Highlights: Jump to the part of this article you are most interested in with these links:
- What the Keto Experiment Was and Why I Did It
- How the Experiment Was Run
- The Keto Experiment Results
- Analysis: What Could Explain These Results?
- Summary: The Takeaways
Video Summary of the Keto Experiment
What the Keto Experiment Was and Why I Did It
I had recently started to take a green smoothie daily to increase my micronutrient intake easily. My assumption had been that it was ketogenic. However, I realized this wasn’t a very “safe” assumption.
I had engineered the green smoothie to be more ketogenic by adding fats to it and C8 MCT oil. This was part of why I had assumed it would not kick me out of ketosis or cause blood ketones to drop.
The question I had and where I thought I may have been wrong were:
- Can blending make ketogenic foods non-ketogenic?: The greens and vegetables used wouldn’t affect ketosis taken in their normal state or cooked e.g. steamed. But what if I blended them and broke them down to drink?
- Does adding fat to a green smoothie make it more ketogenic? for everyone? for me?: Adding fat to a meal slows down its digestion and helps to reduce the glucose spike from a meal’s intake. As a result it helps to make it more ketogenic. But this effect has been shown to differ from person to person. Was it relevant to me? Would it work with a green smoothie?
- Does adding C8 MCT oil offset any non-ketogenic effects of a green smoothie? and keep blood ketones stable?: The added C8 MCT oil would increase my blood ketones. But would it be sufficient to offset any reduction from the blended vegetables themselves?
I needed to answer these questions in order to know if the smoothie was helping or hindering my ketogenic diet. Then based on this, I would stop using it or continue using it – knowing if it truly was ketogenic. Rather than just guessing or hoping.
In order to find out if my modified green smoothie was interrupting my ketosis or supporting my ketogenic diet I ran a controlled experiment one day. In this experiment, I tested my blood glucose and blood ketones before, and for 2 hours after drinking the smoothie.
How the Experiment Was Run
What Was Tested
For the experiment I used a green smoothie that combined the following ingredients. All the ingredients were blended together:
- Vegetables: Celery, lettuce, kale, sweet peppers, broccoli, mint and parsley.
- Fats: Avocado oil, macadamia oil and butter.
- Supplements: C8 MCT oil, Hydrolyzed collagen powder, CalMag (Calcium Magnesium) and Stevia.
How Was the Experiment Organized to Make it Valid
Figure 1: Overview of the Keto Experiment
The experiment was designed to remove some ‘confounders’ that could interfere with accurate results. These are factors that could ‘bias’ the results so they looked more or less positive than they really would be. The two main ‘confounders’ that were accounted for were A) Eating or drinking any other food around the same time, and B) Any exercise or strenuous activity.
Both of these would interfere with the results, so from 12pm (2 hours before the experiment) I made sure to do neither. I made sure not to eat/ drink or doing anything strenuous for the rest of the experimment duration. That was through to 4pm (i.e. 2 hours after drinking the smoothie) when the last blood readings were taken.
Because I was interested in the effect over time on both glucose and ketones, I took readings every 15 minutes. In some cases, I took several readings to double-check the readings. This is necessary sometimes with blood glucose monitors, as they have some variation in the readings they provide.
Keto Experiment Results
The results were more extreme than I had expected for this experiment.
The blood glucose rose by over 30mg/dl within 15 minutes of drinking the smoothie. And the blood ketones dropped slightly by around 0.25 mmol over the experiment duration.
Figure 2: Blood Glucose and Ketone Readings for 2 Hours After Green Smoothie
For my personal goals, I aim to keep my blood ketones over 1 mmol (shown on chart by blue dotted line). Which was in fact achieved.
Another of my goals, which is usually relevant to staying in ketosis, is to keep my blood glucose under 120 mg/dl at all times – and as much as possible under 100 mg/dl. In this experiment, the blood glucose shot up to 115 mg/dl, so didn’t surpass my hard ceiling of 120 mg/dl. But with such a rapid and strong rise it was not ideal (surpassing the red dotted line on the chart).
Analysis: What Could Explain These Results?
Is Blending or Juicing Always Anti-Keto?
Greens are a mix of carbohydrates and fibres. When you blend them to create a smoothie it breaks their structures down and makes them much easier and quicker to digest. In fact, this process can convert some fibers into carbohydrates in terms of their glycemic impact.
So what? The easier carbohydrates are fibers are to digest, the quicker they get into the bloodstream. When a whole smoothie does this it can increase glucose all in one go.
This compares to the greens in their original form – unbroken down – where the gut will digest them and release the glucose slowly. And where their fiber is not broken down, and not easily converted into glucose.
The end result? Glucose is released into bloodstream slowly over many hours and as a result the increase in blood glucose is hardly noticeable if at all.
So this result is most likely driven by the ‘blending’ of the food.
The other question this raises is whether liquid foods are always ‘less ketogenic’? Maybe not anti-ketogenic, but just less ketogenic. One keto product on the market known as “Keto Chow” is basically a keto smoothie? Is this as ketogenic as a solid food form of the same ingredients?
Does Adding Fat to a Food Always Improve its Ketogenic Profile?
In research carried out in 2015 by Eran Segal’s team, they established that fats in food affect people differently in terms of how it ‘inhibits’ increases in glucose.1. (Note: I discussed this research and its implications with him in this Quantified Body podcast episode).
So it is likely that for some people (perhaps you) fat can reduce glucose increases considerably. In the extreme, some people can eat ice cream without large blood glucose spikes due to the high-fat content. The only way to know – is to test your own response.
I have run microbiome tests like DayTwo to better understand my own situation. But to be frank, they have not provided clarity on this topic. So I currently believe my response to fat – is moderate. It may inhibit blood glucose somewhat, but not as much as others. I need to do more testing win specific scenarios (e.g. ice cream because it’s a high sugar – high-fat food) to understand my fat metabolism.
I can also run an experiment with a green smoothie that does not have any fat to see if the glucose response is higher. I would need to carefully control the portions to make sure everything is identical.
So for now – this point is inconclusive. I don’t know if the fat helped.
C8 MCT Increases Ketones: Are These Ketone Results Positively Biased?
The C8 MCT oil consistently increases my blood ketones. I know this from many previous experiments where I’ve tested different doses of it.
So we know that it would increase my blood ketones. But how did it impact this green smoothie and experiment?
My assumptions are that it did the following:
- It increased my blood ketones within the first 30 minutes. This is the only explanation for the rise of around 0.25 mmol in that first half an hour.
- It likely offset some of the drop in my blood ketones. With an increase of blood glucose of 30mg/dl in that first 30 minutes, I would expect a significant drop in ketones – perhaps 1 mmol or more.
- However, only a drop of 0.25 mmol from baseline ketones was seen. So my assumption is that the C8 MCT (adding exogenous ketones) offset a lot of the drop in endogenous ketone production (the body’s own production of ketones).
In a way I wish that I had not been adding ‘fat’ and ‘C8 MCT’ to my green smoothies. This would’ve helped me understand much clearer the impact of green smoothies on my ketogenic diet (and give a much clearer signal/ and takeaways on whether to use them or not).
Accuracy of Blood Glucose Meters
In the video you’ll see that one of the glucose readings I took was 2 mmol (40mg/dl) different than expected. I had to retest, and got the more reasonable and consistent blood reading I was expecting.
This is an important point for using glucose meters. In their technical documentation, they typically specify that the variation in readings can be up to 0.6 mmol (10mg/ dl) or so. In some cases, I’ve seen quite a bit worse.
I follow a simple process to improve the accuracy of glucose meter readings (and neutralize this issue).
- I take the first reading. And then I always take a second reading immediately to compare.
- If the readings are nearly the same (e.g. less than 0.2 mmol difference), I average them and use that value.
- If the readings are very different (e.g. over 0.2 mmol difference), I take a third reading. I then average all 3 readings to get my final estimate.
- Lastly, if I get any really inconsistent readings like I did in this experiment with the value 2 mmol different to the previous one – I just discard it and take a replacement reading.
The process above has allowed me to get more consistent readings in my experiment. So we use this process in all experiments at Ketosource now to remove glucose meter reading biases.
Note: Blood ketone readings don’t have this issue. They are very consistent – so we just take one reading each time.
What Changes Did I Make After this Experiment?
I stopped drinking green smoothies after this experiment because most of the evidence was against them. I’m also a lot more careful with liquid calories and blending in general.
If I was to start drinking some new drink or liquid form of food, I would test my blood glucose and ketones now. As I definitely don’t think we can assume that they are ketogenic.
Summary: The Takeaways
- This n=1 result suggests that “green smoothies” are not ketogenic. If you are on a ketogenic diet and using any type of smoothie, you need to test your blood ketone response to it to ensure it’s not dropping your ketones significantly (and potentially kicking you out of ketosis).
- We need to follow up this experiment with two further experiments to clarify better:
- Test a green smoothie without added C8 MCT oil and fats. These additions would have likely offset the glucose increase, and helped maintain the blood ketone levels. To understand by how much more a green smoothie by itself, would increase glucose and decrease ketones – we need to test separately.
- Test the green smoothie with other people on a ketogenic diet to confirm this is not a unique result for me only. With any N=1 (one person only experiment) it is possible that the results are driven not very common characteristics of my biology. So would not be experienced by the majority of people the same way. Examples could be my personal microbiome or something unique about my metabolism driven by genetics or epigenetics.
- Zeevi D, Segal E, (2015). Personalized Nutrition by Prediction of Glycemic Responses