracy and can easily yield less in practice, and that’s enough that it could easily mask an effect.

    If you want to post the full dataset, I could try analyzing it for you.

    1. Hi Gwern, great to hear from you again. Let me see if I can answer a couple of your questions.

      1. I always took the Bitter Melon at 6:45 – 7:15am with breakfast.
      2. When I exclude the last 13 days of non-randomized data and run a TTest on 21 data points where I was randomizing the intake for the 3PM reading, I get a higher level of glucose when I took the supplement (avg 108) vs not (avg 98) with p=.06. I think that is odd as the Bitter Melon is supposed to decrease the amount of glucose.
      3. The actual increase from 70 to 110 was from the beginning of ketogenic diet so more like 90 days on that chart. The range of the Bitter Melon period was high 90s to mid 110’s with only two high readings above 115.

      One potential confounder is exercise. I would often exercise before breakfast. And I always agree that having few data points runs the risk of underpowered analysis. I try to balance the strength of the statistics with pragmatic “no go” decisions on supplements and other mechanisms.

      I have sent you an invite to your email sharing the data sheet. Look forward to learning from your comments.


  2. This is awesome Paul and the first thought I had when I saw this was how may exercise be impacting this. If you are increasing your exercise effort over this period of time, perhaps you’re also increasing your level of food consumption that would cause increases in glucose at a higher rate. On the flip side, perhaps you were exercising less and still maintaining a certain food consumption level that would increase your glucose over time.

    The trend also shows your glucose level increasing before the test. So what’s been happening to cause that? Also I can’t tell if the chart is the morning reading or an average or what…and I think glucose level can vary a lot over a day depending on food consumption and exercise, so that info would be helpful to understand what’s happening.

    Even more awesome gwern has popped in here to evaluate! Hope you both share what’s learned…

    1. Thanks Samir, good to hear from you. Anecdotally I cut all supplements and have seen an immediate drop in glucose level, so there is potentially something in the combination of diet and supplements that was driving the rise. I’ll keep reporting back on the progress. Right now I am split testing carbohydrate levels, then will reintroduce different supplements to see what the culprit was. I too am looking forward to learning more. Cheers

    1. Ben, thanks for the question. You make a good point. The brand I used was Vitamin World. I should have been clearer about the brand as the each have different efficacies. I will have a look at mpx100 once I work through eliminating and reintroducing a few other supplements. Cheers

  3. When you aren’t fully adapted only mild exercise is recommended as it can drive you into gluconeogenesis. Also how is your sleep and stress levels?

  4. I had the same results when I bought bitter melon tablets from a healt food store (don’t remember the brand). My A1C was originally measured at 6.7 and remained the same after taking one pill at dinner for a 2 months. I’ve decided to transition to the actual fruit. I have just started eating one per day, and am anxious to see the results when my A1C is measured again in mid-January.

    The fruit tastes lousy, but is tolerable. However, at $1.59 per pound it is more cost effective than the pills at $15 for a bottle of 30. I’ll keep you posted on my results when I have my glucose levels retested in January. Maybe, I can share some good news (hopefully).

      1. Keith,

        Thanks for confirming my experience. I have been eating the fruit for the last few months. It isn’t so bad, certainly not as horrible as some have stated. I’m due for a complete physical on March 28th, and will post the results then. I won’t be doing least squares regression curves like the scientists on this thread:)
        The only somewhat negative comment I have heard, is that bitter melon can make one hypoglycemic. If that’s the case, I’ll balance it out with banana splits!

Leave a Reply

Your email address will not be published. Required fields are marked *