CGMs and Personalised Nutrition. Why Your Blood Sugar May Respond Differently to the Same Meal.
- Alastair Hunt
- 4 hours ago
- 8 min read
Continuous glucose monitors (CGMs) have become increasingly popular among people without diabetes who want to understand how food affects their blood sugar. The idea behind personalised or “precision” nutrition is appealing: measure your response to different foods, identify the meals that produce the smallest glucose rises, and use that information to tailor your diet. But research from the US National Institutes of Health suggests there is an important complication, your glucose response to the same meal may be surprisingly different from one occasion to another.
The study, Hengist et al, published in The American Journal of Clinical Nutrition, examined 30 adults without diabetes who ate duplicate meals approximately one week apart while staying in a highly controlled inpatient research environment. Researchers analysed 1,189 glucose responses using two different CGM systems.
Their central finding was striking: responses to identical meals were not reliably reproduced, and the amount of variation was about as large as the variation seen when people ate different meals.
As ever, please talk to your doctor or medical practitioner most familiar with your medical history before implementing any changes in diet, exercise or lifestyle, especially if you are under treatment. Links to all studies at bottom of page.
The promise and problem of personalised nutrition
Personalised nutrition assumes that people respond differently to food. That part is well established. Two people can eat the same meal and experience different post-meal glucose patterns.
The more important question is whether an individual person's response is sufficiently consistent to make useful predictions. Imagine that a CGM suggests a particular lunch produces a relatively low glucose response for you. If that result is reliable, you might reasonably conclude that this meal is a good choice for you. But if eating the identical lunch on another day produces a substantially different response, using one measurement to rank foods becomes much less useful.
The researchers therefore wanted to test a fundamental assumption of precision nutrition:
Does the same person respond similarly when presented with the same meal on separate occasions?
What the researchers did
The researchers analysed data from two previous inpatient feeding studies conducted at the NIH Clinical Center in Bethesda, Maryland. Participants were adults aged 18 to 50 without diabetes. They followed different dietary patterns, including minimally processed plant-based, very-low-carbohydrate ketogenic, minimally processed and ultra-processed diets.
Meals were provided according to seven-day rotating menus. The same meals were therefore presented again during the following week, allowing researchers to compare glucose responses to duplicate meals.
Participants wore either an Abbott FreeStyle Libre Pro or Dexcom G4 Platinum CGM. The researchers then calculated the incremental area under the glucose curve during the two hours after eating. In simple terms, this provided an overall measure of how much glucose rose above the person's starting level during the post-meal period.
Importantly, this was not a typical free-living experiment. Participants were living in an inpatient research environment, where meals and much of their routine were controlled. That makes the findings particularly interesting because substantial variability remained even when many potential sources of variation were reduced.
The same meal did not always produce the same response
The researchers found only weak-to-moderate relationships between the glucose responses to duplicate meals. The repeatability was particularly low according to the statistical measure used to assess whether measurements within the same person were consistent. The study reported reliability values of 0.28 with the Abbott CGM and 0.17 with Dexcom, both considered poor reliability.
The researchers also found wide differences between repeated measurements. Although the average bias between the two measurements was small, individual responses could differ considerably.
Perhaps the most revealing finding was that the variability after eating duplicate meals was similar to the variability observed when participants ate different meals. For Abbott, the average standard deviation was approximately 10.1 mg/dL for duplicate meals, compared with 11.7 mg/dL and 10.6 mg/dL for different meals during the first and second weeks.
With Dexcom, the corresponding figures were approximately 11.2 mg/dL for duplicate meals and 10.9 and 11.0 mg/dL for different meals. In other words, repeating the meal did not dramatically reduce the amount of glucose-response variation.
Why might this happen?
The researchers investigated several possible explanation:
One important factor was the glucose level a person had before eating. Differences in baseline glucose between two occasions were associated with differences in the subsequent glucose response.
Carbohydrate intake also contributed. Differences in the amount of carbohydrate consumed between duplicate meals were associated with differences in glucose responses.
Exercise appeared to matter as well. In particular, light-to-moderate exercise during the two hours after eating was associated with lower glucose responses in the Dexcom data.
Other factors included snack consumption and, in some analyses, the amount of time taken to eat the meal.
However, these factors did not explain most of the variation. The statistical models accounted for less than 33% of the differences in glucose responses. That leaves a substantial amount of unexplained variability. The researchers suggest several possibilities, including the order in which foods were eaten, physical activity that was not fully captured, sleep quality and other day-to-day physiological differences. There may also be technical variability associated with CGMs themselves.
Your CGM isn't necessarily telling you the whole story
The researchers compared CGM measurements with conventional venous blood glucose measurements during standardised glucose and mixed-meal tolerance tests.
The average glucose responses measured by the two approaches were broadly similar, suggesting that CGMs can provide useful information about post-meal glucose patterns.
However, the researchers also found evidence that CGM measurement imprecision could contribute to some of the variability observed between duplicate meals. This is important because it changes how we should interpret a single CGM reading.
A noticeable glucose rise after a particular meal does not necessarily mean that meal will produce exactly the same response every time you eat it. Equally, a relatively modest response on one occasion does not guarantee the same outcome on another day.
The researchers also point out that even laboratory measurements of blood glucose do not produce perfectly reproducible responses to repeated meals. The issue may therefore be broader than CGM technology alone.
What this means for meal rankings
One of the most interesting implications concerns the practice of ranking foods according to their glucose response. The researchers looked at meals that appeared to produce particularly low or high responses during the first week. When those same meals were eaten again, their responses moved closer towards the average. This is an example of what statisticians call regression towards the mean: unusually high or low measurements tend to be less extreme when repeated.
For someone using a CGM, this could create an easy trap. You might eat a meal once and see an unusually small glucose response, concluding that it is an especially good choice. You might then eat the same meal again and see a substantially larger response. That doesn't necessarily mean your metabolism has suddenly changed or that the CGM has “failed”. It may simply demonstrate that glucose responses naturally fluctuate.
Practical recommendations
The study does not suggest that CGMs are useless. Rather, it raises questions about how their data should be interpreted, particularly when making detailed dietary recommendations.
A sensible approach is to avoid treating one glucose reading as a definitive verdict on a particular food or meal.
Instead:
Look for patterns rather than isolated readings. A single response may not represent your usual response.
Consider the circumstances. Meal size, carbohydrate intake, physical activity, sleep and your glucose level before eating can all potentially influence the result.
Be cautious about rigid food rankings. A meal that produces a low response once may not always do so.
Prioritise overall dietary quality. This study does not overturn established principles around balanced, nutritious eating.
Remember the study population. These participants did not have diabetes, so the findings should not automatically be applied to people with diabetes or other metabolic conditions.
Discuss concerning glucose readings with a healthcare professional. Consumer CGM data should not be used on its own to diagnose a medical condition or determine treatment.
For people interested in personalised nutrition, the study's most important message may be that more measurements are likely to be better than fewer. The researchers specifically concluded that two measurements were insufficient to reliably estimate an individual's response to repeated meals, even under controlled inpatient conditions.
Exactly how many repeated measurements are needed remains an important question for future research.
Final Thoughts
Personalised nutrition remains an intriguing area of research, but this study provides a useful reality check. Knowing how your body responds to food can potentially provide interesting information, yet a single glucose measurement may not tell the complete story. The researchers found that people without diabetes could show substantial differences in their glucose response when eating the same meal on separate occasions and that this variability was surprisingly similar to the variation seen between different meals.
That does not mean you should ignore your body's responses to food. Instead, it suggests that those responses should be interpreted in context and, where personalised decisions are being made, based on repeated observations rather than one-off results.
Ultimately, the study shifts the focus away from finding the “perfect” meal from a single glucose reading and towards understanding the complexity of human metabolism. Personalised nutrition may still have considerable potential, but reliable personal recommendations are likely to require more comprehensive data than a handful of CGM measurements.
As always, the best health strategy is one you can stick with - one that fits your personal lifestyle profile. Improving health is about finding motivation, prioritising self-care and taking action. If you want to take effective and targeted steps to that fit into your unique lifestyle, The Whole Health Practice is here to help. Whether you want to improve eating practices, beat chronic illness or enhance your overall well-being, our consultations and programs deliver results that are tailored to your needs.
FOUNDATIONS - our premier whole person health consult - identifies and prioritises the key factors, known and unknown, that affect health and wellbeing. It provides targeted recommendations tailored to you, the individual, and your unique lifestyle.
Stay Healthy,
Alastair
Another CGM focused article - reviewing more studies - can be found here.
Join us, or scroll down to contact us and learn more about our services:
Achieve your Health Goals
Our health, physical – mental – social - is complex and affected by multiple factors within and outside of your control. Our consults and programmes address the whole person, the root causes of ill health and maximising your health, performance & vitality.
Take the first step. Contact us to arrange an introductory call, to discuss how we can support your journey to health. We are based in Singapore and work with clients globally.
Book a Whole Health Consult to assess, identify and prioritise key factors (known and unknown) that affect your health. And receive personalised recommendations on how to address them.
Want to put recommendations into action? Learn more about our programmes for individuals or teams.
Studies and Resources
Hengist A, Ong JA, McNeel K, Guo J, Hall KD. Imprecision nutrition? Intraindividual variability of glucose responses to duplicate presented meals in adults without diabetes. Am J Clin Nutr. 2025 Jan;121(1):74-82. doi: 10.1016/j.ajcnut.2024.10.007. Epub 2024 Dec 2. PMID: 39755436; PMCID: PMC11747189.
Other
Hutchins KM, Betts JA, Thompson D, Hengist A, Gonzalez JT. Continuous glucose monitor overestimates glycemia, with the magnitude of bias varying by postprandial test and individual - a randomized crossover trial. Am J Clin Nutr. 2025 May;121(5):1025-1034. doi: 10.1016/j.ajcnut.2025.02.024. Epub 2025 Feb 26. PMID: 40021059; PMCID: PMC12107490.
Spartano NL, Prescott B, Walker ME, Shi E, Venkatesan G, Fei D, Lin H, Murabito JM, Ahn D, Battelino T, Edelman SV, Fleming GA, Freckmann G, Galindo RJ, Joubert M, Lansang MC, Mader JK, Mankovsky B, Mathioudakis NN, Mohan V, Peters AL, Shah VN, Spanakis EK, Waki K, Wright EE, Zilbermint M, Wolpert HA, Steenkamp DW. Expert Clinical Interpretation of Continuous Glucose Monitor Reports From Individuals Without Diabetes. J Diabetes Sci Technol. 2025 Feb 12:19322968251315171. doi: 10.1177/19322968251315171. Epub ahead of print. PMID: 39936548; PMCID: PMC11822776.
Rodriguez JA, Palermo NE, Song W, Lipsitz S, Caballero AE, Samal L, Spartano NL. Lack of Association Between Hemoglobin A1c and Continuous Glucose Monitor Metrics Among Individuals with Prediabetes and Normoglycemia. Diabetes Technol Ther. 2025 Oct 1. doi: 10.1177/15209156251379506. Epub ahead of print. PMID: 41027845.
Oganesova Z, Pemberton J, Brown A. Innovative solution or cause for concern? The use of continuous glucose monitors in people not living with diabetes: A narrative review. Diabet Med. 2024 Sep;41(9):e15369. doi: 10.1111/dme.15369. Epub 2024 Jun 26. PMID: 38925143.
Shah VN, Vigers T, Pyle L, Calhoun P, Bergenstal RM. Discordance Between Glucose Management Indicator and Glycated Hemoglobin in People Without Diabetes. Diabetes Technol Ther. 2023 May;25(5):324-328. doi: 10.1089/dia.2022.0544. Epub 2023 Mar 3. PMID: 36790875.
Sofizadeh S, Pehrsson A, Ólafsdóttir AF, Lind M. Evaluation of Reference Metrics for Continuous Glucose Monitoring in Persons Without Diabetes and Prediabetes. J Diabetes Sci Technol. 2022 Mar;16(2):373-382. doi: 10.1177/1932296820965599. Epub 2020 Oct 26. PMID: 33100059; PMCID: PMC8861786.
Comments