CONTINUOUS GLUCOSE MONITORING FOR NON-DIABETICS: A SCIENTIFICALLY CRITICAL FAQ
Q. Don’t traditional clinical tests miss the glucose variability that can only be detected by CGM?
A. True. However, we do not know whether or how this glucose variability detected exclusively by CGM predicts outcomes, because this has not been studied. We only know that this variability not detected by other tests exists. But we don’t know what to do with that information. If we were to guess what to do with that information, we might in fact get things wrong, and CGM could end up being less predictive of disease than traditional clinical measures of glucose metabolism, making CGM worse than useless, but harmful.
Q. Doesn’t high blood glucose predict heart disease, cancer, Alzheimer’s disease, etc.? So, this is a reason for non-diabetics to use continuous glucose monitoring (CGM), right?
A. Yes, high blood glucose as measured by traditional clinical tests, such as fasting blood glucose levels and HbAa1c, predicts the development of these diseases. These tests have been validated as important predictors of these diseases. CGM data has not. Because CGM data has not been validated for this purpose, we do not know exactly how we should use CGM data to do what these other tests are doing. What are the appropriate cutoffs for disease risk prediction to use with CGM, and will they necessarily correspond to the equivalent cutoffs using HbA1c? We do not know.
Q. Doesn’t high blood glucose variability predict heart disease, cancer, Alzheimer’s disease, etc.?
A. Yes, high blood glucose variability as measured by traditional clinical tests, such as the oral glucose tolerance test (OGTT), predicts the development of these diseases. OGTT has been validated as an important predictor of these diseases. (OGTT entails drinking a load of glucose and measuring blood glucose levels two hours later.) CGM data has not. Because CGM data has not been validated for this purpose, we do not know exactly how we should use OGTT to do what these other tests are doing. For example, the OGTT measures glucose levels two hours after a standardized glucose load. Above 140 mg/dL indicates prediabetes. Yet many people will interpret any glucose spike above 140 mg/dL at any time as harmful. Yet a blood glucose level of 140 mg/dL two hours after a glucose load means something drastically different from a blood glucose level of 140 mg/dL immediately after a glucose load. We do not know the relevance to disease of a blood glucose level 140 mg/dL immediately after a glucose load. It hasn’t been studied yet. It may be totally harmless, while a blood glucose level of 140 mg/dL may indicate serious pathology. Both may be true. We don’t know.
Q. OK, but although we don’t know exactly what the cutoffs should be for blood glucose spikes, doesn’t simply reducing blood glucose provide a benefit?
A. We do not have any good reason for believing this is the case. There are two reasons for this. First, we do not know why blood glucose spikes as recorded by OGTT predict outcomes. Is it because of the blood glucose itself or the underlying pathology that causes blood glucose to remain high two hours after a glucose load? How much does the glucose contribute, and how much do the other dysregulated processes in the body that are dysregulated at the same time as blood glucose contribute? This question has not been answered. The second reason we do not have any good reason for believing that simply reducing blood glucose spikes is healthful is that after-meal triglycerides increase when replacing carbohydrates with fat. This means that simply reducing blood glucose spikes by replacing carbohydrates with fat, while effective, may increase triglycerides in the blood while lowering glucose. Like glucose, after-meal triglycerides have been shown to predict heart disease. Do we therefore want to reduce glucose by increasing triglycerides? It is not clear that this is beneficial and might be harmful.
Q. Wait, I thought that CGM simply helped people to eat in a healthier way? Lowering carbohydrate is not necessary right?
A. In a trial comparing an ultraprocessed to an unprocessed diet, there was no difference in either the average blood glucose or variability in glucose between the diets, despite a 25% higher carbohydrate intake. This shows that degree of processing does not affect CGM, and neither do substantial differences in carbohydrate intake. In fact, only in another trial comparing a very-low-fat to a ketogenic diet do we see any substantial difference in average blood glucose and glucose variability, and these differences are still modest. In other words, on average, only large reductions in glycemic load “improve” glucose readings on CGM, not “healthier eating”. Therefore the only real dietary interventions that can be expected to meaningfully lower glucose excursions on CGM are a substantially lower carbohydrate diet, weight loss, and additional physical activity. Ironically, since we already know this from clinical trials, we don’t need CGM to implement these changes.
Q. But don’t responses to each particular food differ from person to person?
A. This is true. And perhaps some CGM can provide some insight into this difference. As above, there is however no good evidence that knowing these differences will provide any meaningful health benefit. It is not even clear that such knowledge will meaningfully impact average blood glucose or glucose variability. This has never been demonstrated in any study, and such hypotheses notoriously commonly fail in clinical trials.
Q. Doesn’t reducing glucose spikes reduce hunger?
A. In a recent study, after-meal glucose dips correlated with hunger with an r=0.16. This is almost random noise. While statistically significant, it is clear that after-meal hunger is overwhelmingly related to factors other than blood glucose. “Hacking” blood glucose to reduce after-meal hunger, in other words, is like doing physical exercise to become a better chess player. Physical exercise may help to maintain overall physical and mental health and have some positive effect on chess performance, but the main determinant of chess performance will be playing chess. Using CGM for hunger is not “playing chess”.
Q. But I feel that CGM benefits me personally. Are you saying I cannot use CGM?
A. Nobody is saying that you cannot use CGM if you feel it benefits you personally. Go for it. Just know that there is no evidence of benefit and that perception of benefit can be distorted or inaccurate, or due to nonspecific effects of intervention or placebo, as is often the case for personal health anecdotes.
Q. What harm could there be from using CGM as a nondiabetic?
A. There are several potential harms of using CGM as a nondiabetic.
- Wasted time and money
- Prioritizing lower glycemia over higher-value lifestyle changes
- Prioritizing lower glycemia over healthier diet
- Increasing CVD risk via after-meal trigs (or other mechanism specific to the new diet)
- Risk of compulsive, disordered eating behavior
The first is obvious. If there is no benefit, one risks time and money that one will never get back.
The second is about opportunity costs. How much time could you spend exercising or sleeping or with friends that you spent analyzing glucose data and concerning yourself with it?
The third is the classic “replacing an apple with bacon.” Replacing an apple with bacon will provide better glycemic control. However, to know not to do that, one must know that apples are healthier than bacon. However, if one believes that controlling blood glucose is paramount, one might indeed replace the apple with bacon. Thus, the healthfulness of CGM may depend more on pre-existing nutrition knowledge and less on whether or not one reduces blood glucose. But without this knowledge, CGM can end up pointing dieters in harmful directions, sacrificing health to achieve lower glucose readings—similar to how nutritional illiteracy with low-carb diets can undermine the healthfulness of that LC diet.
The fourth has already been discussed previously in this FAQ. But to recap, after-meal triglycerides, like glucose, are also associated with worse health outcomes. And after-meal triglycerides rise on a higher-fat, lower-carbohydrate diet.
And finally, for the fifth: Every dietary intervention, risk of developing compulsive, disordered eating behavior always exists. If CGM for nondiabetics is not beneficial, then many people will be exposed to risk and some harmed—for no benefit.