The Glucose Monitor on Your Arm Might Be Lying to You

TL;DR

Continuous glucose monitors are now on the arms of fitness trainers, bestselling wellness authors, and a growing number of otherwise-healthy consumers. Dexcom's Stelo and Abbott's Lingo can be bought without a prescription in the US; other markets are following.
The best peer-reviewed head-to-head test in healthy adults — University of Bath, published in Diabetic Medicine — found the Abbott Freestyle Libre 2 overestimated time spent above the recommended blood-sugar threshold by nearly 400% compared with finger-prick blood testing. In healthy people. That is not a rounding error.
Endocrinologists at Yale, Cedars-Sinai, and NHS-aligned GP guidelines all land in roughly the same place: for healthy adults with no metabolic risk factors, there is no validated definition of an "abnormal" CGM reading, no evidence that acting on the data improves long-term health, and a real risk of overreacting to noise.
Genuine candidates for a short trial exist — prediabetes, prior gestational diabetes, PCOS, cardiometabolic risk. For everyone else, the honest answer is that the data are pretty, expensive, and largely uninterpretable.

The number the marketing doesn't put on the box

Start with one figure.

Researchers at the University of Bath's Center for Nutrition, Exercise and Metabolism put an Abbott Freestyle Libre 2 — one of the most common CGMs in circulation, and the sensor sitting inside several of the new consumer-facing products — on the arms of healthy, non-diabetic adults within a normal BMI range. They then measured the same glucose responses with the finger-prick capillary test that remains the clinical gold standard.

The CGM consistently reported higher glucose than the blood test. More importantly, it overestimated the time each participant spent above the Diabetes UK recommended threshold by close to 400%.

Read that again. In a population that by every clinical definition does not have a glucose problem, the sensor was telling them, four hundred percent more often than reality, that they were having a glucose problem.

Professor Javier Gonzalez, who led the work, put it plainly: "For healthy individuals, relying on CGMs could lead to unnecessary food restrictions or poor dietary choices."

This is the number you need to hold in your head while you scroll through the rest of the pitch.

What the trend actually is

The mechanics are simple. In 2024 the US FDA cleared the first over-the-counter CGM — Dexcom's Stelo — for adults 18+ with type 2 diabetes not using insulin, or with prediabetes. Abbott's Lingo followed. Neither product is officially marketed to healthy people. In practice, that is who is buying them.

The influencer layer sits on top of that regulatory shift. In the trigger piece for this cycle — the New York Post's 13 July feature — fitness trainer Ryan Fischer posts a Reel to more than 700,000 followers titled "What I Learned From Wearing a Continuous Glucose Monitor as a Non-Diabetic." Fischer is one of dozens. The content template is now familiar: strap it on, film the graph, react to the "spike," recommend a supplement or protocol against the spike, repeat.

The trend is not confined to the US. In Australia and the UK, wellness-clinic subscription models bundle a CGM with a nutritionist consult for roughly A$300–$450 per fortnight. In India, Chinese-made sensors from Sibionics have driven per-sensor costs down to the ₹2,500–₹4,000 range and turned CGM into a middle-class Instagram artefact rather than a medical device. In China itself, Weibo #血糖 (blood sugar) content has been trending intermittently since May.

This is not going away. It is a category, not a moment.

What the sensor is actually doing

A CGM does not measure your blood sugar. This matters.

The sensor sits in the interstitial fluid — the fluid between your cells, a few layers under your skin — and estimates blood glucose from it. There is a lag of five to fifteen minutes between what your blood is doing and what the interstitial fluid reads. There is also a systematic offset that varies by hydration, temperature, and how firmly the sensor is pressed against your skin overnight.

For a person with type 1 diabetes, none of this matters much. Their glucose swings are so large that a lagged, slightly-off estimate is enormously more useful than no measurement at all. Gonzalez, the Bath researcher, is explicit about this: "CGMs are fantastic tools for people with diabetes because even if a measurement isn't perfectly accurate, it's still better than not having a measurement at all."

For a healthy person, the maths inverts. Their true glucose barely moves outside a tight band — the largest CGM study in healthy people to date, cited in an October 2025 scoping review in Diabetes Research and Clinical Practice, found the median time spent within the standard normal range (3.9–7.8 mmol/L) was 96%. When your real signal is small and your sensor error is not, the sensor is mostly measuring itself.

That is what "400% overestimation of time above threshold" is telling you. The signal was almost entirely noise.

What the evidence base actually says

Set aside the marketing. Here is what clinicians and guideline bodies are on the record saying in the last twelve months.

Yale School of Medicine (Dr Silvio Inzucchi / Dr Sarah Elisseou): "Like a lot of other wellness devices, they give you tons of data, but we don't have good evidence that wearing one actually improves long-term health outcomes in people without diabetes."
Cedars-Sinai (Dr Anne Peter): For a healthy person without diabetes risk factors, "I'm not sure what people would do with that kind of information. It could be easy to overreact if you don't have the context."
NHS / NICE (UK, current guidance): CGM is funded only for insulin-controlled diabetes. From 7 November 2025 the DVLA accepts CGM for glucose checks in group-2 drivers, but this is an operational accommodation, not an endorsement for wellness use.
The multi-author Medscape guideline review (November 2025): "validated glycaemic targets do not exist for populations without diabetes… there is a risk of overmedicalisation, and guideline-based frameworks are absent in this area."
PMC scoping review (October 2025): the medical literature and the "grey literature" (wellness blogs, influencer content) diverge sharply — the grey literature attributes cancer, mood, sleep, and energy consequences to glucose spikes that the peer-reviewed evidence does not support.

There is no group of clinicians, anywhere in the peer-reviewed literature, making the case that healthy adults should wear a CGM continuously. There is a wellness-industry position, and there is a clinical position. They are not the same document.

Who benefits from the noise

Follow the money and the anxiety.

Device manufacturers — Dexcom and Abbott have both explicitly framed the non-diabetic wellness market as their largest addressable growth vector. OTC clearance opened it. Every healthy consumer buying a two-week sensor is a customer the diabetes-only market could never have supplied.
The wellness creator economy — glucose content is unusually well-suited to short-form video. A graph goes up. The creator eats something. The graph goes up more. This is visually legible content in a way that most nutritional science is not. That visual legibility is what makes it viral. It is not what makes it true.
Supplement and "glucose control" adjacencies — berberine, inositol, apple-cider-vinegar-before-meals, "spike-shielding" protein powders. Each of these has a small, real evidence base and a much larger marketing scaffold that the CGM makes credible by generating the graph that seems to endorse it.
The consumer — pays. In dollars. In anxiety. And, if the Bath data is correct, in dietary choices made in response to a signal that wasn't there.

The person who does not benefit is the reader who assumed the sensor was a small, personal, quiet piece of technology. It is not. It is one end of a supply chain that runs through a manufacturer, a creator, a supplement brand, and a nutrition-coaching subscription.

Where CGM does earn a role

Hype deconstruction shouldn't dismiss the tool. Some populations genuinely benefit from a short, structured CGM trial — typically two to four weeks, interpreted with a clinician:

Prediabetes (HbA1c 5.7–6.4% / 39–47 mmol/mol). Real elevated risk; visible glucose data can sustain behaviour change.
Prior gestational diabetes. Sharply elevated long-term diabetes risk; short CGM trials are used clinically to guide postpartum diet.
PCOS with insulin-resistance features. A recognised metabolic phenotype where individual glucose response varies enough to inform choices.
Established cardiometabolic risk — elevated ApoB, high visceral adiposity, family history of early cardiovascular disease. Not a solo indication, but a legitimate adjunct.
Elite endurance athletes managing fuelling and hypoglycaemia risk in long-duration events. Small population, real use case.

Note the pattern. In every case the CGM is a short, time-limited diagnostic-adjacent tool interpreted alongside clinical context. It is not a wearable. It is a two-week experiment with a specific question attached.

What to do — plainly

If you are a healthy adult under 50 with no metabolic risk factors and no family history of type 2 diabetes: you do not need a CGM. If you want to spend the money — roughly US$99 for a Stelo two-pack, £89 for a Lingo starter — do it once, for curiosity, for two weeks, and then stop. Do not restrict foods based on spike graphs. Do not draw dietary conclusions from a device that, in your population, overestimated its own core metric by up to 400% in the best available head-to-head test.

If you have prediabetes, prior gestational diabetes, PCOS, or established cardiometabolic risk: a two- to four-week CGM trial is a reasonable conversation to have with a GP or endocrinologist. Ask them explicitly: what question are we asking, and what would the answer change? If they can't answer that in a sentence, don't buy it yet.

If you are already wearing one and worrying about spikes: the best evidence says that in healthy people, most postprandial glucose excursions are physiological, transient, and do not accumulate into long-term risk. The variability itself is not the enemy. Sustained elevation is.

If you're a GP or primary-care clinician: patients are increasingly bringing you CGM traces from third-party wellness platforms and asking you to interpret them. The current NHS/NICE and RACGP position is that these traces are not clinical instruments. You are not obliged to interpret them, and the Medscape guideline authors explicitly warn against overinterpreting isolated excursions. A short, calm conversation about what the device can and cannot say is often more valuable than a formal readout.

Uncertainty ledger

Sensor accuracy is improving. The next generation — Dexcom G8, Abbott Libre 4, Sibionics GS4 — may narrow the gap in healthy populations. The 400% figure is current-generation, not permanent.
Multi-analyte sensors (glucose + lactate + ketones) are in late trials at several manufacturers. If they arrive, the calculus around wearable metabolic monitoring in healthy adults may shift genuinely.
The evidence base for CGM in prediabetes as a behaviour-change lever — separate from the accuracy question — is still thin. Short trials are promising; long-term outcome data are not yet in.
Individual variability in glucose response to identical foods is real (the ZOE studies, the Weizmann PREDICT work). Whether CGMs are the right tool to characterise that variability, or whether they are too noisy to do so reliably outside diabetes, is genuinely unsettled.

Bottom line

A continuous glucose monitor is a medical device that was designed for a disease you do not have. In the population currently buying them — healthy adults, curious, mildly anxious — the best available evidence says the sensors are systematically wrong, the "abnormal" pattern being flagged is not defined, and the behavioural response the graph provokes is more likely to make eating harder than to make you healthier. If you have a specific metabolic question and a clinician to interpret the answer, a two-week trial can be useful. If you don't, the graph is a story the device is telling you about itself.

Sources

Tier 1 — University of Bath Center for Nutrition, Exercise and Metabolism, Diabetic Medicine, "Continuous glucose monitors overestimate blood sugar in healthy adults" (2025).
Tier 1 — Yale Medicine, "Who Should Use a Continuous Glucose Monitor?" endocrinology commentary (updated 2024).
Tier 1 — Journal of Diabetes Science and Technology / PMC, "Use of Continuous Glucose Monitors by People Without Diabetes" (2022, expert consensus review).
Tier 1 — PMC scoping review, "Glucose Spikes in People Without Diabetes" (October 2025).
Tier 2 — Medscape / multi-guideline expert insight, "Continuous Glucose Monitoring in People Without Diabetes" (November 2025).
Tier 2 — MedTech Dive, "Top diabetes tech stories from the first half of 2026" (7 July 2026).
Tier 3 — New York Post, "Influencers are hopping on a surprising wearable trend…" (13 July 2026) — used as trigger and for named-influencer sourcing, not as evidence-base.