There has recently been a lot of attention paid to the sugar alcohol, erythritol. Used as a zero-calorie sugar replacement, erythritol is considered as GRAS (generally recognized as safe) by both the US FDA and Europe’s EFSA. However, a study, covered by mainstream media, calls that status into question. Or does it? In this article we will dig deep into that study, and conclude with guidance on how you should think about erythritol consumption.
The study, published on Monday, February 27, 2023 in the journal Nature Medicine, has a provocative title: “The artificial sweetener erythritol and cardiovascular event risk.” Here are the details:
Discovery cohort: n=1,157 people
- More than 70% of the subjects already had coronary artery disease (CAD) and hypertension.
- Almost half already had a history of myocardial infarction
The subjects were then followed for 3 years and scored for major adverse cardiovascular events (MACE).
They controlled for multiple hypotheses but did not adjust for anything else (gender, race, age, disease status, the level of other metabolites, etc.).
From metabolomics data, it showed that the group that had MACE had a higher average level of erythritol than the MACE-free sub-population.
The authors then followed up with survival analysis. For this purpose they used the discovery cohort and two other cohorts: one American and the other European.
The goal is to see if there is a strong relationship, which would be detected if it is robust to variation in the particulars of any given cohort. In that case, we could feel that much more confident in the hypothesis.
In survival analysis they divided each cohort into quartiles according to the concentration of erythritol. In survival analysis, not only the frequency of MACE per quartile is scored, but also the time of their event. This timing is crucial because something that causes MACE in the beginning of the three year follow-up period is more concerning than something that causes the same frequency of MACE, but three years later.
This time the authors did perform the proper statistical analysis and adjusted for demographics and other obvious and typical characteristics like BMI, blood pressure, current smoking status, etc. This is crucial, because if the group with the highest concentration of erythritol was also the one with the largest number of smokers, then smoking could be the actual cause of MACE, and the correlation with erythritol levels a false positive.
After controlling for these confounders the results were consistent across the cohorts: if you are in the top-25% of erythritol consumption (the scientists measured the concentration in plasma, but this is the extrapolation), then you are more likely to have a MACE in the next 3-years. But there are a lot more considerations to make before arriving at a position on erythritol and heart health.
Crucial Considerations to Take Into Account
There are several other considerations that can be made:
1 – If you are in the bottom 75% you are completely safe (the data could not be more clear). It is exactly the same “no risk” for everyone in this group.
2 – None of the cohorts in the study were healthy populations. For example, the US population exhibited over 70% CAD and over 60% hypertension at baseline (before the study began).
3 – This is not adjusted for other metabolites, so there is the possibility that one or more metabolites are the actual cause for MACEs. Or that the effect of erythritol (whether good, bad or both) is dependent on the levels of other metabolites. Since biology is a complicated intricate system this is often what we find to be the case.
4 – In the supplementary information (which people rarely read or consider on social or news media) the authors did the same survival analysis but this time stratified by gender. This means that they further divided each cohort by gender before adjusting for everything. The consilience presented in the main text was partially robust to this. In 2 out of 3 cohorts, the females in the top-quartile (75-100%) do not display increased MACE. On the other end, the males do so, for all the cohorts. So we can conclude, that with this additional scrutiny, the findings seem to be of potential relevance only to males.
To complement the previous analysis, the authors also conducted a regression analysis. This treats the levels of erythritol as a continuous variable. This is similar to the survival analysis, but does not divide the population in man-made groups like quartiles, and instead considers the entire spectrum of blood levels as a continuum.
In all of the cohorts, the concentration of erythritol was linearly correlated with MACE even while adjusting for other factors. It’s critical to note that the relationship is linear because the authors “forced” the computer to fit a line. Considering the absolute lack of effect at all quartiles except the last, the data is most likely the shape of an exponential curve or piece-wise behavior. In other words, we suspect that the linearity is a statistical artifact/bad modeling. The authors never plotted the individual data, so we cannot scrutinize that it is the case.
A scientific journal of high-caliber like Nature Medicine, usually requires a proposed biochemical mechanism behind the observations made, to deem a finding worthy of dissemination. The authors tried to tackle this by conducting their own study on the consumption of oral artificial sweeteners on platelet aggregation. This study was not randomized, not placebo controlled, and not blinded. Therefore, it is not the gold-standard of a double-blinded randomized and controlled clinical trial. This means that it is located much lower on the hierarchy of evidence
And now, the great mystery: the authors aimed to enroll 40 people (among two groups, one given xylitol and the other erythritol), but on the manuscript there is only data for 8 people. No explanation for what happened to the other 40 is given in the manuscript.
The solution to the mystery is in the reporting summary. The authors did plan for a study of at least 10 in each group, which according to their calculations would provide the minimum statistical power to test their hypothesis. They stopped recruiting at eight, because, “Since the first 8 subjects all showed similar post prandial erythritol peak levels and time course of elimination, we did not enroll more volunteers.”
Fundamental Study Issues
With this, two violations of a good study were incurred. The study design was changed during the course of the study, and the power calculations (the statistical methods that the scientists used to calculate the sample size necessary to test their hypothesis) became invalid.
Changing study design is a large misstep in science. So much so, that pre-registration of the study, that is publishing a manuscript with all the details of the intended study to go on the record, is required.
We could not find the pre-registration for this study. The importance of pre-registration lies mainly in that it is the only way to prove that the scientists did not change their hypothesis when observing preliminary or interim results of the study in order to advance their careers (guaranteed new finding), and/or that they did not change the study to prove their hypothesis.
Regarding the statistical procedure being invalid, the mathematics are complicated but one facet might be explainable with a simple example:
Let’s image that I have a new coin and I want to test if it is a fair coin. My hypothesis is that it is not. To test my hypothesis I say that I will flip it 10 times and note down the results (head or tail). I start flipping and on my first 4 tries, I got all tails. I then tell you, “Well, I got 4 tails so it is clear that the coin is not fair. This is enough to prove my hypothesis.” Would you be convinced? A professional scientist wouldn’t.
Regardless of what the authors found, not only is the finding only based on eight people, it is also based on one of the weakest forms of clinical trials on the hierarchy of evidence, and with some violations of normal procedure. For the sake of completeness we must mention that the authors also provide in vivo evidence in mice.
Author & Journal Acknowledgements of Broader Picture
It is also worth being aware of limitations of the study acknowledged by the authors themselves:
“Another limitation of our clinical observational studies is that by design, these studies can only show association and not causation. We also recognize the possibility of unmodelled confounding (for example, diet) that may have (directly or indirectly) impacted our results by factors that are not included in our models.”
Finally, although social and traditional media tend to over-hype for clicks, respected journals such as Nature Medicine, would not allow for hyperbolic conclusions. Therefore, this is how the manuscript ends:
“In summary, the present studies suggest that trials investigating the impact of erythritol specifically, and artificial sweeteners in general, with appropriate duration of follow-up for clinically relevant outcomes, are needed. Following exposure to dietary erythritol, a prolonged period of potentially heightened thrombotic risk may occur. This is of concern given that the very subjects for whom artificial sweeteners are marketed (patients with diabetes, obesity, history of CVD and impaired kidney function) are those typically at higher risk for future CVD events.”
So, is erythritol good or bad for me?
When faced with some evidence pointing in one direction it is alway necessary to analyze the pre-existing literature and weigh the quality of all the available evidence, in order to create a framework of interpretation. Even as described in the manuscript by the authors themselves, which let’s not forget are experts in this area, there is currently literature demonstrating that artificial sweeteners, including erythritol, are beneficial or have no effect, good or bad.
For example, erythritol has also been found to exhibit health benefits, such as:
- Reduces blood glucose levels and increases glucose tolerance (R, R). High blood glucose can lead to many health issues.
- Weight management and reduction of insulin secretion.
- May have antioxidant properties that counter free radicals (R).
- Reduces small intestine inflammation (R).
- Improved endothelial health in diabetes (R, R).
- Oral health benefits (R).
Therefore, without clear trend, here is the NOVOS working framework based on this study:
The studies show that the vast majority of people should not be concerned about their erythritol intake. For those who have cardiovascular health issues, are male, and are in the top 25% of erythritol intake, we suggest you reduce your intake; there is no evidence that complete removal for this small subset of the population is necessary.
Do you have a cardiovascular disease or is your MD worried about your cardiovascular health? If not, you have nothing to worry about.
Are you a male? If not, you likely have nothing to worry about.
Is your diet *consistently* very high in erythritol (a lot of zero-calorie and keto-friendly food)? For this subset of the population, we recommend that you consider reducing your intake to be in the middle 50% or lower.
NOVOS will keep an eye out for you on future large-scale, well-designed erythritol studies to provide you with further guidance on the topic.