The CES Las Vegas booth was bright and had shiny counters.
The healthcare company filled the booth with its products and set up a section for its over-the-counter continuous glucose monitor (CGM). Staff gave out little treats and packets of electrolyte powder.
I had just left their CES panel, “Tracking Your Body’s Signals for Better Health,” at the Venetian Expo. Tunde Oyeneyin talked about using glucose data for cardio and preparation. I know her from Peloton and like her as a personality. Alexis Ohanian, Reddit’s co-founder, brought in a tech and fatherhood perspective, which I find interesting. He also appeared in the product’s “Spike Sessions” campaign, which is worth mentioning when considering his role on the company’s stage [1][2].
After 3 years of doubt, I wondered whether I should try one and later speak from experience.
The booth was big. My estimate is six figures, probably more. I spoke with someone I remember as a brand manager or salesperson. They kept explaining how eating order affects glucose, and how levels rise after eating carbs. I kept asking what a healthy person should do after a normal increase.
I never got an answer I could use. I left with a few samples, but not a glucose monitor. I still haven’t tried one.
My doctor’s take begins with me
That’s a fair point. I’m judging a device I haven’t tried myself.
I’m a physician with medical and dental degrees, but I don’t specialize in endocrinology or diabetes. I look at health devices like any other treatment: identify the problem, decide what action to take based on the result, and see if that action improves health.
I use the same approach for myself. Twice a year, I check my fasting glucose and HbA1c, which shows long-term glucose levels. My results are 92 mg/dL (about 5.1 mmol/L) and 4.9%.
I also take valsartan with hydrochlorothiazide for high blood pressure, which is well controlled. I take my medication and check my blood pressure regularly because there’s a clear target, and lowering it reduces health risks.
Hydrochlorothiazide belongs to a family of blood-pressure drugs called thiazides. Across 95 randomized trials in which treatment was assigned by chance, these drugs raised fasting glucose by 3.6 mg/dL on average, or 2.7 mg/dL in trials lasting at least 6 months. The authors called the class effect too small to change care [3]. It cannot predict my result without the drug.
Blood-pressure treatment also counts as 1 of the 5 markers used to define metabolic syndrome, a cluster of linked health risks. The diagnosis requires 3 markers. My treated hypertension is a managed cardiovascular risk factor; by itself, it does not meet that definition [4].
Where the evidence holds
For the right patient with diabetes, continuous glucose monitoring earns its place.
A review published in JAMA Internal Medicine on 3 August 2026 summarized randomized evidence in type 2 diabetes. Average HbA1c fell by about 0.3 percentage points compared with finger-prick checks or usual care, and some patients gained more. Much of that evidence came from people taking insulin or other glucose-lowering drugs [5].
This was a narrative review, meaning the authors chose and explained existing research without the strict process of a systematic review. Published at the journal’s initiative, it found only limited and indirect support for CGM in people with type 2 diabetes not taking glucose-lowering treatment, or with prediabetes or obesity. No CGM-manufacturer conflict was disclosed.
Prediabetes is a different case, since the evidence is more promising there. Here, I’m focusing on healthy adults without symptoms.
Minna Johansson set the boundary in a university press statement: “This technology is tremendously valuable for certain groups of patients. But it is increasingly being used by groups where we simply do not know whether it provides any benefit at all—or if it could even be harmful” [6]. The review didn’t measure a harm rate. The authors suggested using CGM only when a specific patient problem gives the readings a purpose. This is not a clinical guideline.
The FDA cleared Lingo for over-the-counter sale to adults aged 18 or older who do not use insulin. This includes some people with diabetes. Its screen covers 55 to 200 mg/dL, and the agency says users should consult a qualified health professional before taking medical action based on the output [7].
Even precise numbers can be misleading
Hutchins and colleagues ran a randomized crossover study in 15 healthy adults. Each person completed the same set of meal tests, with the order chosen by chance, so the researchers could compare each person with themselves.
The continuous monitor read 0.9 mmol/L, about 16 mg/dL, higher than a finger-prick test both before and after meals. For one smoothie, the monitor gave a glycemic index of 69 and the finger-prick method gave 53. Glycemic index is a score for how strongly a food raises glucose [8].
Those estimates had wide confidence intervals. A confidence interval is the range that still fits the data. The monitor’s range was 48 to 99, while the finger-prick range was 40 to 69. These ranges overlap a lot.
The p-value was .05. A p-value asks how often a difference this big could happen by chance if both methods worked the same. Here, it was about 5 times in 100, which is the usual cutoff researchers use. This example comes from one smoothie and 15 people, not a general error rate.
After adjusting for each person’s starting point, the monitor still showed about twice as much time above 140 mg/dL compared to the finger-prick method. The p-value was below .01, meaning a gap this large would happen less than once in 100 times if there was no real difference. They measured time above 140, not the actual glucose concentration.
Innocent Drinks funded the study with an unrestricted grant and provided the test products. “Unrestricted” means the company didn’t control how the study was run. The authors also disclosed connections to ZOE, PepsiCo, and the European Fruit Juice Association.
Then 18 CGM experts looked at 20 reports from adults without diabetes and decided whether each person needed medical follow-up. The researchers had deliberately chosen difficult reports. Dexcom supplied discounted sensors, and several authors had ties to the manufacturer [9].
Fleiss kappa measures how much a group agrees beyond what chance alone would produce. A score of 0 means chance-level agreement; 1 means perfect agreement. These experts scored 0.36, which the authors called low. All 18 gave the same answer on only 3 of the 20 reports. They all wanted follow-up for 2 and all declined it for 1. This measured agreement in a small selected set. It did not test whether their decisions were medically correct.
One rough pattern appeared. When more than 2 out of every 100 readings were above 180 mg/dL, between 56% and 100% of the experts recommended follow-up, depending on the report. Simply put, the same cutoff convinced just over half of them in some cases and all of them in others. The study didn’t prove that 2% was the right alarm line. I think this number comes from diabetes care. The authors said there are no guidelines for interpreting these reports in people without diabetes. Diabetes care has its own guidelines.
My doctor’s take after the evidence
This fits into the Activate part of the Upward ARC, which focuses on food and movement. It also relates to Capacity, since our attention is limited. Every new number asks you to look, interpret, and decide what to do.
Before buying a monitor, a healthy person should think about what reading would make them change their behavior.
Try this today
Write down the exact number that would make you change your actions. Add what you would do next and what result you expect. If you can’t fill the page, that tells you something.
If you have diabetes, prediabetes, symptoms, pregnancy, medication concerns, or another metabolic risk, take the question to a clinician. This newsletter doesn’t cover you.
Try the habit without tracking it on a graph. Eat protein before carbs if you like. Take a 10- or 20-minute walk after eating. See if you can keep it up.
Consider what the experiment costs, both in money and attention. Lingo launched at $89 for two 14-day sensors in the US. On 21 August 2026, its US 4-week subscription showed an $89 standard price, and Boots in the UK listed two sensors for £118 [10][11]. Decide before you start when you’ll stop.
The unused patches
About a year ago, I noticed an opened FreeStyle Libre package next to a monitor at the office. A blister pack of patches was still sealed. Since the desks were shared, I didn’t know whose it was. Later that day, a colleague told me they saw an influencer wearing one and bought it with a discount code. They showed me the sensor on their arm.
They learned that eating protein before carbohydrates seemed to flatten the curve. A 10- or 20-minute walk after eating had a similar effect. Fruit or something sweet made glucose rise and then fall.
I could have told them all of that without a sensor.
A few weeks later, I asked if they were still using it. They had stopped. I inferred that it had been an expensive way to learn basic lessons. They never said why they quit.
Normal is broader than it seems
The Framingham study followed 560 normoglycemic adults. That simply means their standard blood tests placed them in the normal glucose range [12].
Even so, their readings sat between 140 and 180 for almost 3 hours a day, then above 180 for another 19 minutes. Total time above 140 averaged roughly 3.3 hours. This was a group average, so it doesn’t predict one person’s day. Participants averaged 58.8 years old and had an average body mass index of 26.5, which indicates overweight. The group was 93.2% non-Hispanic White, and Dexcom supplied discounted sensors.
A second study followed 153 healthy, non-obese people aged 7 to 80 [13]. The sample averaged 31 years old. Average glucose was 98 to 99 mg/dL below age 60 and 104 at 60 or older. The median person (the middle value after ranking) spent 96% of the day between 70 and 140. The interquartile range was 93% to 98%, meaning the middle half of participants fell between those values. The sample was 93% White, and an author disclosed ties to Dexcom and Sanofi.
I think the difference comes down to age and who was in each study. The younger group didn’t include people with obesity, while the Framingham study looked at an older community group. Neither paper tested this, so this is just my opinion. Both studies show that healthy glucose levels can vary. Seeing a colored line can make normal changes look like a problem.
Curiosity still needs a purpose
In PREDICT 1, 1,002 generally healthy UK adults ate identical standard meals. Their population coefficient of variation was 68% [14]. This compares the spread of responses with the average; 68% signals wide variation. It is neither the share of people who reacted differently nor a treatment effect. Several authors worked for or consulted with ZOE.
A 2026 systematic review used preset rules to find and assess 23 studies [15]. It found signs of benefit in prediabetes and no appreciable improvement in healthy groups. Only 5 studies could be combined numerically. Most randomized studies lasted 12 weeks or less and paired the sensor with education or coaching, leaving its separate effect unclear.
Another review covered 25 randomized behavior-change trials [16]. Only 3 involved people without diabetes, all with obesity, and every trial added other support. Eight supplied weight data. The combined result was 0.7 kg less, with P = .066. If the true effect were zero, a result this large could appear by chance about 6.6 times in 100. It missed the usual cutoff and was not statistically significant. The analysis could not reliably separate a small loss from random variation, though it didn’t prove zero effect. Eleven trials reported CGM-industry conflicts, which proves neither funding nor bias.
When I visited the booth, I was open to trying a CGM. But after talking with them, I kept coming back to the same question: what should you do after seeing a normal rise in glucose from eating a banana?
I still don’t have a good answer for healthy people without symptoms. My first thought remains: for this group, a CGM is mostly a waste of money.
A monitor might encourage you to take a walk, but it can also turn normal body changes into warnings that even experts can’t interpret. I already have enough information to decide when to take a walk.
I took the electrolyte powder and left the monitor behind.
If you’ve worn a CGM without a diagnosis, what decisions are you still making differently six months later? Reply and tell me exactly what’s changed. I want to know if I’m missing something in my conclusion.
Stay healthy.
Andre
PS: This argument is for healthy, asymptomatic adults. If you know someone who uses a CGM, please forward this to them. That’s how this newsletter grows, and it’s the only way I want it to.
PPS: If someone forwarded this to you, you can get Under Load, my Sunday edition, here:
https://www.andreheeg.com
References
[1] Consumer Technology Association. (2025, January 7). Tracking your body’s signals for better health, presented by Abbott [Conference session]. CES 2025. https://www.ces.tech/videos/tracking-your-body-s-signals-for-better-health-presented-by-abbott/
[2] RQ Agency. (n.d.). Lingo. https://rqagency.com/work/lingo
[3] Hall, J. J., Eurich, D. T., Nagy, D., Tjosvold, L., & Gamble, J.-M. (2020). Thiazide diuretic-induced change in fasting plasma glucose: A meta-analysis of randomized clinical trials. Journal of General Internal Medicine, 35(6), 1849-1860. https://doi.org/10.1007/s11606-020-05731-3
[4] Alberti, K. G. M. M., Eckel, R. H., Grundy, S. M., Zimmet, P. Z., Cleeman, J. I., Donato, K. A., Fruchart, J.-C., James, W. P. T., Loria, C. M., & Smith, S. C., Jr. (2009). Harmonizing the metabolic syndrome: A joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity. Circulation, 120(16), 1640-1645. https://doi.org/10.1161/CIRCULATIONAHA.109.192644
[5] Dower, J. A., Johansson, M., Camp, A. W., Montori, V. M., & Lipska, K. J. (2026). Continuous glucose monitoring in type 2 diabetes and beyond: A review. JAMA Internal Medicine. Advance online publication. https://doi.org/10.1001/jamainternmed.2026.2772
[6] University of Gothenburg. (2026, August 3). No evidence that continuous glucose monitors improve health in people without diabetes [Press release]. https://www.gu.se/en/news/no-evidence-that-continuous-glucose-monitors-improve-health-in-people-without-diabetes
[7] U.S. Food and Drug Administration. (2024, May 29). 510(k) premarket notification K233655: Lingo Glucose System. https://www.accessdata.fda.gov/cdrh_docs/pdf23/K233655.pdf
[8] Hutchins, K., Betts, J., Thompson, D., Hengist, A., & Gonzalez, J. (2025). Continuous glucose monitor overestimates glycemia, with the magnitude of bias varying by postprandial test and individual: A randomized crossover trial. American Journal of Clinical Nutrition, 121(5), 1025-1034. https://doi.org/10.1016/j.ajcnut.2025.02.024
[9] Spartano, N. L., Prescott, B., Walker, M. E., et al. (2025). Expert clinical interpretation of continuous glucose monitor reports from individuals without diabetes. Journal of Diabetes Science and Technology, 20(3), 727-735. https://doi.org/10.1177/19322968251315171
[10] Abbott. (2024, September 5). Abbott’s Lingo continuous glucose monitor for health and wellness now available in the U.S. https://abbott.mediaroom.com/2024-09-05-Abbotts-Lingo-TM-Continuous-Glucose-Monitor-for-Health-and-Wellness-Now-Available-in-the-U-S
[11] Boots. (2026). Lingo continuous glucose monitor bundle. Retrieved August 21, 2026, from https://www.boots.com/lingo-continuous-glucose-monitor-bundle-10390678
[12] Spartano, N. L., Sultana, N., Lin, H., et al. (2025). Defining continuous glucose monitor time in range in a large, community-based cohort without diabetes. The Journal of Clinical Endocrinology & Metabolism, 110(4), 1128-1134. https://doi.org/10.1210/clinem/dgae626
[13] Shah, V. N., DuBose, S. N., Li, Z., et al. (2019). Continuous glucose monitoring profiles in healthy nondiabetic participants: A multicenter prospective study. The Journal of Clinical Endocrinology & Metabolism, 104(10), 4356-4364. https://doi.org/10.1210/jc.2018-02763
[14] Berry, S. E., et al. (2020). Human postprandial responses to food and potential for precision nutrition. Nature Medicine, 26, 964-973. https://doi.org/10.1038/s41591-020-0934-0
[15] Liao, X., Li, Y., Tang, S., Xiao, Y., Yu, X., Huang, R., & Zhong, T. (2026). Continuous glucose monitoring in non-diabetic populations: A systematic review of observational and interventional studies with meta-analysis. European Journal of Medical Research, 31, Article 397. https://doi.org/10.1186/s40001-026-03920-0
[16] Richardson, K. M., Jospe, M. R., Bohlen, L. C., et al. (2024). The efficacy of using continuous glucose monitoring as a behaviour change tool in populations with and without diabetes: A systematic review and meta-analysis of randomised controlled trials. International Journal of Behavioral Nutrition and Physical Activity, 21, Article 145. https://doi.org/10.1186/s12966-024-01692-6